diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31518 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 5248,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2658227848101266e-07,
+      "loss": 1.0128,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5316455696202533e-07,
+      "loss": 1.0117,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.79746835443038e-07,
+      "loss": 0.9927,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.063291139240507e-07,
+      "loss": 1.0017,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.329113924050634e-07,
+      "loss": 0.9911,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.59493670886076e-07,
+      "loss": 0.9905,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.860759493670887e-07,
+      "loss": 0.9593,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0126582278481013e-06,
+      "loss": 0.9224,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.139240506329114e-06,
+      "loss": 0.8917,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2658227848101267e-06,
+      "loss": 0.6819,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3924050632911392e-06,
+      "loss": 0.6585,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.518987341772152e-06,
+      "loss": 0.6423,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6455696202531647e-06,
+      "loss": 0.3323,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7721518987341774e-06,
+      "loss": 0.3209,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 0.3025,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0253164556962026e-06,
+      "loss": 0.2882,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1518987341772153e-06,
+      "loss": 0.2563,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.278481012658228e-06,
+      "loss": 0.1371,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4050632911392408e-06,
+      "loss": 0.1438,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5316455696202535e-06,
+      "loss": 0.1441,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6582278481012658e-06,
+      "loss": 0.1328,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7848101265822785e-06,
+      "loss": 0.1279,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9113924050632912e-06,
+      "loss": 0.1275,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.037974683544304e-06,
+      "loss": 0.128,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.164556962025317e-06,
+      "loss": 0.1512,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2911392405063294e-06,
+      "loss": 0.1434,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.417721518987342e-06,
+      "loss": 0.1533,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.544303797468355e-06,
+      "loss": 0.1593,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6708860759493675e-06,
+      "loss": 0.1405,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 0.1337,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.924050632911393e-06,
+      "loss": 0.136,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.050632911392405e-06,
+      "loss": 0.1205,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.177215189873418e-06,
+      "loss": 0.1333,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.303797468354431e-06,
+      "loss": 0.1243,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.430379746835443e-06,
+      "loss": 0.1197,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.556962025316456e-06,
+      "loss": 0.1421,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.683544303797468e-06,
+      "loss": 0.1368,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8101265822784815e-06,
+      "loss": 0.1278,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.936708860759495e-06,
+      "loss": 0.1129,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.063291139240507e-06,
+      "loss": 0.1169,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.189873417721519e-06,
+      "loss": 0.115,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3164556962025316e-06,
+      "loss": 0.1277,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.443037974683545e-06,
+      "loss": 0.1196,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.569620253164557e-06,
+      "loss": 0.1183,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.69620253164557e-06,
+      "loss": 0.114,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8227848101265824e-06,
+      "loss": 0.1175,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.949367088607595e-06,
+      "loss": 0.1151,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.075949367088608e-06,
+      "loss": 0.1165,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.20253164556962e-06,
+      "loss": 0.1152,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.329113924050634e-06,
+      "loss": 0.1145,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4556962025316464e-06,
+      "loss": 0.118,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.582278481012659e-06,
+      "loss": 0.1182,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.708860759493672e-06,
+      "loss": 0.1159,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.835443037974684e-06,
+      "loss": 0.1148,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.962025316455697e-06,
+      "loss": 0.1134,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.08860759493671e-06,
+      "loss": 0.1155,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.215189873417722e-06,
+      "loss": 0.1128,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.341772151898735e-06,
+      "loss": 0.1146,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.468354430379747e-06,
+      "loss": 0.1138,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5949367088607605e-06,
+      "loss": 0.1157,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.721518987341773e-06,
+      "loss": 0.109,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.848101265822786e-06,
+      "loss": 0.1088,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.974683544303799e-06,
+      "loss": 0.1157,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.10126582278481e-06,
+      "loss": 0.1106,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.227848101265824e-06,
+      "loss": 0.1124,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.354430379746837e-06,
+      "loss": 0.111,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.481012658227848e-06,
+      "loss": 0.1047,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.607594936708861e-06,
+      "loss": 0.111,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.734177215189874e-06,
+      "loss": 0.1144,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.860759493670886e-06,
+      "loss": 0.1059,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.987341772151899e-06,
+      "loss": 0.1124,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.113924050632912e-06,
+      "loss": 0.1122,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.240506329113925e-06,
+      "loss": 0.1078,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.367088607594937e-06,
+      "loss": 0.107,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.49367088607595e-06,
+      "loss": 0.1019,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.620253164556963e-06,
+      "loss": 0.0991,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.746835443037975e-06,
+      "loss": 0.0945,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.87341772151899e-06,
+      "loss": 0.1111,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.0965,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0126582278481014e-05,
+      "loss": 0.0975,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0253164556962025e-05,
+      "loss": 0.0974,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0379746835443039e-05,
+      "loss": 0.093,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0506329113924052e-05,
+      "loss": 0.0941,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0632911392405063e-05,
+      "loss": 0.0949,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0759493670886076e-05,
+      "loss": 0.0924,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.088607594936709e-05,
+      "loss": 0.0956,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1012658227848103e-05,
+      "loss": 0.0903,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1139240506329114e-05,
+      "loss": 0.0859,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1265822784810127e-05,
+      "loss": 0.088,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.139240506329114e-05,
+      "loss": 0.0771,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1518987341772152e-05,
+      "loss": 0.0797,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1645569620253165e-05,
+      "loss": 0.0769,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1772151898734178e-05,
+      "loss": 0.0756,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.189873417721519e-05,
+      "loss": 0.0836,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2025316455696203e-05,
+      "loss": 0.0884,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2151898734177216e-05,
+      "loss": 0.0762,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.227848101265823e-05,
+      "loss": 0.0847,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.240506329113924e-05,
+      "loss": 0.0767,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2531645569620255e-05,
+      "loss": 0.0918,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2658227848101268e-05,
+      "loss": 0.0761,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2784810126582278e-05,
+      "loss": 0.0738,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2911392405063293e-05,
+      "loss": 0.077,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3037974683544306e-05,
+      "loss": 0.0779,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3164556962025317e-05,
+      "loss": 0.0778,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.329113924050633e-05,
+      "loss": 0.0729,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3417721518987344e-05,
+      "loss": 0.0693,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3544303797468355e-05,
+      "loss": 0.0691,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3670886075949368e-05,
+      "loss": 0.0675,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3797468354430381e-05,
+      "loss": 0.0696,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3924050632911395e-05,
+      "loss": 0.0739,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4050632911392406e-05,
+      "loss": 0.0735,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.417721518987342e-05,
+      "loss": 0.0719,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4303797468354432e-05,
+      "loss": 0.0733,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4430379746835444e-05,
+      "loss": 0.0719,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4556962025316457e-05,
+      "loss": 0.0799,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.468354430379747e-05,
+      "loss": 0.0679,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4810126582278482e-05,
+      "loss": 0.0664,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4936708860759495e-05,
+      "loss": 0.0707,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5063291139240508e-05,
+      "loss": 0.0596,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5189873417721521e-05,
+      "loss": 0.0708,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.531645569620253e-05,
+      "loss": 0.0693,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5443037974683546e-05,
+      "loss": 0.0699,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.556962025316456e-05,
+      "loss": 0.0593,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5696202531645572e-05,
+      "loss": 0.063,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5822784810126583e-05,
+      "loss": 0.0647,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5949367088607598e-05,
+      "loss": 0.0618,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.607594936708861e-05,
+      "loss": 0.0651,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.620253164556962e-05,
+      "loss": 0.0605,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6329113924050636e-05,
+      "loss": 0.0663,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6455696202531647e-05,
+      "loss": 0.0705,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.658227848101266e-05,
+      "loss": 0.076,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6708860759493674e-05,
+      "loss": 0.0739,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6835443037974685e-05,
+      "loss": 0.0625,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6962025316455696e-05,
+      "loss": 0.072,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.708860759493671e-05,
+      "loss": 0.0626,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7215189873417723e-05,
+      "loss": 0.0667,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7341772151898734e-05,
+      "loss": 0.0809,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.746835443037975e-05,
+      "loss": 0.0618,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.759493670886076e-05,
+      "loss": 0.0724,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7721518987341772e-05,
+      "loss": 0.0705,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7848101265822787e-05,
+      "loss": 0.0736,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7974683544303798e-05,
+      "loss": 0.0799,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8101265822784813e-05,
+      "loss": 0.0661,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8227848101265824e-05,
+      "loss": 0.0598,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8354430379746836e-05,
+      "loss": 0.0646,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.848101265822785e-05,
+      "loss": 0.0738,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8607594936708862e-05,
+      "loss": 0.0608,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8734177215189874e-05,
+      "loss": 0.0603,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.886075949367089e-05,
+      "loss": 0.066,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.89873417721519e-05,
+      "loss": 0.0639,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.911392405063291e-05,
+      "loss": 0.08,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9240506329113926e-05,
+      "loss": 0.0763,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.936708860759494e-05,
+      "loss": 0.0608,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.949367088607595e-05,
+      "loss": 0.0636,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9620253164556964e-05,
+      "loss": 0.0566,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.974683544303798e-05,
+      "loss": 0.0649,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9873417721518987e-05,
+      "loss": 0.0677,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999998095266717e-05,
+      "loss": 0.0532,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999992381067583e-05,
+      "loss": 0.0606,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999982857404783e-05,
+      "loss": 0.0608,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999969524281942e-05,
+      "loss": 0.0545,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999995238170414e-05,
+      "loss": 0.0544,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999931429677906e-05,
+      "loss": 0.0558,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999990666821122e-05,
+      "loss": 0.0628,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999987809731352e-05,
+      "loss": 0.0662,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999845716995688e-05,
+      "loss": 0.0563,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999809527270053e-05,
+      "loss": 0.0683,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999976952815041e-05,
+      "loss": 0.066,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999725719651995e-05,
+      "loss": 0.0632,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999967810179149e-05,
+      "loss": 0.0612,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999626674587046e-05,
+      "loss": 0.0638,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999571438058242e-05,
+      "loss": 0.0552,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999512392226128e-05,
+      "loss": 0.0561,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999449537113195e-05,
+      "loss": 0.0642,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999382872743387e-05,
+      "loss": 0.0704,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99993123991421e-05,
+      "loss": 0.0611,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999923811633618e-05,
+      "loss": 0.0554,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999916002435393e-05,
+      "loss": 0.0489,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999078123225092e-05,
+      "loss": 0.0592,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999899241298087e-05,
+      "loss": 0.0522,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999890289365391e-05,
+      "loss": 0.0525,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999880956527832e-05,
+      "loss": 0.0546,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998712427889652e-05,
+      "loss": 0.053,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998611481524907e-05,
+      "loss": 0.0468,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999850672622254e-05,
+      "loss": 0.0555,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998398162022464e-05,
+      "loss": 0.0596,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998285788966027e-05,
+      "loss": 0.0503,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998169607096045e-05,
+      "loss": 0.0479,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999804961645677e-05,
+      "loss": 0.0598,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997925817093923e-05,
+      "loss": 0.0496,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997798209054648e-05,
+      "loss": 0.0611,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997666792387572e-05,
+      "loss": 0.063,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997531567142752e-05,
+      "loss": 0.0482,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99973925333717e-05,
+      "loss": 0.0523,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999724969112739e-05,
+      "loss": 0.0506,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997103040464218e-05,
+      "loss": 0.0653,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999695258143807e-05,
+      "loss": 0.0557,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999679831410625e-05,
+      "loss": 0.0577,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996640238527534e-05,
+      "loss": 0.0576,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996478354762132e-05,
+      "loss": 0.0534,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999631266287172e-05,
+      "loss": 0.0673,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996143162919416e-05,
+      "loss": 0.067,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995969854969786e-05,
+      "loss": 0.0504,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999579273908886e-05,
+      "loss": 0.0613,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995611815344102e-05,
+      "loss": 0.0616,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999542708380444e-05,
+      "loss": 0.0554,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999523854454024e-05,
+      "loss": 0.0539,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999504619762333e-05,
+      "loss": 0.051,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994850043126985e-05,
+      "loss": 0.0569,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994650081125926e-05,
+      "loss": 0.0551,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999444631169633e-05,
+      "loss": 0.0504,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999423873491582e-05,
+      "loss": 0.0475,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994027350863475e-05,
+      "loss": 0.0544,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999381215961982e-05,
+      "loss": 0.0609,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999359316126683e-05,
+      "loss": 0.044,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993370355887932e-05,
+      "loss": 0.0676,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993143743568e-05,
+      "loss": 0.0576,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999291332439337e-05,
+      "loss": 0.0553,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999267909845181e-05,
+      "loss": 0.0655,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992441065832553e-05,
+      "loss": 0.0521,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992199226626276e-05,
+      "loss": 0.0455,
+      "step": 222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9991953580925108e-05,
+      "loss": 0.0515,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991704128822623e-05,
+      "loss": 0.048,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991450870413848e-05,
+      "loss": 0.0507,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999119380579527e-05,
+      "loss": 0.0483,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999093293506481e-05,
+      "loss": 0.0521,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999066825832184e-05,
+      "loss": 0.0487,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99903997756672e-05,
+      "loss": 0.0529,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999012748720316e-05,
+      "loss": 0.0533,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998985139303345e-05,
+      "loss": 0.045,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998957149326325e-05,
+      "loss": 0.0461,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998928778799918e-05,
+      "loss": 0.0477,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998900027734932e-05,
+      "loss": 0.0536,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988708961423198e-05,
+      "loss": 0.0458,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988413840331784e-05,
+      "loss": 0.0644,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998811491418751e-05,
+      "loss": 0.0513,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998781218310425e-05,
+      "loss": 0.0644,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987505647197324e-05,
+      "loss": 0.0465,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998719530658351e-05,
+      "loss": 0.0432,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986881161381028e-05,
+      "loss": 0.0514,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986563211709554e-05,
+      "loss": 0.0428,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986241457690208e-05,
+      "loss": 0.0528,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985915899445563e-05,
+      "loss": 0.0515,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985586537099633e-05,
+      "loss": 0.0548,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985253370777896e-05,
+      "loss": 0.0584,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984916400607268e-05,
+      "loss": 0.058,
+      "step": 247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984575626716113e-05,
+      "loss": 0.0504,
+      "step": 248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998423104923425e-05,
+      "loss": 0.0604,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983882668292943e-05,
+      "loss": 0.0451,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998353048402491e-05,
+      "loss": 0.0501,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983174496564315e-05,
+      "loss": 0.0466,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982814706046766e-05,
+      "loss": 0.0506,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982451112609324e-05,
+      "loss": 0.0541,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982083716390502e-05,
+      "loss": 0.0502,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981712517530258e-05,
+      "loss": 0.0417,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998133751617e-05,
+      "loss": 0.0477,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980958712452577e-05,
+      "loss": 0.0487,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980576106522302e-05,
+      "loss": 0.0489,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998018969852492e-05,
+      "loss": 0.0443,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997979948860763e-05,
+      "loss": 0.0508,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979405476919093e-05,
+      "loss": 0.0505,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979007663609398e-05,
+      "loss": 0.0499,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997860604883009e-05,
+      "loss": 0.0482,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978200632734165e-05,
+      "loss": 0.0458,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977791415476067e-05,
+      "loss": 0.0549,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977378397211682e-05,
+      "loss": 0.0403,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976961578098352e-05,
+      "loss": 0.0455,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976540958294855e-05,
+      "loss": 0.037,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976116537961437e-05,
+      "loss": 0.0489,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975688317259768e-05,
+      "loss": 0.0429,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975256296352984e-05,
+      "loss": 0.0556,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974820475405656e-05,
+      "loss": 0.0433,
+      "step": 273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974380854583817e-05,
+      "loss": 0.039,
+      "step": 274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9973937434054932e-05,
+      "loss": 0.0522,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973490213987926e-05,
+      "loss": 0.048,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973039194553158e-05,
+      "loss": 0.0431,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972584375922453e-05,
+      "loss": 0.043,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997212575826906e-05,
+      "loss": 0.0409,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99716633417677e-05,
+      "loss": 0.0387,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971197126594523e-05,
+      "loss": 0.0433,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970727112927133e-05,
+      "loss": 0.0467,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997025330094458e-05,
+      "loss": 0.0448,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996977569082736e-05,
+      "loss": 0.0427,
+      "step": 284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996929428275742e-05,
+      "loss": 0.0405,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968809076918146e-05,
+      "loss": 0.0421,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968320073494384e-05,
+      "loss": 0.0467,
+      "step": 287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967827272672407e-05,
+      "loss": 0.0464,
+      "step": 288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967330674639957e-05,
+      "loss": 0.0403,
+      "step": 289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99668302795862e-05,
+      "loss": 0.0477,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996632608770177e-05,
+      "loss": 0.0443,
+      "step": 291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965818099178732e-05,
+      "loss": 0.0551,
+      "step": 292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965306314210602e-05,
+      "loss": 0.0466,
+      "step": 293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964790732992346e-05,
+      "loss": 0.0471,
+      "step": 294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964271355720372e-05,
+      "loss": 0.0474,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963748182592532e-05,
+      "loss": 0.0373,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963221213808127e-05,
+      "loss": 0.0604,
+      "step": 297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996269044956791e-05,
+      "loss": 0.0463,
+      "step": 298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996215589007407e-05,
+      "loss": 0.0538,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961617535530244e-05,
+      "loss": 0.0489,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996107538614152e-05,
+      "loss": 0.0517,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960529442114426e-05,
+      "loss": 0.0453,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995997970365694e-05,
+      "loss": 0.0438,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959426170978474e-05,
+      "loss": 0.0544,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958868844289908e-05,
+      "loss": 0.0423,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958307723803542e-05,
+      "loss": 0.0457,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957742809733142e-05,
+      "loss": 0.0557,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99571741022939e-05,
+      "loss": 0.0438,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956601601702476e-05,
+      "loss": 0.0397,
+      "step": 309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995602530817695e-05,
+      "loss": 0.058,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9955445221936868e-05,
+      "loss": 0.0584,
+      "step": 311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995486134320321e-05,
+      "loss": 0.0495,
+      "step": 312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954273672198397e-05,
+      "loss": 0.0468,
+      "step": 313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953682209146305e-05,
+      "loss": 0.0821,
+      "step": 314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995308695427225e-05,
+      "loss": 0.0541,
+      "step": 315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9952487907802994e-05,
+      "loss": 0.0478,
+      "step": 316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951885069966734e-05,
+      "loss": 0.0515,
+      "step": 317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951278440993128e-05,
+      "loss": 0.053,
+      "step": 318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950668021113263e-05,
+      "loss": 0.0461,
+      "step": 319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950053810559684e-05,
+      "loss": 0.057,
+      "step": 320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949435809566364e-05,
+      "loss": 0.0589,
+      "step": 321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948814018368732e-05,
+      "loss": 0.0528,
+      "step": 322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948188437203656e-05,
+      "loss": 0.0403,
+      "step": 323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9947559066309452e-05,
+      "loss": 0.0509,
+      "step": 324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946925905925876e-05,
+      "loss": 0.0425,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946288956294126e-05,
+      "loss": 0.0469,
+      "step": 326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9945648217656848e-05,
+      "loss": 0.0472,
+      "step": 327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9945003690258127e-05,
+      "loss": 0.0455,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99443553743435e-05,
+      "loss": 0.0427,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994370327015993e-05,
+      "loss": 0.0473,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994304737795584e-05,
+      "loss": 0.0569,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994238769798109e-05,
+      "loss": 0.0499,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941724230486984e-05,
+      "loss": 0.0492,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994105697572627e-05,
+      "loss": 0.0488,
+      "step": 334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9940385933953122e-05,
+      "loss": 0.0483,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9939711105423193e-05,
+      "loss": 0.0552,
+      "step": 336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9939032490393537e-05,
+      "loss": 0.0532,
+      "step": 337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993835008912268e-05,
+      "loss": 0.0524,
+      "step": 338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937663901870585e-05,
+      "loss": 0.045,
+      "step": 339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9936973928898643e-05,
+      "loss": 0.0433,
+      "step": 340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99362801704697e-05,
+      "loss": 0.0432,
+      "step": 341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9935582626848044e-05,
+      "loss": 0.0509,
+      "step": 342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99348812982994e-05,
+      "loss": 0.045,
+      "step": 343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9934176185090933e-05,
+      "loss": 0.0464,
+      "step": 344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9933467287491257e-05,
+      "loss": 0.0491,
+      "step": 345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932754605770426e-05,
+      "loss": 0.0409,
+      "step": 346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932038140199933e-05,
+      "loss": 0.0431,
+      "step": 347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931317891052707e-05,
+      "loss": 0.0357,
+      "step": 348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930593858603132e-05,
+      "loss": 0.0491,
+      "step": 349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9929866043127025e-05,
+      "loss": 0.0373,
+      "step": 350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992913444490164e-05,
+      "loss": 0.0374,
+      "step": 351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9928399064205685e-05,
+      "loss": 0.0482,
+      "step": 352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9927659901319292e-05,
+      "loss": 0.0441,
+      "step": 353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9926916956524046e-05,
+      "loss": 0.0408,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9926170230102972e-05,
+      "loss": 0.0527,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992541972234053e-05,
+      "loss": 0.043,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9924665433522623e-05,
+      "loss": 0.0431,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923907363936593e-05,
+      "loss": 0.0342,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923145513871232e-05,
+      "loss": 0.0465,
+      "step": 359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9922379883616758e-05,
+      "loss": 0.0495,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9921610473464837e-05,
+      "loss": 0.0512,
+      "step": 361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992083728370857e-05,
+      "loss": 0.0483,
+      "step": 362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9920060314642507e-05,
+      "loss": 0.0399,
+      "step": 363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991927956656263e-05,
+      "loss": 0.0461,
+      "step": 364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9918495039766356e-05,
+      "loss": 0.0414,
+      "step": 365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9917706734552553e-05,
+      "loss": 0.0452,
+      "step": 366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9916914651221523e-05,
+      "loss": 0.0544,
+      "step": 367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991611879007501e-05,
+      "loss": 0.0408,
+      "step": 368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9915319151416193e-05,
+      "loss": 0.0476,
+      "step": 369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9914515735549688e-05,
+      "loss": 0.0426,
+      "step": 370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991370854278156e-05,
+      "loss": 0.0441,
+      "step": 371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912897573419298e-05,
+      "loss": 0.0402,
+      "step": 372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912082827771844e-05,
+      "loss": 0.0388,
+      "step": 373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9911264306149574e-05,
+      "loss": 0.0526,
+      "step": 374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9910442008864298e-05,
+      "loss": 0.0352,
+      "step": 375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9909615936229268e-05,
+      "loss": 0.0357,
+      "step": 376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908786088559175e-05,
+      "loss": 0.0419,
+      "step": 377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990795246617014e-05,
+      "loss": 0.047,
+      "step": 378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9907115069379736e-05,
+      "loss": 0.0406,
+      "step": 379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9906273898506966e-05,
+      "loss": 0.0425,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990542895387227e-05,
+      "loss": 0.0528,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9904580235797528e-05,
+      "loss": 0.057,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9903727744606048e-05,
+      "loss": 0.0365,
+      "step": 383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902871480622598e-05,
+      "loss": 0.0424,
+      "step": 384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902011444173358e-05,
+      "loss": 0.0501,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990114763558596e-05,
+      "loss": 0.0441,
+      "step": 386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990028005518947e-05,
+      "loss": 0.045,
+      "step": 387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9899408703314383e-05,
+      "loss": 0.0477,
+      "step": 388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9898533580292646e-05,
+      "loss": 0.0406,
+      "step": 389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897654686457634e-05,
+      "loss": 0.0348,
+      "step": 390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9896772022144154e-05,
+      "loss": 0.041,
+      "step": 391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9895885587688456e-05,
+      "loss": 0.046,
+      "step": 392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989499538342822e-05,
+      "loss": 0.0464,
+      "step": 393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989410140970257e-05,
+      "loss": 0.0428,
+      "step": 394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989320366685207e-05,
+      "loss": 0.0437,
+      "step": 395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9892302155218698e-05,
+      "loss": 0.0412,
+      "step": 396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9891396875145893e-05,
+      "loss": 0.0507,
+      "step": 397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989048782697851e-05,
+      "loss": 0.0518,
+      "step": 398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9889575011062855e-05,
+      "loss": 0.0409,
+      "step": 399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888658427746658e-05,
+      "loss": 0.0487,
+      "step": 400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988773807737909e-05,
+      "loss": 0.0579,
+      "step": 401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9886813960310752e-05,
+      "loss": 0.0472,
+      "step": 402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988588607689369e-05,
+      "loss": 0.0345,
+      "step": 403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988495442748137e-05,
+      "loss": 0.0341,
+      "step": 404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9884019012428708e-05,
+      "loss": 0.0468,
+      "step": 405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9883079832092044e-05,
+      "loss": 0.0454,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9882136886829156e-05,
+      "loss": 0.048,
+      "step": 407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9881190176999255e-05,
+      "loss": 0.0426,
+      "step": 408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9880239702962987e-05,
+      "loss": 0.0437,
+      "step": 409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987928546508243e-05,
+      "loss": 0.0418,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9878327463721106e-05,
+      "loss": 0.0495,
+      "step": 411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987736569924395e-05,
+      "loss": 0.0417,
+      "step": 412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9876400172017357e-05,
+      "loss": 0.0402,
+      "step": 413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9875430882409127e-05,
+      "loss": 0.0447,
+      "step": 414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987445783078852e-05,
+      "loss": 0.035,
+      "step": 415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9873481017526206e-05,
+      "loss": 0.0358,
+      "step": 416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9872500442994308e-05,
+      "loss": 0.0421,
+      "step": 417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9871516107566366e-05,
+      "loss": 0.0508,
+      "step": 418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9870528011617364e-05,
+      "loss": 0.0499,
+      "step": 419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986953615552371e-05,
+      "loss": 0.049,
+      "step": 420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986854053966325e-05,
+      "loss": 0.0366,
+      "step": 421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9867541164415266e-05,
+      "loss": 0.0505,
+      "step": 422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9866538030160456e-05,
+      "loss": 0.0431,
+      "step": 423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9865531137280967e-05,
+      "loss": 0.0443,
+      "step": 424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986452048616037e-05,
+      "loss": 0.0416,
+      "step": 425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986350607718367e-05,
+      "loss": 0.0432,
+      "step": 426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9862487910737302e-05,
+      "loss": 0.0503,
+      "step": 427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986146598720913e-05,
+      "loss": 0.0476,
+      "step": 428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9860440306988464e-05,
+      "loss": 0.0389,
+      "step": 429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985941087046602e-05,
+      "loss": 0.048,
+      "step": 430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9858377678033964e-05,
+      "loss": 0.0459,
+      "step": 431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9857340730085886e-05,
+      "loss": 0.0504,
+      "step": 432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985630002701681e-05,
+      "loss": 0.0423,
+      "step": 433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9855255569223186e-05,
+      "loss": 0.0577,
+      "step": 434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9854207357102897e-05,
+      "loss": 0.0487,
+      "step": 435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9853155391055256e-05,
+      "loss": 0.0413,
+      "step": 436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985209967148101e-05,
+      "loss": 0.0434,
+      "step": 437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985104019878233e-05,
+      "loss": 0.042,
+      "step": 438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9849976973362808e-05,
+      "loss": 0.0393,
+      "step": 439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848909995627494e-05,
+      "loss": 0.0553,
+      "step": 440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9847839265982836e-05,
+      "loss": 0.0506,
+      "step": 441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984676478483673e-05,
+      "loss": 0.0452,
+      "step": 442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845686552598495e-05,
+      "loss": 0.0397,
+      "step": 443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9844604569678884e-05,
+      "loss": 0.0415,
+      "step": 444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9843518836490067e-05,
+      "loss": 0.0376,
+      "step": 445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984242935344566e-05,
+      "loss": 0.0483,
+      "step": 446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9841336120960687e-05,
+      "loss": 0.0464,
+      "step": 447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9840239139451622e-05,
+      "loss": 0.034,
+      "step": 448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9839138409336348e-05,
+      "loss": 0.0356,
+      "step": 449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983803393103419e-05,
+      "loss": 0.0431,
+      "step": 450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9836925704965893e-05,
+      "loss": 0.035,
+      "step": 451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9835813731553628e-05,
+      "loss": 0.0385,
+      "step": 452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9834698011221008e-05,
+      "loss": 0.0369,
+      "step": 453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983357854439305e-05,
+      "loss": 0.0464,
+      "step": 454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9832455331496224e-05,
+      "loss": 0.0404,
+      "step": 455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9831328372958406e-05,
+      "loss": 0.0513,
+      "step": 456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9830197669208907e-05,
+      "loss": 0.0428,
+      "step": 457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.982906322067847e-05,
+      "loss": 0.0372,
+      "step": 458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9827925027799254e-05,
+      "loss": 0.0321,
+      "step": 459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9826783091004852e-05,
+      "loss": 0.0337,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9825637410730284e-05,
+      "loss": 0.0407,
+      "step": 461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982448798741199e-05,
+      "loss": 0.0414,
+      "step": 462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9823334821487833e-05,
+      "loss": 0.0399,
+      "step": 463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982217791339712e-05,
+      "loss": 0.0402,
+      "step": 464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982101726358056e-05,
+      "loss": 0.0347,
+      "step": 465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9819852872480312e-05,
+      "loss": 0.0439,
+      "step": 466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981868474053993e-05,
+      "loss": 0.0435,
+      "step": 467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9817512868204425e-05,
+      "loss": 0.036,
+      "step": 468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9816337255920207e-05,
+      "loss": 0.0431,
+      "step": 469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981515790413513e-05,
+      "loss": 0.0352,
+      "step": 470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9813974813298457e-05,
+      "loss": 0.0446,
+      "step": 471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9812787983860887e-05,
+      "loss": 0.0306,
+      "step": 472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981159741627454e-05,
+      "loss": 0.0513,
+      "step": 473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981040311099295e-05,
+      "loss": 0.0406,
+      "step": 474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980920506847109e-05,
+      "loss": 0.0381,
+      "step": 475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9808003289165355e-05,
+      "loss": 0.0553,
+      "step": 476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980679777353355e-05,
+      "loss": 0.0376,
+      "step": 477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980558852203492e-05,
+      "loss": 0.0373,
+      "step": 478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9804375535130118e-05,
+      "loss": 0.0337,
+      "step": 479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980315881328123e-05,
+      "loss": 0.0412,
+      "step": 480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9801938356951764e-05,
+      "loss": 0.0465,
+      "step": 481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9800714166606647e-05,
+      "loss": 0.0446,
+      "step": 482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979948624271223e-05,
+      "loss": 0.0499,
+      "step": 483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979825458573629e-05,
+      "loss": 0.0374,
+      "step": 484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9797019196148018e-05,
+      "loss": 0.0345,
+      "step": 485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9795780074418034e-05,
+      "loss": 0.0531,
+      "step": 486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9794537221018377e-05,
+      "loss": 0.0563,
+      "step": 487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9793290636422503e-05,
+      "loss": 0.0454,
+      "step": 488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9792040321105304e-05,
+      "loss": 0.0493,
+      "step": 489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9790786275543076e-05,
+      "loss": 0.04,
+      "step": 490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9789528500213543e-05,
+      "loss": 0.0423,
+      "step": 491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9788266995595857e-05,
+      "loss": 0.0375,
+      "step": 492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9787001762170578e-05,
+      "loss": 0.0444,
+      "step": 493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9785732800419698e-05,
+      "loss": 0.0504,
+      "step": 494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9784460110826615e-05,
+      "loss": 0.0438,
+      "step": 495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9783183693876164e-05,
+      "loss": 0.0414,
+      "step": 496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9781903550054586e-05,
+      "loss": 0.0388,
+      "step": 497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9780619679849552e-05,
+      "loss": 0.0356,
+      "step": 498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9779332083750145e-05,
+      "loss": 0.0356,
+      "step": 499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9778040762246872e-05,
+      "loss": 0.0441,
+      "step": 500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9776745715831657e-05,
+      "loss": 0.0308,
+      "step": 501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9775446944997847e-05,
+      "loss": 0.0399,
+      "step": 502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97741444502402e-05,
+      "loss": 0.0347,
+      "step": 503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9772838232054897e-05,
+      "loss": 0.0319,
+      "step": 504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977152829093954e-05,
+      "loss": 0.0404,
+      "step": 505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9770214627393144e-05,
+      "loss": 0.0432,
+      "step": 506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9768897241916145e-05,
+      "loss": 0.045,
+      "step": 507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97675761350104e-05,
+      "loss": 0.0412,
+      "step": 508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9766251307179175e-05,
+      "loss": 0.0338,
+      "step": 509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9764922758927167e-05,
+      "loss": 0.0295,
+      "step": 510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9763590490760472e-05,
+      "loss": 0.0323,
+      "step": 511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.976225450318662e-05,
+      "loss": 0.0405,
+      "step": 512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.976091479671455e-05,
+      "loss": 0.0417,
+      "step": 513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9759571371854617e-05,
+      "loss": 0.0287,
+      "step": 514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9758224229118593e-05,
+      "loss": 0.0333,
+      "step": 515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975687336901967e-05,
+      "loss": 0.0336,
+      "step": 516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9755518792072452e-05,
+      "loss": 0.0389,
+      "step": 517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9754160498792964e-05,
+      "loss": 0.0408,
+      "step": 518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975279848969864e-05,
+      "loss": 0.0374,
+      "step": 519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9751432765308335e-05,
+      "loss": 0.037,
+      "step": 520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9750063326142316e-05,
+      "loss": 0.0519,
+      "step": 521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974869017272226e-05,
+      "loss": 0.0369,
+      "step": 522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9747313305571278e-05,
+      "loss": 0.04,
+      "step": 523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9745932725213873e-05,
+      "loss": 0.0456,
+      "step": 524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9744548432175978e-05,
+      "loss": 0.0396,
+      "step": 525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974316042698493e-05,
+      "loss": 0.0399,
+      "step": 526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9741768710169486e-05,
+      "loss": 0.0419,
+      "step": 527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974037328225982e-05,
+      "loss": 0.0364,
+      "step": 528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973897414378751e-05,
+      "loss": 0.0432,
+      "step": 529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973757129528556e-05,
+      "loss": 0.0335,
+      "step": 530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9736164737288374e-05,
+      "loss": 0.0375,
+      "step": 531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9734754470331778e-05,
+      "loss": 0.0393,
+      "step": 532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9733340494953012e-05,
+      "loss": 0.0433,
+      "step": 533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9731922811690718e-05,
+      "loss": 0.0517,
+      "step": 534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973050142108496e-05,
+      "loss": 0.0487,
+      "step": 535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9729076323677217e-05,
+      "loss": 0.0402,
+      "step": 536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9727647520010368e-05,
+      "loss": 0.0503,
+      "step": 537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9726215010628717e-05,
+      "loss": 0.0471,
+      "step": 538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972477879607797e-05,
+      "loss": 0.0401,
+      "step": 539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972333887690525e-05,
+      "loss": 0.0407,
+      "step": 540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9721895253659088e-05,
+      "loss": 0.0371,
+      "step": 541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972044792688943e-05,
+      "loss": 0.034,
+      "step": 542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9718996897147626e-05,
+      "loss": 0.0419,
+      "step": 543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9717542164986447e-05,
+      "loss": 0.035,
+      "step": 544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971608373096006e-05,
+      "loss": 0.0319,
+      "step": 545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9714621595624058e-05,
+      "loss": 0.0337,
+      "step": 546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9713155759535433e-05,
+      "loss": 0.0413,
+      "step": 547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971168622325259e-05,
+      "loss": 0.0336,
+      "step": 548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9710212987335345e-05,
+      "loss": 0.0373,
+      "step": 549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9708736052344927e-05,
+      "loss": 0.0349,
+      "step": 550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970725541884396e-05,
+      "loss": 0.0364,
+      "step": 551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970577108739649e-05,
+      "loss": 0.0375,
+      "step": 552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9704283058567972e-05,
+      "loss": 0.0365,
+      "step": 553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9702791332925262e-05,
+      "loss": 0.035,
+      "step": 554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970129591103663e-05,
+      "loss": 0.0341,
+      "step": 555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969979679347175e-05,
+      "loss": 0.0324,
+      "step": 556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9698293980801708e-05,
+      "loss": 0.0406,
+      "step": 557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9696787473598993e-05,
+      "loss": 0.0319,
+      "step": 558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9695277272437507e-05,
+      "loss": 0.0291,
+      "step": 559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9693763377892548e-05,
+      "loss": 0.0426,
+      "step": 560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969224579054084e-05,
+      "loss": 0.0339,
+      "step": 561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9690724510960496e-05,
+      "loss": 0.0357,
+      "step": 562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.968919953973105e-05,
+      "loss": 0.0397,
+      "step": 563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9687670877433423e-05,
+      "loss": 0.0382,
+      "step": 564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968613852464996e-05,
+      "loss": 0.0407,
+      "step": 565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9684602481964407e-05,
+      "loss": 0.0418,
+      "step": 566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9683062749961915e-05,
+      "loss": 0.0413,
+      "step": 567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968151932922903e-05,
+      "loss": 0.0397,
+      "step": 568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967997222035373e-05,
+      "loss": 0.0454,
+      "step": 569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9678421423925366e-05,
+      "loss": 0.0433,
+      "step": 570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9676866940534714e-05,
+      "loss": 0.0405,
+      "step": 571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9675308770773954e-05,
+      "loss": 0.0349,
+      "step": 572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9673746915236657e-05,
+      "loss": 0.0383,
+      "step": 573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967218137451781e-05,
+      "loss": 0.0427,
+      "step": 574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9670612149213802e-05,
+      "loss": 0.0439,
+      "step": 575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9669039239922422e-05,
+      "loss": 0.0442,
+      "step": 576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966746264724287e-05,
+      "loss": 0.0484,
+      "step": 577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9665882371775735e-05,
+      "loss": 0.0374,
+      "step": 578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9664298414123023e-05,
+      "loss": 0.0444,
+      "step": 579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966271077488814e-05,
+      "loss": 0.0322,
+      "step": 580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9661119454675884e-05,
+      "loss": 0.0417,
+      "step": 581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9659524454092472e-05,
+      "loss": 0.0395,
+      "step": 582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965792577374551e-05,
+      "loss": 0.0439,
+      "step": 583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9656323414244004e-05,
+      "loss": 0.0459,
+      "step": 584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9654717376198377e-05,
+      "loss": 0.0427,
+      "step": 585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965310766022044e-05,
+      "loss": 0.0387,
+      "step": 586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965149426692341e-05,
+      "loss": 0.035,
+      "step": 587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96498771969219e-05,
+      "loss": 0.0459,
+      "step": 588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964825645083193e-05,
+      "loss": 0.0484,
+      "step": 589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9646632029270918e-05,
+      "loss": 0.05,
+      "step": 590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9645003932857684e-05,
+      "loss": 0.0451,
+      "step": 591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9643372162212443e-05,
+      "loss": 0.0338,
+      "step": 592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964173671795681e-05,
+      "loss": 0.0454,
+      "step": 593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9640097600713808e-05,
+      "loss": 0.0447,
+      "step": 594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963845481110785e-05,
+      "loss": 0.0465,
+      "step": 595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9636808349764752e-05,
+      "loss": 0.0423,
+      "step": 596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9635158217311727e-05,
+      "loss": 0.0528,
+      "step": 597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963350441437739e-05,
+      "loss": 0.0337,
+      "step": 598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9631846941591748e-05,
+      "loss": 0.0387,
+      "step": 599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9630185799586213e-05,
+      "loss": 0.0431,
+      "step": 600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962852098899359e-05,
+      "loss": 0.0402,
+      "step": 601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962685251044808e-05,
+      "loss": 0.0363,
+      "step": 602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962518036458529e-05,
+      "loss": 0.0376,
+      "step": 603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9623504552042214e-05,
+      "loss": 0.0463,
+      "step": 604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962182507345725e-05,
+      "loss": 0.0388,
+      "step": 605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9620141929470193e-05,
+      "loss": 0.0354,
+      "step": 606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961845512072222e-05,
+      "loss": 0.0396,
+      "step": 607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9616764647855924e-05,
+      "loss": 0.0461,
+      "step": 608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9615070511515285e-05,
+      "loss": 0.0331,
+      "step": 609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9613372712345674e-05,
+      "loss": 0.0351,
+      "step": 610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9611671250993868e-05,
+      "loss": 0.034,
+      "step": 611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9609966128108028e-05,
+      "loss": 0.0462,
+      "step": 612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9608257344337713e-05,
+      "loss": 0.0426,
+      "step": 613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9606544900333884e-05,
+      "loss": 0.0326,
+      "step": 614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960482879674889e-05,
+      "loss": 0.0421,
+      "step": 615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9603109034236472e-05,
+      "loss": 0.0382,
+      "step": 616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9601385613451768e-05,
+      "loss": 0.0436,
+      "step": 617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9599658535051312e-05,
+      "loss": 0.0325,
+      "step": 618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959792779969303e-05,
+      "loss": 0.0276,
+      "step": 619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9596193408036233e-05,
+      "loss": 0.0359,
+      "step": 620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959445536074164e-05,
+      "loss": 0.035,
+      "step": 621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9592713658471347e-05,
+      "loss": 0.0405,
+      "step": 622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9590968301888856e-05,
+      "loss": 0.031,
+      "step": 623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958921929165905e-05,
+      "loss": 0.0434,
+      "step": 624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9587466628448206e-05,
+      "loss": 0.0442,
+      "step": 625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9585710312924005e-05,
+      "loss": 0.033,
+      "step": 626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9583950345755502e-05,
+      "loss": 0.0411,
+      "step": 627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9582186727613152e-05,
+      "loss": 0.0442,
+      "step": 628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9580419459168804e-05,
+      "loss": 0.0397,
+      "step": 629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9578648541095682e-05,
+      "loss": 0.0403,
+      "step": 630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9576873974068426e-05,
+      "loss": 0.038,
+      "step": 631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9575095758763042e-05,
+      "loss": 0.0412,
+      "step": 632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9573313895856933e-05,
+      "loss": 0.0446,
+      "step": 633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95715283860289e-05,
+      "loss": 0.0454,
+      "step": 634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9569739229959125e-05,
+      "loss": 0.05,
+      "step": 635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956794642832918e-05,
+      "loss": 0.0385,
+      "step": 636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9566149981822028e-05,
+      "loss": 0.0519,
+      "step": 637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9564349891122017e-05,
+      "loss": 0.0444,
+      "step": 638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956254615691489e-05,
+      "loss": 0.0409,
+      "step": 639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956073877988777e-05,
+      "loss": 0.0379,
+      "step": 640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955892776072917e-05,
+      "loss": 0.049,
+      "step": 641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9557113100128996e-05,
+      "loss": 0.0409,
+      "step": 642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9555294798778533e-05,
+      "loss": 0.0414,
+      "step": 643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955347285737046e-05,
+      "loss": 0.0359,
+      "step": 644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955164727659884e-05,
+      "loss": 0.0421,
+      "step": 645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9549818057159114e-05,
+      "loss": 0.0372,
+      "step": 646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954798519974813e-05,
+      "loss": 0.0348,
+      "step": 647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9546148705064097e-05,
+      "loss": 0.0405,
+      "step": 648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9544308573806624e-05,
+      "loss": 0.0349,
+      "step": 649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9542464806676713e-05,
+      "loss": 0.0378,
+      "step": 650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9540617404376724e-05,
+      "loss": 0.0357,
+      "step": 651
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9538766367610433e-05,
+      "loss": 0.0391,
+      "step": 652
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953691169708298e-05,
+      "loss": 0.0365,
+      "step": 653
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95350533935009e-05,
+      "loss": 0.0349,
+      "step": 654
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9533191457572097e-05,
+      "loss": 0.0367,
+      "step": 655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9531325890005882e-05,
+      "loss": 0.0315,
+      "step": 656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952945669151293e-05,
+      "loss": 0.0488,
+      "step": 657
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9527583862805303e-05,
+      "loss": 0.0398,
+      "step": 658
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9525707404596455e-05,
+      "loss": 0.0362,
+      "step": 659
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9523827317601216e-05,
+      "loss": 0.0367,
+      "step": 660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95219436025358e-05,
+      "loss": 0.0397,
+      "step": 661
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9520056260117793e-05,
+      "loss": 0.0407,
+      "step": 662
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9518165291066182e-05,
+      "loss": 0.029,
+      "step": 663
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951627069610132e-05,
+      "loss": 0.0326,
+      "step": 664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9514372475944948e-05,
+      "loss": 0.0314,
+      "step": 665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951247063132019e-05,
+      "loss": 0.0314,
+      "step": 666
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.0321,
+      "step": 667
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508656071564883e-05,
+      "loss": 0.0399,
+      "step": 668
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9506743357887482e-05,
+      "loss": 0.0344,
+      "step": 669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9504827022647978e-05,
+      "loss": 0.0345,
+      "step": 670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9502907066576395e-05,
+      "loss": 0.0332,
+      "step": 671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9500983490404133e-05,
+      "loss": 0.0308,
+      "step": 672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9499056294863967e-05,
+      "loss": 0.0368,
+      "step": 673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949712548069006e-05,
+      "loss": 0.0403,
+      "step": 674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9495191048617954e-05,
+      "loss": 0.0316,
+      "step": 675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9493252999384556e-05,
+      "loss": 0.0403,
+      "step": 676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949131133372816e-05,
+      "loss": 0.0352,
+      "step": 677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9489366052388443e-05,
+      "loss": 0.0427,
+      "step": 678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9487417156106447e-05,
+      "loss": 0.0315,
+      "step": 679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9485464645624603e-05,
+      "loss": 0.0336,
+      "step": 680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948350852168671e-05,
+      "loss": 0.0421,
+      "step": 681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948154878503795e-05,
+      "loss": 0.0508,
+      "step": 682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947958543642487e-05,
+      "loss": 0.0341,
+      "step": 683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947761847659541e-05,
+      "loss": 0.0427,
+      "step": 684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9475647906298876e-05,
+      "loss": 0.0357,
+      "step": 685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9473673726285947e-05,
+      "loss": 0.0305,
+      "step": 686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947169593730868e-05,
+      "loss": 0.0334,
+      "step": 687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946971454012051e-05,
+      "loss": 0.034,
+      "step": 688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9467729535476238e-05,
+      "loss": 0.0376,
+      "step": 689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9465740924132053e-05,
+      "loss": 0.0389,
+      "step": 690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9463748706845504e-05,
+      "loss": 0.036,
+      "step": 691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946175288437552e-05,
+      "loss": 0.0369,
+      "step": 692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9459753457482408e-05,
+      "loss": 0.0479,
+      "step": 693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9457750426927833e-05,
+      "loss": 0.0463,
+      "step": 694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9455743793474852e-05,
+      "loss": 0.041,
+      "step": 695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9453733557887883e-05,
+      "loss": 0.0261,
+      "step": 696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9451719720932717e-05,
+      "loss": 0.0336,
+      "step": 697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9449702283376516e-05,
+      "loss": 0.037,
+      "step": 698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9447681245987822e-05,
+      "loss": 0.0464,
+      "step": 699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944565660953654e-05,
+      "loss": 0.0417,
+      "step": 700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9443628374793947e-05,
+      "loss": 0.0464,
+      "step": 701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9441596542532696e-05,
+      "loss": 0.0469,
+      "step": 702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9439561113526802e-05,
+      "loss": 0.0403,
+      "step": 703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943752208855166e-05,
+      "loss": 0.0369,
+      "step": 704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9435479468384022e-05,
+      "loss": 0.0345,
+      "step": 705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9433433253802025e-05,
+      "loss": 0.0372,
+      "step": 706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9431383445585163e-05,
+      "loss": 0.0275,
+      "step": 707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9429330044514305e-05,
+      "loss": 0.0339,
+      "step": 708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942727305137169e-05,
+      "loss": 0.0332,
+      "step": 709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9425212466940916e-05,
+      "loss": 0.0432,
+      "step": 710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9423148292006963e-05,
+      "loss": 0.0441,
+      "step": 711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942108052735617e-05,
+      "loss": 0.0326,
+      "step": 712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941900917377624e-05,
+      "loss": 0.0448,
+      "step": 713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9416934232056253e-05,
+      "loss": 0.0356,
+      "step": 714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941485570298665e-05,
+      "loss": 0.0415,
+      "step": 715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941277358735924e-05,
+      "loss": 0.0365,
+      "step": 716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9410687885967198e-05,
+      "loss": 0.0407,
+      "step": 717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9408598599605062e-05,
+      "loss": 0.0338,
+      "step": 718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9406505729068744e-05,
+      "loss": 0.0274,
+      "step": 719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940440927515551e-05,
+      "loss": 0.0422,
+      "step": 720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9402309238664e-05,
+      "loss": 0.0395,
+      "step": 721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.940020562039422e-05,
+      "loss": 0.047,
+      "step": 722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939809842114753e-05,
+      "loss": 0.0319,
+      "step": 723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9395987641726665e-05,
+      "loss": 0.026,
+      "step": 724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9393873282935717e-05,
+      "loss": 0.047,
+      "step": 725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9391755345580142e-05,
+      "loss": 0.0368,
+      "step": 726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9389633830466767e-05,
+      "loss": 0.0343,
+      "step": 727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938750873840377e-05,
+      "loss": 0.0368,
+      "step": 728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.93853800702007e-05,
+      "loss": 0.0321,
+      "step": 729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9383247826668466e-05,
+      "loss": 0.0424,
+      "step": 730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938111200861934e-05,
+      "loss": 0.0361,
+      "step": 731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9378972616866952e-05,
+      "loss": 0.0447,
+      "step": 732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9376829652226303e-05,
+      "loss": 0.0456,
+      "step": 733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937468311551374e-05,
+      "loss": 0.0416,
+      "step": 734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9372533007546984e-05,
+      "loss": 0.0309,
+      "step": 735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9370379329145107e-05,
+      "loss": 0.0458,
+      "step": 736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9368222081128554e-05,
+      "loss": 0.0356,
+      "step": 737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9366061264319112e-05,
+      "loss": 0.0353,
+      "step": 738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936389687953994e-05,
+      "loss": 0.0426,
+      "step": 739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9361728927615557e-05,
+      "loss": 0.0341,
+      "step": 740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9359557409371832e-05,
+      "loss": 0.046,
+      "step": 741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9357382325636e-05,
+      "loss": 0.0359,
+      "step": 742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935520367723665e-05,
+      "loss": 0.0335,
+      "step": 743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9353021465003734e-05,
+      "loss": 0.0428,
+      "step": 744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935083568976856e-05,
+      "loss": 0.0417,
+      "step": 745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9348646352363784e-05,
+      "loss": 0.0299,
+      "step": 746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.934645345362343e-05,
+      "loss": 0.0281,
+      "step": 747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934425699438288e-05,
+      "loss": 0.0323,
+      "step": 748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9342056975478865e-05,
+      "loss": 0.0357,
+      "step": 749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9339853397749473e-05,
+      "loss": 0.0243,
+      "step": 750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933764626203415e-05,
+      "loss": 0.0328,
+      "step": 751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93354355691737e-05,
+      "loss": 0.0362,
+      "step": 752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9333221320010275e-05,
+      "loss": 0.0356,
+      "step": 753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9331003515387388e-05,
+      "loss": 0.0335,
+      "step": 754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9328782156149905e-05,
+      "loss": 0.0358,
+      "step": 755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9326557243144048e-05,
+      "loss": 0.0308,
+      "step": 756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9324328777217382e-05,
+      "loss": 0.0427,
+      "step": 757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9322096759218838e-05,
+      "loss": 0.0444,
+      "step": 758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93198611899987e-05,
+      "loss": 0.0382,
+      "step": 759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9317622070408594e-05,
+      "loss": 0.037,
+      "step": 760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9315379401301507e-05,
+      "loss": 0.0326,
+      "step": 761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9313133183531775e-05,
+      "loss": 0.0416,
+      "step": 762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9310883417955094e-05,
+      "loss": 0.0318,
+      "step": 763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93086301054285e-05,
+      "loss": 0.0426,
+      "step": 764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9306373246810383e-05,
+      "loss": 0.0368,
+      "step": 765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9304112842960484e-05,
+      "loss": 0.0359,
+      "step": 766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9301848894739906e-05,
+      "loss": 0.0324,
+      "step": 767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9299581403011082e-05,
+      "loss": 0.0348,
+      "step": 768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9297310368637815e-05,
+      "loss": 0.031,
+      "step": 769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9295035792485238e-05,
+      "loss": 0.0335,
+      "step": 770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.929275767541985e-05,
+      "loss": 0.032,
+      "step": 771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9290476018309487e-05,
+      "loss": 0.0326,
+      "step": 772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9288190822023343e-05,
+      "loss": 0.0293,
+      "step": 773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9285902087431957e-05,
+      "loss": 0.0344,
+      "step": 774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9283609815407208e-05,
+      "loss": 0.0341,
+      "step": 775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9281314006822338e-05,
+      "loss": 0.0367,
+      "step": 776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9279014662551924e-05,
+      "loss": 0.0348,
+      "step": 777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9276711783471888e-05,
+      "loss": 0.0356,
+      "step": 778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9274405370459512e-05,
+      "loss": 0.0287,
+      "step": 779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927209542439341e-05,
+      "loss": 0.0383,
+      "step": 780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9269781946153557e-05,
+      "loss": 0.0362,
+      "step": 781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9267464936621253e-05,
+      "loss": 0.0245,
+      "step": 782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9265144396679166e-05,
+      "loss": 0.028,
+      "step": 783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926282032721129e-05,
+      "loss": 0.0372,
+      "step": 784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9260492729102974e-05,
+      "loss": 0.0272,
+      "step": 785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.925816160324091e-05,
+      "loss": 0.0289,
+      "step": 786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.925582695051314e-05,
+      "loss": 0.0331,
+      "step": 787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9253488771809024e-05,
+      "loss": 0.0298,
+      "step": 788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9251147068019298e-05,
+      "loss": 0.0488,
+      "step": 789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9248801840036022e-05,
+      "loss": 0.0427,
+      "step": 790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.92464530887526e-05,
+      "loss": 0.0467,
+      "step": 791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9244100815063787e-05,
+      "loss": 0.0297,
+      "step": 792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9241745019865666e-05,
+      "loss": 0.0322,
+      "step": 793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9239385704055676e-05,
+      "loss": 0.0437,
+      "step": 794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9237022868532587e-05,
+      "loss": 0.0309,
+      "step": 795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9234656514196514e-05,
+      "loss": 0.039,
+      "step": 796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9232286641948917e-05,
+      "loss": 0.0417,
+      "step": 797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.922991325269258e-05,
+      "loss": 0.0377,
+      "step": 798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9227536347331647e-05,
+      "loss": 0.0342,
+      "step": 799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.922515592677159e-05,
+      "loss": 0.0303,
+      "step": 800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9222771991919216e-05,
+      "loss": 0.0435,
+      "step": 801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9220384543682686e-05,
+      "loss": 0.0353,
+      "step": 802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9217993582971485e-05,
+      "loss": 0.0379,
+      "step": 803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9215599110696442e-05,
+      "loss": 0.0396,
+      "step": 804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9213201127769726e-05,
+      "loss": 0.0328,
+      "step": 805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921079963510484e-05,
+      "loss": 0.043,
+      "step": 806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9208394633616624e-05,
+      "loss": 0.0413,
+      "step": 807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920598612422125e-05,
+      "loss": 0.03,
+      "step": 808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920357410783624e-05,
+      "loss": 0.0382,
+      "step": 809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9201158585380438e-05,
+      "loss": 0.0405,
+      "step": 810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9198739557774035e-05,
+      "loss": 0.0276,
+      "step": 811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9196317025938545e-05,
+      "loss": 0.0352,
+      "step": 812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9193890990796826e-05,
+      "loss": 0.0392,
+      "step": 813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9191461453273067e-05,
+      "loss": 0.0394,
+      "step": 814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9189028414292795e-05,
+      "loss": 0.039,
+      "step": 815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9186591874782864e-05,
+      "loss": 0.0338,
+      "step": 816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918415183567147e-05,
+      "loss": 0.0315,
+      "step": 817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9181708297888133e-05,
+      "loss": 0.0344,
+      "step": 818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9179261262363717e-05,
+      "loss": 0.0306,
+      "step": 819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9176810730030406e-05,
+      "loss": 0.0299,
+      "step": 820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9174356701821722e-05,
+      "loss": 0.0402,
+      "step": 821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9171899178672523e-05,
+      "loss": 0.0487,
+      "step": 822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9169438161518995e-05,
+      "loss": 0.0352,
+      "step": 823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9166973651298647e-05,
+      "loss": 0.0298,
+      "step": 824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9164505648950335e-05,
+      "loss": 0.0262,
+      "step": 825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9162034155414225e-05,
+      "loss": 0.0382,
+      "step": 826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9159559171631836e-05,
+      "loss": 0.0333,
+      "step": 827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9157080698546e-05,
+      "loss": 0.0268,
+      "step": 828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9154598737100884e-05,
+      "loss": 0.0337,
+      "step": 829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9152113288241976e-05,
+      "loss": 0.0321,
+      "step": 830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914962435291611e-05,
+      "loss": 0.0329,
+      "step": 831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914713193207143e-05,
+      "loss": 0.0415,
+      "step": 832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914463602665742e-05,
+      "loss": 0.0306,
+      "step": 833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9142136637624884e-05,
+      "loss": 0.0292,
+      "step": 834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9139633765925957e-05,
+      "loss": 0.0405,
+      "step": 835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.91371274125141e-05,
+      "loss": 0.0284,
+      "step": 836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.91346175783441e-05,
+      "loss": 0.0334,
+      "step": 837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9132104264372065e-05,
+      "loss": 0.0348,
+      "step": 838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912958747155544e-05,
+      "loss": 0.0394,
+      "step": 839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9127067200852987e-05,
+      "loss": 0.0322,
+      "step": 840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9124543453224796e-05,
+      "loss": 0.0323,
+      "step": 841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912201622963228e-05,
+      "loss": 0.0419,
+      "step": 842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.911948553103817e-05,
+      "loss": 0.0436,
+      "step": 843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9116951358406534e-05,
+      "loss": 0.0307,
+      "step": 844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9114413712702755e-05,
+      "loss": 0.0419,
+      "step": 845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9111872594893542e-05,
+      "loss": 0.0392,
+      "step": 846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9109328005946924e-05,
+      "loss": 0.0298,
+      "step": 847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.910677994683225e-05,
+      "loss": 0.0302,
+      "step": 848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9104228418520198e-05,
+      "loss": 0.0328,
+      "step": 849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9101673421982768e-05,
+      "loss": 0.0365,
+      "step": 850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9099114958193272e-05,
+      "loss": 0.0291,
+      "step": 851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9096553028126347e-05,
+      "loss": 0.0376,
+      "step": 852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9093987632757957e-05,
+      "loss": 0.0443,
+      "step": 853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.909141877306538e-05,
+      "loss": 0.0364,
+      "step": 854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908884645002721e-05,
+      "loss": 0.0313,
+      "step": 855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9086270664623367e-05,
+      "loss": 0.0294,
+      "step": 856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9083691417835092e-05,
+      "loss": 0.0286,
+      "step": 857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9081108710644933e-05,
+      "loss": 0.0394,
+      "step": 858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907852254403677e-05,
+      "loss": 0.0255,
+      "step": 859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907593291899579e-05,
+      "loss": 0.0281,
+      "step": 860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9073339836508508e-05,
+      "loss": 0.0374,
+      "step": 861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907074329756274e-05,
+      "loss": 0.0342,
+      "step": 862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.906814330314764e-05,
+      "loss": 0.0379,
+      "step": 863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.906553985425366e-05,
+      "loss": 0.0419,
+      "step": 864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9062932951872578e-05,
+      "loss": 0.0271,
+      "step": 865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9060322596997486e-05,
+      "loss": 0.0285,
+      "step": 866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9057708790622786e-05,
+      "loss": 0.0319,
+      "step": 867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9055091533744204e-05,
+      "loss": 0.0333,
+      "step": 868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9052470827358766e-05,
+      "loss": 0.0274,
+      "step": 869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9049846672464835e-05,
+      "loss": 0.0431,
+      "step": 870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904721907006206e-05,
+      "loss": 0.0394,
+      "step": 871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9044588021151427e-05,
+      "loss": 0.0366,
+      "step": 872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904195352673522e-05,
+      "loss": 0.0411,
+      "step": 873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9039315587817047e-05,
+      "loss": 0.0233,
+      "step": 874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9036674205401816e-05,
+      "loss": 0.0341,
+      "step": 875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9034029380495752e-05,
+      "loss": 0.0377,
+      "step": 876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9031381114106397e-05,
+      "loss": 0.0353,
+      "step": 877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9028729407242598e-05,
+      "loss": 0.0315,
+      "step": 878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9026074260914513e-05,
+      "loss": 0.0472,
+      "step": 879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.902341567613361e-05,
+      "loss": 0.0342,
+      "step": 880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.902075365391267e-05,
+      "loss": 0.0364,
+      "step": 881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901808819526578e-05,
+      "loss": 0.0353,
+      "step": 882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901541930120834e-05,
+      "loss": 0.0269,
+      "step": 883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9012746972757056e-05,
+      "loss": 0.0498,
+      "step": 884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901007121092994e-05,
+      "loss": 0.0387,
+      "step": 885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9007392016746314e-05,
+      "loss": 0.0348,
+      "step": 886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900470939122681e-05,
+      "loss": 0.0406,
+      "step": 887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9002023335393366e-05,
+      "loss": 0.0273,
+      "step": 888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8999333850269223e-05,
+      "loss": 0.0344,
+      "step": 889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8996640936878935e-05,
+      "loss": 0.0401,
+      "step": 890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8993944596248357e-05,
+      "loss": 0.0331,
+      "step": 891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8991244829404646e-05,
+      "loss": 0.0239,
+      "step": 892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8988541637376282e-05,
+      "loss": 0.0335,
+      "step": 893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898583502119302e-05,
+      "loss": 0.0347,
+      "step": 894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8983124981885953e-05,
+      "loss": 0.0411,
+      "step": 895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8980411520487447e-05,
+      "loss": 0.0336,
+      "step": 896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8977694638031196e-05,
+      "loss": 0.0334,
+      "step": 897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897497433555218e-05,
+      "loss": 0.0318,
+      "step": 898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8972250614086697e-05,
+      "loss": 0.035,
+      "step": 899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8969523474672335e-05,
+      "loss": 0.034,
+      "step": 900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896679291834799e-05,
+      "loss": 0.0378,
+      "step": 901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8964058946153856e-05,
+      "loss": 0.0334,
+      "step": 902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896132155913143e-05,
+      "loss": 0.0337,
+      "step": 903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8958580758323516e-05,
+      "loss": 0.0341,
+      "step": 904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8955836544774208e-05,
+      "loss": 0.0306,
+      "step": 905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.895308891952891e-05,
+      "loss": 0.0225,
+      "step": 906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.895033788363431e-05,
+      "loss": 0.0261,
+      "step": 907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894758343813842e-05,
+      "loss": 0.0337,
+      "step": 908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8944825584090526e-05,
+      "loss": 0.0277,
+      "step": 909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8942064322541226e-05,
+      "loss": 0.029,
+      "step": 910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8939299654542417e-05,
+      "loss": 0.0334,
+      "step": 911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8936531581147285e-05,
+      "loss": 0.0316,
+      "step": 912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.893376010341032e-05,
+      "loss": 0.0431,
+      "step": 913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8930985222387312e-05,
+      "loss": 0.0292,
+      "step": 914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8928206939135338e-05,
+      "loss": 0.0325,
+      "step": 915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8925425254712774e-05,
+      "loss": 0.0345,
+      "step": 916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8922640170179297e-05,
+      "loss": 0.0295,
+      "step": 917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8919851686595875e-05,
+      "loss": 0.0319,
+      "step": 918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891705980502477e-05,
+      "loss": 0.0317,
+      "step": 919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891426452652954e-05,
+      "loss": 0.036,
+      "step": 920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8911465852175038e-05,
+      "loss": 0.0304,
+      "step": 921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890866378302741e-05,
+      "loss": 0.0339,
+      "step": 922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8905858320154093e-05,
+      "loss": 0.0359,
+      "step": 923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8903049464623825e-05,
+      "loss": 0.0258,
+      "step": 924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890023721750662e-05,
+      "loss": 0.0294,
+      "step": 925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.88974215798738e-05,
+      "loss": 0.0393,
+      "step": 926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8894602552797973e-05,
+      "loss": 0.0304,
+      "step": 927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8891780137353036e-05,
+      "loss": 0.0338,
+      "step": 928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8888954334614178e-05,
+      "loss": 0.0324,
+      "step": 929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8886125145657887e-05,
+      "loss": 0.0286,
+      "step": 930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.888329257156192e-05,
+      "loss": 0.033,
+      "step": 931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8880456613405346e-05,
+      "loss": 0.0299,
+      "step": 932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8877617272268512e-05,
+      "loss": 0.036,
+      "step": 933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8874774549233056e-05,
+      "loss": 0.0249,
+      "step": 934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.88719284453819e-05,
+      "loss": 0.0292,
+      "step": 935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8869078961799263e-05,
+      "loss": 0.0335,
+      "step": 936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.886622609957064e-05,
+      "loss": 0.0296,
+      "step": 937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8863369859782824e-05,
+      "loss": 0.0314,
+      "step": 938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.886051024352389e-05,
+      "loss": 0.0367,
+      "step": 939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8857647251883196e-05,
+      "loss": 0.0259,
+      "step": 940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8854780885951397e-05,
+      "loss": 0.0249,
+      "step": 941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8851911146820417e-05,
+      "loss": 0.0352,
+      "step": 942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8849038035583474e-05,
+      "loss": 0.0313,
+      "step": 943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8846161553335073e-05,
+      "loss": 0.0315,
+      "step": 944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8843281701171e-05,
+      "loss": 0.0277,
+      "step": 945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8840398480188326e-05,
+      "loss": 0.0338,
+      "step": 946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8837511891485403e-05,
+      "loss": 0.0262,
+      "step": 947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883462193616187e-05,
+      "loss": 0.0284,
+      "step": 948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883172861531864e-05,
+      "loss": 0.0296,
+      "step": 949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8828831930057917e-05,
+      "loss": 0.0351,
+      "step": 950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8825931881483184e-05,
+      "loss": 0.0272,
+      "step": 951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8823028470699206e-05,
+      "loss": 0.0331,
+      "step": 952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8820121698812028e-05,
+      "loss": 0.0406,
+      "step": 953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8817211566928967e-05,
+      "loss": 0.0308,
+      "step": 954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8814298076158638e-05,
+      "loss": 0.0392,
+      "step": 955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8811381227610925e-05,
+      "loss": 0.0389,
+      "step": 956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8808461022396986e-05,
+      "loss": 0.0312,
+      "step": 957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8805537461629266e-05,
+      "loss": 0.0199,
+      "step": 958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8802610546421485e-05,
+      "loss": 0.0346,
+      "step": 959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879968027788864e-05,
+      "loss": 0.0313,
+      "step": 960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8796746657147014e-05,
+      "loss": 0.0326,
+      "step": 961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879380968531415e-05,
+      "loss": 0.0341,
+      "step": 962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879086936350889e-05,
+      "loss": 0.0346,
+      "step": 963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8787925692851326e-05,
+      "loss": 0.0396,
+      "step": 964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.878497867446285e-05,
+      "loss": 0.031,
+      "step": 965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8782028309466117e-05,
+      "loss": 0.0429,
+      "step": 966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877907459898505e-05,
+      "loss": 0.0358,
+      "step": 967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8776117544144866e-05,
+      "loss": 0.0309,
+      "step": 968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8773157146072042e-05,
+      "loss": 0.0307,
+      "step": 969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8770193405894325e-05,
+      "loss": 0.0326,
+      "step": 970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8767226324740753e-05,
+      "loss": 0.0393,
+      "step": 971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876425590374162e-05,
+      "loss": 0.0309,
+      "step": 972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8761282144028493e-05,
+      "loss": 0.0306,
+      "step": 973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.875830504673423e-05,
+      "loss": 0.0299,
+      "step": 974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8755324612992928e-05,
+      "loss": 0.0196,
+      "step": 975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.875234084393999e-05,
+      "loss": 0.0328,
+      "step": 976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874935374071206e-05,
+      "loss": 0.0376,
+      "step": 977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8746363304447073e-05,
+      "loss": 0.0308,
+      "step": 978
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8743369536284222e-05,
+      "loss": 0.0255,
+      "step": 979
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8740372437363973e-05,
+      "loss": 0.0324,
+      "step": 980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8737372008828063e-05,
+      "loss": 0.0279,
+      "step": 981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8734368251819494e-05,
+      "loss": 0.0233,
+      "step": 982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873136116748254e-05,
+      "loss": 0.0274,
+      "step": 983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872835075696273e-05,
+      "loss": 0.0263,
+      "step": 984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8725337021406887e-05,
+      "loss": 0.0246,
+      "step": 985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8722319961963067e-05,
+      "loss": 0.0334,
+      "step": 986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8719299579780614e-05,
+      "loss": 0.0253,
+      "step": 987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8716275876010135e-05,
+      "loss": 0.041,
+      "step": 988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.87132488518035e-05,
+      "loss": 0.0279,
+      "step": 989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8710218508313838e-05,
+      "loss": 0.0253,
+      "step": 990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8707184846695556e-05,
+      "loss": 0.0256,
+      "step": 991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8704147868104313e-05,
+      "loss": 0.0303,
+      "step": 992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8701107573697033e-05,
+      "loss": 0.0299,
+      "step": 993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8698063964631917e-05,
+      "loss": 0.0371,
+      "step": 994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8695017042068402e-05,
+      "loss": 0.0299,
+      "step": 995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8691966807167216e-05,
+      "loss": 0.0359,
+      "step": 996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868891326109033e-05,
+      "loss": 0.0329,
+      "step": 997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8685856405000984e-05,
+      "loss": 0.0437,
+      "step": 998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8682796240063675e-05,
+      "loss": 0.0268,
+      "step": 999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.867973276744416e-05,
+      "loss": 0.0344,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8676665988309465e-05,
+      "loss": 0.0407,
+      "step": 1001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8673595903827868e-05,
+      "loss": 0.0256,
+      "step": 1002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.867052251516891e-05,
+      "loss": 0.033,
+      "step": 1003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8667445823503375e-05,
+      "loss": 0.0283,
+      "step": 1004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8664365830003334e-05,
+      "loss": 0.0422,
+      "step": 1005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866128253584209e-05,
+      "loss": 0.0348,
+      "step": 1006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8658195942194222e-05,
+      "loss": 0.0328,
+      "step": 1007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.865510605023555e-05,
+      "loss": 0.0326,
+      "step": 1008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8652012861143164e-05,
+      "loss": 0.0418,
+      "step": 1009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8648916376095396e-05,
+      "loss": 0.0378,
+      "step": 1010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.864581659627185e-05,
+      "loss": 0.0339,
+      "step": 1011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8642713522853373e-05,
+      "loss": 0.0389,
+      "step": 1012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863960715702207e-05,
+      "loss": 0.0342,
+      "step": 1013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8636497499961304e-05,
+      "loss": 0.039,
+      "step": 1014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8633384552855684e-05,
+      "loss": 0.0374,
+      "step": 1015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8630268316891078e-05,
+      "loss": 0.0255,
+      "step": 1016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8627148793254606e-05,
+      "loss": 0.0324,
+      "step": 1017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8624025983134643e-05,
+      "loss": 0.0285,
+      "step": 1018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862089988772081e-05,
+      "loss": 0.0319,
+      "step": 1019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8617770508203984e-05,
+      "loss": 0.0371,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.861463784577629e-05,
+      "loss": 0.034,
+      "step": 1021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8611501901631104e-05,
+      "loss": 0.0255,
+      "step": 1022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860836267696306e-05,
+      "loss": 0.0306,
+      "step": 1023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8605220172968028e-05,
+      "loss": 0.0346,
+      "step": 1024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8602074390843138e-05,
+      "loss": 0.0346,
+      "step": 1025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8598925331786764e-05,
+      "loss": 0.0332,
+      "step": 1026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859577299699853e-05,
+      "loss": 0.0348,
+      "step": 1027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8592617387679304e-05,
+      "loss": 0.0331,
+      "step": 1028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8589458505031215e-05,
+      "loss": 0.0275,
+      "step": 1029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8586296350257615e-05,
+      "loss": 0.0231,
+      "step": 1030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8583130924563127e-05,
+      "loss": 0.0411,
+      "step": 1031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8579962229153608e-05,
+      "loss": 0.026,
+      "step": 1032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8576790265236154e-05,
+      "loss": 0.0244,
+      "step": 1033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8573615034019122e-05,
+      "loss": 0.0301,
+      "step": 1034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8570436536712103e-05,
+      "loss": 0.0337,
+      "step": 1035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8567254774525936e-05,
+      "loss": 0.0308,
+      "step": 1036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8564069748672702e-05,
+      "loss": 0.0338,
+      "step": 1037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8560881460365726e-05,
+      "loss": 0.0311,
+      "step": 1038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8557689910819573e-05,
+      "loss": 0.0261,
+      "step": 1039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8554495101250056e-05,
+      "loss": 0.0379,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8551297032874228e-05,
+      "loss": 0.0385,
+      "step": 1041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.854809570691038e-05,
+      "loss": 0.0303,
+      "step": 1042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.854489112457805e-05,
+      "loss": 0.0393,
+      "step": 1043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.854168328709801e-05,
+      "loss": 0.024,
+      "step": 1044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853847219569227e-05,
+      "loss": 0.0294,
+      "step": 1045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.85352578515841e-05,
+      "loss": 0.0338,
+      "step": 1046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8532040255997972e-05,
+      "loss": 0.0328,
+      "step": 1047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8528819410159638e-05,
+      "loss": 0.0356,
+      "step": 1048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852559531529606e-05,
+      "loss": 0.028,
+      "step": 1049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8522367972635442e-05,
+      "loss": 0.0291,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.851913738340724e-05,
+      "loss": 0.0301,
+      "step": 1051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8515903548842125e-05,
+      "loss": 0.0325,
+      "step": 1052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8512666470172024e-05,
+      "loss": 0.0271,
+      "step": 1053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8509426148630085e-05,
+      "loss": 0.0346,
+      "step": 1054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8506182585450704e-05,
+      "loss": 0.035,
+      "step": 1055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8502935781869498e-05,
+      "loss": 0.0317,
+      "step": 1056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.849968573912333e-05,
+      "loss": 0.0338,
+      "step": 1057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8496432458450297e-05,
+      "loss": 0.035,
+      "step": 1058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8493175941089718e-05,
+      "loss": 0.0294,
+      "step": 1059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8489916188282157e-05,
+      "loss": 0.0323,
+      "step": 1060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8486653201269406e-05,
+      "loss": 0.031,
+      "step": 1061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8483386981294488e-05,
+      "loss": 0.028,
+      "step": 1062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8480117529601654e-05,
+      "loss": 0.0239,
+      "step": 1063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.84768448474364e-05,
+      "loss": 0.0324,
+      "step": 1064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8473568936045435e-05,
+      "loss": 0.0292,
+      "step": 1065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.847028979667671e-05,
+      "loss": 0.0351,
+      "step": 1066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.84670074305794e-05,
+      "loss": 0.0257,
+      "step": 1067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8463721839003917e-05,
+      "loss": 0.0327,
+      "step": 1068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8460433023201893e-05,
+      "loss": 0.0341,
+      "step": 1069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8457140984426187e-05,
+      "loss": 0.0296,
+      "step": 1070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8453845723930897e-05,
+      "loss": 0.0274,
+      "step": 1071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8450547242971337e-05,
+      "loss": 0.0266,
+      "step": 1072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8447245542804054e-05,
+      "loss": 0.0319,
+      "step": 1073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8443940624686823e-05,
+      "loss": 0.0254,
+      "step": 1074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8440632489878637e-05,
+      "loss": 0.027,
+      "step": 1075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8437321139639715e-05,
+      "loss": 0.022,
+      "step": 1076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8434006575231516e-05,
+      "loss": 0.0228,
+      "step": 1077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8430688797916702e-05,
+      "loss": 0.0295,
+      "step": 1078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8427367808959174e-05,
+      "loss": 0.0362,
+      "step": 1079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842404360962405e-05,
+      "loss": 0.0257,
+      "step": 1080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8420716201177672e-05,
+      "loss": 0.0252,
+      "step": 1081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8417385584887607e-05,
+      "loss": 0.0381,
+      "step": 1082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8414051762022643e-05,
+      "loss": 0.031,
+      "step": 1083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.841071473385279e-05,
+      "loss": 0.0314,
+      "step": 1084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840737450164927e-05,
+      "loss": 0.0353,
+      "step": 1085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840403106668454e-05,
+      "loss": 0.0274,
+      "step": 1086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840068443023227e-05,
+      "loss": 0.0298,
+      "step": 1087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8397334593567347e-05,
+      "loss": 0.0483,
+      "step": 1088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8393981557965883e-05,
+      "loss": 0.0401,
+      "step": 1089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8390625324705206e-05,
+      "loss": 0.0352,
+      "step": 1090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8387265895063856e-05,
+      "loss": 0.0308,
+      "step": 1091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83839032703216e-05,
+      "loss": 0.0298,
+      "step": 1092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8380537451759424e-05,
+      "loss": 0.0355,
+      "step": 1093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.837716844065952e-05,
+      "loss": 0.0356,
+      "step": 1094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8373796238305302e-05,
+      "loss": 0.0261,
+      "step": 1095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8370420845981397e-05,
+      "loss": 0.0338,
+      "step": 1096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8367042264973655e-05,
+      "loss": 0.0313,
+      "step": 1097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.836366049656913e-05,
+      "loss": 0.032,
+      "step": 1098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8360275542056098e-05,
+      "loss": 0.0354,
+      "step": 1099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8356887402724046e-05,
+      "loss": 0.0273,
+      "step": 1100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8353496079863672e-05,
+      "loss": 0.0399,
+      "step": 1101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8350101574766893e-05,
+      "loss": 0.0274,
+      "step": 1102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8346703888726833e-05,
+      "loss": 0.0261,
+      "step": 1103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8343303023037825e-05,
+      "loss": 0.0263,
+      "step": 1104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8339898978995424e-05,
+      "loss": 0.0337,
+      "step": 1105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8336491757896385e-05,
+      "loss": 0.031,
+      "step": 1106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833308136103868e-05,
+      "loss": 0.038,
+      "step": 1107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8329667789721487e-05,
+      "loss": 0.0329,
+      "step": 1108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832625104524519e-05,
+      "loss": 0.0336,
+      "step": 1109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8322831128911396e-05,
+      "loss": 0.028,
+      "step": 1110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8319408042022905e-05,
+      "loss": 0.0338,
+      "step": 1111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8315981785883728e-05,
+      "loss": 0.0315,
+      "step": 1112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831255236179909e-05,
+      "loss": 0.0274,
+      "step": 1113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8309119771075416e-05,
+      "loss": 0.0291,
+      "step": 1114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8305684015020343e-05,
+      "loss": 0.0378,
+      "step": 1115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8302245094942712e-05,
+      "loss": 0.026,
+      "step": 1116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829880301215256e-05,
+      "loss": 0.04,
+      "step": 1117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8295357767961144e-05,
+      "loss": 0.0271,
+      "step": 1118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8291909363680923e-05,
+      "loss": 0.0267,
+      "step": 1119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8288457800625542e-05,
+      "loss": 0.0353,
+      "step": 1120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8285003080109872e-05,
+      "loss": 0.0236,
+      "step": 1121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.828154520344997e-05,
+      "loss": 0.0264,
+      "step": 1122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.827808417196311e-05,
+      "loss": 0.0265,
+      "step": 1123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8274619986967755e-05,
+      "loss": 0.0285,
+      "step": 1124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8271152649783583e-05,
+      "loss": 0.0339,
+      "step": 1125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826768216173145e-05,
+      "loss": 0.0244,
+      "step": 1126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826420852413344e-05,
+      "loss": 0.0306,
+      "step": 1127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8260731738312817e-05,
+      "loss": 0.0291,
+      "step": 1128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8257251805594054e-05,
+      "loss": 0.0261,
+      "step": 1129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8253768727302818e-05,
+      "loss": 0.0322,
+      "step": 1130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8250282504765976e-05,
+      "loss": 0.0248,
+      "step": 1131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8246793139311597e-05,
+      "loss": 0.0322,
+      "step": 1132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8243300632268935e-05,
+      "loss": 0.0304,
+      "step": 1133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8239804984968453e-05,
+      "loss": 0.0341,
+      "step": 1134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823630619874181e-05,
+      "loss": 0.0327,
+      "step": 1135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8232804274921853e-05,
+      "loss": 0.029,
+      "step": 1136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8229299214842625e-05,
+      "loss": 0.0279,
+      "step": 1137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8225791019839375e-05,
+      "loss": 0.0273,
+      "step": 1138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.822227969124853e-05,
+      "loss": 0.0253,
+      "step": 1139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8218765230407727e-05,
+      "loss": 0.0279,
+      "step": 1140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.821524763865578e-05,
+      "loss": 0.0337,
+      "step": 1141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.821172691733271e-05,
+      "loss": 0.0378,
+      "step": 1142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.820820306777972e-05,
+      "loss": 0.0306,
+      "step": 1143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.820467609133921e-05,
+      "loss": 0.0272,
+      "step": 1144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8201145989354768e-05,
+      "loss": 0.0309,
+      "step": 1145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819761276317118e-05,
+      "loss": 0.0316,
+      "step": 1146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819407641413441e-05,
+      "loss": 0.0277,
+      "step": 1147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8190536943591627e-05,
+      "loss": 0.0374,
+      "step": 1148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8186994352891172e-05,
+      "loss": 0.0357,
+      "step": 1149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8183448643382585e-05,
+      "loss": 0.0307,
+      "step": 1150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8179899816416598e-05,
+      "loss": 0.0339,
+      "step": 1151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8176347873345114e-05,
+      "loss": 0.0314,
+      "step": 1152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8172792815521246e-05,
+      "loss": 0.0218,
+      "step": 1153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8169234644299272e-05,
+      "loss": 0.0312,
+      "step": 1154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816567336103467e-05,
+      "loss": 0.0311,
+      "step": 1155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8162108967084098e-05,
+      "loss": 0.0363,
+      "step": 1156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.81585414638054e-05,
+      "loss": 0.0294,
+      "step": 1157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8154970852557604e-05,
+      "loss": 0.0227,
+      "step": 1158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.815139713470092e-05,
+      "loss": 0.0338,
+      "step": 1159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8147820311596747e-05,
+      "loss": 0.0259,
+      "step": 1160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8144240384607667e-05,
+      "loss": 0.0321,
+      "step": 1161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8140657355097438e-05,
+      "loss": 0.03,
+      "step": 1162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8137071224431e-05,
+      "loss": 0.0305,
+      "step": 1163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813348199397448e-05,
+      "loss": 0.0356,
+      "step": 1164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.812988966509518e-05,
+      "loss": 0.0277,
+      "step": 1165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8126294239161596e-05,
+      "loss": 0.0374,
+      "step": 1166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8122695717543383e-05,
+      "loss": 0.0248,
+      "step": 1167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811909410161139e-05,
+      "loss": 0.0289,
+      "step": 1168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811548939273764e-05,
+      "loss": 0.0349,
+      "step": 1169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8111881592295337e-05,
+      "loss": 0.0264,
+      "step": 1170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8108270701658853e-05,
+      "loss": 0.0353,
+      "step": 1171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8104656722203756e-05,
+      "loss": 0.0266,
+      "step": 1172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.810103965530677e-05,
+      "loss": 0.0291,
+      "step": 1173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.809741950234581e-05,
+      "loss": 0.0325,
+      "step": 1174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.809379626469996e-05,
+      "loss": 0.0283,
+      "step": 1175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.0298,
+      "step": 1176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8086540540875795e-05,
+      "loss": 0.042,
+      "step": 1177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8082908057461534e-05,
+      "loss": 0.0383,
+      "step": 1178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8079272494890463e-05,
+      "loss": 0.0287,
+      "step": 1179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8075633854547546e-05,
+      "loss": 0.0302,
+      "step": 1180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8071992137818902e-05,
+      "loss": 0.0344,
+      "step": 1181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8068347346091845e-05,
+      "loss": 0.0287,
+      "step": 1182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.806469948075483e-05,
+      "loss": 0.0276,
+      "step": 1183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8061048543197516e-05,
+      "loss": 0.0391,
+      "step": 1184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80573945348107e-05,
+      "loss": 0.0398,
+      "step": 1185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805373745698637e-05,
+      "loss": 0.0378,
+      "step": 1186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805007731111768e-05,
+      "loss": 0.0338,
+      "step": 1187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8046414098598947e-05,
+      "loss": 0.033,
+      "step": 1188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8042747820825664e-05,
+      "loss": 0.0336,
+      "step": 1189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8039078479194485e-05,
+      "loss": 0.0314,
+      "step": 1190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.803540607510323e-05,
+      "loss": 0.0421,
+      "step": 1191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8031730609950896e-05,
+      "loss": 0.0257,
+      "step": 1192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.802805208513763e-05,
+      "loss": 0.0295,
+      "step": 1193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8024370502064764e-05,
+      "loss": 0.0304,
+      "step": 1194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8020685862134776e-05,
+      "loss": 0.0306,
+      "step": 1195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8016998166751322e-05,
+      "loss": 0.0367,
+      "step": 1196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8013307417319213e-05,
+      "loss": 0.0261,
+      "step": 1197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8009613615244438e-05,
+      "loss": 0.0269,
+      "step": 1198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8005916761934126e-05,
+      "loss": 0.0236,
+      "step": 1199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8002216858796587e-05,
+      "loss": 0.0273,
+      "step": 1200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7998513907241285e-05,
+      "loss": 0.0343,
+      "step": 1201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.799480790867885e-05,
+      "loss": 0.0335,
+      "step": 1202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7991098864521066e-05,
+      "loss": 0.0344,
+      "step": 1203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7987386776180882e-05,
+      "loss": 0.0321,
+      "step": 1204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.798367164507241e-05,
+      "loss": 0.0263,
+      "step": 1205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7979953472610912e-05,
+      "loss": 0.0319,
+      "step": 1206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7976232260212814e-05,
+      "loss": 0.037,
+      "step": 1207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79725080092957e-05,
+      "loss": 0.0233,
+      "step": 1208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.796878072127831e-05,
+      "loss": 0.0363,
+      "step": 1209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.796505039758054e-05,
+      "loss": 0.0293,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7961317039623453e-05,
+      "loss": 0.0357,
+      "step": 1211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.795758064882925e-05,
+      "loss": 0.0219,
+      "step": 1212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79538412266213e-05,
+      "loss": 0.0332,
+      "step": 1213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7950098774424125e-05,
+      "loss": 0.0316,
+      "step": 1214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7946353293663395e-05,
+      "loss": 0.0372,
+      "step": 1215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7942604785765944e-05,
+      "loss": 0.0301,
+      "step": 1216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.793885325215975e-05,
+      "loss": 0.0287,
+      "step": 1217
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.793509869427395e-05,
+      "loss": 0.0306,
+      "step": 1218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7931341113538825e-05,
+      "loss": 0.0295,
+      "step": 1219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7927580511385818e-05,
+      "loss": 0.0352,
+      "step": 1220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7923816889247517e-05,
+      "loss": 0.0352,
+      "step": 1221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7920050248557656e-05,
+      "loss": 0.0281,
+      "step": 1222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.791628059075113e-05,
+      "loss": 0.0362,
+      "step": 1223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.791250791726398e-05,
+      "loss": 0.0325,
+      "step": 1224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7908732229533387e-05,
+      "loss": 0.0324,
+      "step": 1225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.790495352899769e-05,
+      "loss": 0.0214,
+      "step": 1226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7901171817096368e-05,
+      "loss": 0.038,
+      "step": 1227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7897387095270058e-05,
+      "loss": 0.0287,
+      "step": 1228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7893599364960535e-05,
+      "loss": 0.0325,
+      "step": 1229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7889808627610717e-05,
+      "loss": 0.0283,
+      "step": 1230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.788601488466468e-05,
+      "loss": 0.0434,
+      "step": 1231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.788221813756763e-05,
+      "loss": 0.0337,
+      "step": 1232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7878418387765934e-05,
+      "loss": 0.0287,
+      "step": 1233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7874615636707088e-05,
+      "loss": 0.0305,
+      "step": 1234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.787080988583974e-05,
+      "loss": 0.0259,
+      "step": 1235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7867001136613678e-05,
+      "loss": 0.0337,
+      "step": 1236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7863189390479832e-05,
+      "loss": 0.0329,
+      "step": 1237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785937464889027e-05,
+      "loss": 0.0394,
+      "step": 1238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785555691329821e-05,
+      "loss": 0.0334,
+      "step": 1239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7851736185158003e-05,
+      "loss": 0.0267,
+      "step": 1240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7847912465925145e-05,
+      "loss": 0.0251,
+      "step": 1241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7844085757056266e-05,
+      "loss": 0.0314,
+      "step": 1242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.784025606000914e-05,
+      "loss": 0.0282,
+      "step": 1243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7836423376242678e-05,
+      "loss": 0.033,
+      "step": 1244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7832587707216922e-05,
+      "loss": 0.0322,
+      "step": 1245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.782874905439307e-05,
+      "loss": 0.0302,
+      "step": 1246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.782490741923343e-05,
+      "loss": 0.0239,
+      "step": 1247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.782106280320147e-05,
+      "loss": 0.0286,
+      "step": 1248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7817215207761777e-05,
+      "loss": 0.0329,
+      "step": 1249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7813364634380084e-05,
+      "loss": 0.0274,
+      "step": 1250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7809511084523256e-05,
+      "loss": 0.0362,
+      "step": 1251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.780565455965928e-05,
+      "loss": 0.0301,
+      "step": 1252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7801795061257293e-05,
+      "loss": 0.0328,
+      "step": 1253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7797932590787563e-05,
+      "loss": 0.0299,
+      "step": 1254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7794067149721475e-05,
+      "loss": 0.0268,
+      "step": 1255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7790198739531567e-05,
+      "loss": 0.0279,
+      "step": 1256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7786327361691485e-05,
+      "loss": 0.0329,
+      "step": 1257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7782453017676025e-05,
+      "loss": 0.0293,
+      "step": 1258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7778575708961106e-05,
+      "loss": 0.0244,
+      "step": 1259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777469543702377e-05,
+      "loss": 0.0321,
+      "step": 1260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7770812203342198e-05,
+      "loss": 0.0336,
+      "step": 1261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.776692600939569e-05,
+      "loss": 0.0282,
+      "step": 1262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7763036856664692e-05,
+      "loss": 0.0453,
+      "step": 1263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7759144746630745e-05,
+      "loss": 0.0343,
+      "step": 1264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.775524968077655e-05,
+      "loss": 0.0312,
+      "step": 1265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7751351660585912e-05,
+      "loss": 0.0226,
+      "step": 1266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7747450687543767e-05,
+      "loss": 0.0278,
+      "step": 1267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7743546763136187e-05,
+      "loss": 0.038,
+      "step": 1268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7739639888850353e-05,
+      "loss": 0.0299,
+      "step": 1269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7735730066174573e-05,
+      "loss": 0.0214,
+      "step": 1270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7731817296598285e-05,
+      "loss": 0.0326,
+      "step": 1271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7727901581612042e-05,
+      "loss": 0.0312,
+      "step": 1272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.772398292270753e-05,
+      "loss": 0.0338,
+      "step": 1273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.772006132137754e-05,
+      "loss": 0.0232,
+      "step": 1274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7716136779116e-05,
+      "loss": 0.0287,
+      "step": 1275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.771220929741794e-05,
+      "loss": 0.0322,
+      "step": 1276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7708278877779536e-05,
+      "loss": 0.027,
+      "step": 1277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7704345521698057e-05,
+      "loss": 0.0316,
+      "step": 1278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7700409230671912e-05,
+      "loss": 0.0272,
+      "step": 1279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.769647000620061e-05,
+      "loss": 0.0294,
+      "step": 1280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7692527849784785e-05,
+      "loss": 0.0307,
+      "step": 1281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.768858276292619e-05,
+      "loss": 0.03,
+      "step": 1282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7684634747127698e-05,
+      "loss": 0.0247,
+      "step": 1283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.768068380389328e-05,
+      "loss": 0.0274,
+      "step": 1284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767672993472805e-05,
+      "loss": 0.0348,
+      "step": 1285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767277314113821e-05,
+      "loss": 0.0385,
+      "step": 1286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7668813424631093e-05,
+      "loss": 0.0315,
+      "step": 1287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.766485078671514e-05,
+      "loss": 0.0235,
+      "step": 1288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7660885228899895e-05,
+      "loss": 0.0271,
+      "step": 1289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7656916752696032e-05,
+      "loss": 0.0281,
+      "step": 1290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7652945359615334e-05,
+      "loss": 0.0263,
+      "step": 1291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.764897105117068e-05,
+      "loss": 0.027,
+      "step": 1292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7644993828876077e-05,
+      "loss": 0.034,
+      "step": 1293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7641013694246626e-05,
+      "loss": 0.0266,
+      "step": 1294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7637030648798556e-05,
+      "loss": 0.0257,
+      "step": 1295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7633044694049187e-05,
+      "loss": 0.0295,
+      "step": 1296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7629055831516958e-05,
+      "loss": 0.0287,
+      "step": 1297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7625064062721414e-05,
+      "loss": 0.0283,
+      "step": 1298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7621069389183204e-05,
+      "loss": 0.0371,
+      "step": 1299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7617071812424087e-05,
+      "loss": 0.025,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7613071333966927e-05,
+      "loss": 0.0355,
+      "step": 1301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.760906795533569e-05,
+      "loss": 0.031,
+      "step": 1302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7605061678055453e-05,
+      "loss": 0.0229,
+      "step": 1303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7601052503652393e-05,
+      "loss": 0.0321,
+      "step": 1304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7597040433653788e-05,
+      "loss": 0.0311,
+      "step": 1305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7593025469588026e-05,
+      "loss": 0.0325,
+      "step": 1306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7589007612984598e-05,
+      "loss": 0.0302,
+      "step": 1307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7584986865374084e-05,
+      "loss": 0.0193,
+      "step": 1308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.758096322828818e-05,
+      "loss": 0.0248,
+      "step": 1309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757693670325968e-05,
+      "loss": 0.04,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7572907291822466e-05,
+      "loss": 0.0349,
+      "step": 1311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7568874995511537e-05,
+      "loss": 0.0264,
+      "step": 1312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.756483981586298e-05,
+      "loss": 0.0316,
+      "step": 1313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7560801754413983e-05,
+      "loss": 0.02,
+      "step": 1314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7556760812702832e-05,
+      "loss": 0.0345,
+      "step": 1315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7552716992268912e-05,
+      "loss": 0.03,
+      "step": 1316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7548670294652704e-05,
+      "loss": 0.0347,
+      "step": 1317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754462072139578e-05,
+      "loss": 0.0279,
+      "step": 1318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754056827404081e-05,
+      "loss": 0.0276,
+      "step": 1319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7536512954131563e-05,
+      "loss": 0.0279,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7532454763212904e-05,
+      "loss": 0.0302,
+      "step": 1321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.752839370283078e-05,
+      "loss": 0.0299,
+      "step": 1322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.752432977453224e-05,
+      "loss": 0.0233,
+      "step": 1323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7520262979865425e-05,
+      "loss": 0.0365,
+      "step": 1324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.751619332037957e-05,
+      "loss": 0.0263,
+      "step": 1325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7512120797624993e-05,
+      "loss": 0.0334,
+      "step": 1326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.750804541315311e-05,
+      "loss": 0.0297,
+      "step": 1327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7503967168516426e-05,
+      "loss": 0.0263,
+      "step": 1328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7499886065268534e-05,
+      "loss": 0.026,
+      "step": 1329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7495802104964118e-05,
+      "loss": 0.0353,
+      "step": 1330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7491715289158947e-05,
+      "loss": 0.0292,
+      "step": 1331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.748762561940988e-05,
+      "loss": 0.0342,
+      "step": 1332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.748353309727486e-05,
+      "loss": 0.0263,
+      "step": 1333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7479437724312925e-05,
+      "loss": 0.0288,
+      "step": 1334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7475339502084196e-05,
+      "loss": 0.0255,
+      "step": 1335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7471238432149867e-05,
+      "loss": 0.031,
+      "step": 1336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7467134516072235e-05,
+      "loss": 0.0325,
+      "step": 1337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.746302775541467e-05,
+      "loss": 0.0334,
+      "step": 1338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.745891815174163e-05,
+      "loss": 0.0279,
+      "step": 1339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7454805706618657e-05,
+      "loss": 0.0345,
+      "step": 1340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7450690421612365e-05,
+      "loss": 0.0288,
+      "step": 1341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.744657229829047e-05,
+      "loss": 0.0277,
+      "step": 1342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7442451338221747e-05,
+      "loss": 0.0332,
+      "step": 1343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7438327542976064e-05,
+      "loss": 0.0334,
+      "step": 1344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.743420091412437e-05,
+      "loss": 0.0237,
+      "step": 1345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7430071453238692e-05,
+      "loss": 0.0318,
+      "step": 1346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.742593916189213e-05,
+      "loss": 0.0248,
+      "step": 1347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7421804041658867e-05,
+      "loss": 0.0238,
+      "step": 1348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7417666094114164e-05,
+      "loss": 0.0289,
+      "step": 1349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7413525320834363e-05,
+      "loss": 0.0306,
+      "step": 1350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.740938172339687e-05,
+      "loss": 0.0249,
+      "step": 1351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7405235303380178e-05,
+      "loss": 0.0316,
+      "step": 1352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.740108606236385e-05,
+      "loss": 0.0253,
+      "step": 1353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.739693400192853e-05,
+      "loss": 0.0174,
+      "step": 1354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7392779123655927e-05,
+      "loss": 0.0302,
+      "step": 1355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.738862142912883e-05,
+      "loss": 0.0318,
+      "step": 1356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7384460919931103e-05,
+      "loss": 0.0225,
+      "step": 1357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.738029759764767e-05,
+      "loss": 0.0331,
+      "step": 1358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.737613146386454e-05,
+      "loss": 0.0198,
+      "step": 1359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7371962520168786e-05,
+      "loss": 0.0291,
+      "step": 1360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.736779076814855e-05,
+      "loss": 0.0348,
+      "step": 1361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7363616209393055e-05,
+      "loss": 0.0358,
+      "step": 1362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7359438845492576e-05,
+      "loss": 0.0305,
+      "step": 1363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7355258678038475e-05,
+      "loss": 0.0301,
+      "step": 1364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7351075708623168e-05,
+      "loss": 0.0339,
+      "step": 1365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7346889938840143e-05,
+      "loss": 0.0352,
+      "step": 1366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7342701370283955e-05,
+      "loss": 0.0231,
+      "step": 1367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7338510004550225e-05,
+      "loss": 0.0276,
+      "step": 1368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733431584323564e-05,
+      "loss": 0.0277,
+      "step": 1369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733011888793795e-05,
+      "loss": 0.0371,
+      "step": 1370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.732591914025598e-05,
+      "loss": 0.0347,
+      "step": 1371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7321716601789598e-05,
+      "loss": 0.0243,
+      "step": 1372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7317511274139752e-05,
+      "loss": 0.0329,
+      "step": 1373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7313303158908447e-05,
+      "loss": 0.0229,
+      "step": 1374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7309092257698754e-05,
+      "loss": 0.024,
+      "step": 1375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7304878572114798e-05,
+      "loss": 0.0387,
+      "step": 1376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7300662103761768e-05,
+      "loss": 0.0282,
+      "step": 1377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7296442854245915e-05,
+      "loss": 0.0246,
+      "step": 1378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.729222082517455e-05,
+      "loss": 0.0279,
+      "step": 1379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7287996018156033e-05,
+      "loss": 0.0281,
+      "step": 1380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.72837684347998e-05,
+      "loss": 0.0354,
+      "step": 1381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.727953807671633e-05,
+      "loss": 0.0257,
+      "step": 1382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.727530494551716e-05,
+      "loss": 0.0308,
+      "step": 1383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.72710690428149e-05,
+      "loss": 0.0281,
+      "step": 1384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7266830370223188e-05,
+      "loss": 0.0303,
+      "step": 1385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7262588929356737e-05,
+      "loss": 0.0271,
+      "step": 1386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7258344721831314e-05,
+      "loss": 0.0329,
+      "step": 1387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7254097749263735e-05,
+      "loss": 0.0259,
+      "step": 1388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7249848013271865e-05,
+      "loss": 0.0325,
+      "step": 1389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.724559551547463e-05,
+      "loss": 0.026,
+      "step": 1390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7241340257492005e-05,
+      "loss": 0.035,
+      "step": 1391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7237082240945016e-05,
+      "loss": 0.0279,
+      "step": 1392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7232821467455738e-05,
+      "loss": 0.029,
+      "step": 1393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.72285579386473e-05,
+      "loss": 0.0301,
+      "step": 1394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.722429165614388e-05,
+      "loss": 0.0263,
+      "step": 1395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7220022621570708e-05,
+      "loss": 0.0226,
+      "step": 1396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7215750836554047e-05,
+      "loss": 0.0291,
+      "step": 1397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.721147630272123e-05,
+      "loss": 0.0264,
+      "step": 1398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7207199021700622e-05,
+      "loss": 0.0354,
+      "step": 1399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.720291899512164e-05,
+      "loss": 0.0316,
+      "step": 1400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7198636224614744e-05,
+      "loss": 0.0265,
+      "step": 1401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719435071181144e-05,
+      "loss": 0.033,
+      "step": 1402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719006245834429e-05,
+      "loss": 0.0361,
+      "step": 1403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7185771465846875e-05,
+      "loss": 0.0275,
+      "step": 1404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.718147773595384e-05,
+      "loss": 0.0291,
+      "step": 1405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717718127030087e-05,
+      "loss": 0.0306,
+      "step": 1406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717288207052469e-05,
+      "loss": 0.0315,
+      "step": 1407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7168580138263064e-05,
+      "loss": 0.0209,
+      "step": 1408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7164275475154793e-05,
+      "loss": 0.0267,
+      "step": 1409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.715996808283973e-05,
+      "loss": 0.0328,
+      "step": 1410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7155657962958762e-05,
+      "loss": 0.0259,
+      "step": 1411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7151345117153814e-05,
+      "loss": 0.0246,
+      "step": 1412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.714702954706785e-05,
+      "loss": 0.0311,
+      "step": 1413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.714271125434487e-05,
+      "loss": 0.0284,
+      "step": 1414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7138390240629915e-05,
+      "loss": 0.0243,
+      "step": 1415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.713406650756906e-05,
+      "loss": 0.026,
+      "step": 1416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7129740056809415e-05,
+      "loss": 0.0346,
+      "step": 1417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7125410889999135e-05,
+      "loss": 0.0318,
+      "step": 1418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7121079008787395e-05,
+      "loss": 0.0328,
+      "step": 1419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7116744414824406e-05,
+      "loss": 0.032,
+      "step": 1420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7112407109761428e-05,
+      "loss": 0.0273,
+      "step": 1421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7108067095250735e-05,
+      "loss": 0.0313,
+      "step": 1422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.710372437294564e-05,
+      "loss": 0.0308,
+      "step": 1423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7099378944500497e-05,
+      "loss": 0.025,
+      "step": 1424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7095030811570673e-05,
+      "loss": 0.0296,
+      "step": 1425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7090679975812582e-05,
+      "loss": 0.0325,
+      "step": 1426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.708632643888366e-05,
+      "loss": 0.0303,
+      "step": 1427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7081970202442363e-05,
+      "loss": 0.0288,
+      "step": 1428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.70776112681482e-05,
+      "loss": 0.0299,
+      "step": 1429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7073249637661678e-05,
+      "loss": 0.0264,
+      "step": 1430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.706888531264435e-05,
+      "loss": 0.0369,
+      "step": 1431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7064518294758797e-05,
+      "loss": 0.0232,
+      "step": 1432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7060148585668613e-05,
+      "loss": 0.0232,
+      "step": 1433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7055776187038425e-05,
+      "loss": 0.0314,
+      "step": 1434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.705140110053389e-05,
+      "loss": 0.0339,
+      "step": 1435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.704702332782167e-05,
+      "loss": 0.0293,
+      "step": 1436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7042642870569475e-05,
+      "loss": 0.0238,
+      "step": 1437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.703825973044602e-05,
+      "loss": 0.0238,
+      "step": 1438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.703387390912105e-05,
+      "loss": 0.0287,
+      "step": 1439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7029485408265326e-05,
+      "loss": 0.0273,
+      "step": 1440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7025094229550636e-05,
+      "loss": 0.032,
+      "step": 1441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7020700374649782e-05,
+      "loss": 0.0315,
+      "step": 1442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.701630384523659e-05,
+      "loss": 0.0221,
+      "step": 1443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7011904642985905e-05,
+      "loss": 0.0274,
+      "step": 1444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.700750276957358e-05,
+      "loss": 0.0303,
+      "step": 1445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7003098226676507e-05,
+      "loss": 0.0286,
+      "step": 1446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6998691015972574e-05,
+      "loss": 0.0341,
+      "step": 1447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.699428113914069e-05,
+      "loss": 0.0321,
+      "step": 1448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.698986859786079e-05,
+      "loss": 0.032,
+      "step": 1449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.698545339381381e-05,
+      "loss": 0.0342,
+      "step": 1450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6981035528681714e-05,
+      "loss": 0.0265,
+      "step": 1451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6976615004147474e-05,
+      "loss": 0.0244,
+      "step": 1452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6972191821895065e-05,
+      "loss": 0.0262,
+      "step": 1453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6967765983609486e-05,
+      "loss": 0.0347,
+      "step": 1454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.696333749097675e-05,
+      "loss": 0.0306,
+      "step": 1455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6958906345683877e-05,
+      "loss": 0.0303,
+      "step": 1456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6954472549418894e-05,
+      "loss": 0.0315,
+      "step": 1457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.695003610387084e-05,
+      "loss": 0.027,
+      "step": 1458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6945597010729764e-05,
+      "loss": 0.0251,
+      "step": 1459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6941155271686726e-05,
+      "loss": 0.0273,
+      "step": 1460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693671088843379e-05,
+      "loss": 0.0281,
+      "step": 1461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693226386266403e-05,
+      "loss": 0.0222,
+      "step": 1462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6927814196071528e-05,
+      "loss": 0.0251,
+      "step": 1463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6923361890351362e-05,
+      "loss": 0.0335,
+      "step": 1464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691890694719963e-05,
+      "loss": 0.0333,
+      "step": 1465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6914449368313425e-05,
+      "loss": 0.0308,
+      "step": 1466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6909989155390846e-05,
+      "loss": 0.0241,
+      "step": 1467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6905526310131e-05,
+      "loss": 0.032,
+      "step": 1468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6901060834233992e-05,
+      "loss": 0.031,
+      "step": 1469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6896592729400923e-05,
+      "loss": 0.0253,
+      "step": 1470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.689212199733391e-05,
+      "loss": 0.025,
+      "step": 1471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6887648639736066e-05,
+      "loss": 0.0263,
+      "step": 1472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6883172658311494e-05,
+      "loss": 0.0283,
+      "step": 1473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6878694054765303e-05,
+      "loss": 0.0326,
+      "step": 1474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.687421283080361e-05,
+      "loss": 0.0266,
+      "step": 1475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6869728988133517e-05,
+      "loss": 0.0265,
+      "step": 1476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6865242528463138e-05,
+      "loss": 0.0403,
+      "step": 1477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686075345350156e-05,
+      "loss": 0.0231,
+      "step": 1478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6856261764958887e-05,
+      "loss": 0.0326,
+      "step": 1479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685176746454622e-05,
+      "loss": 0.0322,
+      "step": 1480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.684727055397564e-05,
+      "loss": 0.0268,
+      "step": 1481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.684277103496023e-05,
+      "loss": 0.0298,
+      "step": 1482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.683826890921407e-05,
+      "loss": 0.0341,
+      "step": 1483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6833764178452227e-05,
+      "loss": 0.031,
+      "step": 1484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6829256844390768e-05,
+      "loss": 0.028,
+      "step": 1485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6824746908746742e-05,
+      "loss": 0.039,
+      "step": 1486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6820234373238192e-05,
+      "loss": 0.0268,
+      "step": 1487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.681571923958416e-05,
+      "loss": 0.0336,
+      "step": 1488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6811201509504665e-05,
+      "loss": 0.031,
+      "step": 1489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6806681184720726e-05,
+      "loss": 0.0283,
+      "step": 1490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6802158266954344e-05,
+      "loss": 0.0416,
+      "step": 1491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6797632757928507e-05,
+      "loss": 0.0313,
+      "step": 1492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6793104659367195e-05,
+      "loss": 0.0285,
+      "step": 1493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6788573972995376e-05,
+      "loss": 0.0256,
+      "step": 1494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678404070053899e-05,
+      "loss": 0.0367,
+      "step": 1495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6779504843724977e-05,
+      "loss": 0.0336,
+      "step": 1496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6774966404281258e-05,
+      "loss": 0.0337,
+      "step": 1497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6770425383936734e-05,
+      "loss": 0.0339,
+      "step": 1498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6765881784421295e-05,
+      "loss": 0.031,
+      "step": 1499
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6761335607465806e-05,
+      "loss": 0.0331,
+      "step": 1500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6756786854802123e-05,
+      "loss": 0.0407,
+      "step": 1501
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.675223552816307e-05,
+      "loss": 0.0302,
+      "step": 1502
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6747681629282468e-05,
+      "loss": 0.0312,
+      "step": 1503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.67431251598951e-05,
+      "loss": 0.0321,
+      "step": 1504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673856612173675e-05,
+      "loss": 0.0234,
+      "step": 1505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673400451654416e-05,
+      "loss": 0.034,
+      "step": 1506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.672944034605506e-05,
+      "loss": 0.0341,
+      "step": 1507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6724873612008155e-05,
+      "loss": 0.0322,
+      "step": 1508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.672030431614313e-05,
+      "loss": 0.0314,
+      "step": 1509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6715732460200638e-05,
+      "loss": 0.031,
+      "step": 1510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6711158045922317e-05,
+      "loss": 0.0347,
+      "step": 1511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6706581075050773e-05,
+      "loss": 0.0269,
+      "step": 1512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6702001549329584e-05,
+      "loss": 0.0313,
+      "step": 1513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6697419470503315e-05,
+      "loss": 0.03,
+      "step": 1514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6692834840317484e-05,
+      "loss": 0.0325,
+      "step": 1515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6688247660518592e-05,
+      "loss": 0.0249,
+      "step": 1516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6683657932854115e-05,
+      "loss": 0.0329,
+      "step": 1517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6679065659072486e-05,
+      "loss": 0.03,
+      "step": 1518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6674470840923128e-05,
+      "loss": 0.0255,
+      "step": 1519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6669873480156412e-05,
+      "loss": 0.0289,
+      "step": 1520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6665273578523687e-05,
+      "loss": 0.0279,
+      "step": 1521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6660671137777277e-05,
+      "loss": 0.0278,
+      "step": 1522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6656066159670462e-05,
+      "loss": 0.032,
+      "step": 1523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6651458645957493e-05,
+      "loss": 0.025,
+      "step": 1524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.664684859839359e-05,
+      "loss": 0.0385,
+      "step": 1525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6642236018734932e-05,
+      "loss": 0.0239,
+      "step": 1526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6637620908738667e-05,
+      "loss": 0.0273,
+      "step": 1527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6633003270162903e-05,
+      "loss": 0.0309,
+      "step": 1528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662838310476672e-05,
+      "loss": 0.0212,
+      "step": 1529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6623760414310146e-05,
+      "loss": 0.0257,
+      "step": 1530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661913520055419e-05,
+      "loss": 0.0242,
+      "step": 1531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6614507465260802e-05,
+      "loss": 0.0232,
+      "step": 1532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6609877210192912e-05,
+      "loss": 0.0322,
+      "step": 1533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.660524443711439e-05,
+      "loss": 0.0237,
+      "step": 1534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6600609147790078e-05,
+      "loss": 0.0262,
+      "step": 1535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.659597134398578e-05,
+      "loss": 0.032,
+      "step": 1536
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6591331027468248e-05,
+      "loss": 0.0358,
+      "step": 1537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6586688200005193e-05,
+      "loss": 0.0304,
+      "step": 1538
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6582042863365287e-05,
+      "loss": 0.0332,
+      "step": 1539
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.657739501931816e-05,
+      "loss": 0.024,
+      "step": 1540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6572744669634384e-05,
+      "loss": 0.0266,
+      "step": 1541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6568091816085498e-05,
+      "loss": 0.0344,
+      "step": 1542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6563436460443986e-05,
+      "loss": 0.0394,
+      "step": 1543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6558778604483302e-05,
+      "loss": 0.0241,
+      "step": 1544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.655411824997783e-05,
+      "loss": 0.0308,
+      "step": 1545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6549455398702923e-05,
+      "loss": 0.0291,
+      "step": 1546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6544790052434874e-05,
+      "loss": 0.0296,
+      "step": 1547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654012221295093e-05,
+      "loss": 0.0324,
+      "step": 1548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6535451882029296e-05,
+      "loss": 0.032,
+      "step": 1549
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6530779061449113e-05,
+      "loss": 0.0397,
+      "step": 1550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.652610375299048e-05,
+      "loss": 0.0362,
+      "step": 1551
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6521425958434435e-05,
+      "loss": 0.0375,
+      "step": 1552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6516745679562977e-05,
+      "loss": 0.0322,
+      "step": 1553
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6512062918159032e-05,
+      "loss": 0.0233,
+      "step": 1554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6507377676006486e-05,
+      "loss": 0.0248,
+      "step": 1555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6502689954890173e-05,
+      "loss": 0.0281,
+      "step": 1556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6497999756595854e-05,
+      "loss": 0.0379,
+      "step": 1557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.649330708291025e-05,
+      "loss": 0.0371,
+      "step": 1558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6488611935621023e-05,
+      "loss": 0.0307,
+      "step": 1559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6483914316516766e-05,
+      "loss": 0.0288,
+      "step": 1560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6479214227387026e-05,
+      "loss": 0.0221,
+      "step": 1561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6474511670022285e-05,
+      "loss": 0.032,
+      "step": 1562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6469806646213967e-05,
+      "loss": 0.0308,
+      "step": 1563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6465099157754435e-05,
+      "loss": 0.0228,
+      "step": 1564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6460389206436992e-05,
+      "loss": 0.0256,
+      "step": 1565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6455676794055878e-05,
+      "loss": 0.034,
+      "step": 1566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6450961922406264e-05,
+      "loss": 0.0258,
+      "step": 1567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6446244593284277e-05,
+      "loss": 0.0273,
+      "step": 1568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644152480848696e-05,
+      "loss": 0.0284,
+      "step": 1569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6436802569812295e-05,
+      "loss": 0.0215,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.643207787905921e-05,
+      "loss": 0.0218,
+      "step": 1571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.642735073802756e-05,
+      "loss": 0.025,
+      "step": 1572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6422621148518135e-05,
+      "loss": 0.0279,
+      "step": 1573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.641788911233265e-05,
+      "loss": 0.0297,
+      "step": 1574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6413154631273762e-05,
+      "loss": 0.0308,
+      "step": 1575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6408417707145056e-05,
+      "loss": 0.0263,
+      "step": 1576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6403678341751048e-05,
+      "loss": 0.0267,
+      "step": 1577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6398936536897182e-05,
+      "loss": 0.0409,
+      "step": 1578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6394192294389832e-05,
+      "loss": 0.0299,
+      "step": 1579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6389445616036304e-05,
+      "loss": 0.0253,
+      "step": 1580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6384696503644826e-05,
+      "loss": 0.0293,
+      "step": 1581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6379944959024556e-05,
+      "loss": 0.0296,
+      "step": 1582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6375190983985586e-05,
+      "loss": 0.0189,
+      "step": 1583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.637043458033892e-05,
+      "loss": 0.0307,
+      "step": 1584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6365675749896496e-05,
+      "loss": 0.0295,
+      "step": 1585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6360914494471173e-05,
+      "loss": 0.0358,
+      "step": 1586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.635615081587674e-05,
+      "loss": 0.0224,
+      "step": 1587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6351384715927897e-05,
+      "loss": 0.0283,
+      "step": 1588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634661619644028e-05,
+      "loss": 0.0331,
+      "step": 1589
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6341845259230442e-05,
+      "loss": 0.0179,
+      "step": 1590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.633707190611585e-05,
+      "loss": 0.037,
+      "step": 1591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.63322961389149e-05,
+      "loss": 0.0313,
+      "step": 1592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6327517959446903e-05,
+      "loss": 0.0292,
+      "step": 1593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6322737369532092e-05,
+      "loss": 0.0291,
+      "step": 1594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6317954370991618e-05,
+      "loss": 0.0241,
+      "step": 1595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6313168965647542e-05,
+      "loss": 0.0252,
+      "step": 1596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6308381155322857e-05,
+      "loss": 0.0338,
+      "step": 1597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6303590941841457e-05,
+      "loss": 0.021,
+      "step": 1598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629879832702816e-05,
+      "loss": 0.0248,
+      "step": 1599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6294003312708696e-05,
+      "loss": 0.0213,
+      "step": 1600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628920590070971e-05,
+      "loss": 0.0228,
+      "step": 1601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628440609285876e-05,
+      "loss": 0.0293,
+      "step": 1602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6279603890984315e-05,
+      "loss": 0.0225,
+      "step": 1603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6274799296915764e-05,
+      "loss": 0.0315,
+      "step": 1604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6269992312483393e-05,
+      "loss": 0.0233,
+      "step": 1605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.626518293951841e-05,
+      "loss": 0.0212,
+      "step": 1606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6260371179852933e-05,
+      "loss": 0.0279,
+      "step": 1607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625555703531998e-05,
+      "loss": 0.0291,
+      "step": 1608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625074050775349e-05,
+      "loss": 0.0261,
+      "step": 1609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6245921598988294e-05,
+      "loss": 0.0259,
+      "step": 1610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6241100310860146e-05,
+      "loss": 0.0212,
+      "step": 1611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6236276645205697e-05,
+      "loss": 0.0266,
+      "step": 1612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.623145060386251e-05,
+      "loss": 0.0302,
+      "step": 1613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6226622188669042e-05,
+      "loss": 0.0261,
+      "step": 1614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.622179140146467e-05,
+      "loss": 0.0224,
+      "step": 1615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.621695824408966e-05,
+      "loss": 0.0222,
+      "step": 1616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6212122718385187e-05,
+      "loss": 0.0327,
+      "step": 1617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6207284826193334e-05,
+      "loss": 0.0209,
+      "step": 1618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6202444569357078e-05,
+      "loss": 0.0413,
+      "step": 1619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.619760194972029e-05,
+      "loss": 0.0264,
+      "step": 1620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6192756969127764e-05,
+      "loss": 0.0332,
+      "step": 1621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.618790962942517e-05,
+      "loss": 0.0345,
+      "step": 1622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6183059932459092e-05,
+      "loss": 0.0293,
+      "step": 1623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6178207880076996e-05,
+      "loss": 0.0307,
+      "step": 1624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6173353474127263e-05,
+      "loss": 0.027,
+      "step": 1625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.616849671645916e-05,
+      "loss": 0.0282,
+      "step": 1626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6163637608922855e-05,
+      "loss": 0.0309,
+      "step": 1627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6158776153369406e-05,
+      "loss": 0.0208,
+      "step": 1628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.615391235165077e-05,
+      "loss": 0.0291,
+      "step": 1629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6149046205619792e-05,
+      "loss": 0.0333,
+      "step": 1630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.614417771713022e-05,
+      "loss": 0.0279,
+      "step": 1631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6139306888036687e-05,
+      "loss": 0.0331,
+      "step": 1632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6134433720194716e-05,
+      "loss": 0.0315,
+      "step": 1633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6129558215460727e-05,
+      "loss": 0.0364,
+      "step": 1634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6124680375692028e-05,
+      "loss": 0.0289,
+      "step": 1635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.611980020274681e-05,
+      "loss": 0.0277,
+      "step": 1636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6114917698484165e-05,
+      "loss": 0.0264,
+      "step": 1637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.611003286476406e-05,
+      "loss": 0.0321,
+      "step": 1638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6105145703447367e-05,
+      "loss": 0.0315,
+      "step": 1639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6100256216395822e-05,
+      "loss": 0.0383,
+      "step": 1640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6095364405472066e-05,
+      "loss": 0.0221,
+      "step": 1641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6090470272539613e-05,
+      "loss": 0.0396,
+      "step": 1642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608557381946287e-05,
+      "loss": 0.029,
+      "step": 1643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6080675048107126e-05,
+      "loss": 0.0321,
+      "step": 1644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6075773960338548e-05,
+      "loss": 0.0252,
+      "step": 1645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607087055802419e-05,
+      "loss": 0.0233,
+      "step": 1646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6065964843031987e-05,
+      "loss": 0.0334,
+      "step": 1647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6061056817230754e-05,
+      "loss": 0.0277,
+      "step": 1648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.605614648249019e-05,
+      "loss": 0.0197,
+      "step": 1649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6051233840680868e-05,
+      "loss": 0.0318,
+      "step": 1650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6046318893674242e-05,
+      "loss": 0.0235,
+      "step": 1651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6041401643342643e-05,
+      "loss": 0.0266,
+      "step": 1652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6036482091559287e-05,
+      "loss": 0.0259,
+      "step": 1653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6031560240198257e-05,
+      "loss": 0.0269,
+      "step": 1654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6026636091134516e-05,
+      "loss": 0.03,
+      "step": 1655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60217096462439e-05,
+      "loss": 0.0296,
+      "step": 1656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6016780907403124e-05,
+      "loss": 0.0413,
+      "step": 1657
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6011849876489777e-05,
+      "loss": 0.0362,
+      "step": 1658
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6006916555382317e-05,
+      "loss": 0.0281,
+      "step": 1659
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6001980945960076e-05,
+      "loss": 0.0232,
+      "step": 1660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5997043050103253e-05,
+      "loss": 0.0331,
+      "step": 1661
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5992102869692933e-05,
+      "loss": 0.0262,
+      "step": 1662
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.598716040661105e-05,
+      "loss": 0.0308,
+      "step": 1663
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5982215662740427e-05,
+      "loss": 0.0319,
+      "step": 1664
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5977268639964745e-05,
+      "loss": 0.0274,
+      "step": 1665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5972319340168555e-05,
+      "loss": 0.0248,
+      "step": 1666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5967367765237278e-05,
+      "loss": 0.0358,
+      "step": 1667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.59624139170572e-05,
+      "loss": 0.0243,
+      "step": 1668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.595745779751547e-05,
+      "loss": 0.0327,
+      "step": 1669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5952499408500105e-05,
+      "loss": 0.0253,
+      "step": 1670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5947538751899994e-05,
+      "loss": 0.023,
+      "step": 1671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5942575829604875e-05,
+      "loss": 0.021,
+      "step": 1672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5937610643505355e-05,
+      "loss": 0.0301,
+      "step": 1673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5932643195492915e-05,
+      "loss": 0.0272,
+      "step": 1674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5927673487459878e-05,
+      "loss": 0.0242,
+      "step": 1675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5922701521299443e-05,
+      "loss": 0.0278,
+      "step": 1676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591772729890566e-05,
+      "loss": 0.0318,
+      "step": 1677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5912750822173446e-05,
+      "loss": 0.0248,
+      "step": 1678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5907772092998573e-05,
+      "loss": 0.0266,
+      "step": 1679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5902791113277668e-05,
+      "loss": 0.027,
+      "step": 1680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5897807884908223e-05,
+      "loss": 0.0304,
+      "step": 1681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.589282240978858e-05,
+      "loss": 0.03,
+      "step": 1682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5887834689817937e-05,
+      "loss": 0.0223,
+      "step": 1683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.588284472689635e-05,
+      "loss": 0.0319,
+      "step": 1684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.0218,
+      "step": 1685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5872858079804846e-05,
+      "loss": 0.031,
+      "step": 1686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5867861399439302e-05,
+      "loss": 0.0298,
+      "step": 1687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5862862483731574e-05,
+      "loss": 0.0236,
+      "step": 1688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5857861334585986e-05,
+      "loss": 0.0234,
+      "step": 1689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.58528579539077e-05,
+      "loss": 0.0257,
+      "step": 1690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5847852343602747e-05,
+      "loss": 0.0277,
+      "step": 1691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5842844505577985e-05,
+      "loss": 0.0268,
+      "step": 1692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5837834441741145e-05,
+      "loss": 0.0241,
+      "step": 1693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.583282215400079e-05,
+      "loss": 0.0326,
+      "step": 1694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5827807644266335e-05,
+      "loss": 0.0264,
+      "step": 1695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5822790914448035e-05,
+      "loss": 0.0214,
+      "step": 1696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5817771966457004e-05,
+      "loss": 0.0239,
+      "step": 1697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5812750802205187e-05,
+      "loss": 0.0265,
+      "step": 1698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5807727423605387e-05,
+      "loss": 0.0417,
+      "step": 1699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.580270183257124e-05,
+      "loss": 0.0323,
+      "step": 1700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5797674031017223e-05,
+      "loss": 0.0354,
+      "step": 1701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5792644020858666e-05,
+      "loss": 0.0293,
+      "step": 1702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5787611804011735e-05,
+      "loss": 0.0289,
+      "step": 1703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5782577382393435e-05,
+      "loss": 0.0249,
+      "step": 1704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.577754075792161e-05,
+      "loss": 0.0296,
+      "step": 1705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5772501932514948e-05,
+      "loss": 0.023,
+      "step": 1706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5767460908092972e-05,
+      "loss": 0.0326,
+      "step": 1707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.576241768657604e-05,
+      "loss": 0.0257,
+      "step": 1708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5757372269885354e-05,
+      "loss": 0.0312,
+      "step": 1709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.575232465994295e-05,
+      "loss": 0.0324,
+      "step": 1710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5747274858671695e-05,
+      "loss": 0.0234,
+      "step": 1711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5742222867995295e-05,
+      "loss": 0.0371,
+      "step": 1712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5737168689838287e-05,
+      "loss": 0.0262,
+      "step": 1713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5732112326126046e-05,
+      "loss": 0.0295,
+      "step": 1714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.572705377878478e-05,
+      "loss": 0.0268,
+      "step": 1715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.572199304974152e-05,
+      "loss": 0.0254,
+      "step": 1716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5716930140924128e-05,
+      "loss": 0.0295,
+      "step": 1717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.571186505426132e-05,
+      "loss": 0.0278,
+      "step": 1718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.570679779168261e-05,
+      "loss": 0.0222,
+      "step": 1719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5701728355118355e-05,
+      "loss": 0.0314,
+      "step": 1720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5696656746499746e-05,
+      "loss": 0.0176,
+      "step": 1721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5691582967758793e-05,
+      "loss": 0.0268,
+      "step": 1722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5686507020828332e-05,
+      "loss": 0.0231,
+      "step": 1723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5681428907642033e-05,
+      "loss": 0.0284,
+      "step": 1724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5676348630134385e-05,
+      "loss": 0.0354,
+      "step": 1725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.56712661902407e-05,
+      "loss": 0.0289,
+      "step": 1726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.566618158989712e-05,
+      "loss": 0.0303,
+      "step": 1727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.56610948310406e-05,
+      "loss": 0.0335,
+      "step": 1728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5656005915608932e-05,
+      "loss": 0.0328,
+      "step": 1729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.565091484554072e-05,
+      "loss": 0.0252,
+      "step": 1730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5645821622775383e-05,
+      "loss": 0.0216,
+      "step": 1731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5640726249253172e-05,
+      "loss": 0.0335,
+      "step": 1732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5635628726915154e-05,
+      "loss": 0.0318,
+      "step": 1733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5630529057703213e-05,
+      "loss": 0.027,
+      "step": 1734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5625427243560053e-05,
+      "loss": 0.0349,
+      "step": 1735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.562032328642918e-05,
+      "loss": 0.0227,
+      "step": 1736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5615217188254945e-05,
+      "loss": 0.0352,
+      "step": 1737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5610108950982494e-05,
+      "loss": 0.0296,
+      "step": 1738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.560499857655779e-05,
+      "loss": 0.0327,
+      "step": 1739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.559988606692761e-05,
+      "loss": 0.0276,
+      "step": 1740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.559477142403956e-05,
+      "loss": 0.0319,
+      "step": 1741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5589654649842036e-05,
+      "loss": 0.0245,
+      "step": 1742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.558453574628426e-05,
+      "loss": 0.0328,
+      "step": 1743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5579414715316258e-05,
+      "loss": 0.028,
+      "step": 1744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5574291558888872e-05,
+      "loss": 0.0268,
+      "step": 1745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.556916627895375e-05,
+      "loss": 0.0253,
+      "step": 1746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.556403887746335e-05,
+      "loss": 0.0338,
+      "step": 1747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5558909356370944e-05,
+      "loss": 0.0267,
+      "step": 1748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.55537777176306e-05,
+      "loss": 0.0263,
+      "step": 1749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5548643963197197e-05,
+      "loss": 0.0219,
+      "step": 1750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5543508095026424e-05,
+      "loss": 0.0189,
+      "step": 1751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.553837011507478e-05,
+      "loss": 0.0281,
+      "step": 1752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5533230025299547e-05,
+      "loss": 0.0255,
+      "step": 1753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5528087827658836e-05,
+      "loss": 0.0213,
+      "step": 1754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5522943524111547e-05,
+      "loss": 0.0201,
+      "step": 1755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5517797116617383e-05,
+      "loss": 0.0325,
+      "step": 1756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5512648607136854e-05,
+      "loss": 0.0236,
+      "step": 1757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5507497997631267e-05,
+      "loss": 0.0245,
+      "step": 1758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5502345290062728e-05,
+      "loss": 0.0248,
+      "step": 1759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5497190486394146e-05,
+      "loss": 0.03,
+      "step": 1760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5492033588589223e-05,
+      "loss": 0.0261,
+      "step": 1761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.548687459861246e-05,
+      "loss": 0.0269,
+      "step": 1762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.548171351842917e-05,
+      "loss": 0.026,
+      "step": 1763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.547655035000543e-05,
+      "loss": 0.0185,
+      "step": 1764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5471385095308142e-05,
+      "loss": 0.0336,
+      "step": 1765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5466217756304996e-05,
+      "loss": 0.0259,
+      "step": 1766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546104833496447e-05,
+      "loss": 0.0189,
+      "step": 1767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.545587683325583e-05,
+      "loss": 0.0215,
+      "step": 1768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5450703253149155e-05,
+      "loss": 0.0245,
+      "step": 1769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5445527596615294e-05,
+      "loss": 0.0261,
+      "step": 1770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.54403498656259e-05,
+      "loss": 0.0241,
+      "step": 1771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.543517006215341e-05,
+      "loss": 0.0268,
+      "step": 1772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5429988188171053e-05,
+      "loss": 0.0284,
+      "step": 1773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.542480424565285e-05,
+      "loss": 0.022,
+      "step": 1774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5419618236573598e-05,
+      "loss": 0.0239,
+      "step": 1775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5414430162908903e-05,
+      "loss": 0.0227,
+      "step": 1776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5409240026635135e-05,
+      "loss": 0.0236,
+      "step": 1777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.540404782972946e-05,
+      "loss": 0.0304,
+      "step": 1778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.539885357416983e-05,
+      "loss": 0.0283,
+      "step": 1779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5393657261934978e-05,
+      "loss": 0.0206,
+      "step": 1780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.538845889500442e-05,
+      "loss": 0.0311,
+      "step": 1781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5383258475358465e-05,
+      "loss": 0.0272,
+      "step": 1782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5378056004978183e-05,
+      "loss": 0.0299,
+      "step": 1783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5372851485845446e-05,
+      "loss": 0.0321,
+      "step": 1784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5367644919942896e-05,
+      "loss": 0.0244,
+      "step": 1785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.536243630925396e-05,
+      "loss": 0.0217,
+      "step": 1786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5357225655762836e-05,
+      "loss": 0.0301,
+      "step": 1787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.535201296145451e-05,
+      "loss": 0.026,
+      "step": 1788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.534679822831473e-05,
+      "loss": 0.0258,
+      "step": 1789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5341581458330043e-05,
+      "loss": 0.0253,
+      "step": 1790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5336362653487756e-05,
+      "loss": 0.0277,
+      "step": 1791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5331141815775953e-05,
+      "loss": 0.024,
+      "step": 1792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5325918947183498e-05,
+      "loss": 0.0234,
+      "step": 1793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5320694049700025e-05,
+      "loss": 0.0269,
+      "step": 1794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.531546712531594e-05,
+      "loss": 0.0229,
+      "step": 1795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5310238176022417e-05,
+      "loss": 0.03,
+      "step": 1796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.530500720381142e-05,
+      "loss": 0.0326,
+      "step": 1797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.529977421067566e-05,
+      "loss": 0.0332,
+      "step": 1798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5294539198608627e-05,
+      "loss": 0.0245,
+      "step": 1799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.528930216960459e-05,
+      "loss": 0.0323,
+      "step": 1800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.528406312565857e-05,
+      "loss": 0.0243,
+      "step": 1801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5278822068766367e-05,
+      "loss": 0.0321,
+      "step": 1802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5273579000924545e-05,
+      "loss": 0.0255,
+      "step": 1803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.526833392413043e-05,
+      "loss": 0.0229,
+      "step": 1804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5263086840382117e-05,
+      "loss": 0.031,
+      "step": 1805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5257837751678467e-05,
+      "loss": 0.0287,
+      "step": 1806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5252586660019101e-05,
+      "loss": 0.0342,
+      "step": 1807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5247333567404407e-05,
+      "loss": 0.0224,
+      "step": 1808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.524207847583553e-05,
+      "loss": 0.0222,
+      "step": 1809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.523682138731438e-05,
+      "loss": 0.0324,
+      "step": 1810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5231562303843631e-05,
+      "loss": 0.0278,
+      "step": 1811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.522630122742671e-05,
+      "loss": 0.0233,
+      "step": 1812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.522103816006781e-05,
+      "loss": 0.0306,
+      "step": 1813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5215773103771872e-05,
+      "loss": 0.0193,
+      "step": 1814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5210506060544605e-05,
+      "loss": 0.0336,
+      "step": 1815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5205237032392477e-05,
+      "loss": 0.0308,
+      "step": 1816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5199966021322698e-05,
+      "loss": 0.0285,
+      "step": 1817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5194693029343249e-05,
+      "loss": 0.0238,
+      "step": 1818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5189418058462853e-05,
+      "loss": 0.033,
+      "step": 1819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5184141110690995e-05,
+      "loss": 0.025,
+      "step": 1820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.517886218803791e-05,
+      "loss": 0.0336,
+      "step": 1821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5173581292514591e-05,
+      "loss": 0.028,
+      "step": 1822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.516829842613277e-05,
+      "loss": 0.0264,
+      "step": 1823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.516301359090494e-05,
+      "loss": 0.0217,
+      "step": 1824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5157726788844342e-05,
+      "loss": 0.0362,
+      "step": 1825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5152438021964968e-05,
+      "loss": 0.0264,
+      "step": 1826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.514714729228155e-05,
+      "loss": 0.0318,
+      "step": 1827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5141854601809583e-05,
+      "loss": 0.027,
+      "step": 1828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5136559952565288e-05,
+      "loss": 0.0262,
+      "step": 1829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5131263346565654e-05,
+      "loss": 0.0182,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5125964785828395e-05,
+      "loss": 0.0254,
+      "step": 1831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5120664272371991e-05,
+      "loss": 0.0226,
+      "step": 1832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5115361808215651e-05,
+      "loss": 0.0226,
+      "step": 1833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5110057395379327e-05,
+      "loss": 0.0197,
+      "step": 1834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5104751035883723e-05,
+      "loss": 0.027,
+      "step": 1835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5099442731750275e-05,
+      "loss": 0.0275,
+      "step": 1836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5094132485001165e-05,
+      "loss": 0.03,
+      "step": 1837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5088820297659314e-05,
+      "loss": 0.0261,
+      "step": 1838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5083506171748383e-05,
+      "loss": 0.0241,
+      "step": 1839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5078190109292768e-05,
+      "loss": 0.025,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5072872112317607e-05,
+      "loss": 0.0237,
+      "step": 1841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5067552182848775e-05,
+      "loss": 0.0244,
+      "step": 1842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5062230322912875e-05,
+      "loss": 0.021,
+      "step": 1843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5056906534537256e-05,
+      "loss": 0.0242,
+      "step": 1844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.505158081975e-05,
+      "loss": 0.0342,
+      "step": 1845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5046253180579915e-05,
+      "loss": 0.0177,
+      "step": 1846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5040923619056554e-05,
+      "loss": 0.0228,
+      "step": 1847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5035592137210188e-05,
+      "loss": 0.0263,
+      "step": 1848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5030258737071827e-05,
+      "loss": 0.018,
+      "step": 1849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5024923420673223e-05,
+      "loss": 0.0214,
+      "step": 1850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5019586190046837e-05,
+      "loss": 0.0326,
+      "step": 1851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5014247047225868e-05,
+      "loss": 0.029,
+      "step": 1852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5008905994244255e-05,
+      "loss": 0.0304,
+      "step": 1853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.500356303313664e-05,
+      "loss": 0.0169,
+      "step": 1854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4998218165938419e-05,
+      "loss": 0.0277,
+      "step": 1855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4992871394685695e-05,
+      "loss": 0.026,
+      "step": 1856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4987522721415303e-05,
+      "loss": 0.0365,
+      "step": 1857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4982172148164804e-05,
+      "loss": 0.0242,
+      "step": 1858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4976819676972478e-05,
+      "loss": 0.0317,
+      "step": 1859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4971465309877334e-05,
+      "loss": 0.0327,
+      "step": 1860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4966109048919099e-05,
+      "loss": 0.0278,
+      "step": 1861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4960750896138223e-05,
+      "loss": 0.0206,
+      "step": 1862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4955390853575875e-05,
+      "loss": 0.0188,
+      "step": 1863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.495002892327395e-05,
+      "loss": 0.0269,
+      "step": 1864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.494466510727505e-05,
+      "loss": 0.033,
+      "step": 1865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.493929940762251e-05,
+      "loss": 0.0216,
+      "step": 1866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4933931826360369e-05,
+      "loss": 0.0278,
+      "step": 1867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4928562365533393e-05,
+      "loss": 0.0183,
+      "step": 1868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.492319102718706e-05,
+      "loss": 0.0281,
+      "step": 1869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4917817813367563e-05,
+      "loss": 0.0227,
+      "step": 1870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4912442726121807e-05,
+      "loss": 0.0199,
+      "step": 1871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.490706576749742e-05,
+      "loss": 0.0296,
+      "step": 1872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4901686939542731e-05,
+      "loss": 0.0174,
+      "step": 1873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4896306244306789e-05,
+      "loss": 0.0227,
+      "step": 1874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4890923683839349e-05,
+      "loss": 0.0225,
+      "step": 1875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4885539260190883e-05,
+      "loss": 0.0239,
+      "step": 1876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4880152975412565e-05,
+      "loss": 0.0226,
+      "step": 1877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4874764831556285e-05,
+      "loss": 0.0234,
+      "step": 1878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4869374830674641e-05,
+      "loss": 0.0263,
+      "step": 1879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4863982974820931e-05,
+      "loss": 0.0256,
+      "step": 1880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4858589266049168e-05,
+      "loss": 0.0247,
+      "step": 1881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4853193706414065e-05,
+      "loss": 0.0253,
+      "step": 1882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.484779629797104e-05,
+      "loss": 0.0177,
+      "step": 1883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4842397042776225e-05,
+      "loss": 0.0261,
+      "step": 1884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4836995942886443e-05,
+      "loss": 0.0235,
+      "step": 1885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4831593000359226e-05,
+      "loss": 0.0326,
+      "step": 1886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4826188217252803e-05,
+      "loss": 0.0303,
+      "step": 1887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4820781595626116e-05,
+      "loss": 0.0266,
+      "step": 1888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4815373137538796e-05,
+      "loss": 0.0247,
+      "step": 1889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4809962845051173e-05,
+      "loss": 0.0341,
+      "step": 1890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4804550720224287e-05,
+      "loss": 0.0239,
+      "step": 1891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4799136765119864e-05,
+      "loss": 0.0214,
+      "step": 1892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4793720981800331e-05,
+      "loss": 0.0242,
+      "step": 1893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4788303372328816e-05,
+      "loss": 0.0286,
+      "step": 1894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4782883938769138e-05,
+      "loss": 0.0273,
+      "step": 1895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4777462683185811e-05,
+      "loss": 0.0273,
+      "step": 1896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4772039607644047e-05,
+      "loss": 0.0189,
+      "step": 1897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4766614714209748e-05,
+      "loss": 0.0192,
+      "step": 1898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4761188004949502e-05,
+      "loss": 0.032,
+      "step": 1899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.475575948193061e-05,
+      "loss": 0.0209,
+      "step": 1900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4750329147221034e-05,
+      "loss": 0.0251,
+      "step": 1901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4744897002889451e-05,
+      "loss": 0.0333,
+      "step": 1902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4739463051005221e-05,
+      "loss": 0.0311,
+      "step": 1903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4734027293638378e-05,
+      "loss": 0.0348,
+      "step": 1904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4728589732859663e-05,
+      "loss": 0.0222,
+      "step": 1905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4723150370740502e-05,
+      "loss": 0.0262,
+      "step": 1906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4717709209352993e-05,
+      "loss": 0.0302,
+      "step": 1907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4712266250769932e-05,
+      "loss": 0.028,
+      "step": 1908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4706821497064794e-05,
+      "loss": 0.0308,
+      "step": 1909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4701374950311742e-05,
+      "loss": 0.0243,
+      "step": 1910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4695926612585615e-05,
+      "loss": 0.0231,
+      "step": 1911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4690476485961949e-05,
+      "loss": 0.029,
+      "step": 1912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4685024572516943e-05,
+      "loss": 0.0221,
+      "step": 1913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4679570874327487e-05,
+      "loss": 0.0254,
+      "step": 1914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.467411539347115e-05,
+      "loss": 0.028,
+      "step": 1915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4668658132026178e-05,
+      "loss": 0.0271,
+      "step": 1916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4663199092071497e-05,
+      "loss": 0.0286,
+      "step": 1917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4657738275686711e-05,
+      "loss": 0.0277,
+      "step": 1918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4652275684952097e-05,
+      "loss": 0.0284,
+      "step": 1919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4646811321948618e-05,
+      "loss": 0.0292,
+      "step": 1920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4641345188757894e-05,
+      "loss": 0.0255,
+      "step": 1921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4635877287462239e-05,
+      "loss": 0.0261,
+      "step": 1922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4630407620144627e-05,
+      "loss": 0.021,
+      "step": 1923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4624936188888713e-05,
+      "loss": 0.0243,
+      "step": 1924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4619462995778815e-05,
+      "loss": 0.0318,
+      "step": 1925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4613988042899933e-05,
+      "loss": 0.0322,
+      "step": 1926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.460851133233773e-05,
+      "loss": 0.0278,
+      "step": 1927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.460303286617854e-05,
+      "loss": 0.0248,
+      "step": 1928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4597552646509367e-05,
+      "loss": 0.0267,
+      "step": 1929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4592070675417882e-05,
+      "loss": 0.0297,
+      "step": 1930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4586586954992426e-05,
+      "loss": 0.0209,
+      "step": 1931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4581101487322e-05,
+      "loss": 0.0209,
+      "step": 1932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4575614274496275e-05,
+      "loss": 0.025,
+      "step": 1933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4570125318605593e-05,
+      "loss": 0.0269,
+      "step": 1934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.456463462174094e-05,
+      "loss": 0.0219,
+      "step": 1935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.455914218599399e-05,
+      "loss": 0.0237,
+      "step": 1936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4553648013457072e-05,
+      "loss": 0.0248,
+      "step": 1937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4548152106223157e-05,
+      "loss": 0.0253,
+      "step": 1938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4542654466385902e-05,
+      "loss": 0.024,
+      "step": 1939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4537155096039617e-05,
+      "loss": 0.0295,
+      "step": 1940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4531653997279262e-05,
+      "loss": 0.0289,
+      "step": 1941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4526151172200466e-05,
+      "loss": 0.0292,
+      "step": 1942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4520646622899515e-05,
+      "loss": 0.0296,
+      "step": 1943
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.451514035147334e-05,
+      "loss": 0.0262,
+      "step": 1944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4509632360019545e-05,
+      "loss": 0.0288,
+      "step": 1945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4504122650636374e-05,
+      "loss": 0.0208,
+      "step": 1946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.449861122542274e-05,
+      "loss": 0.0271,
+      "step": 1947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4493098086478196e-05,
+      "loss": 0.0284,
+      "step": 1948
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4487583235902955e-05,
+      "loss": 0.0215,
+      "step": 1949
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4482066675797883e-05,
+      "loss": 0.0323,
+      "step": 1950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4476548408264494e-05,
+      "loss": 0.0235,
+      "step": 1951
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.447102843540495e-05,
+      "loss": 0.0229,
+      "step": 1952
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4465506759322074e-05,
+      "loss": 0.0326,
+      "step": 1953
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4459983382119325e-05,
+      "loss": 0.0251,
+      "step": 1954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4454458305900811e-05,
+      "loss": 0.0241,
+      "step": 1955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4448931532771303e-05,
+      "loss": 0.0397,
+      "step": 1956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4443403064836194e-05,
+      "loss": 0.026,
+      "step": 1957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4437872904201542e-05,
+      "loss": 0.0264,
+      "step": 1958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4432341052974047e-05,
+      "loss": 0.0273,
+      "step": 1959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.442680751326104e-05,
+      "loss": 0.0295,
+      "step": 1960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4421272287170509e-05,
+      "loss": 0.0204,
+      "step": 1961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4415735376811084e-05,
+      "loss": 0.0223,
+      "step": 1962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4410196784292025e-05,
+      "loss": 0.0273,
+      "step": 1963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4404656511723243e-05,
+      "loss": 0.0263,
+      "step": 1964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.439911456121529e-05,
+      "loss": 0.0247,
+      "step": 1965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4393570934879346e-05,
+      "loss": 0.0292,
+      "step": 1966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4388025634827245e-05,
+      "loss": 0.0202,
+      "step": 1967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4382478663171449e-05,
+      "loss": 0.0284,
+      "step": 1968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4376930022025054e-05,
+      "loss": 0.0249,
+      "step": 1969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4371379713501795e-05,
+      "loss": 0.0241,
+      "step": 1970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4365827739716053e-05,
+      "loss": 0.0253,
+      "step": 1971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4360274102782823e-05,
+      "loss": 0.0322,
+      "step": 1972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4354718804817753e-05,
+      "loss": 0.0274,
+      "step": 1973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4349161847937109e-05,
+      "loss": 0.0228,
+      "step": 1974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.43436032342578e-05,
+      "loss": 0.0222,
+      "step": 1975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4338042965897356e-05,
+      "loss": 0.0171,
+      "step": 1976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.433248104497395e-05,
+      "loss": 0.0378,
+      "step": 1977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4326917473606368e-05,
+      "loss": 0.0252,
+      "step": 1978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.432135225391404e-05,
+      "loss": 0.0224,
+      "step": 1979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4315785388017016e-05,
+      "loss": 0.0345,
+      "step": 1980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4310216878035979e-05,
+      "loss": 0.0262,
+      "step": 1981
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4304646726092229e-05,
+      "loss": 0.0271,
+      "step": 1982
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4299074934307697e-05,
+      "loss": 0.024,
+      "step": 1983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4293501504804944e-05,
+      "loss": 0.0305,
+      "step": 1984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4287926439707141e-05,
+      "loss": 0.0214,
+      "step": 1985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4282349741138096e-05,
+      "loss": 0.0188,
+      "step": 1986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4276771411222235e-05,
+      "loss": 0.0265,
+      "step": 1987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4271191452084598e-05,
+      "loss": 0.0216,
+      "step": 1988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4265609865850856e-05,
+      "loss": 0.0241,
+      "step": 1989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4260026654647296e-05,
+      "loss": 0.0257,
+      "step": 1990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4254441820600822e-05,
+      "loss": 0.0372,
+      "step": 1991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4248855365838956e-05,
+      "loss": 0.0265,
+      "step": 1992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.424326729248985e-05,
+      "loss": 0.024,
+      "step": 1993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4237677602682244e-05,
+      "loss": 0.0267,
+      "step": 1994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4232086298545525e-05,
+      "loss": 0.0218,
+      "step": 1995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4226493382209681e-05,
+      "loss": 0.0245,
+      "step": 1996
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4220898855805306e-05,
+      "loss": 0.0306,
+      "step": 1997
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4215302721463624e-05,
+      "loss": 0.0282,
+      "step": 1998
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4209704981316463e-05,
+      "loss": 0.0275,
+      "step": 1999
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4204105637496261e-05,
+      "loss": 0.0189,
+      "step": 2000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.419850469213607e-05,
+      "loss": 0.031,
+      "step": 2001
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4192902147369552e-05,
+      "loss": 0.0193,
+      "step": 2002
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4187298005330976e-05,
+      "loss": 0.0263,
+      "step": 2003
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4181692268155223e-05,
+      "loss": 0.0243,
+      "step": 2004
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4176084937977778e-05,
+      "loss": 0.0299,
+      "step": 2005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4170476016934738e-05,
+      "loss": 0.0242,
+      "step": 2006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4164865507162801e-05,
+      "loss": 0.0322,
+      "step": 2007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4159253410799272e-05,
+      "loss": 0.0284,
+      "step": 2008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4153639729982057e-05,
+      "loss": 0.025,
+      "step": 2009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4148024466849673e-05,
+      "loss": 0.0219,
+      "step": 2010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4142407623541237e-05,
+      "loss": 0.0288,
+      "step": 2011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.413678920219646e-05,
+      "loss": 0.0203,
+      "step": 2012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4131169204955667e-05,
+      "loss": 0.0236,
+      "step": 2013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4125547633959777e-05,
+      "loss": 0.0186,
+      "step": 2014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4119924491350307e-05,
+      "loss": 0.0246,
+      "step": 2015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4114299779269372e-05,
+      "loss": 0.02,
+      "step": 2016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4108673499859689e-05,
+      "loss": 0.0225,
+      "step": 2017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4103045655264576e-05,
+      "loss": 0.0232,
+      "step": 2018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4097416247627933e-05,
+      "loss": 0.0236,
+      "step": 2019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4091785279094266e-05,
+      "loss": 0.0194,
+      "step": 2020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.408615275180868e-05,
+      "loss": 0.0237,
+      "step": 2021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4080518667916857e-05,
+      "loss": 0.017,
+      "step": 2022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4074883029565092e-05,
+      "loss": 0.0254,
+      "step": 2023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.406924583890026e-05,
+      "loss": 0.0233,
+      "step": 2024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4063607098069825e-05,
+      "loss": 0.0194,
+      "step": 2025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4057966809221847e-05,
+      "loss": 0.0212,
+      "step": 2026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4052324974504987e-05,
+      "loss": 0.0297,
+      "step": 2027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4046681596068468e-05,
+      "loss": 0.0183,
+      "step": 2028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4041036676062125e-05,
+      "loss": 0.0332,
+      "step": 2029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.403539021663637e-05,
+      "loss": 0.026,
+      "step": 2030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4029742219942197e-05,
+      "loss": 0.0267,
+      "step": 2031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.40240926881312e-05,
+      "loss": 0.0185,
+      "step": 2032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4018441623355546e-05,
+      "loss": 0.025,
+      "step": 2033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4012789027767985e-05,
+      "loss": 0.0226,
+      "step": 2034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4007134903521862e-05,
+      "loss": 0.0183,
+      "step": 2035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4001479252771092e-05,
+      "loss": 0.0244,
+      "step": 2036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3995822077670174e-05,
+      "loss": 0.0244,
+      "step": 2037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3990163380374195e-05,
+      "loss": 0.0262,
+      "step": 2038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3984503163038816e-05,
+      "loss": 0.0288,
+      "step": 2039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3978841427820273e-05,
+      "loss": 0.0228,
+      "step": 2040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3973178176875392e-05,
+      "loss": 0.0334,
+      "step": 2041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3967513412361568e-05,
+      "loss": 0.0237,
+      "step": 2042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.396184713643677e-05,
+      "loss": 0.0232,
+      "step": 2043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3956179351259548e-05,
+      "loss": 0.0252,
+      "step": 2044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3950510058989028e-05,
+      "loss": 0.0211,
+      "step": 2045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.394483926178491e-05,
+      "loss": 0.0195,
+      "step": 2046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3939166961807457e-05,
+      "loss": 0.0278,
+      "step": 2047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3933493161217521e-05,
+      "loss": 0.0242,
+      "step": 2048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3927817862176516e-05,
+      "loss": 0.0306,
+      "step": 2049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3922141066846425e-05,
+      "loss": 0.0185,
+      "step": 2050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3916462777389802e-05,
+      "loss": 0.0241,
+      "step": 2051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3910782995969783e-05,
+      "loss": 0.0251,
+      "step": 2052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.390510172475005e-05,
+      "loss": 0.0203,
+      "step": 2053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.389941896589487e-05,
+      "loss": 0.0231,
+      "step": 2054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3893734721569074e-05,
+      "loss": 0.0285,
+      "step": 2055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3888048993938048e-05,
+      "loss": 0.0254,
+      "step": 2056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3882361785167757e-05,
+      "loss": 0.0291,
+      "step": 2057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.387667309742472e-05,
+      "loss": 0.0273,
+      "step": 2058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3870982932876027e-05,
+      "loss": 0.0195,
+      "step": 2059
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3865291293689324e-05,
+      "loss": 0.0259,
+      "step": 2060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3859598182032827e-05,
+      "loss": 0.0256,
+      "step": 2061
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3853903600075302e-05,
+      "loss": 0.0294,
+      "step": 2062
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3848207549986085e-05,
+      "loss": 0.0194,
+      "step": 2063
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3842510033935068e-05,
+      "loss": 0.0279,
+      "step": 2064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3836811054092694e-05,
+      "loss": 0.0224,
+      "step": 2065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3831110612629979e-05,
+      "loss": 0.0266,
+      "step": 2066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.382540871171848e-05,
+      "loss": 0.0228,
+      "step": 2067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3819705353530322e-05,
+      "loss": 0.0266,
+      "step": 2068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3814000540238176e-05,
+      "loss": 0.0213,
+      "step": 2069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3808294274015277e-05,
+      "loss": 0.0243,
+      "step": 2070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3802586557035404e-05,
+      "loss": 0.0211,
+      "step": 2071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.379687739147289e-05,
+      "loss": 0.0256,
+      "step": 2072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3791166779502632e-05,
+      "loss": 0.0209,
+      "step": 2073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3785454723300058e-05,
+      "loss": 0.0224,
+      "step": 2074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3779741225041164e-05,
+      "loss": 0.0347,
+      "step": 2075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3774026286902484e-05,
+      "loss": 0.0213,
+      "step": 2076
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.376830991106111e-05,
+      "loss": 0.0185,
+      "step": 2077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3762592099694666e-05,
+      "loss": 0.0284,
+      "step": 2078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3756872854981342e-05,
+      "loss": 0.0186,
+      "step": 2079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3751152179099865e-05,
+      "loss": 0.0245,
+      "step": 2080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.37454300742295e-05,
+      "loss": 0.0205,
+      "step": 2081
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3739706542550073e-05,
+      "loss": 0.0211,
+      "step": 2082
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.373398158624194e-05,
+      "loss": 0.0175,
+      "step": 2083
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3728255207486001e-05,
+      "loss": 0.0241,
+      "step": 2084
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3722527408463705e-05,
+      "loss": 0.0231,
+      "step": 2085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3716798191357038e-05,
+      "loss": 0.0233,
+      "step": 2086
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3711067558348523e-05,
+      "loss": 0.027,
+      "step": 2087
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3705335511621229e-05,
+      "loss": 0.0183,
+      "step": 2088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3699602053358758e-05,
+      "loss": 0.0359,
+      "step": 2089
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.369386718574525e-05,
+      "loss": 0.0271,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3688130910965387e-05,
+      "loss": 0.0234,
+      "step": 2091
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3682393231204382e-05,
+      "loss": 0.02,
+      "step": 2092
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3676654148647985e-05,
+      "loss": 0.0254,
+      "step": 2093
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.367091366548248e-05,
+      "loss": 0.0294,
+      "step": 2094
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3665171783894691e-05,
+      "loss": 0.0221,
+      "step": 2095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3659428506071957e-05,
+      "loss": 0.0282,
+      "step": 2096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3653683834202168e-05,
+      "loss": 0.0291,
+      "step": 2097
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3647937770473739e-05,
+      "loss": 0.0409,
+      "step": 2098
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3642190317075607e-05,
+      "loss": 0.03,
+      "step": 2099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.363644147619725e-05,
+      "loss": 0.0257,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.363069125002867e-05,
+      "loss": 0.0238,
+      "step": 2101
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3624939640760395e-05,
+      "loss": 0.0262,
+      "step": 2102
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.361918665058348e-05,
+      "loss": 0.0288,
+      "step": 2103
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.361343228168951e-05,
+      "loss": 0.0243,
+      "step": 2104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3607676536270594e-05,
+      "loss": 0.0257,
+      "step": 2105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3601919416519356e-05,
+      "loss": 0.0339,
+      "step": 2106
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.359616092462896e-05,
+      "loss": 0.0245,
+      "step": 2107
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3590401062793084e-05,
+      "loss": 0.028,
+      "step": 2108
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3584639833205919e-05,
+      "loss": 0.0165,
+      "step": 2109
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3578877238062198e-05,
+      "loss": 0.0305,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3573113279557156e-05,
+      "loss": 0.026,
+      "step": 2111
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.356734795988655e-05,
+      "loss": 0.0281,
+      "step": 2112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3561581281246667e-05,
+      "loss": 0.0294,
+      "step": 2113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3555813245834303e-05,
+      "loss": 0.0238,
+      "step": 2114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3550043855846766e-05,
+      "loss": 0.0305,
+      "step": 2115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3544273113481891e-05,
+      "loss": 0.022,
+      "step": 2116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3538501020938025e-05,
+      "loss": 0.0252,
+      "step": 2117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3532727580414018e-05,
+      "loss": 0.0361,
+      "step": 2118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3526952794109251e-05,
+      "loss": 0.0246,
+      "step": 2119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3521176664223608e-05,
+      "loss": 0.0252,
+      "step": 2120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3515399192957483e-05,
+      "loss": 0.0277,
+      "step": 2121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.350962038251179e-05,
+      "loss": 0.0251,
+      "step": 2122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3503840235087944e-05,
+      "loss": 0.0195,
+      "step": 2123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.349805875288787e-05,
+      "loss": 0.0237,
+      "step": 2124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.349227593811401e-05,
+      "loss": 0.0272,
+      "step": 2125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3486491792969307e-05,
+      "loss": 0.0269,
+      "step": 2126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3480706319657208e-05,
+      "loss": 0.0277,
+      "step": 2127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3474919520381673e-05,
+      "loss": 0.0254,
+      "step": 2128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3469131397347163e-05,
+      "loss": 0.0232,
+      "step": 2129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3463341952758647e-05,
+      "loss": 0.0259,
+      "step": 2130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3457551188821586e-05,
+      "loss": 0.0235,
+      "step": 2131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3451759107741962e-05,
+      "loss": 0.0302,
+      "step": 2132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3445965711726243e-05,
+      "loss": 0.0227,
+      "step": 2133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3440171002981403e-05,
+      "loss": 0.0273,
+      "step": 2134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3434374983714924e-05,
+      "loss": 0.0257,
+      "step": 2135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3428577656134773e-05,
+      "loss": 0.0223,
+      "step": 2136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3422779022449426e-05,
+      "loss": 0.0198,
+      "step": 2137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3416979084867851e-05,
+      "loss": 0.0256,
+      "step": 2138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3411177845599519e-05,
+      "loss": 0.0247,
+      "step": 2139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3405375306854384e-05,
+      "loss": 0.0273,
+      "step": 2140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3399571470842915e-05,
+      "loss": 0.0292,
+      "step": 2141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3393766339776058e-05,
+      "loss": 0.0244,
+      "step": 2142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3387959915865256e-05,
+      "loss": 0.0198,
+      "step": 2143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3382152201322451e-05,
+      "loss": 0.0271,
+      "step": 2144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3376343198360073e-05,
+      "loss": 0.0253,
+      "step": 2145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3370532909191036e-05,
+      "loss": 0.0279,
+      "step": 2146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3364721336028758e-05,
+      "loss": 0.0291,
+      "step": 2147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3358908481087133e-05,
+      "loss": 0.0216,
+      "step": 2148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.335309434658055e-05,
+      "loss": 0.0219,
+      "step": 2149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3347278934723888e-05,
+      "loss": 0.024,
+      "step": 2150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3341462247732504e-05,
+      "loss": 0.0239,
+      "step": 2151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3335644287822245e-05,
+      "loss": 0.0204,
+      "step": 2152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3329825057209446e-05,
+      "loss": 0.0279,
+      "step": 2153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3324004558110923e-05,
+      "loss": 0.0238,
+      "step": 2154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3318182792743975e-05,
+      "loss": 0.0202,
+      "step": 2155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3312359763326384e-05,
+      "loss": 0.0208,
+      "step": 2156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3306535472076415e-05,
+      "loss": 0.0298,
+      "step": 2157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.330070992121281e-05,
+      "loss": 0.0259,
+      "step": 2158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3294883112954793e-05,
+      "loss": 0.0272,
+      "step": 2159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.328905504952207e-05,
+      "loss": 0.0249,
+      "step": 2160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3283225733134824e-05,
+      "loss": 0.0258,
+      "step": 2161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3277395166013705e-05,
+      "loss": 0.0182,
+      "step": 2162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3271563350379857e-05,
+      "loss": 0.0193,
+      "step": 2163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3265730288454886e-05,
+      "loss": 0.0252,
+      "step": 2164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3259895982460877e-05,
+      "loss": 0.0246,
+      "step": 2165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3254060434620392e-05,
+      "loss": 0.0238,
+      "step": 2166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3248223647156464e-05,
+      "loss": 0.0346,
+      "step": 2167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3242385622292593e-05,
+      "loss": 0.0221,
+      "step": 2168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.323654636225276e-05,
+      "loss": 0.0213,
+      "step": 2169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3230705869261408e-05,
+      "loss": 0.0249,
+      "step": 2170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3224864145543456e-05,
+      "loss": 0.0276,
+      "step": 2171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3219021193324288e-05,
+      "loss": 0.0245,
+      "step": 2172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.321317701482976e-05,
+      "loss": 0.0302,
+      "step": 2173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3207331612286183e-05,
+      "loss": 0.0219,
+      "step": 2174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3201484987920354e-05,
+      "loss": 0.0249,
+      "step": 2175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3195637143959523e-05,
+      "loss": 0.0287,
+      "step": 2176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.31897880826314e-05,
+      "loss": 0.0155,
+      "step": 2177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3183937806164174e-05,
+      "loss": 0.0206,
+      "step": 2178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.317808631678648e-05,
+      "loss": 0.0214,
+      "step": 2179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3172233616727431e-05,
+      "loss": 0.0264,
+      "step": 2180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3166379708216588e-05,
+      "loss": 0.0256,
+      "step": 2181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3160524593483981e-05,
+      "loss": 0.0253,
+      "step": 2182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3154668274760094e-05,
+      "loss": 0.0279,
+      "step": 2183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3148810754275876e-05,
+      "loss": 0.0217,
+      "step": 2184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3142952034262726e-05,
+      "loss": 0.0281,
+      "step": 2185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3137092116952505e-05,
+      "loss": 0.021,
+      "step": 2186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3131231004577527e-05,
+      "loss": 0.0246,
+      "step": 2187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3125368699370567e-05,
+      "loss": 0.0187,
+      "step": 2188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3119505203564849e-05,
+      "loss": 0.0145,
+      "step": 2189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3113640519394051e-05,
+      "loss": 0.0263,
+      "step": 2190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3107774649092307e-05,
+      "loss": 0.0183,
+      "step": 2191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3101907594894197e-05,
+      "loss": 0.0219,
+      "step": 2192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3096039359034759e-05,
+      "loss": 0.0216,
+      "step": 2193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.0335,
+      "step": 2194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3084299351274284e-05,
+      "loss": 0.0121,
+      "step": 2195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3078427583845564e-05,
+      "loss": 0.0227,
+      "step": 2196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3072554643700145e-05,
+      "loss": 0.0295,
+      "step": 2197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3066680533075312e-05,
+      "loss": 0.0237,
+      "step": 2198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3060805254208773e-05,
+      "loss": 0.0333,
+      "step": 2199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.305492880933871e-05,
+      "loss": 0.0298,
+      "step": 2200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3049051200703726e-05,
+      "loss": 0.025,
+      "step": 2201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.304317243054288e-05,
+      "loss": 0.0241,
+      "step": 2202
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3037292501095674e-05,
+      "loss": 0.0202,
+      "step": 2203
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3031411414602038e-05,
+      "loss": 0.0231,
+      "step": 2204
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3025529173302358e-05,
+      "loss": 0.0277,
+      "step": 2205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3019645779437456e-05,
+      "loss": 0.0204,
+      "step": 2206
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3013761235248585e-05,
+      "loss": 0.024,
+      "step": 2207
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3007875542977448e-05,
+      "loss": 0.0202,
+      "step": 2208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3001988704866177e-05,
+      "loss": 0.0197,
+      "step": 2209
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2996100723157346e-05,
+      "loss": 0.0308,
+      "step": 2210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2990211600093957e-05,
+      "loss": 0.0221,
+      "step": 2211
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2984321337919457e-05,
+      "loss": 0.0191,
+      "step": 2212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2978429938877721e-05,
+      "loss": 0.0218,
+      "step": 2213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2972537405213055e-05,
+      "loss": 0.0241,
+      "step": 2214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.29666437391702e-05,
+      "loss": 0.0279,
+      "step": 2215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.296074894299433e-05,
+      "loss": 0.0232,
+      "step": 2216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.295485301893105e-05,
+      "loss": 0.0235,
+      "step": 2217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2948955969226384e-05,
+      "loss": 0.0254,
+      "step": 2218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2943057796126803e-05,
+      "loss": 0.0317,
+      "step": 2219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2937158501879195e-05,
+      "loss": 0.0195,
+      "step": 2220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2931258088730871e-05,
+      "loss": 0.0297,
+      "step": 2221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2925356558929579e-05,
+      "loss": 0.0156,
+      "step": 2222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2919453914723486e-05,
+      "loss": 0.0203,
+      "step": 2223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2913550158361181e-05,
+      "loss": 0.0217,
+      "step": 2224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2907645292091683e-05,
+      "loss": 0.0217,
+      "step": 2225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2901739318164432e-05,
+      "loss": 0.0232,
+      "step": 2226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2895832238829286e-05,
+      "loss": 0.0285,
+      "step": 2227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2889924056336531e-05,
+      "loss": 0.0312,
+      "step": 2228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2884014772936869e-05,
+      "loss": 0.0311,
+      "step": 2229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2878104390881414e-05,
+      "loss": 0.0289,
+      "step": 2230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2872192912421717e-05,
+      "loss": 0.0258,
+      "step": 2231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2866280339809731e-05,
+      "loss": 0.0256,
+      "step": 2232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2860366675297829e-05,
+      "loss": 0.0206,
+      "step": 2233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2854451921138806e-05,
+      "loss": 0.0249,
+      "step": 2234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2848536079585866e-05,
+      "loss": 0.0165,
+      "step": 2235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2842619152892625e-05,
+      "loss": 0.0202,
+      "step": 2236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2836701143313126e-05,
+      "loss": 0.0297,
+      "step": 2237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2830782053101807e-05,
+      "loss": 0.023,
+      "step": 2238
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2824861884513525e-05,
+      "loss": 0.0267,
+      "step": 2239
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2818940639803555e-05,
+      "loss": 0.0295,
+      "step": 2240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.281301832122757e-05,
+      "loss": 0.0264,
+      "step": 2241
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.280709493104166e-05,
+      "loss": 0.0256,
+      "step": 2242
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2801170471502315e-05,
+      "loss": 0.0208,
+      "step": 2243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.279524494486645e-05,
+      "loss": 0.0226,
+      "step": 2244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2789318353391362e-05,
+      "loss": 0.0259,
+      "step": 2245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2783390699334772e-05,
+      "loss": 0.0186,
+      "step": 2246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2777461984954803e-05,
+      "loss": 0.0229,
+      "step": 2247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2771532212509974e-05,
+      "loss": 0.0261,
+      "step": 2248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2765601384259212e-05,
+      "loss": 0.0307,
+      "step": 2249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.275966950246185e-05,
+      "loss": 0.0241,
+      "step": 2250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2753736569377615e-05,
+      "loss": 0.0218,
+      "step": 2251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2747802587266637e-05,
+      "loss": 0.0189,
+      "step": 2252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.274186755838945e-05,
+      "loss": 0.0234,
+      "step": 2253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2735931485006986e-05,
+      "loss": 0.0205,
+      "step": 2254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2729994369380565e-05,
+      "loss": 0.0334,
+      "step": 2255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2724056213771917e-05,
+      "loss": 0.0207,
+      "step": 2256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.271811702044316e-05,
+      "loss": 0.0332,
+      "step": 2257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2712176791656807e-05,
+      "loss": 0.0191,
+      "step": 2258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2706235529675778e-05,
+      "loss": 0.0308,
+      "step": 2259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2700293236763366e-05,
+      "loss": 0.0294,
+      "step": 2260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2694349915183272e-05,
+      "loss": 0.0216,
+      "step": 2261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2688405567199586e-05,
+      "loss": 0.0202,
+      "step": 2262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2682460195076787e-05,
+      "loss": 0.0264,
+      "step": 2263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2676513801079741e-05,
+      "loss": 0.0235,
+      "step": 2264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2670566387473713e-05,
+      "loss": 0.0187,
+      "step": 2265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2664617956524348e-05,
+      "loss": 0.0252,
+      "step": 2266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2658668510497674e-05,
+      "loss": 0.0176,
+      "step": 2267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2652718051660122e-05,
+      "loss": 0.0317,
+      "step": 2268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2646766582278498e-05,
+      "loss": 0.0254,
+      "step": 2269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2640814104619991e-05,
+      "loss": 0.0255,
+      "step": 2270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2634860620952177e-05,
+      "loss": 0.0261,
+      "step": 2271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2628906133543018e-05,
+      "loss": 0.0275,
+      "step": 2272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2622950644660856e-05,
+      "loss": 0.0313,
+      "step": 2273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2616994156574411e-05,
+      "loss": 0.0198,
+      "step": 2274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2611036671552795e-05,
+      "loss": 0.0193,
+      "step": 2275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2605078191865487e-05,
+      "loss": 0.0287,
+      "step": 2276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2599118719782348e-05,
+      "loss": 0.0238,
+      "step": 2277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.259315825757362e-05,
+      "loss": 0.0203,
+      "step": 2278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2587196807509927e-05,
+      "loss": 0.0218,
+      "step": 2279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2581234371862253e-05,
+      "loss": 0.0214,
+      "step": 2280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2575270952901978e-05,
+      "loss": 0.0278,
+      "step": 2281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2569306552900843e-05,
+      "loss": 0.018,
+      "step": 2282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.256334117413096e-05,
+      "loss": 0.0219,
+      "step": 2283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2557374818864834e-05,
+      "loss": 0.0238,
+      "step": 2284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2551407489375314e-05,
+      "loss": 0.0218,
+      "step": 2285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.254543918793564e-05,
+      "loss": 0.0273,
+      "step": 2286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.253946991681942e-05,
+      "loss": 0.0268,
+      "step": 2287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2533499678300618e-05,
+      "loss": 0.0259,
+      "step": 2288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2527528474653586e-05,
+      "loss": 0.0265,
+      "step": 2289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2521556308153033e-05,
+      "loss": 0.0237,
+      "step": 2290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.251558318107403e-05,
+      "loss": 0.0293,
+      "step": 2291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2509609095692023e-05,
+      "loss": 0.0268,
+      "step": 2292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2503634054282828e-05,
+      "loss": 0.0264,
+      "step": 2293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2497658059122602e-05,
+      "loss": 0.0256,
+      "step": 2294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2491681112487885e-05,
+      "loss": 0.045,
+      "step": 2295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2485703216655584e-05,
+      "loss": 0.0207,
+      "step": 2296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2479724373902951e-05,
+      "loss": 0.0213,
+      "step": 2297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2473744586507606e-05,
+      "loss": 0.0285,
+      "step": 2298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2467763856747527e-05,
+      "loss": 0.0286,
+      "step": 2299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2461782186901061e-05,
+      "loss": 0.0222,
+      "step": 2300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.24557995792469e-05,
+      "loss": 0.018,
+      "step": 2301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2449816036064096e-05,
+      "loss": 0.0245,
+      "step": 2302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2443831559632065e-05,
+      "loss": 0.025,
+      "step": 2303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2437846152230571e-05,
+      "loss": 0.0209,
+      "step": 2304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2431859816139731e-05,
+      "loss": 0.021,
+      "step": 2305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2425872553640025e-05,
+      "loss": 0.0265,
+      "step": 2306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2419884367012282e-05,
+      "loss": 0.02,
+      "step": 2307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2413895258537676e-05,
+      "loss": 0.0234,
+      "step": 2308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.240790523049774e-05,
+      "loss": 0.024,
+      "step": 2309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2401914285174357e-05,
+      "loss": 0.0268,
+      "step": 2310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2395922424849756e-05,
+      "loss": 0.0265,
+      "step": 2311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2389929651806516e-05,
+      "loss": 0.0194,
+      "step": 2312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2383935968327566e-05,
+      "loss": 0.0268,
+      "step": 2313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2377941376696176e-05,
+      "loss": 0.0187,
+      "step": 2314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2371945879195973e-05,
+      "loss": 0.0199,
+      "step": 2315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2365949478110909e-05,
+      "loss": 0.0241,
+      "step": 2316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2359952175725303e-05,
+      "loss": 0.0201,
+      "step": 2317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2353953974323807e-05,
+      "loss": 0.0207,
+      "step": 2318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.234795487619141e-05,
+      "loss": 0.0281,
+      "step": 2319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2341954883613451e-05,
+      "loss": 0.0142,
+      "step": 2320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2335953998875614e-05,
+      "loss": 0.0199,
+      "step": 2321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2329952224263905e-05,
+      "loss": 0.0217,
+      "step": 2322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2323949562064685e-05,
+      "loss": 0.0325,
+      "step": 2323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.231794601456465e-05,
+      "loss": 0.0271,
+      "step": 2324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2311941584050829e-05,
+      "loss": 0.0276,
+      "step": 2325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2305936272810586e-05,
+      "loss": 0.0233,
+      "step": 2326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2299930083131635e-05,
+      "loss": 0.0286,
+      "step": 2327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2293923017302004e-05,
+      "loss": 0.0271,
+      "step": 2328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2287915077610067e-05,
+      "loss": 0.0277,
+      "step": 2329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.228190626634453e-05,
+      "loss": 0.0197,
+      "step": 2330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2275896585794428e-05,
+      "loss": 0.0288,
+      "step": 2331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2269886038249132e-05,
+      "loss": 0.0218,
+      "step": 2332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2263874625998336e-05,
+      "loss": 0.0202,
+      "step": 2333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2257862351332068e-05,
+      "loss": 0.0325,
+      "step": 2334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2251849216540687e-05,
+      "loss": 0.0254,
+      "step": 2335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2245835223914872e-05,
+      "loss": 0.0246,
+      "step": 2336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2239820375745639e-05,
+      "loss": 0.0284,
+      "step": 2337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.223380467432432e-05,
+      "loss": 0.0298,
+      "step": 2338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2227788121942579e-05,
+      "loss": 0.0212,
+      "step": 2339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.22217707208924e-05,
+      "loss": 0.0275,
+      "step": 2340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2215752473466089e-05,
+      "loss": 0.0206,
+      "step": 2341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2209733381956282e-05,
+      "loss": 0.0281,
+      "step": 2342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2203713448655932e-05,
+      "loss": 0.0224,
+      "step": 2343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2197692675858309e-05,
+      "loss": 0.024,
+      "step": 2344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2191671065857004e-05,
+      "loss": 0.0194,
+      "step": 2345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.218564862094594e-05,
+      "loss": 0.0213,
+      "step": 2346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2179625343419335e-05,
+      "loss": 0.0263,
+      "step": 2347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.217360123557174e-05,
+      "loss": 0.0269,
+      "step": 2348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2167576299698027e-05,
+      "loss": 0.0254,
+      "step": 2349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2161550538093362e-05,
+      "loss": 0.0206,
+      "step": 2350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2155523953053248e-05,
+      "loss": 0.0241,
+      "step": 2351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2149496546873492e-05,
+      "loss": 0.0249,
+      "step": 2352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.214346832185021e-05,
+      "loss": 0.021,
+      "step": 2353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2137439280279833e-05,
+      "loss": 0.0263,
+      "step": 2354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2131409424459113e-05,
+      "loss": 0.0194,
+      "step": 2355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2125378756685093e-05,
+      "loss": 0.0287,
+      "step": 2356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2119347279255144e-05,
+      "loss": 0.0256,
+      "step": 2357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.211331499446693e-05,
+      "loss": 0.0243,
+      "step": 2358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2107281904618438e-05,
+      "loss": 0.0236,
+      "step": 2359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2101248012007944e-05,
+      "loss": 0.0234,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2095213318934049e-05,
+      "loss": 0.0226,
+      "step": 2361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2089177827695639e-05,
+      "loss": 0.0234,
+      "step": 2362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2083141540591923e-05,
+      "loss": 0.0196,
+      "step": 2363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.20771044599224e-05,
+      "loss": 0.0224,
+      "step": 2364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2071066587986874e-05,
+      "loss": 0.0168,
+      "step": 2365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2065027927085458e-05,
+      "loss": 0.0232,
+      "step": 2366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2058988479518556e-05,
+      "loss": 0.0167,
+      "step": 2367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2052948247586872e-05,
+      "loss": 0.0281,
+      "step": 2368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2046907233591421e-05,
+      "loss": 0.0311,
+      "step": 2369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2040865439833496e-05,
+      "loss": 0.0251,
+      "step": 2370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.203482286861471e-05,
+      "loss": 0.0201,
+      "step": 2371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.202877952223695e-05,
+      "loss": 0.0199,
+      "step": 2372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2022735403002415e-05,
+      "loss": 0.03,
+      "step": 2373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.201669051321359e-05,
+      "loss": 0.0236,
+      "step": 2374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2010644855173254e-05,
+      "loss": 0.032,
+      "step": 2375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2004598431184481e-05,
+      "loss": 0.0246,
+      "step": 2376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1998551243550643e-05,
+      "loss": 0.0237,
+      "step": 2377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1992503294575385e-05,
+      "loss": 0.0288,
+      "step": 2378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1986454586562654e-05,
+      "loss": 0.0183,
+      "step": 2379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1980405121816696e-05,
+      "loss": 0.025,
+      "step": 2380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1974354902642023e-05,
+      "loss": 0.0224,
+      "step": 2381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1968303931343448e-05,
+      "loss": 0.0197,
+      "step": 2382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1962252210226076e-05,
+      "loss": 0.0219,
+      "step": 2383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1956199741595278e-05,
+      "loss": 0.0237,
+      "step": 2384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1950146527756733e-05,
+      "loss": 0.0289,
+      "step": 2385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1944092571016383e-05,
+      "loss": 0.0218,
+      "step": 2386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1938037873680468e-05,
+      "loss": 0.0252,
+      "step": 2387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1931982438055506e-05,
+      "loss": 0.0212,
+      "step": 2388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.192592626644829e-05,
+      "loss": 0.0194,
+      "step": 2389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1919869361165898e-05,
+      "loss": 0.0254,
+      "step": 2390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1913811724515696e-05,
+      "loss": 0.0238,
+      "step": 2391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1907753358805314e-05,
+      "loss": 0.022,
+      "step": 2392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1901694266342666e-05,
+      "loss": 0.0236,
+      "step": 2393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1895634449435946e-05,
+      "loss": 0.0282,
+      "step": 2394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1889573910393619e-05,
+      "loss": 0.0196,
+      "step": 2395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1883512651524424e-05,
+      "loss": 0.0289,
+      "step": 2396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1877450675137385e-05,
+      "loss": 0.0206,
+      "step": 2397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1871387983541789e-05,
+      "loss": 0.0232,
+      "step": 2398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1865324579047193e-05,
+      "loss": 0.026,
+      "step": 2399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1859260463963434e-05,
+      "loss": 0.0176,
+      "step": 2400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1853195640600616e-05,
+      "loss": 0.0221,
+      "step": 2401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1847130111269117e-05,
+      "loss": 0.0184,
+      "step": 2402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1841063878279572e-05,
+      "loss": 0.0227,
+      "step": 2403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1834996943942898e-05,
+      "loss": 0.0261,
+      "step": 2404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1828929310570279e-05,
+      "loss": 0.0193,
+      "step": 2405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1822860980473146e-05,
+      "loss": 0.0214,
+      "step": 2406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1816791955963216e-05,
+      "loss": 0.0266,
+      "step": 2407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1810722239352468e-05,
+      "loss": 0.0182,
+      "step": 2408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1804651832953132e-05,
+      "loss": 0.0262,
+      "step": 2409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1798580739077713e-05,
+      "loss": 0.0242,
+      "step": 2410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1792508960038977e-05,
+      "loss": 0.0168,
+      "step": 2411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1786436498149942e-05,
+      "loss": 0.02,
+      "step": 2412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1780363355723899e-05,
+      "loss": 0.0198,
+      "step": 2413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1774289535074384e-05,
+      "loss": 0.0262,
+      "step": 2414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1768215038515201e-05,
+      "loss": 0.0172,
+      "step": 2415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1762139868360412e-05,
+      "loss": 0.0217,
+      "step": 2416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.175606402692433e-05,
+      "loss": 0.0265,
+      "step": 2417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1749987516521523e-05,
+      "loss": 0.0255,
+      "step": 2418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.174391033946683e-05,
+      "loss": 0.0151,
+      "step": 2419
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1737832498075316e-05,
+      "loss": 0.0221,
+      "step": 2420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1731753994662323e-05,
+      "loss": 0.0231,
+      "step": 2421
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1725674831543436e-05,
+      "loss": 0.0236,
+      "step": 2422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.171959501103449e-05,
+      "loss": 0.0225,
+      "step": 2423
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1713514535451572e-05,
+      "loss": 0.0233,
+      "step": 2424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.170743340711102e-05,
+      "loss": 0.0239,
+      "step": 2425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.170135162832942e-05,
+      "loss": 0.0191,
+      "step": 2426
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1695269201423606e-05,
+      "loss": 0.0231,
+      "step": 2427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1689186128710654e-05,
+      "loss": 0.0239,
+      "step": 2428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1683102412507897e-05,
+      "loss": 0.0241,
+      "step": 2429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1677018055132898e-05,
+      "loss": 0.0279,
+      "step": 2430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1670933058903478e-05,
+      "loss": 0.0232,
+      "step": 2431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1664847426137696e-05,
+      "loss": 0.0258,
+      "step": 2432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1658761159153851e-05,
+      "loss": 0.0246,
+      "step": 2433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1652674260270487e-05,
+      "loss": 0.0178,
+      "step": 2434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1646586731806391e-05,
+      "loss": 0.0313,
+      "step": 2435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1640498576080583e-05,
+      "loss": 0.0196,
+      "step": 2436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1634409795412323e-05,
+      "loss": 0.024,
+      "step": 2437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1628320392121118e-05,
+      "loss": 0.0155,
+      "step": 2438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1622230368526704e-05,
+      "loss": 0.0236,
+      "step": 2439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1616139726949051e-05,
+      "loss": 0.026,
+      "step": 2440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1610048469708372e-05,
+      "loss": 0.0266,
+      "step": 2441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1603956599125115e-05,
+      "loss": 0.0151,
+      "step": 2442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1597864117519947e-05,
+      "loss": 0.0218,
+      "step": 2443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.159177102721379e-05,
+      "loss": 0.0285,
+      "step": 2444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1585677330527777e-05,
+      "loss": 0.0246,
+      "step": 2445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1579583029783288e-05,
+      "loss": 0.0232,
+      "step": 2446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1573488127301924e-05,
+      "loss": 0.0186,
+      "step": 2447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.156739262540552e-05,
+      "loss": 0.0251,
+      "step": 2448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.156129652641613e-05,
+      "loss": 0.0229,
+      "step": 2449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1555199832656052e-05,
+      "loss": 0.0232,
+      "step": 2450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1549102546447795e-05,
+      "loss": 0.0298,
+      "step": 2451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.15430046701141e-05,
+      "loss": 0.0225,
+      "step": 2452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1536906205977936e-05,
+      "loss": 0.0189,
+      "step": 2453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1530807156362491e-05,
+      "loss": 0.0196,
+      "step": 2454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1524707523591174e-05,
+      "loss": 0.0257,
+      "step": 2455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1518607309987624e-05,
+      "loss": 0.0219,
+      "step": 2456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1512506517875697e-05,
+      "loss": 0.0216,
+      "step": 2457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.150640514957947e-05,
+      "loss": 0.0257,
+      "step": 2458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1500303207423231e-05,
+      "loss": 0.0228,
+      "step": 2459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1494200693731506e-05,
+      "loss": 0.0177,
+      "step": 2460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1488097610829019e-05,
+      "loss": 0.0236,
+      "step": 2461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1481993961040721e-05,
+      "loss": 0.0217,
+      "step": 2462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1475889746691779e-05,
+      "loss": 0.0225,
+      "step": 2463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.146978497010757e-05,
+      "loss": 0.0213,
+      "step": 2464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1463679633613692e-05,
+      "loss": 0.0128,
+      "step": 2465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.145757373953595e-05,
+      "loss": 0.0174,
+      "step": 2466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1451467290200364e-05,
+      "loss": 0.0225,
+      "step": 2467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1445360287933165e-05,
+      "loss": 0.0205,
+      "step": 2468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1439252735060797e-05,
+      "loss": 0.0253,
+      "step": 2469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1433144633909911e-05,
+      "loss": 0.0185,
+      "step": 2470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1427035986807366e-05,
+      "loss": 0.0149,
+      "step": 2471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1420926796080236e-05,
+      "loss": 0.0157,
+      "step": 2472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1414817064055792e-05,
+      "loss": 0.024,
+      "step": 2473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1408706793061513e-05,
+      "loss": 0.0139,
+      "step": 2474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1402595985425093e-05,
+      "loss": 0.0205,
+      "step": 2475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1396484643474421e-05,
+      "loss": 0.0206,
+      "step": 2476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1390372769537592e-05,
+      "loss": 0.0234,
+      "step": 2477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1384260365942905e-05,
+      "loss": 0.0243,
+      "step": 2478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1378147435018858e-05,
+      "loss": 0.0213,
+      "step": 2479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1372033979094153e-05,
+      "loss": 0.0203,
+      "step": 2480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.136592000049769e-05,
+      "loss": 0.0256,
+      "step": 2481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.135980550155857e-05,
+      "loss": 0.0197,
+      "step": 2482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1353690484606091e-05,
+      "loss": 0.0215,
+      "step": 2483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1347574951969742e-05,
+      "loss": 0.0189,
+      "step": 2484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1341458905979224e-05,
+      "loss": 0.0228,
+      "step": 2485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.133534234896442e-05,
+      "loss": 0.0318,
+      "step": 2486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.132922528325541e-05,
+      "loss": 0.0364,
+      "step": 2487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1323107711182474e-05,
+      "loss": 0.03,
+      "step": 2488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1316989635076076e-05,
+      "loss": 0.021,
+      "step": 2489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1310871057266879e-05,
+      "loss": 0.0273,
+      "step": 2490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1304751980085735e-05,
+      "loss": 0.0192,
+      "step": 2491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1298632405863688e-05,
+      "loss": 0.025,
+      "step": 2492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1292512336931965e-05,
+      "loss": 0.0166,
+      "step": 2493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.128639177562199e-05,
+      "loss": 0.0249,
+      "step": 2494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1280270724265368e-05,
+      "loss": 0.0216,
+      "step": 2495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.127414918519389e-05,
+      "loss": 0.0294,
+      "step": 2496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.126802716073954e-05,
+      "loss": 0.0204,
+      "step": 2497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1261904653234486e-05,
+      "loss": 0.0224,
+      "step": 2498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1255781665011069e-05,
+      "loss": 0.0219,
+      "step": 2499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1249658198401827e-05,
+      "loss": 0.0215,
+      "step": 2500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.124353425573947e-05,
+      "loss": 0.0249,
+      "step": 2501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1237409839356896e-05,
+      "loss": 0.0294,
+      "step": 2502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.123128495158718e-05,
+      "loss": 0.0219,
+      "step": 2503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.122515959476358e-05,
+      "loss": 0.0188,
+      "step": 2504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1219033771219523e-05,
+      "loss": 0.0223,
+      "step": 2505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1212907483288629e-05,
+      "loss": 0.0204,
+      "step": 2506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1206780733304686e-05,
+      "loss": 0.0208,
+      "step": 2507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1200653523601652e-05,
+      "loss": 0.0251,
+      "step": 2508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1194525856513675e-05,
+      "loss": 0.0244,
+      "step": 2509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1188397734375064e-05,
+      "loss": 0.0203,
+      "step": 2510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1182269159520309e-05,
+      "loss": 0.0253,
+      "step": 2511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1176140134284073e-05,
+      "loss": 0.0246,
+      "step": 2512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1170010661001184e-05,
+      "loss": 0.0283,
+      "step": 2513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.116388074200664e-05,
+      "loss": 0.0212,
+      "step": 2514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1157750379635622e-05,
+      "loss": 0.0205,
+      "step": 2515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1151619576223467e-05,
+      "loss": 0.0215,
+      "step": 2516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1145488334105682e-05,
+      "loss": 0.0169,
+      "step": 2517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1139356655617945e-05,
+      "loss": 0.0208,
+      "step": 2518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1133224543096102e-05,
+      "loss": 0.026,
+      "step": 2519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1127091998876155e-05,
+      "loss": 0.0209,
+      "step": 2520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.112095902529428e-05,
+      "loss": 0.0197,
+      "step": 2521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1114825624686811e-05,
+      "loss": 0.0202,
+      "step": 2522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.110869179939025e-05,
+      "loss": 0.0309,
+      "step": 2523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1102557551741248e-05,
+      "loss": 0.026,
+      "step": 2524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1096422884076638e-05,
+      "loss": 0.0173,
+      "step": 2525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1090287798733395e-05,
+      "loss": 0.0202,
+      "step": 2526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1084152298048656e-05,
+      "loss": 0.0233,
+      "step": 2527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1078016384359725e-05,
+      "loss": 0.0209,
+      "step": 2528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1071880060004056e-05,
+      "loss": 0.0203,
+      "step": 2529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1065743327319261e-05,
+      "loss": 0.0271,
+      "step": 2530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.105960618864311e-05,
+      "loss": 0.0174,
+      "step": 2531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1053468646313523e-05,
+      "loss": 0.015,
+      "step": 2532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1047330702668576e-05,
+      "loss": 0.0243,
+      "step": 2533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1041192360046498e-05,
+      "loss": 0.0246,
+      "step": 2534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1035053620785673e-05,
+      "loss": 0.0183,
+      "step": 2535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1028914487224629e-05,
+      "loss": 0.0225,
+      "step": 2536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1022774961702051e-05,
+      "loss": 0.0197,
+      "step": 2537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1016635046556773e-05,
+      "loss": 0.02,
+      "step": 2538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.101049474412777e-05,
+      "loss": 0.0238,
+      "step": 2539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1004354056754172e-05,
+      "loss": 0.0336,
+      "step": 2540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0998212986775256e-05,
+      "loss": 0.0194,
+      "step": 2541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0992071536530437e-05,
+      "loss": 0.0246,
+      "step": 2542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0985929708359282e-05,
+      "loss": 0.0209,
+      "step": 2543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0979787504601502e-05,
+      "loss": 0.021,
+      "step": 2544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.097364492759695e-05,
+      "loss": 0.0273,
+      "step": 2545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0967501979685612e-05,
+      "loss": 0.024,
+      "step": 2546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0961358663207633e-05,
+      "loss": 0.0156,
+      "step": 2547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0955214980503284e-05,
+      "loss": 0.0248,
+      "step": 2548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0949070933912979e-05,
+      "loss": 0.0234,
+      "step": 2549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0942926525777278e-05,
+      "loss": 0.0215,
+      "step": 2550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.093678175843687e-05,
+      "loss": 0.0181,
+      "step": 2551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0930636634232577e-05,
+      "loss": 0.0266,
+      "step": 2552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0924491155505375e-05,
+      "loss": 0.0242,
+      "step": 2553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0918345324596356e-05,
+      "loss": 0.0211,
+      "step": 2554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0912199143846751e-05,
+      "loss": 0.0185,
+      "step": 2555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0906052615597938e-05,
+      "loss": 0.0205,
+      "step": 2556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0899905742191406e-05,
+      "loss": 0.0238,
+      "step": 2557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0893758525968789e-05,
+      "loss": 0.0154,
+      "step": 2558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0887610969271853e-05,
+      "loss": 0.0173,
+      "step": 2559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0881463074442485e-05,
+      "loss": 0.0249,
+      "step": 2560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0875314843822702e-05,
+      "loss": 0.0221,
+      "step": 2561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0869166279754657e-05,
+      "loss": 0.0207,
+      "step": 2562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0863017384580623e-05,
+      "loss": 0.0266,
+      "step": 2563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0856868160643002e-05,
+      "loss": 0.0222,
+      "step": 2564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0850718610284318e-05,
+      "loss": 0.0234,
+      "step": 2565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0844568735847227e-05,
+      "loss": 0.0286,
+      "step": 2566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0838418539674496e-05,
+      "loss": 0.0307,
+      "step": 2567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0832268024109025e-05,
+      "loss": 0.0203,
+      "step": 2568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0826117191493835e-05,
+      "loss": 0.0231,
+      "step": 2569
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0819966044172062e-05,
+      "loss": 0.0221,
+      "step": 2570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0813814584486962e-05,
+      "loss": 0.0239,
+      "step": 2571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0807662814781921e-05,
+      "loss": 0.0181,
+      "step": 2572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0801510737400426e-05,
+      "loss": 0.0307,
+      "step": 2573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0795358354686099e-05,
+      "loss": 0.0184,
+      "step": 2574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0789205668982663e-05,
+      "loss": 0.0271,
+      "step": 2575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0783052682633968e-05,
+      "loss": 0.0207,
+      "step": 2576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0776899397983969e-05,
+      "loss": 0.0243,
+      "step": 2577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0770745817376741e-05,
+      "loss": 0.0213,
+      "step": 2578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0764591943156473e-05,
+      "loss": 0.0272,
+      "step": 2579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0758437777667457e-05,
+      "loss": 0.0216,
+      "step": 2580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0752283323254107e-05,
+      "loss": 0.0186,
+      "step": 2581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.074612858226094e-05,
+      "loss": 0.0187,
+      "step": 2582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0739973557032581e-05,
+      "loss": 0.031,
+      "step": 2583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0733818249913774e-05,
+      "loss": 0.02,
+      "step": 2584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0727662663249354e-05,
+      "loss": 0.0255,
+      "step": 2585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0721506799384276e-05,
+      "loss": 0.0218,
+      "step": 2586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0715350660663595e-05,
+      "loss": 0.025,
+      "step": 2587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0709194249432471e-05,
+      "loss": 0.0272,
+      "step": 2588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0703037568036168e-05,
+      "loss": 0.0175,
+      "step": 2589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0696880618820053e-05,
+      "loss": 0.0194,
+      "step": 2590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0690723404129597e-05,
+      "loss": 0.0276,
+      "step": 2591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0684565926310368e-05,
+      "loss": 0.0215,
+      "step": 2592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.067840818770804e-05,
+      "loss": 0.0195,
+      "step": 2593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0672250190668381e-05,
+      "loss": 0.019,
+      "step": 2594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0666091937537255e-05,
+      "loss": 0.0204,
+      "step": 2595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0659933430660634e-05,
+      "loss": 0.0231,
+      "step": 2596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0653774672384581e-05,
+      "loss": 0.0216,
+      "step": 2597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0647615665055251e-05,
+      "loss": 0.021,
+      "step": 2598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0641456411018895e-05,
+      "loss": 0.0161,
+      "step": 2599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0635296912621866e-05,
+      "loss": 0.0217,
+      "step": 2600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0629137172210605e-05,
+      "loss": 0.0199,
+      "step": 2601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0622977192131637e-05,
+      "loss": 0.0206,
+      "step": 2602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.061681697473159e-05,
+      "loss": 0.0222,
+      "step": 2603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0610656522357181e-05,
+      "loss": 0.024,
+      "step": 2604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0604495837355209e-05,
+      "loss": 0.0268,
+      "step": 2605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0598334922072571e-05,
+      "loss": 0.0222,
+      "step": 2606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0592173778856242e-05,
+      "loss": 0.029,
+      "step": 2607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0586012410053293e-05,
+      "loss": 0.0257,
+      "step": 2608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0579850818010874e-05,
+      "loss": 0.0294,
+      "step": 2609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0573689005076226e-05,
+      "loss": 0.0189,
+      "step": 2610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0567526973596667e-05,
+      "loss": 0.025,
+      "step": 2611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.056136472591961e-05,
+      "loss": 0.0189,
+      "step": 2612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.055520226439253e-05,
+      "loss": 0.0233,
+      "step": 2613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0549039591363006e-05,
+      "loss": 0.0241,
+      "step": 2614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0542876709178687e-05,
+      "loss": 0.016,
+      "step": 2615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0536713620187298e-05,
+      "loss": 0.0181,
+      "step": 2616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0530550326736651e-05,
+      "loss": 0.0267,
+      "step": 2617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0524386831174629e-05,
+      "loss": 0.02,
+      "step": 2618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0518223135849199e-05,
+      "loss": 0.0219,
+      "step": 2619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0512059243108395e-05,
+      "loss": 0.0187,
+      "step": 2620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0505895155300337e-05,
+      "loss": 0.0185,
+      "step": 2621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0499730874773208e-05,
+      "loss": 0.0178,
+      "step": 2622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0493566403875273e-05,
+      "loss": 0.0264,
+      "step": 2623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0487401744954869e-05,
+      "loss": 0.0195,
+      "step": 2624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0481236900360399e-05,
+      "loss": 0.0148,
+      "step": 2625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0475071872440339e-05,
+      "loss": 0.0139,
+      "step": 2626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0468906663543236e-05,
+      "loss": 0.0216,
+      "step": 2627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0462741276017711e-05,
+      "loss": 0.0155,
+      "step": 2628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0456575712212442e-05,
+      "loss": 0.014,
+      "step": 2629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0450409974476181e-05,
+      "loss": 0.0161,
+      "step": 2630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0444244065157746e-05,
+      "loss": 0.0199,
+      "step": 2631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0438077986606021e-05,
+      "loss": 0.0205,
+      "step": 2632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0431911741169949e-05,
+      "loss": 0.0239,
+      "step": 2633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0425745331198545e-05,
+      "loss": 0.0223,
+      "step": 2634
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0419578759040879e-05,
+      "loss": 0.0216,
+      "step": 2635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0413412027046086e-05,
+      "loss": 0.0153,
+      "step": 2636
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0407245137563367e-05,
+      "loss": 0.0213,
+      "step": 2637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0401078092941972e-05,
+      "loss": 0.0165,
+      "step": 2638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0394910895531216e-05,
+      "loss": 0.0156,
+      "step": 2639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.038874354768048e-05,
+      "loss": 0.0187,
+      "step": 2640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0382576051739185e-05,
+      "loss": 0.0283,
+      "step": 2641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0376408410056822e-05,
+      "loss": 0.0183,
+      "step": 2642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0370240624982937e-05,
+      "loss": 0.0185,
+      "step": 2643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.036407269886712e-05,
+      "loss": 0.022,
+      "step": 2644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0357904634059024e-05,
+      "loss": 0.0195,
+      "step": 2645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.035173643290836e-05,
+      "loss": 0.0222,
+      "step": 2646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0345568097764874e-05,
+      "loss": 0.0236,
+      "step": 2647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0339399630978373e-05,
+      "loss": 0.0227,
+      "step": 2648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0333231034898722e-05,
+      "loss": 0.0252,
+      "step": 2649
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0327062311875821e-05,
+      "loss": 0.0199,
+      "step": 2650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0320893464259625e-05,
+      "loss": 0.0129,
+      "step": 2651
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0314724494400136e-05,
+      "loss": 0.0213,
+      "step": 2652
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0308555404647407e-05,
+      "loss": 0.019,
+      "step": 2653
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0302386197351524e-05,
+      "loss": 0.0288,
+      "step": 2654
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.029621687486263e-05,
+      "loss": 0.0258,
+      "step": 2655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.029004743953091e-05,
+      "loss": 0.0141,
+      "step": 2656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0283877893706586e-05,
+      "loss": 0.017,
+      "step": 2657
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0277708239739925e-05,
+      "loss": 0.0173,
+      "step": 2658
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0271538479981242e-05,
+      "loss": 0.0178,
+      "step": 2659
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0265368616780884e-05,
+      "loss": 0.019,
+      "step": 2660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0259198652489233e-05,
+      "loss": 0.0216,
+      "step": 2661
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0253028589456726e-05,
+      "loss": 0.0175,
+      "step": 2662
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0246858430033823e-05,
+      "loss": 0.0212,
+      "step": 2663
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0240688176571025e-05,
+      "loss": 0.0169,
+      "step": 2664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0234517831418874e-05,
+      "loss": 0.0189,
+      "step": 2665
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0228347396927935e-05,
+      "loss": 0.012,
+      "step": 2666
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.022217687544882e-05,
+      "loss": 0.0262,
+      "step": 2667
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.021600626933217e-05,
+      "loss": 0.0236,
+      "step": 2668
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.020983558092865e-05,
+      "loss": 0.0269,
+      "step": 2669
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0203664812588967e-05,
+      "loss": 0.0192,
+      "step": 2670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0197493966663858e-05,
+      "loss": 0.0214,
+      "step": 2671
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0191323045504078e-05,
+      "loss": 0.016,
+      "step": 2672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0185152051460425e-05,
+      "loss": 0.0237,
+      "step": 2673
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.017898098688372e-05,
+      "loss": 0.0216,
+      "step": 2674
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0172809854124802e-05,
+      "loss": 0.0182,
+      "step": 2675
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0166638655534548e-05,
+      "loss": 0.02,
+      "step": 2676
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.016046739346386e-05,
+      "loss": 0.0184,
+      "step": 2677
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0154296070263649e-05,
+      "loss": 0.0155,
+      "step": 2678
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0148124688284869e-05,
+      "loss": 0.0191,
+      "step": 2679
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0141953249878484e-05,
+      "loss": 0.0239,
+      "step": 2680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0135781757395482e-05,
+      "loss": 0.0221,
+      "step": 2681
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0129610213186876e-05,
+      "loss": 0.0186,
+      "step": 2682
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0123438619603689e-05,
+      "loss": 0.021,
+      "step": 2683
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0117266978996973e-05,
+      "loss": 0.022,
+      "step": 2684
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0111095293717797e-05,
+      "loss": 0.0145,
+      "step": 2685
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0104923566117235e-05,
+      "loss": 0.0238,
+      "step": 2686
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0098751798546393e-05,
+      "loss": 0.0299,
+      "step": 2687
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0092579993356386e-05,
+      "loss": 0.0174,
+      "step": 2688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0086408152898336e-05,
+      "loss": 0.0165,
+      "step": 2689
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0080236279523392e-05,
+      "loss": 0.0151,
+      "step": 2690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0074064375582707e-05,
+      "loss": 0.0178,
+      "step": 2691
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.006789244342744e-05,
+      "loss": 0.0135,
+      "step": 2692
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0061720485408778e-05,
+      "loss": 0.0167,
+      "step": 2693
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0055548503877905e-05,
+      "loss": 0.0187,
+      "step": 2694
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.004937650118601e-05,
+      "loss": 0.0188,
+      "step": 2695
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0043204479684307e-05,
+      "loss": 0.0176,
+      "step": 2696
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0037032441723999e-05,
+      "loss": 0.015,
+      "step": 2697
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0030860389656306e-05,
+      "loss": 0.0212,
+      "step": 2698
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0024688325832454e-05,
+      "loss": 0.0233,
+      "step": 2699
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0018516252603663e-05,
+      "loss": 0.0256,
+      "step": 2700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0012344172321166e-05,
+      "loss": 0.0177,
+      "step": 2701
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0006172087336203e-05,
+      "loss": 0.021,
+      "step": 2702
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1e-05,
+      "loss": 0.0201,
+      "step": 2703
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.993827912663799e-06,
+      "loss": 0.0218,
+      "step": 2704
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.987655827678833e-06,
+      "loss": 0.0215,
+      "step": 2705
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.981483747396342e-06,
+      "loss": 0.0243,
+      "step": 2706
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.97531167416755e-06,
+      "loss": 0.0259,
+      "step": 2707
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.969139610343696e-06,
+      "loss": 0.017,
+      "step": 2708
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.962967558276003e-06,
+      "loss": 0.0236,
+      "step": 2709
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.956795520315694e-06,
+      "loss": 0.0139,
+      "step": 2710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.95062349881399e-06,
+      "loss": 0.0204,
+      "step": 2711
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.9444514961221e-06,
+      "loss": 0.0233,
+      "step": 2712
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.938279514591223e-06,
+      "loss": 0.0235,
+      "step": 2713
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.932107556572561e-06,
+      "loss": 0.0129,
+      "step": 2714
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.925935624417296e-06,
+      "loss": 0.0159,
+      "step": 2715
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.919763720476608e-06,
+      "loss": 0.018,
+      "step": 2716
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.913591847101663e-06,
+      "loss": 0.0212,
+      "step": 2717
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.907420006643619e-06,
+      "loss": 0.0188,
+      "step": 2718
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.901248201453609e-06,
+      "loss": 0.0221,
+      "step": 2719
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.895076433882768e-06,
+      "loss": 0.0148,
+      "step": 2720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.888904706282208e-06,
+      "loss": 0.0232,
+      "step": 2721
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.882733021003027e-06,
+      "loss": 0.0232,
+      "step": 2722
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.876561380396316e-06,
+      "loss": 0.0148,
+      "step": 2723
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.870389786813131e-06,
+      "loss": 0.0151,
+      "step": 2724
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.864218242604521e-06,
+      "loss": 0.0176,
+      "step": 2725
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.858046750121518e-06,
+      "loss": 0.0297,
+      "step": 2726
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.851875311715133e-06,
+      "loss": 0.0153,
+      "step": 2727
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.845703929736351e-06,
+      "loss": 0.0175,
+      "step": 2728
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.839532606536145e-06,
+      "loss": 0.0147,
+      "step": 2729
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.833361344465455e-06,
+      "loss": 0.0285,
+      "step": 2730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.8271901458752e-06,
+      "loss": 0.0215,
+      "step": 2731
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.821019013116283e-06,
+      "loss": 0.0229,
+      "step": 2732
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.814847948539576e-06,
+      "loss": 0.022,
+      "step": 2733
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.808676954495927e-06,
+      "loss": 0.0267,
+      "step": 2734
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.802506033336147e-06,
+      "loss": 0.0183,
+      "step": 2735
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.796335187411036e-06,
+      "loss": 0.0172,
+      "step": 2736
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.790164419071353e-06,
+      "loss": 0.0197,
+      "step": 2737
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.783993730667833e-06,
+      "loss": 0.0169,
+      "step": 2738
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.77782312455118e-06,
+      "loss": 0.0186,
+      "step": 2739
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.771652603072068e-06,
+      "loss": 0.0174,
+      "step": 2740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.76548216858113e-06,
+      "loss": 0.0183,
+      "step": 2741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.759311823428978e-06,
+      "loss": 0.0199,
+      "step": 2742
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.753141569966178e-06,
+      "loss": 0.0204,
+      "step": 2743
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.746971410543274e-06,
+      "loss": 0.019,
+      "step": 2744
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.740801347510767e-06,
+      "loss": 0.02,
+      "step": 2745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.734631383219121e-06,
+      "loss": 0.0189,
+      "step": 2746
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.728461520018761e-06,
+      "loss": 0.0208,
+      "step": 2747
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.722291760260077e-06,
+      "loss": 0.017,
+      "step": 2748
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.716122106293416e-06,
+      "loss": 0.0272,
+      "step": 2749
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.709952560469091e-06,
+      "loss": 0.0272,
+      "step": 2750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.703783125137373e-06,
+      "loss": 0.0141,
+      "step": 2751
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.69761380264848e-06,
+      "loss": 0.0189,
+      "step": 2752
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.691444595352596e-06,
+      "loss": 0.0201,
+      "step": 2753
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.685275505599865e-06,
+      "loss": 0.017,
+      "step": 2754
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.679106535740377e-06,
+      "loss": 0.0249,
+      "step": 2755
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.672937688124179e-06,
+      "loss": 0.0208,
+      "step": 2756
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.666768965101281e-06,
+      "loss": 0.0229,
+      "step": 2757
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.66060036902163e-06,
+      "loss": 0.0188,
+      "step": 2758
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.654431902235131e-06,
+      "loss": 0.024,
+      "step": 2759
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.648263567091644e-06,
+      "loss": 0.0118,
+      "step": 2760
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.642095365940976e-06,
+      "loss": 0.0244,
+      "step": 2761
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.635927301132885e-06,
+      "loss": 0.0166,
+      "step": 2762
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.629759375017068e-06,
+      "loss": 0.0197,
+      "step": 2763
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.623591589943182e-06,
+      "loss": 0.0152,
+      "step": 2764
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.617423948260818e-06,
+      "loss": 0.0207,
+      "step": 2765
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.611256452319521e-06,
+      "loss": 0.0175,
+      "step": 2766
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.605089104468784e-06,
+      "loss": 0.0193,
+      "step": 2767
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.598921907058033e-06,
+      "loss": 0.0199,
+      "step": 2768
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.592754862436638e-06,
+      "loss": 0.0186,
+      "step": 2769
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.586587972953917e-06,
+      "loss": 0.0184,
+      "step": 2770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.580421240959123e-06,
+      "loss": 0.0164,
+      "step": 2771
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.574254668801457e-06,
+      "loss": 0.0192,
+      "step": 2772
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.568088258830053e-06,
+      "loss": 0.0174,
+      "step": 2773
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.561922013393986e-06,
+      "loss": 0.0166,
+      "step": 2774
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.555755934842257e-06,
+      "loss": 0.0186,
+      "step": 2775
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.549590025523824e-06,
+      "loss": 0.0147,
+      "step": 2776
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.54342428778756e-06,
+      "loss": 0.0171,
+      "step": 2777
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.53725872398229e-06,
+      "loss": 0.0196,
+      "step": 2778
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.531093336456766e-06,
+      "loss": 0.0222,
+      "step": 2779
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.524928127559664e-06,
+      "loss": 0.0172,
+      "step": 2780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.518763099639604e-06,
+      "loss": 0.0195,
+      "step": 2781
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.512598255045133e-06,
+      "loss": 0.0171,
+      "step": 2782
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.506433596124727e-06,
+      "loss": 0.0192,
+      "step": 2783
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.500269125226792e-06,
+      "loss": 0.0196,
+      "step": 2784
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.494104844699667e-06,
+      "loss": 0.018,
+      "step": 2785
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.487940756891608e-06,
+      "loss": 0.0258,
+      "step": 2786
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.481776864150804e-06,
+      "loss": 0.0206,
+      "step": 2787
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.475613168825374e-06,
+      "loss": 0.0166,
+      "step": 2788
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.46944967326335e-06,
+      "loss": 0.014,
+      "step": 2789
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.463286379812706e-06,
+      "loss": 0.0185,
+      "step": 2790
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.457123290821318e-06,
+      "loss": 0.0273,
+      "step": 2791
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.450960408636997e-06,
+      "loss": 0.0107,
+      "step": 2792
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.444797735607472e-06,
+      "loss": 0.0145,
+      "step": 2793
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.438635274080394e-06,
+      "loss": 0.0194,
+      "step": 2794
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.432473026403334e-06,
+      "loss": 0.0197,
+      "step": 2795
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.426310994923779e-06,
+      "loss": 0.0177,
+      "step": 2796
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.420149181989129e-06,
+      "loss": 0.022,
+      "step": 2797
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.41398758994671e-06,
+      "loss": 0.0199,
+      "step": 2798
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.407826221143761e-06,
+      "loss": 0.0126,
+      "step": 2799
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.40166507792743e-06,
+      "loss": 0.0181,
+      "step": 2800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.395504162644793e-06,
+      "loss": 0.0236,
+      "step": 2801
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.389343477642824e-06,
+      "loss": 0.0252,
+      "step": 2802
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.383183025268411e-06,
+      "loss": 0.0164,
+      "step": 2803
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.377022807868368e-06,
+      "loss": 0.0237,
+      "step": 2804
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.370862827789399e-06,
+      "loss": 0.0184,
+      "step": 2805
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.364703087378133e-06,
+      "loss": 0.0142,
+      "step": 2806
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.358543588981107e-06,
+      "loss": 0.022,
+      "step": 2807
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.352384334944754e-06,
+      "loss": 0.0153,
+      "step": 2808
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.346225327615422e-06,
+      "loss": 0.0254,
+      "step": 2809
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.340066569339368e-06,
+      "loss": 0.0224,
+      "step": 2810
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.333908062462746e-06,
+      "loss": 0.0148,
+      "step": 2811
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.32774980933162e-06,
+      "loss": 0.0208,
+      "step": 2812
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.321591812291963e-06,
+      "loss": 0.0281,
+      "step": 2813
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.315434073689634e-06,
+      "loss": 0.0248,
+      "step": 2814
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.309276595870404e-06,
+      "loss": 0.0248,
+      "step": 2815
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.303119381179949e-06,
+      "loss": 0.0179,
+      "step": 2816
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.296962431963834e-06,
+      "loss": 0.0268,
+      "step": 2817
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.290805750567532e-06,
+      "loss": 0.0169,
+      "step": 2818
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.284649339336408e-06,
+      "loss": 0.0231,
+      "step": 2819
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.278493200615727e-06,
+      "loss": 0.0208,
+      "step": 2820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.272337336750647e-06,
+      "loss": 0.0173,
+      "step": 2821
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.26618175008623e-06,
+      "loss": 0.0225,
+      "step": 2822
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.260026442967419e-06,
+      "loss": 0.0143,
+      "step": 2823
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.253871417739064e-06,
+      "loss": 0.0221,
+      "step": 2824
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.247716676745894e-06,
+      "loss": 0.0192,
+      "step": 2825
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.241562222332544e-06,
+      "loss": 0.0197,
+      "step": 2826
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.23540805684353e-06,
+      "loss": 0.0225,
+      "step": 2827
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.22925418262326e-06,
+      "loss": 0.0184,
+      "step": 2828
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.223100602016033e-06,
+      "loss": 0.0234,
+      "step": 2829
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.216947317366037e-06,
+      "loss": 0.0147,
+      "step": 2830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.21079433101734e-06,
+      "loss": 0.0182,
+      "step": 2831
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.204641645313905e-06,
+      "loss": 0.0245,
+      "step": 2832
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.198489262599575e-06,
+      "loss": 0.0184,
+      "step": 2833
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.19233718521808e-06,
+      "loss": 0.0267,
+      "step": 2834
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.18618541551304e-06,
+      "loss": 0.0194,
+      "step": 2835
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.180033955827943e-06,
+      "loss": 0.0232,
+      "step": 2836
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.173882808506168e-06,
+      "loss": 0.0176,
+      "step": 2837
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.167731975890977e-06,
+      "loss": 0.0234,
+      "step": 2838
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.161581460325505e-06,
+      "loss": 0.018,
+      "step": 2839
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.155431264152775e-06,
+      "loss": 0.0163,
+      "step": 2840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.149281389715683e-06,
+      "loss": 0.0158,
+      "step": 2841
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.143131839357001e-06,
+      "loss": 0.0173,
+      "step": 2842
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.136982615419379e-06,
+      "loss": 0.0198,
+      "step": 2843
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.130833720245345e-06,
+      "loss": 0.014,
+      "step": 2844
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.1246851561773e-06,
+      "loss": 0.023,
+      "step": 2845
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.118536925557522e-06,
+      "loss": 0.0184,
+      "step": 2846
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.11238903072815e-06,
+      "loss": 0.0185,
+      "step": 2847
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.106241474031213e-06,
+      "loss": 0.0193,
+      "step": 2848
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.100094257808596e-06,
+      "loss": 0.0143,
+      "step": 2849
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.093947384402067e-06,
+      "loss": 0.034,
+      "step": 2850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.087800856153249e-06,
+      "loss": 0.0169,
+      "step": 2851
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.08165467540365e-06,
+      "loss": 0.0227,
+      "step": 2852
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.07550884449463e-06,
+      "loss": 0.0234,
+      "step": 2853
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.069363365767424e-06,
+      "loss": 0.0226,
+      "step": 2854
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.063218241563135e-06,
+      "loss": 0.021,
+      "step": 2855
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.057073474222723e-06,
+      "loss": 0.0162,
+      "step": 2856
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.05092906608702e-06,
+      "loss": 0.0183,
+      "step": 2857
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.04478501949672e-06,
+      "loss": 0.0203,
+      "step": 2858
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.038641336792369e-06,
+      "loss": 0.0171,
+      "step": 2859
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.032498020314391e-06,
+      "loss": 0.0243,
+      "step": 2860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.026355072403054e-06,
+      "loss": 0.0207,
+      "step": 2861
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.0202124953985e-06,
+      "loss": 0.0236,
+      "step": 2862
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.01407029164072e-06,
+      "loss": 0.0137,
+      "step": 2863
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.007928463469568e-06,
+      "loss": 0.0263,
+      "step": 2864
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.001787013224748e-06,
+      "loss": 0.018,
+      "step": 2865
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.99564594324583e-06,
+      "loss": 0.0149,
+      "step": 2866
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.989505255872232e-06,
+      "loss": 0.0187,
+      "step": 2867
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.983364953443227e-06,
+      "loss": 0.0158,
+      "step": 2868
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.97722503829795e-06,
+      "loss": 0.017,
+      "step": 2869
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.971085512775374e-06,
+      "loss": 0.0203,
+      "step": 2870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.96494637921433e-06,
+      "loss": 0.0203,
+      "step": 2871
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.958807639953506e-06,
+      "loss": 0.0217,
+      "step": 2872
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.952669297331426e-06,
+      "loss": 0.0233,
+      "step": 2873
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.946531353686482e-06,
+      "loss": 0.0248,
+      "step": 2874
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.940393811356893e-06,
+      "loss": 0.0181,
+      "step": 2875
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.93425667268074e-06,
+      "loss": 0.0207,
+      "step": 2876
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.928119939995945e-06,
+      "loss": 0.0129,
+      "step": 2877
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.921983615640277e-06,
+      "loss": 0.02,
+      "step": 2878
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.915847701951345e-06,
+      "loss": 0.0243,
+      "step": 2879
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.909712201266612e-06,
+      "loss": 0.0235,
+      "step": 2880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.903577115923365e-06,
+      "loss": 0.0197,
+      "step": 2881
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.897442448258754e-06,
+      "loss": 0.0169,
+      "step": 2882
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.891308200609754e-06,
+      "loss": 0.0142,
+      "step": 2883
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.88517437531319e-06,
+      "loss": 0.0176,
+      "step": 2884
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.87904097470572e-06,
+      "loss": 0.0189,
+      "step": 2885
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.872908001123848e-06,
+      "loss": 0.0251,
+      "step": 2886
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.8667754569039e-06,
+      "loss": 0.0189,
+      "step": 2887
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.860643344382057e-06,
+      "loss": 0.022,
+      "step": 2888
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.85451166589432e-06,
+      "loss": 0.0196,
+      "step": 2889
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.848380423776537e-06,
+      "loss": 0.0204,
+      "step": 2890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.842249620364383e-06,
+      "loss": 0.0226,
+      "step": 2891
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.836119257993363e-06,
+      "loss": 0.0172,
+      "step": 2892
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.829989338998821e-06,
+      "loss": 0.022,
+      "step": 2893
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.82385986571593e-06,
+      "loss": 0.0206,
+      "step": 2894
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.817730840479691e-06,
+      "loss": 0.0215,
+      "step": 2895
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.811602265624937e-06,
+      "loss": 0.0207,
+      "step": 2896
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.805474143486329e-06,
+      "loss": 0.0131,
+      "step": 2897
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.799346476398351e-06,
+      "loss": 0.0219,
+      "step": 2898
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.79321926669532e-06,
+      "loss": 0.0241,
+      "step": 2899
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.787092516711373e-06,
+      "loss": 0.024,
+      "step": 2900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.780966228780478e-06,
+      "loss": 0.0196,
+      "step": 2901
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.774840405236422e-06,
+      "loss": 0.0278,
+      "step": 2902
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.768715048412823e-06,
+      "loss": 0.0224,
+      "step": 2903
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.762590160643108e-06,
+      "loss": 0.0224,
+      "step": 2904
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.756465744260531e-06,
+      "loss": 0.0265,
+      "step": 2905
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.750341801598177e-06,
+      "loss": 0.0217,
+      "step": 2906
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.744218334988933e-06,
+      "loss": 0.0186,
+      "step": 2907
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.738095346765519e-06,
+      "loss": 0.0161,
+      "step": 2908
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.731972839260462e-06,
+      "loss": 0.0204,
+      "step": 2909
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.725850814806114e-06,
+      "loss": 0.023,
+      "step": 2910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.719729275734637e-06,
+      "loss": 0.0206,
+      "step": 2911
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.713608224378014e-06,
+      "loss": 0.0242,
+      "step": 2912
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.707487663068035e-06,
+      "loss": 0.0218,
+      "step": 2913
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.701367594136317e-06,
+      "loss": 0.0237,
+      "step": 2914
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.695248019914267e-06,
+      "loss": 0.022,
+      "step": 2915
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.689128942733124e-06,
+      "loss": 0.0271,
+      "step": 2916
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.683010364923927e-06,
+      "loss": 0.0206,
+      "step": 2917
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.676892288817531e-06,
+      "loss": 0.016,
+      "step": 2918
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.670774716744596e-06,
+      "loss": 0.0127,
+      "step": 2919
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.664657651035586e-06,
+      "loss": 0.0211,
+      "step": 2920
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.65854109402078e-06,
+      "loss": 0.0184,
+      "step": 2921
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.65242504803026e-06,
+      "loss": 0.0212,
+      "step": 2922
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.646309515393914e-06,
+      "loss": 0.0262,
+      "step": 2923
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.640194498441433e-06,
+      "loss": 0.0201,
+      "step": 2924
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.634079999502315e-06,
+      "loss": 0.0176,
+      "step": 2925
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.62796602090585e-06,
+      "loss": 0.0167,
+      "step": 2926
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.621852564981144e-06,
+      "loss": 0.0203,
+      "step": 2927
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.615739634057098e-06,
+      "loss": 0.0212,
+      "step": 2928
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.60962723046241e-06,
+      "loss": 0.018,
+      "step": 2929
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.60351535652558e-06,
+      "loss": 0.0196,
+      "step": 2930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.597404014574912e-06,
+      "loss": 0.0211,
+      "step": 2931
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.59129320693849e-06,
+      "loss": 0.0219,
+      "step": 2932
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.585182935944211e-06,
+      "loss": 0.0168,
+      "step": 2933
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.579073203919769e-06,
+      "loss": 0.0156,
+      "step": 2934
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.572964013192634e-06,
+      "loss": 0.0134,
+      "step": 2935
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.566855366090092e-06,
+      "loss": 0.0298,
+      "step": 2936
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.560747264939205e-06,
+      "loss": 0.0208,
+      "step": 2937
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.554639712066837e-06,
+      "loss": 0.0125,
+      "step": 2938
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.548532709799638e-06,
+      "loss": 0.0192,
+      "step": 2939
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.542426260464053e-06,
+      "loss": 0.0094,
+      "step": 2940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.53632036638631e-06,
+      "loss": 0.0185,
+      "step": 2941
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.530215029892434e-06,
+      "loss": 0.0162,
+      "step": 2942
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.524110253308224e-06,
+      "loss": 0.0163,
+      "step": 2943
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.518006038959282e-06,
+      "loss": 0.0153,
+      "step": 2944
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.511902389170984e-06,
+      "loss": 0.0179,
+      "step": 2945
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.505799306268499e-06,
+      "loss": 0.0137,
+      "step": 2946
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.499696792576774e-06,
+      "loss": 0.0232,
+      "step": 2947
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.493594850420537e-06,
+      "loss": 0.0155,
+      "step": 2948
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.487493482124305e-06,
+      "loss": 0.0189,
+      "step": 2949
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.481392690012378e-06,
+      "loss": 0.0251,
+      "step": 2950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.475292476408827e-06,
+      "loss": 0.0118,
+      "step": 2951
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.469192843637512e-06,
+      "loss": 0.0204,
+      "step": 2952
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.463093794022069e-06,
+      "loss": 0.0246,
+      "step": 2953
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.456995329885902e-06,
+      "loss": 0.0187,
+      "step": 2954
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.450897453552207e-06,
+      "loss": 0.0177,
+      "step": 2955
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.444800167343952e-06,
+      "loss": 0.0106,
+      "step": 2956
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.438703473583869e-06,
+      "loss": 0.0148,
+      "step": 2957
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.432607374594484e-06,
+      "loss": 0.0238,
+      "step": 2958
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.42651187269808e-06,
+      "loss": 0.0188,
+      "step": 2959
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.420416970216714e-06,
+      "loss": 0.0195,
+      "step": 2960
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.414322669472224e-06,
+      "loss": 0.0223,
+      "step": 2961
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.408228972786215e-06,
+      "loss": 0.0124,
+      "step": 2962
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.402135882480054e-06,
+      "loss": 0.0234,
+      "step": 2963
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.396043400874891e-06,
+      "loss": 0.0153,
+      "step": 2964
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.389951530291631e-06,
+      "loss": 0.0224,
+      "step": 2965
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.383860273050952e-06,
+      "loss": 0.0251,
+      "step": 2966
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.3777696314733e-06,
+      "loss": 0.016,
+      "step": 2967
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.371679607878884e-06,
+      "loss": 0.0259,
+      "step": 2968
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.365590204587677e-06,
+      "loss": 0.0174,
+      "step": 2969
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.359501423919422e-06,
+      "loss": 0.0184,
+      "step": 2970
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.353413268193614e-06,
+      "loss": 0.0203,
+      "step": 2971
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.347325739729516e-06,
+      "loss": 0.0122,
+      "step": 2972
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.34123884084615e-06,
+      "loss": 0.0239,
+      "step": 2973
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.335152573862308e-06,
+      "loss": 0.0147,
+      "step": 2974
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.329066941096529e-06,
+      "loss": 0.0176,
+      "step": 2975
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.322981944867107e-06,
+      "loss": 0.0163,
+      "step": 2976
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.316897587492108e-06,
+      "loss": 0.0156,
+      "step": 2977
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.310813871289349e-06,
+      "loss": 0.0223,
+      "step": 2978
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.304730798576397e-06,
+      "loss": 0.0219,
+      "step": 2979
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.298648371670581e-06,
+      "loss": 0.0163,
+      "step": 2980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.292566592888985e-06,
+      "loss": 0.0188,
+      "step": 2981
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.286485464548432e-06,
+      "loss": 0.0201,
+      "step": 2982
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.280404988965512e-06,
+      "loss": 0.0271,
+      "step": 2983
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.274325168456567e-06,
+      "loss": 0.0152,
+      "step": 2984
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.268246005337677e-06,
+      "loss": 0.0196,
+      "step": 2985
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.262167501924686e-06,
+      "loss": 0.0152,
+      "step": 2986
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.256089660533177e-06,
+      "loss": 0.0182,
+      "step": 2987
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.250012483478478e-06,
+      "loss": 0.0173,
+      "step": 2988
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.243935973075675e-06,
+      "loss": 0.0242,
+      "step": 2989
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.237860131639593e-06,
+      "loss": 0.0126,
+      "step": 2990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.2317849614848e-06,
+      "loss": 0.0208,
+      "step": 2991
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.225710464925621e-06,
+      "loss": 0.0247,
+      "step": 2992
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.219636644276108e-06,
+      "loss": 0.0173,
+      "step": 2993
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.21356350185006e-06,
+      "loss": 0.0137,
+      "step": 2994
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.207491039961025e-06,
+      "loss": 0.0332,
+      "step": 2995
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.201419260922288e-06,
+      "loss": 0.0176,
+      "step": 2996
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.195348167046868e-06,
+      "loss": 0.0137,
+      "step": 2997
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.189277760647537e-06,
+      "loss": 0.0117,
+      "step": 2998
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.183208044036787e-06,
+      "loss": 0.0167,
+      "step": 2999
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.177139019526859e-06,
+      "loss": 0.0105,
+      "step": 3000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.171070689429725e-06,
+      "loss": 0.0109,
+      "step": 3001
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.165003056057102e-06,
+      "loss": 0.0195,
+      "step": 3002
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.158936121720433e-06,
+      "loss": 0.0167,
+      "step": 3003
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.152869888730888e-06,
+      "loss": 0.0233,
+      "step": 3004
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.146804359399387e-06,
+      "loss": 0.019,
+      "step": 3005
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.14073953603657e-06,
+      "loss": 0.011,
+      "step": 3006
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.13467542095281e-06,
+      "loss": 0.0174,
+      "step": 3007
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.128612016458214e-06,
+      "loss": 0.0202,
+      "step": 3008
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.122549324862619e-06,
+      "loss": 0.0149,
+      "step": 3009
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.116487348475577e-06,
+      "loss": 0.0186,
+      "step": 3010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.110426089606383e-06,
+      "loss": 0.0131,
+      "step": 3011
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.104365550564057e-06,
+      "loss": 0.0246,
+      "step": 3012
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.098305733657335e-06,
+      "loss": 0.0176,
+      "step": 3013
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.092246641194687e-06,
+      "loss": 0.015,
+      "step": 3014
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.086188275484308e-06,
+      "loss": 0.0179,
+      "step": 3015
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.080130638834103e-06,
+      "loss": 0.0202,
+      "step": 3016
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.074073733551712e-06,
+      "loss": 0.0214,
+      "step": 3017
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.0680175619445e-06,
+      "loss": 0.0189,
+      "step": 3018
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.061962126319534e-06,
+      "loss": 0.0115,
+      "step": 3019
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.055907428983622e-06,
+      "loss": 0.0181,
+      "step": 3020
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.049853472243274e-06,
+      "loss": 0.0238,
+      "step": 3021
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.043800258404725e-06,
+      "loss": 0.0141,
+      "step": 3022
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.037747789773928e-06,
+      "loss": 0.0168,
+      "step": 3023
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.031696068656552e-06,
+      "loss": 0.0207,
+      "step": 3024
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.025645097357978e-06,
+      "loss": 0.0191,
+      "step": 3025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.01959487818331e-06,
+      "loss": 0.0134,
+      "step": 3026
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.013545413437348e-06,
+      "loss": 0.0124,
+      "step": 3027
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.00749670542462e-06,
+      "loss": 0.0128,
+      "step": 3028
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.00144875644936e-06,
+      "loss": 0.0164,
+      "step": 3029
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.995401568815519e-06,
+      "loss": 0.0167,
+      "step": 3030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.98935514482675e-06,
+      "loss": 0.023,
+      "step": 3031
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.983309486786414e-06,
+      "loss": 0.0202,
+      "step": 3032
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.977264596997589e-06,
+      "loss": 0.0153,
+      "step": 3033
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.971220477763053e-06,
+      "loss": 0.031,
+      "step": 3034
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.965177131385291e-06,
+      "loss": 0.0139,
+      "step": 3035
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.959134560166504e-06,
+      "loss": 0.0165,
+      "step": 3036
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.953092766408585e-06,
+      "loss": 0.0165,
+      "step": 3037
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.947051752413131e-06,
+      "loss": 0.0167,
+      "step": 3038
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.941011520481447e-06,
+      "loss": 0.0158,
+      "step": 3039
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.934972072914545e-06,
+      "loss": 0.0219,
+      "step": 3040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.928933412013127e-06,
+      "loss": 0.0158,
+      "step": 3041
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.922895540077604e-06,
+      "loss": 0.0186,
+      "step": 3042
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.916858459408082e-06,
+      "loss": 0.0181,
+      "step": 3043
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.910822172304363e-06,
+      "loss": 0.0173,
+      "step": 3044
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.904786681065955e-06,
+      "loss": 0.0144,
+      "step": 3045
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.898751987992058e-06,
+      "loss": 0.0161,
+      "step": 3046
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.892718095381564e-06,
+      "loss": 0.0109,
+      "step": 3047
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.886685005533073e-06,
+      "loss": 0.0252,
+      "step": 3048
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.88065272074486e-06,
+      "loss": 0.0191,
+      "step": 3049
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.87462124331491e-06,
+      "loss": 0.0167,
+      "step": 3050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.86859057554089e-06,
+      "loss": 0.0148,
+      "step": 3051
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.862560719720169e-06,
+      "loss": 0.0217,
+      "step": 3052
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.856531678149792e-06,
+      "loss": 0.0127,
+      "step": 3053
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.850503453126513e-06,
+      "loss": 0.0238,
+      "step": 3054
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.844476046946756e-06,
+      "loss": 0.0249,
+      "step": 3055
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.83844946190664e-06,
+      "loss": 0.0193,
+      "step": 3056
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.832423700301976e-06,
+      "loss": 0.0166,
+      "step": 3057
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.826398764428261e-06,
+      "loss": 0.0164,
+      "step": 3058
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.820374656580672e-06,
+      "loss": 0.0162,
+      "step": 3059
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.814351379054066e-06,
+      "loss": 0.017,
+      "step": 3060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.808328934142999e-06,
+      "loss": 0.0236,
+      "step": 3061
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.802307324141696e-06,
+      "loss": 0.0201,
+      "step": 3062
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.79628655134407e-06,
+      "loss": 0.0179,
+      "step": 3063
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.79026661804372e-06,
+      "loss": 0.0151,
+      "step": 3064
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.784247526533914e-06,
+      "loss": 0.0208,
+      "step": 3065
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.778229279107604e-06,
+      "loss": 0.0163,
+      "step": 3066
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.772211878057425e-06,
+      "loss": 0.0177,
+      "step": 3067
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.766195325675681e-06,
+      "loss": 0.0197,
+      "step": 3068
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.760179624254361e-06,
+      "loss": 0.0153,
+      "step": 3069
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.754164776085126e-06,
+      "loss": 0.0187,
+      "step": 3070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.748150783459318e-06,
+      "loss": 0.0166,
+      "step": 3071
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.742137648667934e-06,
+      "loss": 0.02,
+      "step": 3072
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.736125374001669e-06,
+      "loss": 0.0162,
+      "step": 3073
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.730113961750871e-06,
+      "loss": 0.0137,
+      "step": 3074
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.724103414205573e-06,
+      "loss": 0.0195,
+      "step": 3075
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.718093733655474e-06,
+      "loss": 0.0201,
+      "step": 3076
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.712084922389938e-06,
+      "loss": 0.0161,
+      "step": 3077
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.706076982698e-06,
+      "loss": 0.0245,
+      "step": 3078
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.700069916868367e-06,
+      "loss": 0.0138,
+      "step": 3079
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.694063727189413e-06,
+      "loss": 0.0144,
+      "step": 3080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.688058415949173e-06,
+      "loss": 0.0189,
+      "step": 3081
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.682053985435355e-06,
+      "loss": 0.0174,
+      "step": 3082
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.676050437935318e-06,
+      "loss": 0.0189,
+      "step": 3083
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.670047775736098e-06,
+      "loss": 0.019,
+      "step": 3084
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.66404600112439e-06,
+      "loss": 0.0113,
+      "step": 3085
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.658045116386549e-06,
+      "loss": 0.019,
+      "step": 3086
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.652045123808596e-06,
+      "loss": 0.018,
+      "step": 3087
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.646046025676198e-06,
+      "loss": 0.0165,
+      "step": 3088
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.640047824274702e-06,
+      "loss": 0.0231,
+      "step": 3089
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.634050521889093e-06,
+      "loss": 0.024,
+      "step": 3090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.628054120804032e-06,
+      "loss": 0.0218,
+      "step": 3091
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.622058623303825e-06,
+      "loss": 0.019,
+      "step": 3092
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.616064031672439e-06,
+      "loss": 0.0238,
+      "step": 3093
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.610070348193486e-06,
+      "loss": 0.0216,
+      "step": 3094
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.604077575150247e-06,
+      "loss": 0.0223,
+      "step": 3095
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.598085714825645e-06,
+      "loss": 0.0221,
+      "step": 3096
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5920947695022606e-06,
+      "loss": 0.0172,
+      "step": 3097
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.586104741462326e-06,
+      "loss": 0.0122,
+      "step": 3098
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.580115632987724e-06,
+      "loss": 0.022,
+      "step": 3099
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.574127446359977e-06,
+      "loss": 0.02,
+      "step": 3100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.568140183860272e-06,
+      "loss": 0.0143,
+      "step": 3101
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.562153847769432e-06,
+      "loss": 0.0183,
+      "step": 3102
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5561684403679355e-06,
+      "loss": 0.0182,
+      "step": 3103
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.550183963935907e-06,
+      "loss": 0.0193,
+      "step": 3104
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5442004207531046e-06,
+      "loss": 0.0212,
+      "step": 3105
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5382178130989405e-06,
+      "loss": 0.0232,
+      "step": 3106
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.532236143252473e-06,
+      "loss": 0.0166,
+      "step": 3107
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.526255413492396e-06,
+      "loss": 0.0149,
+      "step": 3108
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.52027562609705e-06,
+      "loss": 0.0201,
+      "step": 3109
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.514296783344419e-06,
+      "loss": 0.0127,
+      "step": 3110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.508318887512116e-06,
+      "loss": 0.0125,
+      "step": 3111
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.502341940877403e-06,
+      "loss": 0.0163,
+      "step": 3112
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.496365945717176e-06,
+      "loss": 0.0201,
+      "step": 3113
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.490390904307976e-06,
+      "loss": 0.0141,
+      "step": 3114
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.48441681892597e-06,
+      "loss": 0.0195,
+      "step": 3115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4784436918469716e-06,
+      "loss": 0.0172,
+      "step": 3116
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.472471525346418e-06,
+      "loss": 0.0162,
+      "step": 3117
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4665003216993835e-06,
+      "loss": 0.0202,
+      "step": 3118
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.460530083180584e-06,
+      "loss": 0.0161,
+      "step": 3119
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.454560812064362e-06,
+      "loss": 0.0194,
+      "step": 3120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.448592510624692e-06,
+      "loss": 0.0137,
+      "step": 3121
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4426251811351715e-06,
+      "loss": 0.0148,
+      "step": 3122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.436658825869041e-06,
+      "loss": 0.0182,
+      "step": 3123
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.43069344709916e-06,
+      "loss": 0.0193,
+      "step": 3124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.424729047098022e-06,
+      "loss": 0.0196,
+      "step": 3125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4187656281377465e-06,
+      "loss": 0.021,
+      "step": 3126
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.412803192490079e-06,
+      "loss": 0.0232,
+      "step": 3127
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.40684174242638e-06,
+      "loss": 0.0222,
+      "step": 3128
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.400881280217656e-06,
+      "loss": 0.0169,
+      "step": 3129
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.394921808134516e-06,
+      "loss": 0.0199,
+      "step": 3130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.388963328447205e-06,
+      "loss": 0.0166,
+      "step": 3131
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3830058434255904e-06,
+      "loss": 0.0116,
+      "step": 3132
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.377049355339149e-06,
+      "loss": 0.0177,
+      "step": 3133
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3710938664569845e-06,
+      "loss": 0.0215,
+      "step": 3134
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3651393790478275e-06,
+      "loss": 0.0171,
+      "step": 3135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.359185895380012e-06,
+      "loss": 0.0159,
+      "step": 3136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.353233417721504e-06,
+      "loss": 0.02,
+      "step": 3137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.34728194833988e-06,
+      "loss": 0.0247,
+      "step": 3138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.341331489502328e-06,
+      "loss": 0.0148,
+      "step": 3139
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.335382043475656e-06,
+      "loss": 0.0201,
+      "step": 3140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3294336125262896e-06,
+      "loss": 0.0224,
+      "step": 3141
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3234861989202585e-06,
+      "loss": 0.0225,
+      "step": 3142
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.317539804923212e-06,
+      "loss": 0.0194,
+      "step": 3143
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.311594432800416e-06,
+      "loss": 0.018,
+      "step": 3144
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.305650084816731e-06,
+      "loss": 0.0167,
+      "step": 3145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.299706763236636e-06,
+      "loss": 0.0182,
+      "step": 3146
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.293764470324225e-06,
+      "loss": 0.0124,
+      "step": 3147
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.287823208343192e-06,
+      "loss": 0.021,
+      "step": 3148
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2818829795568444e-06,
+      "loss": 0.0197,
+      "step": 3149
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.275943786228086e-06,
+      "loss": 0.0172,
+      "step": 3150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.270005630619438e-06,
+      "loss": 0.0202,
+      "step": 3151
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.264068514993017e-06,
+      "loss": 0.0213,
+      "step": 3152
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.258132441610548e-06,
+      "loss": 0.0174,
+      "step": 3153
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.252197412733364e-06,
+      "loss": 0.0201,
+      "step": 3154
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2462634306223914e-06,
+      "loss": 0.0206,
+      "step": 3155
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.240330497538154e-06,
+      "loss": 0.009,
+      "step": 3156
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.234398615740792e-06,
+      "loss": 0.0155,
+      "step": 3157
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2284677874900285e-06,
+      "loss": 0.0136,
+      "step": 3158
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.222538015045197e-06,
+      "loss": 0.0178,
+      "step": 3159
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.216609300665229e-06,
+      "loss": 0.0214,
+      "step": 3160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.210681646608641e-06,
+      "loss": 0.0186,
+      "step": 3161
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.204755055133554e-06,
+      "loss": 0.0139,
+      "step": 3162
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.198829528497686e-06,
+      "loss": 0.0171,
+      "step": 3163
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.192905068958344e-06,
+      "loss": 0.0224,
+      "step": 3164
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.186981678772431e-06,
+      "loss": 0.023,
+      "step": 3165
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1810593601964495e-06,
+      "loss": 0.0242,
+      "step": 3166
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.175138115486478e-06,
+      "loss": 0.0122,
+      "step": 3167
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.169217946898197e-06,
+      "loss": 0.0216,
+      "step": 3168
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.163298856686877e-06,
+      "loss": 0.0157,
+      "step": 3169
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.157380847107373e-06,
+      "loss": 0.0198,
+      "step": 3170
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.151463920414134e-06,
+      "loss": 0.026,
+      "step": 3171
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.145548078861197e-06,
+      "loss": 0.0145,
+      "step": 3172
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.139633324702173e-06,
+      "loss": 0.0197,
+      "step": 3173
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.133719660190271e-06,
+      "loss": 0.0153,
+      "step": 3174
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.127807087578286e-06,
+      "loss": 0.0159,
+      "step": 3175
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.121895609118586e-06,
+      "loss": 0.0161,
+      "step": 3176
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.115985227063138e-06,
+      "loss": 0.0235,
+      "step": 3177
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.110075943663473e-06,
+      "loss": 0.0183,
+      "step": 3178
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.104167761170717e-06,
+      "loss": 0.0179,
+      "step": 3179
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.098260681835571e-06,
+      "loss": 0.0114,
+      "step": 3180
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.09235470790832e-06,
+      "loss": 0.0176,
+      "step": 3181
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.086449841638821e-06,
+      "loss": 0.0146,
+      "step": 3182
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.080546085276519e-06,
+      "loss": 0.0225,
+      "step": 3183
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0746434410704235e-06,
+      "loss": 0.0188,
+      "step": 3184
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.068741911269131e-06,
+      "loss": 0.0186,
+      "step": 3185
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.062841498120807e-06,
+      "loss": 0.013,
+      "step": 3186
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.056942203873196e-06,
+      "loss": 0.0146,
+      "step": 3187
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.051044030773619e-06,
+      "loss": 0.0184,
+      "step": 3188
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.045146981068957e-06,
+      "loss": 0.0204,
+      "step": 3189
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.039251057005674e-06,
+      "loss": 0.0168,
+      "step": 3190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.033356260829805e-06,
+      "loss": 0.0207,
+      "step": 3191
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.027462594786949e-06,
+      "loss": 0.0188,
+      "step": 3192
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.021570061122281e-06,
+      "loss": 0.0186,
+      "step": 3193
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.015678662080545e-06,
+      "loss": 0.015,
+      "step": 3194
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.009788399906046e-06,
+      "loss": 0.0117,
+      "step": 3195
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.003899276842656e-06,
+      "loss": 0.0223,
+      "step": 3196
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.998011295133825e-06,
+      "loss": 0.0092,
+      "step": 3197
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9921244570225536e-06,
+      "loss": 0.0104,
+      "step": 3198
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.986238764751414e-06,
+      "loss": 0.0166,
+      "step": 3199
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.980354220562548e-06,
+      "loss": 0.0131,
+      "step": 3200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.974470826697645e-06,
+      "loss": 0.0158,
+      "step": 3201
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.968588585397964e-06,
+      "loss": 0.0263,
+      "step": 3202
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.962707498904331e-06,
+      "loss": 0.0199,
+      "step": 3203
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9568275694571195e-06,
+      "loss": 0.0221,
+      "step": 3204
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.950948799296278e-06,
+      "loss": 0.016,
+      "step": 3205
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.945071190661294e-06,
+      "loss": 0.0138,
+      "step": 3206
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.939194745791229e-06,
+      "loss": 0.019,
+      "step": 3207
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.933319466924693e-06,
+      "loss": 0.0154,
+      "step": 3208
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.927445356299855e-06,
+      "loss": 0.0231,
+      "step": 3209
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9215724161544385e-06,
+      "loss": 0.0257,
+      "step": 3210
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.91570064872572e-06,
+      "loss": 0.0171,
+      "step": 3211
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.0196,
+      "step": 3212
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.903960640965246e-06,
+      "loss": 0.0145,
+      "step": 3213
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.898092405105805e-06,
+      "loss": 0.0218,
+      "step": 3214
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.892225350907697e-06,
+      "loss": 0.0163,
+      "step": 3215
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.886359480605953e-06,
+      "loss": 0.018,
+      "step": 3216
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.880494796435155e-06,
+      "loss": 0.0189,
+      "step": 3217
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.874631300629435e-06,
+      "loss": 0.019,
+      "step": 3218
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.868768995422476e-06,
+      "loss": 0.0144,
+      "step": 3219
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.862907883047498e-06,
+      "loss": 0.0199,
+      "step": 3220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.857047965737276e-06,
+      "loss": 0.0135,
+      "step": 3221
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.851189245724128e-06,
+      "loss": 0.0109,
+      "step": 3222
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.845331725239909e-06,
+      "loss": 0.0133,
+      "step": 3223
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.83947540651602e-06,
+      "loss": 0.0167,
+      "step": 3224
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.833620291783415e-06,
+      "loss": 0.0154,
+      "step": 3225
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.827766383272572e-06,
+      "loss": 0.011,
+      "step": 3226
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.821913683213519e-06,
+      "loss": 0.0189,
+      "step": 3227
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.81606219383583e-06,
+      "loss": 0.016,
+      "step": 3228
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.810211917368603e-06,
+      "loss": 0.0146,
+      "step": 3229
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.804362856040481e-06,
+      "loss": 0.0159,
+      "step": 3230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.798515012079647e-06,
+      "loss": 0.0217,
+      "step": 3231
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.792668387713817e-06,
+      "loss": 0.0142,
+      "step": 3232
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7868229851702475e-06,
+      "loss": 0.0169,
+      "step": 3233
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.780978806675714e-06,
+      "loss": 0.0174,
+      "step": 3234
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.775135854456547e-06,
+      "loss": 0.0177,
+      "step": 3235
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.769294130738593e-06,
+      "loss": 0.0179,
+      "step": 3236
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.763453637747244e-06,
+      "loss": 0.0161,
+      "step": 3237
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.757614377707409e-06,
+      "loss": 0.0212,
+      "step": 3238
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.751776352843542e-06,
+      "loss": 0.0175,
+      "step": 3239
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.74593956537961e-06,
+      "loss": 0.0174,
+      "step": 3240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.740104017539126e-06,
+      "loss": 0.0183,
+      "step": 3241
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7342697115451165e-06,
+      "loss": 0.0179,
+      "step": 3242
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.728436649620147e-06,
+      "loss": 0.0172,
+      "step": 3243
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7226048339862994e-06,
+      "loss": 0.0109,
+      "step": 3244
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.716774266865182e-06,
+      "loss": 0.0172,
+      "step": 3245
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.710944950477931e-06,
+      "loss": 0.0198,
+      "step": 3246
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.705116887045209e-06,
+      "loss": 0.0244,
+      "step": 3247
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.699290078787193e-06,
+      "loss": 0.0169,
+      "step": 3248
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.693464527923588e-06,
+      "loss": 0.017,
+      "step": 3249
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.687640236673619e-06,
+      "loss": 0.0175,
+      "step": 3250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.68181720725603e-06,
+      "loss": 0.0142,
+      "step": 3251
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.67599544188908e-06,
+      "loss": 0.0184,
+      "step": 3252
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.670174942790557e-06,
+      "loss": 0.0187,
+      "step": 3253
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.664355712177757e-06,
+      "loss": 0.0141,
+      "step": 3254
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.658537752267498e-06,
+      "loss": 0.0253,
+      "step": 3255
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.652721065276115e-06,
+      "loss": 0.0135,
+      "step": 3256
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.646905653419451e-06,
+      "loss": 0.0266,
+      "step": 3257
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6410915189128675e-06,
+      "loss": 0.0221,
+      "step": 3258
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.635278663971245e-06,
+      "loss": 0.018,
+      "step": 3259
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.629467090808963e-06,
+      "loss": 0.0245,
+      "step": 3260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.623656801639933e-06,
+      "loss": 0.0206,
+      "step": 3261
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6178477986775505e-06,
+      "loss": 0.0221,
+      "step": 3262
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6120400841347465e-06,
+      "loss": 0.0167,
+      "step": 3263
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.606233660223944e-06,
+      "loss": 0.0158,
+      "step": 3264
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.600428529157088e-06,
+      "loss": 0.0224,
+      "step": 3265
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.594624693145615e-06,
+      "loss": 0.0207,
+      "step": 3266
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.588822154400487e-06,
+      "loss": 0.0181,
+      "step": 3267
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5830209151321525e-06,
+      "loss": 0.0193,
+      "step": 3268
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.577220977550578e-06,
+      "loss": 0.0215,
+      "step": 3269
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5714223438652295e-06,
+      "loss": 0.021,
+      "step": 3270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5656250162850795e-06,
+      "loss": 0.0178,
+      "step": 3271
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5598289970186e-06,
+      "loss": 0.0143,
+      "step": 3272
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.554034288273763e-06,
+      "loss": 0.0186,
+      "step": 3273
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.548240892258042e-06,
+      "loss": 0.0197,
+      "step": 3274
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.542448811178417e-06,
+      "loss": 0.0238,
+      "step": 3275
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5366580472413565e-06,
+      "loss": 0.0171,
+      "step": 3276
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.530868602652838e-06,
+      "loss": 0.0235,
+      "step": 3277
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.525080479618331e-06,
+      "loss": 0.0147,
+      "step": 3278
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5192936803427954e-06,
+      "loss": 0.0185,
+      "step": 3279
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.513508207030694e-06,
+      "loss": 0.0161,
+      "step": 3280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.507724061885992e-06,
+      "loss": 0.0212,
+      "step": 3281
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.50194124711213e-06,
+      "loss": 0.0202,
+      "step": 3282
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.49615976491206e-06,
+      "loss": 0.0183,
+      "step": 3283
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.490379617488215e-06,
+      "loss": 0.0211,
+      "step": 3284
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.484600807042519e-06,
+      "loss": 0.0113,
+      "step": 3285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.478823335776394e-06,
+      "loss": 0.016,
+      "step": 3286
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.473047205890752e-06,
+      "loss": 0.0223,
+      "step": 3287
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.467272419585984e-06,
+      "loss": 0.016,
+      "step": 3288
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.461498979061982e-06,
+      "loss": 0.0148,
+      "step": 3289
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.45572688651811e-06,
+      "loss": 0.014,
+      "step": 3290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.449956144153235e-06,
+      "loss": 0.0186,
+      "step": 3291
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4441867541657e-06,
+      "loss": 0.0131,
+      "step": 3292
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4384187187533346e-06,
+      "loss": 0.0157,
+      "step": 3293
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.43265204011345e-06,
+      "loss": 0.0132,
+      "step": 3294
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.426886720442851e-06,
+      "loss": 0.0202,
+      "step": 3295
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.421122761937806e-06,
+      "loss": 0.0194,
+      "step": 3296
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.415360166794083e-06,
+      "loss": 0.0177,
+      "step": 3297
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.40959893720692e-06,
+      "loss": 0.0236,
+      "step": 3298
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.403839075371042e-06,
+      "loss": 0.0204,
+      "step": 3299
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.398080583480645e-06,
+      "loss": 0.0179,
+      "step": 3300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.392323463729413e-06,
+      "loss": 0.0186,
+      "step": 3301
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.386567718310493e-06,
+      "loss": 0.0125,
+      "step": 3302
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.380813349416523e-06,
+      "loss": 0.0179,
+      "step": 3303
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.375060359239608e-06,
+      "loss": 0.0204,
+      "step": 3304
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.369308749971332e-06,
+      "loss": 0.0141,
+      "step": 3305
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.363558523802755e-06,
+      "loss": 0.0116,
+      "step": 3306
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.357809682924396e-06,
+      "loss": 0.013,
+      "step": 3307
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.352062229526266e-06,
+      "loss": 0.0143,
+      "step": 3308
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3463161657978345e-06,
+      "loss": 0.0105,
+      "step": 3309
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.340571493928045e-06,
+      "loss": 0.0165,
+      "step": 3310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.334828216105313e-06,
+      "loss": 0.033,
+      "step": 3311
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.329086334517523e-06,
+      "loss": 0.0167,
+      "step": 3312
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.323345851352017e-06,
+      "loss": 0.0193,
+      "step": 3313
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.317606768795619e-06,
+      "loss": 0.013,
+      "step": 3314
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.311869089034615e-06,
+      "loss": 0.0155,
+      "step": 3315
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3061328142547505e-06,
+      "loss": 0.0171,
+      "step": 3316
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.300397946641248e-06,
+      "loss": 0.0202,
+      "step": 3317
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.294664488378777e-06,
+      "loss": 0.0216,
+      "step": 3318
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.28893244165148e-06,
+      "loss": 0.02,
+      "step": 3319
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.283201808642965e-06,
+      "loss": 0.0212,
+      "step": 3320
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.277472591536297e-06,
+      "loss": 0.0086,
+      "step": 3321
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.271744792513999e-06,
+      "loss": 0.0123,
+      "step": 3322
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.266018413758065e-06,
+      "loss": 0.0121,
+      "step": 3323
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.260293457449929e-06,
+      "loss": 0.0177,
+      "step": 3324
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.254569925770502e-06,
+      "loss": 0.0149,
+      "step": 3325
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.248847820900139e-06,
+      "loss": 0.0244,
+      "step": 3326
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.24312714501866e-06,
+      "loss": 0.0136,
+      "step": 3327
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.237407900305334e-06,
+      "loss": 0.0248,
+      "step": 3328
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2316900889388975e-06,
+      "loss": 0.0191,
+      "step": 3329
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.225973713097518e-06,
+      "loss": 0.0184,
+      "step": 3330
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.220258774958841e-06,
+      "loss": 0.022,
+      "step": 3331
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.214545276699944e-06,
+      "loss": 0.0142,
+      "step": 3332
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.208833220497372e-06,
+      "loss": 0.0173,
+      "step": 3333
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.203122608527114e-06,
+      "loss": 0.0157,
+      "step": 3334
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1974134429646015e-06,
+      "loss": 0.0174,
+      "step": 3335
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1917057259847265e-06,
+      "loss": 0.0141,
+      "step": 3336
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.185999459761827e-06,
+      "loss": 0.0155,
+      "step": 3337
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.18029464646968e-06,
+      "loss": 0.0128,
+      "step": 3338
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1745912882815214e-06,
+      "loss": 0.0145,
+      "step": 3339
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.168889387370027e-06,
+      "loss": 0.0168,
+      "step": 3340
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.163188945907308e-06,
+      "loss": 0.0157,
+      "step": 3341
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.157489966064935e-06,
+      "loss": 0.0143,
+      "step": 3342
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.151792450013917e-06,
+      "loss": 0.0155,
+      "step": 3343
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.146096399924698e-06,
+      "loss": 0.0183,
+      "step": 3344
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.140401817967177e-06,
+      "loss": 0.0124,
+      "step": 3345
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.134708706310679e-06,
+      "loss": 0.013,
+      "step": 3346
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.129017067123977e-06,
+      "loss": 0.0171,
+      "step": 3347
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.123326902575282e-06,
+      "loss": 0.0182,
+      "step": 3348
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.117638214832247e-06,
+      "loss": 0.0194,
+      "step": 3349
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.111951006061953e-06,
+      "loss": 0.0187,
+      "step": 3350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.106265278430932e-06,
+      "loss": 0.0202,
+      "step": 3351
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.100581034105133e-06,
+      "loss": 0.0166,
+      "step": 3352
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.094898275249952e-06,
+      "loss": 0.0231,
+      "step": 3353
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.089217004030219e-06,
+      "loss": 0.0163,
+      "step": 3354
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0835372226101974e-06,
+      "loss": 0.0125,
+      "step": 3355
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.077858933153578e-06,
+      "loss": 0.0127,
+      "step": 3356
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.072182137823488e-06,
+      "loss": 0.0211,
+      "step": 3357
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.06650683878248e-06,
+      "loss": 0.0184,
+      "step": 3358
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.060833038192544e-06,
+      "loss": 0.0128,
+      "step": 3359
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.055160738215094e-06,
+      "loss": 0.0149,
+      "step": 3360
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.049489941010973e-06,
+      "loss": 0.0102,
+      "step": 3361
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0438206487404576e-06,
+      "loss": 0.0117,
+      "step": 3362
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.038152863563234e-06,
+      "loss": 0.0296,
+      "step": 3363
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.032486587638435e-06,
+      "loss": 0.0231,
+      "step": 3364
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.026821823124609e-06,
+      "loss": 0.0163,
+      "step": 3365
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.021158572179726e-06,
+      "loss": 0.0172,
+      "step": 3366
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.015496836961188e-06,
+      "loss": 0.018,
+      "step": 3367
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0098366196258095e-06,
+      "loss": 0.0187,
+      "step": 3368
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.004177922329831e-06,
+      "loss": 0.0146,
+      "step": 3369
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.998520747228912e-06,
+      "loss": 0.0118,
+      "step": 3370
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.992865096478142e-06,
+      "loss": 0.0135,
+      "step": 3371
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.987210972232016e-06,
+      "loss": 0.0142,
+      "step": 3372
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.981558376644459e-06,
+      "loss": 0.0239,
+      "step": 3373
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.975907311868805e-06,
+      "loss": 0.0169,
+      "step": 3374
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.970257780057806e-06,
+      "loss": 0.0161,
+      "step": 3375
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.964609783363634e-06,
+      "loss": 0.0144,
+      "step": 3376
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.958963323937878e-06,
+      "loss": 0.0146,
+      "step": 3377
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.953318403931533e-06,
+      "loss": 0.0164,
+      "step": 3378
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.947675025495019e-06,
+      "loss": 0.0239,
+      "step": 3379
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9420331907781534e-06,
+      "loss": 0.0127,
+      "step": 3380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.936392901930179e-06,
+      "loss": 0.0128,
+      "step": 3381
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.930754161099743e-06,
+      "loss": 0.0173,
+      "step": 3382
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.92511697043491e-06,
+      "loss": 0.02,
+      "step": 3383
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.919481332083143e-06,
+      "loss": 0.0162,
+      "step": 3384
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.913847248191325e-06,
+      "loss": 0.0205,
+      "step": 3385
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.908214720905736e-06,
+      "loss": 0.018,
+      "step": 3386
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.902583752372071e-06,
+      "loss": 0.0133,
+      "step": 3387
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.896954344735426e-06,
+      "loss": 0.0168,
+      "step": 3388
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.891326500140311e-06,
+      "loss": 0.0186,
+      "step": 3389
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.885700220730633e-06,
+      "loss": 0.025,
+      "step": 3390
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8800755086496975e-06,
+      "loss": 0.0135,
+      "step": 3391
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.874452366040224e-06,
+      "loss": 0.0138,
+      "step": 3392
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.868830795044335e-06,
+      "loss": 0.0138,
+      "step": 3393
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8632107978035405e-06,
+      "loss": 0.0183,
+      "step": 3394
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.857592376458767e-06,
+      "loss": 0.0145,
+      "step": 3395
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8519755331503315e-06,
+      "loss": 0.016,
+      "step": 3396
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.846360270017947e-06,
+      "loss": 0.0132,
+      "step": 3397
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.840746589200732e-06,
+      "loss": 0.0208,
+      "step": 3398
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8351344928372e-06,
+      "loss": 0.0133,
+      "step": 3399
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.829523983065265e-06,
+      "loss": 0.0145,
+      "step": 3400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.823915062022227e-06,
+      "loss": 0.0201,
+      "step": 3401
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.818307731844781e-06,
+      "loss": 0.0168,
+      "step": 3402
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.812701994669028e-06,
+      "loss": 0.014,
+      "step": 3403
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.807097852630451e-06,
+      "loss": 0.0111,
+      "step": 3404
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8014953078639316e-06,
+      "loss": 0.0149,
+      "step": 3405
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.795894362503742e-06,
+      "loss": 0.0116,
+      "step": 3406
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.790295018683543e-06,
+      "loss": 0.0168,
+      "step": 3407
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.78469727853638e-06,
+      "loss": 0.0175,
+      "step": 3408
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.779101144194697e-06,
+      "loss": 0.0271,
+      "step": 3409
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.773506617790323e-06,
+      "loss": 0.0159,
+      "step": 3410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7679137014544754e-06,
+      "loss": 0.0193,
+      "step": 3411
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.762322397317759e-06,
+      "loss": 0.0182,
+      "step": 3412
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.756732707510159e-06,
+      "loss": 0.0209,
+      "step": 3413
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.751144634161045e-06,
+      "loss": 0.0101,
+      "step": 3414
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.745558179399181e-06,
+      "loss": 0.0241,
+      "step": 3415
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.739973345352705e-06,
+      "loss": 0.0161,
+      "step": 3416
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7343901341491435e-06,
+      "loss": 0.0211,
+      "step": 3417
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.728808547915405e-06,
+      "loss": 0.0221,
+      "step": 3418
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7232285887777715e-06,
+      "loss": 0.0189,
+      "step": 3419
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.717650258861908e-06,
+      "loss": 0.0108,
+      "step": 3420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.712073560292861e-06,
+      "loss": 0.0121,
+      "step": 3421
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.706498495195059e-06,
+      "loss": 0.0188,
+      "step": 3422
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.700925065692302e-06,
+      "loss": 0.0121,
+      "step": 3423
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.695353273907774e-06,
+      "loss": 0.0117,
+      "step": 3424
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.689783121964025e-06,
+      "loss": 0.0099,
+      "step": 3425
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6842146119829854e-06,
+      "loss": 0.014,
+      "step": 3426
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6786477460859615e-06,
+      "loss": 0.0118,
+      "step": 3427
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.673082526393634e-06,
+      "loss": 0.018,
+      "step": 3428
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.667518955026057e-06,
+      "loss": 0.0203,
+      "step": 3429
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.661957034102645e-06,
+      "loss": 0.0142,
+      "step": 3430
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.656396765742207e-06,
+      "loss": 0.0128,
+      "step": 3431
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.650838152062894e-06,
+      "loss": 0.0133,
+      "step": 3432
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.645281195182251e-06,
+      "loss": 0.0138,
+      "step": 3433
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.639725897217177e-06,
+      "loss": 0.0188,
+      "step": 3434
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6341722602839525e-06,
+      "loss": 0.0124,
+      "step": 3435
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.628620286498205e-06,
+      "loss": 0.0096,
+      "step": 3436
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.623069977974953e-06,
+      "loss": 0.0187,
+      "step": 3437
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.617521336828556e-06,
+      "loss": 0.0189,
+      "step": 3438
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6119743651727565e-06,
+      "loss": 0.0137,
+      "step": 3439
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.606429065120653e-06,
+      "loss": 0.0197,
+      "step": 3440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.600885438784714e-06,
+      "loss": 0.0248,
+      "step": 3441
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5953434882767585e-06,
+      "loss": 0.0231,
+      "step": 3442
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.58980321570798e-06,
+      "loss": 0.0138,
+      "step": 3443
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.584264623188921e-06,
+      "loss": 0.0133,
+      "step": 3444
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.578727712829491e-06,
+      "loss": 0.0189,
+      "step": 3445
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5731924867389654e-06,
+      "loss": 0.0156,
+      "step": 3446
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.567658947025958e-06,
+      "loss": 0.0111,
+      "step": 3447
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5621270957984575e-06,
+      "loss": 0.0158,
+      "step": 3448
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.556596935163805e-06,
+      "loss": 0.014,
+      "step": 3449
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.551068467228702e-06,
+      "loss": 0.0206,
+      "step": 3450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.545541694099188e-06,
+      "loss": 0.0166,
+      "step": 3451
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.540016617880681e-06,
+      "loss": 0.0128,
+      "step": 3452
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.53449324067793e-06,
+      "loss": 0.021,
+      "step": 3453
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.52897156459505e-06,
+      "loss": 0.0155,
+      "step": 3454
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.523451591735507e-06,
+      "loss": 0.0156,
+      "step": 3455
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.51793332420212e-06,
+      "loss": 0.0155,
+      "step": 3456
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.512416764097051e-06,
+      "loss": 0.0158,
+      "step": 3457
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.506901913521808e-06,
+      "loss": 0.0151,
+      "step": 3458
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.501388774577263e-06,
+      "loss": 0.02,
+      "step": 3459
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.495877349363625e-06,
+      "loss": 0.0163,
+      "step": 3460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.490367639980456e-06,
+      "loss": 0.0155,
+      "step": 3461
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.484859648526661e-06,
+      "loss": 0.0165,
+      "step": 3462
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.479353377100491e-06,
+      "loss": 0.0168,
+      "step": 3463
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.473848827799536e-06,
+      "loss": 0.0195,
+      "step": 3464
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.468346002720739e-06,
+      "loss": 0.0188,
+      "step": 3465
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.462844903960385e-06,
+      "loss": 0.0109,
+      "step": 3466
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.457345533614096e-06,
+      "loss": 0.015,
+      "step": 3467
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.451847893776845e-06,
+      "loss": 0.0205,
+      "step": 3468
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.446351986542936e-06,
+      "loss": 0.0188,
+      "step": 3469
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.440857814006011e-06,
+      "loss": 0.0178,
+      "step": 3470
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.435365378259061e-06,
+      "loss": 0.0115,
+      "step": 3471
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.429874681394412e-06,
+      "loss": 0.0214,
+      "step": 3472
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4243857255037245e-06,
+      "loss": 0.0175,
+      "step": 3473
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.418898512678004e-06,
+      "loss": 0.0104,
+      "step": 3474
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4134130450075804e-06,
+      "loss": 0.0141,
+      "step": 3475
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.407929324582122e-06,
+      "loss": 0.0165,
+      "step": 3476
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.402447353490635e-06,
+      "loss": 0.0219,
+      "step": 3477
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.396967133821461e-06,
+      "loss": 0.0156,
+      "step": 3478
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.39148866766227e-06,
+      "loss": 0.0116,
+      "step": 3479
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.38601195710007e-06,
+      "loss": 0.0186,
+      "step": 3480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3805370042211895e-06,
+      "loss": 0.017,
+      "step": 3481
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.375063811111292e-06,
+      "loss": 0.0184,
+      "step": 3482
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.369592379855375e-06,
+      "loss": 0.017,
+      "step": 3483
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.364122712537763e-06,
+      "loss": 0.0224,
+      "step": 3484
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.358654811242109e-06,
+      "loss": 0.0176,
+      "step": 3485
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.353188678051387e-06,
+      "loss": 0.021,
+      "step": 3486
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.347724315047906e-06,
+      "loss": 0.0117,
+      "step": 3487
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.342261724313292e-06,
+      "loss": 0.016,
+      "step": 3488
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.336800907928505e-06,
+      "loss": 0.0179,
+      "step": 3489
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.331341867973822e-06,
+      "loss": 0.0146,
+      "step": 3490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.325884606528855e-06,
+      "loss": 0.0124,
+      "step": 3491
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.320429125672515e-06,
+      "loss": 0.0157,
+      "step": 3492
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.314975427483062e-06,
+      "loss": 0.012,
+      "step": 3493
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.309523514038054e-06,
+      "loss": 0.0125,
+      "step": 3494
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.304073387414384e-06,
+      "loss": 0.0092,
+      "step": 3495
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.298625049688259e-06,
+      "loss": 0.016,
+      "step": 3496
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2931785029352105e-06,
+      "loss": 0.0117,
+      "step": 3497
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.287733749230071e-06,
+      "loss": 0.0126,
+      "step": 3498
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.282290790647011e-06,
+      "loss": 0.0113,
+      "step": 3499
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.276849629259502e-06,
+      "loss": 0.0196,
+      "step": 3500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.271410267140336e-06,
+      "loss": 0.012,
+      "step": 3501
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.265972706361626e-06,
+      "loss": 0.0206,
+      "step": 3502
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.260536948994786e-06,
+      "loss": 0.0173,
+      "step": 3503
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.25510299711055e-06,
+      "loss": 0.0149,
+      "step": 3504
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.24967085277897e-06,
+      "loss": 0.0162,
+      "step": 3505
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.244240518069394e-06,
+      "loss": 0.0114,
+      "step": 3506
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.238811995050497e-06,
+      "loss": 0.0132,
+      "step": 3507
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.233385285790258e-06,
+      "loss": 0.0214,
+      "step": 3508
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2279603923559555e-06,
+      "loss": 0.0144,
+      "step": 3509
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.222537316814189e-06,
+      "loss": 0.0135,
+      "step": 3510
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.217116061230867e-06,
+      "loss": 0.0131,
+      "step": 3511
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2116966276711865e-06,
+      "loss": 0.014,
+      "step": 3512
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2062790181996714e-06,
+      "loss": 0.0183,
+      "step": 3513
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.200863234880142e-06,
+      "loss": 0.0125,
+      "step": 3514
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.195449279775717e-06,
+      "loss": 0.0172,
+      "step": 3515
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.190037154948827e-06,
+      "loss": 0.0137,
+      "step": 3516
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.184626862461208e-06,
+      "loss": 0.0115,
+      "step": 3517
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.179218404373886e-06,
+      "loss": 0.0168,
+      "step": 3518
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1738117827472e-06,
+      "loss": 0.0121,
+      "step": 3519
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.16840699964078e-06,
+      "loss": 0.0146,
+      "step": 3520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.163004057113561e-06,
+      "loss": 0.0201,
+      "step": 3521
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.157602957223776e-06,
+      "loss": 0.0184,
+      "step": 3522
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.152203702028958e-06,
+      "loss": 0.0155,
+      "step": 3523
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.146806293585939e-06,
+      "loss": 0.0104,
+      "step": 3524
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.141410733950838e-06,
+      "loss": 0.0137,
+      "step": 3525
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.136017025179072e-06,
+      "loss": 0.0129,
+      "step": 3526
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.13062516932536e-06,
+      "loss": 0.0135,
+      "step": 3527
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.125235168443714e-06,
+      "loss": 0.0149,
+      "step": 3528
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.119847024587434e-06,
+      "loss": 0.0176,
+      "step": 3529
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.11446073980912e-06,
+      "loss": 0.0101,
+      "step": 3530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.109076316160656e-06,
+      "loss": 0.0079,
+      "step": 3531
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.103693755693215e-06,
+      "loss": 0.0135,
+      "step": 3532
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.098313060457272e-06,
+      "loss": 0.0116,
+      "step": 3533
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.09293423250258e-06,
+      "loss": 0.0129,
+      "step": 3534
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.087557273878191e-06,
+      "loss": 0.0155,
+      "step": 3535
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0821821866324395e-06,
+      "loss": 0.0134,
+      "step": 3536
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.076808972812944e-06,
+      "loss": 0.0167,
+      "step": 3537
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0714376344666095e-06,
+      "loss": 0.0174,
+      "step": 3538
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.066068173639634e-06,
+      "loss": 0.018,
+      "step": 3539
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.060700592377492e-06,
+      "loss": 0.0194,
+      "step": 3540
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0553348927249506e-06,
+      "loss": 0.0122,
+      "step": 3541
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.049971076726053e-06,
+      "loss": 0.0219,
+      "step": 3542
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.044609146424129e-06,
+      "loss": 0.0188,
+      "step": 3543
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0392491038617805e-06,
+      "loss": 0.0082,
+      "step": 3544
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.033890951080903e-06,
+      "loss": 0.016,
+      "step": 3545
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.028534690122666e-06,
+      "loss": 0.0141,
+      "step": 3546
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0231803230275255e-06,
+      "loss": 0.0183,
+      "step": 3547
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.017827851835199e-06,
+      "loss": 0.0189,
+      "step": 3548
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0124772785847024e-06,
+      "loss": 0.0153,
+      "step": 3549
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.007128605314309e-06,
+      "loss": 0.0216,
+      "step": 3550
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.001781834061583e-06,
+      "loss": 0.0156,
+      "step": 3551
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9964369668633584e-06,
+      "loss": 0.0164,
+      "step": 3552
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.99109400575575e-06,
+      "loss": 0.0128,
+      "step": 3553
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.985752952774132e-06,
+      "loss": 0.0235,
+      "step": 3554
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.980413809953168e-06,
+      "loss": 0.0166,
+      "step": 3555
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.975076579326781e-06,
+      "loss": 0.0123,
+      "step": 3556
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.969741262928172e-06,
+      "loss": 0.0194,
+      "step": 3557
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.964407862789817e-06,
+      "loss": 0.0185,
+      "step": 3558
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.959076380943451e-06,
+      "loss": 0.0127,
+      "step": 3559
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.953746819420085e-06,
+      "loss": 0.0174,
+      "step": 3560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.948419180250004e-06,
+      "loss": 0.0189,
+      "step": 3561
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9430934654627464e-06,
+      "loss": 0.0109,
+      "step": 3562
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.937769677087126e-06,
+      "loss": 0.0182,
+      "step": 3563
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9324478171512315e-06,
+      "loss": 0.0089,
+      "step": 3564
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.927127887682396e-06,
+      "loss": 0.0177,
+      "step": 3565
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.921809890707233e-06,
+      "loss": 0.0186,
+      "step": 3566
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.916493828251621e-06,
+      "loss": 0.0173,
+      "step": 3567
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9111797023406884e-06,
+      "loss": 0.0143,
+      "step": 3568
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.905867514998835e-06,
+      "loss": 0.0164,
+      "step": 3569
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.900557268249728e-06,
+      "loss": 0.014,
+      "step": 3570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.895248964116279e-06,
+      "loss": 0.0137,
+      "step": 3571
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.889942604620673e-06,
+      "loss": 0.0111,
+      "step": 3572
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.884638191784353e-06,
+      "loss": 0.0088,
+      "step": 3573
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.87933572762801e-06,
+      "loss": 0.0183,
+      "step": 3574
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.874035214171608e-06,
+      "loss": 0.0155,
+      "step": 3575
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.868736653434351e-06,
+      "loss": 0.0171,
+      "step": 3576
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.863440047434714e-06,
+      "loss": 0.0221,
+      "step": 3577
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8581453981904205e-06,
+      "loss": 0.0202,
+      "step": 3578
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.852852707718452e-06,
+      "loss": 0.0143,
+      "step": 3579
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.847561978035033e-06,
+      "loss": 0.0121,
+      "step": 3580
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.842273211155661e-06,
+      "loss": 0.0161,
+      "step": 3581
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.836986409095062e-06,
+      "loss": 0.0238,
+      "step": 3582
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8317015738672325e-06,
+      "loss": 0.0189,
+      "step": 3583
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.826418707485411e-06,
+      "loss": 0.0145,
+      "step": 3584
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.821137811962091e-06,
+      "loss": 0.0136,
+      "step": 3585
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.815858889309007e-06,
+      "loss": 0.0136,
+      "step": 3586
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8105819415371525e-06,
+      "loss": 0.0247,
+      "step": 3587
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8053069706567555e-06,
+      "loss": 0.01,
+      "step": 3588
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8000339786773045e-06,
+      "loss": 0.0155,
+      "step": 3589
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7947629676075245e-06,
+      "loss": 0.0141,
+      "step": 3590
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.789493939455393e-06,
+      "loss": 0.017,
+      "step": 3591
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.784226896228131e-06,
+      "loss": 0.0096,
+      "step": 3592
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.778961839932197e-06,
+      "loss": 0.0198,
+      "step": 3593
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.773698772573293e-06,
+      "loss": 0.0194,
+      "step": 3594
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.76843769615637e-06,
+      "loss": 0.0109,
+      "step": 3595
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.763178612685619e-06,
+      "loss": 0.0198,
+      "step": 3596
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7579215241644705e-06,
+      "loss": 0.0159,
+      "step": 3597
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.752666432595596e-06,
+      "loss": 0.0159,
+      "step": 3598
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.747413339980904e-06,
+      "loss": 0.0118,
+      "step": 3599
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.742162248321537e-06,
+      "loss": 0.0117,
+      "step": 3600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.736913159617886e-06,
+      "loss": 0.0187,
+      "step": 3601
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.731666075869572e-06,
+      "loss": 0.0189,
+      "step": 3602
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7264209990754594e-06,
+      "loss": 0.0169,
+      "step": 3603
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.721177931233635e-06,
+      "loss": 0.0173,
+      "step": 3604
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7159368743414345e-06,
+      "loss": 0.0209,
+      "step": 3605
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.710697830395414e-06,
+      "loss": 0.0172,
+      "step": 3606
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.705460801391374e-06,
+      "loss": 0.0128,
+      "step": 3607
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.700225789324343e-06,
+      "loss": 0.0078,
+      "step": 3608
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.694992796188585e-06,
+      "loss": 0.0117,
+      "step": 3609
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.689761823977582e-06,
+      "loss": 0.0148,
+      "step": 3610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.684532874684067e-06,
+      "loss": 0.0145,
+      "step": 3611
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6793059502999784e-06,
+      "loss": 0.0141,
+      "step": 3612
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.674081052816503e-06,
+      "loss": 0.0137,
+      "step": 3613
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.66885818422405e-06,
+      "loss": 0.0135,
+      "step": 3614
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.663637346512248e-06,
+      "loss": 0.0233,
+      "step": 3615
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.658418541669958e-06,
+      "loss": 0.0136,
+      "step": 3616
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.653201771685273e-06,
+      "loss": 0.0168,
+      "step": 3617
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.647987038545496e-06,
+      "loss": 0.0161,
+      "step": 3618
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.642774344237165e-06,
+      "loss": 0.0136,
+      "step": 3619
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.637563690746044e-06,
+      "loss": 0.0202,
+      "step": 3620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.632355080057105e-06,
+      "loss": 0.013,
+      "step": 3621
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.627148514154554e-06,
+      "loss": 0.009,
+      "step": 3622
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.621943995021821e-06,
+      "loss": 0.0099,
+      "step": 3623
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.616741524641539e-06,
+      "loss": 0.0144,
+      "step": 3624
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6115411049955795e-06,
+      "loss": 0.0112,
+      "step": 3625
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.606342738065027e-06,
+      "loss": 0.0143,
+      "step": 3626
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6011464258301745e-06,
+      "loss": 0.0141,
+      "step": 3627
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.595952170270542e-06,
+      "loss": 0.0195,
+      "step": 3628
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.59075997336487e-06,
+      "loss": 0.0124,
+      "step": 3629
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5855698370911004e-06,
+      "loss": 0.0123,
+      "step": 3630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.580381763426405e-06,
+      "loss": 0.013,
+      "step": 3631
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.575195754347156e-06,
+      "loss": 0.0144,
+      "step": 3632
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.57001181182895e-06,
+      "loss": 0.0095,
+      "step": 3633
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.564829937846592e-06,
+      "loss": 0.0162,
+      "step": 3634
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.559650134374105e-06,
+      "loss": 0.0183,
+      "step": 3635
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.554472403384709e-06,
+      "loss": 0.0116,
+      "step": 3636
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.54929674685085e-06,
+      "loss": 0.015,
+      "step": 3637
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5441231667441724e-06,
+      "loss": 0.0093,
+      "step": 3638
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.538951665035535e-06,
+      "loss": 0.0209,
+      "step": 3639
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5337822436950055e-06,
+      "loss": 0.0114,
+      "step": 3640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.528614904691861e-06,
+      "loss": 0.015,
+      "step": 3641
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.523449649994578e-06,
+      "loss": 0.0124,
+      "step": 3642
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.518286481570841e-06,
+      "loss": 0.0176,
+      "step": 3643
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.513125401387542e-06,
+      "loss": 0.01,
+      "step": 3644
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5079664114107814e-06,
+      "loss": 0.0171,
+      "step": 3645
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.502809513605857e-06,
+      "loss": 0.0137,
+      "step": 3646
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.497654709937276e-06,
+      "loss": 0.0159,
+      "step": 3647
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.492502002368738e-06,
+      "loss": 0.0142,
+      "step": 3648
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.48735139286315e-06,
+      "loss": 0.0105,
+      "step": 3649
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.48220288338262e-06,
+      "loss": 0.0087,
+      "step": 3650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.477056475888456e-06,
+      "loss": 0.0114,
+      "step": 3651
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.471912172341165e-06,
+      "loss": 0.0145,
+      "step": 3652
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4667699747004555e-06,
+      "loss": 0.0149,
+      "step": 3653
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.461629884925228e-06,
+      "loss": 0.0149,
+      "step": 3654
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.456491904973579e-06,
+      "loss": 0.0161,
+      "step": 3655
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4513560368028064e-06,
+      "loss": 0.0235,
+      "step": 3656
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.446222282369404e-06,
+      "loss": 0.018,
+      "step": 3657
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.441090643629057e-06,
+      "loss": 0.0093,
+      "step": 3658
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.43596112253665e-06,
+      "loss": 0.0133,
+      "step": 3659
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.43083372104625e-06,
+      "loss": 0.0146,
+      "step": 3660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.425708441111132e-06,
+      "loss": 0.0142,
+      "step": 3661
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.420585284683744e-06,
+      "loss": 0.0161,
+      "step": 3662
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.415464253715741e-06,
+      "loss": 0.0197,
+      "step": 3663
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.410345350157963e-06,
+      "loss": 0.0133,
+      "step": 3664
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.405228575960442e-06,
+      "loss": 0.0105,
+      "step": 3665
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4001139330723875e-06,
+      "loss": 0.0144,
+      "step": 3666
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.395001423442216e-06,
+      "loss": 0.0097,
+      "step": 3667
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.389891049017511e-06,
+      "loss": 0.0132,
+      "step": 3668
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3847828117450565e-06,
+      "loss": 0.0156,
+      "step": 3669
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.379676713570823e-06,
+      "loss": 0.0139,
+      "step": 3670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3745727564399534e-06,
+      "loss": 0.0152,
+      "step": 3671
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.369470942296788e-06,
+      "loss": 0.0125,
+      "step": 3672
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.364371273084848e-06,
+      "loss": 0.0094,
+      "step": 3673
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.359273750746829e-06,
+      "loss": 0.0179,
+      "step": 3674
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.354178377224618e-06,
+      "loss": 0.0184,
+      "step": 3675
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.349085154459286e-06,
+      "loss": 0.0166,
+      "step": 3676
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3439940843910696e-06,
+      "loss": 0.0161,
+      "step": 3677
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3389051689594e-06,
+      "loss": 0.0143,
+      "step": 3678
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.333818410102884e-06,
+      "loss": 0.0203,
+      "step": 3679
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.328733809759302e-06,
+      "loss": 0.0145,
+      "step": 3680
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.323651369865616e-06,
+      "loss": 0.015,
+      "step": 3681
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.318571092357969e-06,
+      "loss": 0.0081,
+      "step": 3682
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3134929791716685e-06,
+      "loss": 0.022,
+      "step": 3683
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.308417032241208e-06,
+      "loss": 0.007,
+      "step": 3684
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.303343253500258e-06,
+      "loss": 0.0104,
+      "step": 3685
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.298271644881646e-06,
+      "loss": 0.0103,
+      "step": 3686
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.293202208317396e-06,
+      "loss": 0.0177,
+      "step": 3687
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.288134945738684e-06,
+      "loss": 0.0108,
+      "step": 3688
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.283069859075871e-06,
+      "loss": 0.0124,
+      "step": 3689
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.278006950258484e-06,
+      "loss": 0.0115,
+      "step": 3690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.272946221215225e-06,
+      "loss": 0.0152,
+      "step": 3691
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.267887673873955e-06,
+      "loss": 0.0117,
+      "step": 3692
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.262831310161717e-06,
+      "loss": 0.0117,
+      "step": 3693
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.25777713200471e-06,
+      "loss": 0.0164,
+      "step": 3694
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.252725141328308e-06,
+      "loss": 0.0195,
+      "step": 3695
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.247675340057052e-06,
+      "loss": 0.0093,
+      "step": 3696
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.242627730114649e-06,
+      "loss": 0.0094,
+      "step": 3697
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2375823134239624e-06,
+      "loss": 0.0126,
+      "step": 3698
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.232539091907034e-06,
+      "loss": 0.0105,
+      "step": 3699
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2274980674850564e-06,
+      "loss": 0.0068,
+      "step": 3700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.222459242078392e-06,
+      "loss": 0.0145,
+      "step": 3701
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.217422617606566e-06,
+      "loss": 0.0141,
+      "step": 3702
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.212388195988267e-06,
+      "loss": 0.0201,
+      "step": 3703
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.207355979141338e-06,
+      "loss": 0.0151,
+      "step": 3704
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.202325968982781e-06,
+      "loss": 0.0143,
+      "step": 3705
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.197298167428765e-06,
+      "loss": 0.0234,
+      "step": 3706
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.192272576394615e-06,
+      "loss": 0.0145,
+      "step": 3707
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1872491977948125e-06,
+      "loss": 0.0159,
+      "step": 3708
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.182228033543e-06,
+      "loss": 0.0094,
+      "step": 3709
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.177209085551971e-06,
+      "loss": 0.0172,
+      "step": 3710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.172192355733671e-06,
+      "loss": 0.0131,
+      "step": 3711
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.167177845999213e-06,
+      "loss": 0.0104,
+      "step": 3712
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.162165558258855e-06,
+      "loss": 0.0183,
+      "step": 3713
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1571554944220125e-06,
+      "loss": 0.0092,
+      "step": 3714
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.152147656397257e-06,
+      "loss": 0.0195,
+      "step": 3715
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.147142046092304e-06,
+      "loss": 0.0084,
+      "step": 3716
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.142138665414018e-06,
+      "loss": 0.0195,
+      "step": 3717
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1371375162684255e-06,
+      "loss": 0.0122,
+      "step": 3718
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.132138600560699e-06,
+      "loss": 0.0158,
+      "step": 3719
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.127141920195156e-06,
+      "loss": 0.0154,
+      "step": 3720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.0134,
+      "step": 3721
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.117155273103653e-06,
+      "loss": 0.0175,
+      "step": 3722
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.112165310182067e-06,
+      "loss": 0.019,
+      "step": 3723
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.107177590211423e-06,
+      "loss": 0.0167,
+      "step": 3724
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1021921150917775e-06,
+      "loss": 0.0165,
+      "step": 3725
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.09720888672233e-06,
+      "loss": 0.0126,
+      "step": 3726
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.092227907001428e-06,
+      "loss": 0.0155,
+      "step": 3727
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.087249177826553e-06,
+      "loss": 0.0129,
+      "step": 3728
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0822727010943415e-06,
+      "loss": 0.0164,
+      "step": 3729
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0772984787005595e-06,
+      "loss": 0.0158,
+      "step": 3730
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.072326512540122e-06,
+      "loss": 0.0157,
+      "step": 3731
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.067356804507089e-06,
+      "loss": 0.0092,
+      "step": 3732
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0623893564946446e-06,
+      "loss": 0.0308,
+      "step": 3733
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.057424170395128e-06,
+      "loss": 0.0196,
+      "step": 3734
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.052461248100011e-06,
+      "loss": 0.0223,
+      "step": 3735
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.047500591499896e-06,
+      "loss": 0.014,
+      "step": 3736
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.042542202484532e-06,
+      "loss": 0.0146,
+      "step": 3737
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.037586082942805e-06,
+      "loss": 0.013,
+      "step": 3738
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0326322347627235e-06,
+      "loss": 0.0232,
+      "step": 3739
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.027680659831446e-06,
+      "loss": 0.0179,
+      "step": 3740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.022731360035258e-06,
+      "loss": 0.0119,
+      "step": 3741
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.017784337259575e-06,
+      "loss": 0.0112,
+      "step": 3742
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.012839593388954e-06,
+      "loss": 0.0174,
+      "step": 3743
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.007897130307072e-06,
+      "loss": 0.0199,
+      "step": 3744
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.002956949896748e-06,
+      "loss": 0.0123,
+      "step": 3745
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.998019054039927e-06,
+      "loss": 0.0124,
+      "step": 3746
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9930834446176865e-06,
+      "loss": 0.0115,
+      "step": 3747
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.988150123510224e-06,
+      "loss": 0.0177,
+      "step": 3748
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.983219092596878e-06,
+      "loss": 0.0177,
+      "step": 3749
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.978290353756104e-06,
+      "loss": 0.0079,
+      "step": 3750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.973363908865487e-06,
+      "loss": 0.0121,
+      "step": 3751
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9684397598017445e-06,
+      "loss": 0.0154,
+      "step": 3752
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.963517908440716e-06,
+      "loss": 0.0119,
+      "step": 3753
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.958598356657359e-06,
+      "loss": 0.0134,
+      "step": 3754
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9536811063257644e-06,
+      "loss": 0.0133,
+      "step": 3755
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9487661593191366e-06,
+      "loss": 0.0205,
+      "step": 3756
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9438535175098125e-06,
+      "loss": 0.0112,
+      "step": 3757
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9389431827692455e-06,
+      "loss": 0.0104,
+      "step": 3758
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9340351569680155e-06,
+      "loss": 0.0172,
+      "step": 3759
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.929129441975816e-06,
+      "loss": 0.01,
+      "step": 3760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.924226039661456e-06,
+      "loss": 0.012,
+      "step": 3761
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.919324951892876e-06,
+      "loss": 0.012,
+      "step": 3762
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9144261805371305e-06,
+      "loss": 0.013,
+      "step": 3763
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.909529727460387e-06,
+      "loss": 0.0124,
+      "step": 3764
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.904635594527938e-06,
+      "loss": 0.0174,
+      "step": 3765
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.899743783604183e-06,
+      "loss": 0.013,
+      "step": 3766
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.894854296552638e-06,
+      "loss": 0.0209,
+      "step": 3767
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.88996713523594e-06,
+      "loss": 0.0207,
+      "step": 3768
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.885082301515837e-06,
+      "loss": 0.0138,
+      "step": 3769
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8801997972531906e-06,
+      "loss": 0.0191,
+      "step": 3770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.875319624307976e-06,
+      "loss": 0.02,
+      "step": 3771
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.870441784539277e-06,
+      "loss": 0.0129,
+      "step": 3772
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865566279805286e-06,
+      "loss": 0.0088,
+      "step": 3773
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.860693111963315e-06,
+      "loss": 0.0156,
+      "step": 3774
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.85582228286978e-06,
+      "loss": 0.0141,
+      "step": 3775
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.850953794380207e-06,
+      "loss": 0.0073,
+      "step": 3776
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.846087648349234e-06,
+      "loss": 0.0215,
+      "step": 3777
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.841223846630599e-06,
+      "loss": 0.0135,
+      "step": 3778
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.836362391077149e-06,
+      "loss": 0.0181,
+      "step": 3779
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.831503283540841e-06,
+      "loss": 0.013,
+      "step": 3780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8266465258727396e-06,
+      "loss": 0.0106,
+      "step": 3781
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8217921199230045e-06,
+      "loss": 0.0122,
+      "step": 3782
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.816940067540913e-06,
+      "loss": 0.0248,
+      "step": 3783
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8120903705748337e-06,
+      "loss": 0.0163,
+      "step": 3784
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8072430308722387e-06,
+      "loss": 0.007,
+      "step": 3785
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8023980502797086e-06,
+      "loss": 0.018,
+      "step": 3786
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7975554306429253e-06,
+      "loss": 0.0183,
+      "step": 3787
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7927151738066693e-06,
+      "loss": 0.0122,
+      "step": 3788
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7878772816148145e-06,
+      "loss": 0.0127,
+      "step": 3789
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.783041755910346e-06,
+      "loss": 0.0157,
+      "step": 3790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7782085985353343e-06,
+      "loss": 0.0123,
+      "step": 3791
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7733778113309593e-06,
+      "loss": 0.0175,
+      "step": 3792
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.768549396137493e-06,
+      "loss": 0.0075,
+      "step": 3793
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7637233547943054e-06,
+      "loss": 0.0118,
+      "step": 3794
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7588996891398555e-06,
+      "loss": 0.0116,
+      "step": 3795
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7540784010117103e-06,
+      "loss": 0.0199,
+      "step": 3796
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7492594922465144e-06,
+      "loss": 0.0099,
+      "step": 3797
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7444429646800206e-06,
+      "loss": 0.0163,
+      "step": 3798
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.739628820147071e-06,
+      "loss": 0.0172,
+      "step": 3799
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.73481706048159e-06,
+      "loss": 0.0092,
+      "step": 3800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.730007687516608e-06,
+      "loss": 0.0166,
+      "step": 3801
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7252007030842375e-06,
+      "loss": 0.0166,
+      "step": 3802
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.720396109015686e-06,
+      "loss": 0.0107,
+      "step": 3803
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7155939071412416e-06,
+      "loss": 0.0092,
+      "step": 3804
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7107940992902947e-06,
+      "loss": 0.013,
+      "step": 3805
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.705996687291308e-06,
+      "loss": 0.0197,
+      "step": 3806
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7012016729718424e-06,
+      "loss": 0.0085,
+      "step": 3807
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.696409058158544e-06,
+      "loss": 0.01,
+      "step": 3808
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.691618844677146e-06,
+      "loss": 0.0111,
+      "step": 3809
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.686831034352458e-06,
+      "loss": 0.0115,
+      "step": 3810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.682045629008387e-06,
+      "loss": 0.015,
+      "step": 3811
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6772626304679104e-06,
+      "loss": 0.0129,
+      "step": 3812
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.672482040553098e-06,
+      "loss": 0.014,
+      "step": 3813
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6677038610851e-06,
+      "loss": 0.0162,
+      "step": 3814
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6629280938841515e-06,
+      "loss": 0.0175,
+      "step": 3815
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6581547407695626e-06,
+      "loss": 0.0208,
+      "step": 3816
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6533838035597213e-06,
+      "loss": 0.0159,
+      "step": 3817
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6486152840721046e-06,
+      "loss": 0.0185,
+      "step": 3818
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.643849184123264e-06,
+      "loss": 0.0122,
+      "step": 3819
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6390855055288275e-06,
+      "loss": 0.0138,
+      "step": 3820
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.634324250103508e-06,
+      "loss": 0.0148,
+      "step": 3821
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6295654196610863e-06,
+      "loss": 0.0096,
+      "step": 3822
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6248090160144187e-06,
+      "loss": 0.0143,
+      "step": 3823
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6200550409754455e-06,
+      "loss": 0.014,
+      "step": 3824
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.615303496355176e-06,
+      "loss": 0.0152,
+      "step": 3825
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.610554383963698e-06,
+      "loss": 0.0142,
+      "step": 3826
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6058077056101705e-06,
+      "loss": 0.0219,
+      "step": 3827
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.601063463102823e-06,
+      "loss": 0.0183,
+      "step": 3828
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.596321658248956e-06,
+      "loss": 0.0108,
+      "step": 3829
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.591582292854945e-06,
+      "loss": 0.0176,
+      "step": 3830
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5868453687262384e-06,
+      "loss": 0.0117,
+      "step": 3831
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5821108876673495e-06,
+      "loss": 0.0173,
+      "step": 3832
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5773788514818676e-06,
+      "loss": 0.0128,
+      "step": 3833
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5726492619724427e-06,
+      "loss": 0.0125,
+      "step": 3834
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.567922120940791e-06,
+      "loss": 0.0098,
+      "step": 3835
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.563197430187707e-06,
+      "loss": 0.0195,
+      "step": 3836
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5584751915130443e-06,
+      "loss": 0.0142,
+      "step": 3837
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.553755406715724e-06,
+      "loss": 0.0078,
+      "step": 3838
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.549038077593736e-06,
+      "loss": 0.0144,
+      "step": 3839
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.544323205944128e-06,
+      "loss": 0.0147,
+      "step": 3840
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.539610793563011e-06,
+      "loss": 0.0098,
+      "step": 3841
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5349008422455656e-06,
+      "loss": 0.0116,
+      "step": 3842
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.530193353786032e-06,
+      "loss": 0.0167,
+      "step": 3843
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5254883299777177e-06,
+      "loss": 0.0196,
+      "step": 3844
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5207857726129756e-06,
+      "loss": 0.018,
+      "step": 3845
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.516085683483238e-06,
+      "loss": 0.0087,
+      "step": 3846
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.511388064378981e-06,
+      "loss": 0.0144,
+      "step": 3847
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.506692917089751e-06,
+      "loss": 0.0171,
+      "step": 3848
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5020002434041476e-06,
+      "loss": 0.0121,
+      "step": 3849
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.497310045109833e-06,
+      "loss": 0.0103,
+      "step": 3850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4926223239935154e-06,
+      "loss": 0.0143,
+      "step": 3851
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.487937081840974e-06,
+      "loss": 0.0119,
+      "step": 3852
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4832543204370284e-06,
+      "loss": 0.0117,
+      "step": 3853
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4785740415655654e-06,
+      "loss": 0.0114,
+      "step": 3854
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.473896247009524e-06,
+      "loss": 0.0118,
+      "step": 3855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.469220938550889e-06,
+      "loss": 0.0166,
+      "step": 3856
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4645481179707052e-06,
+      "loss": 0.0123,
+      "step": 3857
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4598777870490717e-06,
+      "loss": 0.0109,
+      "step": 3858
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4552099475651304e-06,
+      "loss": 0.0116,
+      "step": 3859
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.45054460129708e-06,
+      "loss": 0.0142,
+      "step": 3860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.445881750022173e-06,
+      "loss": 0.0089,
+      "step": 3861
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.441221395516702e-06,
+      "loss": 0.0155,
+      "step": 3862
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.436563539556014e-06,
+      "loss": 0.0083,
+      "step": 3863
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4319081839145087e-06,
+      "loss": 0.0098,
+      "step": 3864
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4272553303656207e-06,
+      "loss": 0.0162,
+      "step": 3865
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.422604980681843e-06,
+      "loss": 0.0121,
+      "step": 3866
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.417957136634714e-06,
+      "loss": 0.0105,
+      "step": 3867
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4133117999948086e-06,
+      "loss": 0.014,
+      "step": 3868
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4086689725317535e-06,
+      "loss": 0.0073,
+      "step": 3869
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4040286560142222e-06,
+      "loss": 0.0105,
+      "step": 3870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.399390852209923e-06,
+      "loss": 0.0131,
+      "step": 3871
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3947555628856166e-06,
+      "loss": 0.009,
+      "step": 3872
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3901227898070944e-06,
+      "loss": 0.0141,
+      "step": 3873
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3854925347392e-06,
+      "loss": 0.015,
+      "step": 3874
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.380864799445812e-06,
+      "loss": 0.0197,
+      "step": 3875
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3762395856898523e-06,
+      "loss": 0.0128,
+      "step": 3876
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.371616895233284e-06,
+      "loss": 0.0132,
+      "step": 3877
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.366996729837102e-06,
+      "loss": 0.0144,
+      "step": 3878
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3623790912613387e-06,
+      "loss": 0.0129,
+      "step": 3879
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.357763981265071e-06,
+      "loss": 0.0104,
+      "step": 3880
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3531514016064114e-06,
+      "loss": 0.022,
+      "step": 3881
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.348541354042506e-06,
+      "loss": 0.0091,
+      "step": 3882
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3439338403295405e-06,
+      "loss": 0.0124,
+      "step": 3883
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.339328862222727e-06,
+      "loss": 0.0143,
+      "step": 3884
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3347264214763155e-06,
+      "loss": 0.0227,
+      "step": 3885
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.330126519843593e-06,
+      "loss": 0.0147,
+      "step": 3886
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3255291590768748e-06,
+      "loss": 0.0113,
+      "step": 3887
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.320934340927513e-06,
+      "loss": 0.0122,
+      "step": 3888
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.316342067145889e-06,
+      "loss": 0.0203,
+      "step": 3889
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3117523394814122e-06,
+      "loss": 0.0195,
+      "step": 3890
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3071651596825216e-06,
+      "loss": 0.0159,
+      "step": 3891
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.302580529496688e-06,
+      "loss": 0.0136,
+      "step": 3892
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2979984506704153e-06,
+      "loss": 0.0169,
+      "step": 3893
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.293418924949229e-06,
+      "loss": 0.012,
+      "step": 3894
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.288841954077686e-06,
+      "loss": 0.0159,
+      "step": 3895
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2842675397993663e-06,
+      "loss": 0.0148,
+      "step": 3896
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.279695683856875e-06,
+      "loss": 0.0128,
+      "step": 3897
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.275126387991847e-06,
+      "loss": 0.0141,
+      "step": 3898
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.270559653944941e-06,
+      "loss": 0.014,
+      "step": 3899
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.265995483455844e-06,
+      "loss": 0.0109,
+      "step": 3900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2614338782632524e-06,
+      "loss": 0.0133,
+      "step": 3901
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2568748401049024e-06,
+      "loss": 0.0154,
+      "step": 3902
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2523183707175366e-06,
+      "loss": 0.0146,
+      "step": 3903
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2477644718369317e-06,
+      "loss": 0.0172,
+      "step": 3904
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.24321314519788e-06,
+      "loss": 0.0069,
+      "step": 3905
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2386643925341953e-06,
+      "loss": 0.0153,
+      "step": 3906
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2341182155787067e-06,
+      "loss": 0.008,
+      "step": 3907
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.229574616063268e-06,
+      "loss": 0.0187,
+      "step": 3908
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.225033595718744e-06,
+      "loss": 0.011,
+      "step": 3909
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.220495156275024e-06,
+      "loss": 0.0075,
+      "step": 3910
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2159592994610113e-06,
+      "loss": 0.0123,
+      "step": 3911
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2114260270046294e-06,
+      "loss": 0.014,
+      "step": 3912
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.206895340632805e-06,
+      "loss": 0.0199,
+      "step": 3913
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.202367242071496e-06,
+      "loss": 0.0154,
+      "step": 3914
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.197841733045658e-06,
+      "loss": 0.0166,
+      "step": 3915
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.193318815279275e-06,
+      "loss": 0.0154,
+      "step": 3916
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1887984904953373e-06,
+      "loss": 0.0114,
+      "step": 3917
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.184280760415843e-06,
+      "loss": 0.0112,
+      "step": 3918
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1797656267618092e-06,
+      "loss": 0.014,
+      "step": 3919
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1752530912532642e-06,
+      "loss": 0.0116,
+      "step": 3920
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.170743155609236e-06,
+      "loss": 0.0119,
+      "step": 3921
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.166235821547774e-06,
+      "loss": 0.0125,
+      "step": 3922
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1617310907859344e-06,
+      "loss": 0.0179,
+      "step": 3923
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1572289650397724e-06,
+      "loss": 0.0089,
+      "step": 3924
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.152729446024363e-06,
+      "loss": 0.0136,
+      "step": 3925
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1482325354537845e-06,
+      "loss": 0.0097,
+      "step": 3926
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1437382350411137e-06,
+      "loss": 0.0083,
+      "step": 3927
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1392465464984455e-06,
+      "loss": 0.018,
+      "step": 3928
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1347574715368677e-06,
+      "loss": 0.0131,
+      "step": 3929
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1302710118664824e-06,
+      "loss": 0.0122,
+      "step": 3930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1257871691963902e-06,
+      "loss": 0.0116,
+      "step": 3931
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1213059452346992e-06,
+      "loss": 0.0181,
+      "step": 3932
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1168273416885107e-06,
+      "loss": 0.0141,
+      "step": 3933
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.112351360263941e-06,
+      "loss": 0.016,
+      "step": 3934
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1078780026660914e-06,
+      "loss": 0.0149,
+      "step": 3935
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.103407270599078e-06,
+      "loss": 0.0148,
+      "step": 3936
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.098939165766012e-06,
+      "loss": 0.0171,
+      "step": 3937
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.094473689869002e-06,
+      "loss": 0.0107,
+      "step": 3938
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0900108446091535e-06,
+      "loss": 0.0135,
+      "step": 3939
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.085550631686579e-06,
+      "loss": 0.0121,
+      "step": 3940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0810930528003723e-06,
+      "loss": 0.0062,
+      "step": 3941
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0766381096486396e-06,
+      "loss": 0.0132,
+      "step": 3942
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0721858039284746e-06,
+      "loss": 0.0088,
+      "step": 3943
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0677361373359684e-06,
+      "loss": 0.0146,
+      "step": 3944
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.063289111566211e-06,
+      "loss": 0.0176,
+      "step": 3945
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.058844728313277e-06,
+      "loss": 0.0174,
+      "step": 3946
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0544029892702387e-06,
+      "loss": 0.0127,
+      "step": 3947
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0499638961291623e-06,
+      "loss": 0.0132,
+      "step": 3948
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.045527450581107e-06,
+      "loss": 0.0138,
+      "step": 3949
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0410936543161216e-06,
+      "loss": 0.011,
+      "step": 3950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0366625090232493e-06,
+      "loss": 0.0162,
+      "step": 3951
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0322340163905174e-06,
+      "loss": 0.0162,
+      "step": 3952
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0278081781049405e-06,
+      "loss": 0.0131,
+      "step": 3953
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0233849958525306e-06,
+      "loss": 0.0122,
+      "step": 3954
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.018964471318285e-06,
+      "loss": 0.0215,
+      "step": 3955
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.014546606186192e-06,
+      "loss": 0.0105,
+      "step": 3956
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0101314021392136e-06,
+      "loss": 0.014,
+      "step": 3957
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0057188608593146e-06,
+      "loss": 0.019,
+      "step": 3958
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.001308984027431e-06,
+      "loss": 0.0135,
+      "step": 3959
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.996901773323495e-06,
+      "loss": 0.0143,
+      "step": 3960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9924972304264176e-06,
+      "loss": 0.0097,
+      "step": 3961
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.988095357014098e-06,
+      "loss": 0.0121,
+      "step": 3962
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.98369615476341e-06,
+      "loss": 0.0128,
+      "step": 3963
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9792996253502204e-06,
+      "loss": 0.0138,
+      "step": 3964
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9749057704493654e-06,
+      "loss": 0.0096,
+      "step": 3965
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9705145917346735e-06,
+      "loss": 0.0153,
+      "step": 3966
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9661260908789503e-06,
+      "loss": 0.013,
+      "step": 3967
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.961740269553981e-06,
+      "loss": 0.0199,
+      "step": 3968
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9573571294305257e-06,
+      "loss": 0.0109,
+      "step": 3969
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9529766721783326e-06,
+      "loss": 0.016,
+      "step": 3970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9485988994661163e-06,
+      "loss": 0.006,
+      "step": 3971
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9442238129615765e-06,
+      "loss": 0.0148,
+      "step": 3972
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9398514143313916e-06,
+      "loss": 0.0104,
+      "step": 3973
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9354817052412076e-06,
+      "loss": 0.015,
+      "step": 3974
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9311146873556507e-06,
+      "loss": 0.0159,
+      "step": 3975
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9267503623383276e-06,
+      "loss": 0.01,
+      "step": 3976
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.922388731851805e-06,
+      "loss": 0.0136,
+      "step": 3977
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9180297975576368e-06,
+      "loss": 0.0148,
+      "step": 3978
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9136735611163445e-06,
+      "loss": 0.0115,
+      "step": 3979
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9093200241874188e-06,
+      "loss": 0.0151,
+      "step": 3980
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9049691884293264e-06,
+      "loss": 0.0097,
+      "step": 3981
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9006210554995072e-06,
+      "loss": 0.0121,
+      "step": 3982
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.896275627054361e-06,
+      "loss": 0.0084,
+      "step": 3983
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8919329047492715e-06,
+      "loss": 0.0198,
+      "step": 3984
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8875928902385774e-06,
+      "loss": 0.0084,
+      "step": 3985
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.883255585175596e-06,
+      "loss": 0.0128,
+      "step": 3986
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8789209912126083e-06,
+      "loss": 0.0111,
+      "step": 3987
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8745891100008683e-06,
+      "loss": 0.0122,
+      "step": 3988
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.870259943190584e-06,
+      "loss": 0.0092,
+      "step": 3989
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8659334924309444e-06,
+      "loss": 0.0133,
+      "step": 3990
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.861609759370089e-06,
+      "loss": 0.016,
+      "step": 3991
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8572887456551324e-06,
+      "loss": 0.0146,
+      "step": 3992
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.852970452932152e-06,
+      "loss": 0.0118,
+      "step": 3993
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8486548828461893e-06,
+      "loss": 0.0102,
+      "step": 3994
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.844342037041239e-06,
+      "loss": 0.0121,
+      "step": 3995
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.840031917160273e-06,
+      "loss": 0.0139,
+      "step": 3996
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8357245248452104e-06,
+      "loss": 0.0098,
+      "step": 3997
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8314198617369403e-06,
+      "loss": 0.0172,
+      "step": 3998
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8271179294753114e-06,
+      "loss": 0.0215,
+      "step": 3999
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8228187296991316e-06,
+      "loss": 0.015,
+      "step": 4000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8185222640461642e-06,
+      "loss": 0.0151,
+      "step": 4001
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8142285341531305e-06,
+      "loss": 0.011,
+      "step": 4002
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8099375416557163e-06,
+      "loss": 0.0143,
+      "step": 4003
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.80564928818856e-06,
+      "loss": 0.0137,
+      "step": 4004
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8013637753852572e-06,
+      "loss": 0.0127,
+      "step": 4005
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.797081004878364e-06,
+      "loss": 0.0129,
+      "step": 4006
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7928009782993827e-06,
+      "loss": 0.0103,
+      "step": 4007
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7885236972787733e-06,
+      "loss": 0.0169,
+      "step": 4008
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7842491634459545e-06,
+      "loss": 0.0195,
+      "step": 4009
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.779977378429295e-06,
+      "loss": 0.0165,
+      "step": 4010
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7757083438561185e-06,
+      "loss": 0.0136,
+      "step": 4011
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7714420613527014e-06,
+      "loss": 0.0096,
+      "step": 4012
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7671785325442644e-06,
+      "loss": 0.0112,
+      "step": 4013
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7629177590549894e-06,
+      "loss": 0.016,
+      "step": 4014
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7586597425079985e-06,
+      "loss": 0.0137,
+      "step": 4015
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7544044845253726e-06,
+      "loss": 0.0135,
+      "step": 4016
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7501519867281367e-06,
+      "loss": 0.0178,
+      "step": 4017
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7459022507362687e-06,
+      "loss": 0.0134,
+      "step": 4018
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7416552781686867e-06,
+      "loss": 0.014,
+      "step": 4019
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7374110706432656e-06,
+      "loss": 0.009,
+      "step": 4020
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7331696297768162e-06,
+      "loss": 0.0105,
+      "step": 4021
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.728930957185104e-06,
+      "loss": 0.0125,
+      "step": 4022
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.724695054482839e-06,
+      "loss": 0.0132,
+      "step": 4023
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.720461923283674e-06,
+      "loss": 0.0116,
+      "step": 4024
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7162315652002026e-06,
+      "loss": 0.0059,
+      "step": 4025
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7120039818439705e-06,
+      "loss": 0.0082,
+      "step": 4026
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7077791748254554e-06,
+      "loss": 0.0216,
+      "step": 4027
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7035571457540865e-06,
+      "loss": 0.0093,
+      "step": 4028
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6993378962382365e-06,
+      "loss": 0.0144,
+      "step": 4029
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6951214278852055e-06,
+      "loss": 0.0125,
+      "step": 4030
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6909077423012475e-06,
+      "loss": 0.0189,
+      "step": 4031
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.686696841091555e-06,
+      "loss": 0.0084,
+      "step": 4032
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6824887258602507e-06,
+      "loss": 0.0176,
+      "step": 4033
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.678283398210405e-06,
+      "loss": 0.0059,
+      "step": 4034
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.674080859744026e-06,
+      "loss": 0.0132,
+      "step": 4035
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6698811120620504e-06,
+      "loss": 0.0148,
+      "step": 4036
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.665684156764362e-06,
+      "loss": 0.0146,
+      "step": 4037
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6614899954497797e-06,
+      "loss": 0.0147,
+      "step": 4038
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6572986297160486e-06,
+      "loss": 0.0156,
+      "step": 4039
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6531100611598614e-06,
+      "loss": 0.0076,
+      "step": 4040
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6489242913768355e-06,
+      "loss": 0.0193,
+      "step": 4041
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6447413219615257e-06,
+      "loss": 0.0128,
+      "step": 4042
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6405611545074215e-06,
+      "loss": 0.0093,
+      "step": 4043
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.636383790606948e-06,
+      "loss": 0.0108,
+      "step": 4044
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6322092318514503e-06,
+      "loss": 0.0176,
+      "step": 4045
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.628037479831219e-06,
+      "loss": 0.0121,
+      "step": 4046
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6238685361354632e-06,
+      "loss": 0.0181,
+      "step": 4047
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.619702402352332e-06,
+      "loss": 0.0155,
+      "step": 4048
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6155390800688986e-06,
+      "loss": 0.0081,
+      "step": 4049
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6113785708711705e-06,
+      "loss": 0.0131,
+      "step": 4050
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6072208763440742e-06,
+      "loss": 0.0176,
+      "step": 4051
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6030659980714746e-06,
+      "loss": 0.0139,
+      "step": 4052
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.598913937636153e-06,
+      "loss": 0.0113,
+      "step": 4053
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.594764696619826e-06,
+      "loss": 0.0133,
+      "step": 4054
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.590618276603134e-06,
+      "loss": 0.0143,
+      "step": 4055
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5864746791656427e-06,
+      "loss": 0.0146,
+      "step": 4056
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.582333905885841e-06,
+      "loss": 0.0095,
+      "step": 4057
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5781959583411375e-06,
+      "loss": 0.0095,
+      "step": 4058
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.574060838107875e-06,
+      "loss": 0.0141,
+      "step": 4059
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5699285467613113e-06,
+      "loss": 0.011,
+      "step": 4060
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.56579908587563e-06,
+      "loss": 0.0148,
+      "step": 4061
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.561672457023939e-06,
+      "loss": 0.0199,
+      "step": 4062
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5575486617782597e-06,
+      "loss": 0.0112,
+      "step": 4063
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5534277017095353e-06,
+      "loss": 0.0106,
+      "step": 4064
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5493095783876367e-06,
+      "loss": 0.0124,
+      "step": 4065
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5451942933813454e-06,
+      "loss": 0.0105,
+      "step": 4066
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5410818482583687e-06,
+      "loss": 0.0119,
+      "step": 4067
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.536972244585331e-06,
+      "loss": 0.0125,
+      "step": 4068
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.532865483927769e-06,
+      "loss": 0.0104,
+      "step": 4069
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5287615678501355e-06,
+      "loss": 0.0145,
+      "step": 4070
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5246604979158073e-06,
+      "loss": 0.0106,
+      "step": 4071
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.520562275687074e-06,
+      "loss": 0.0088,
+      "step": 4072
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5164669027251385e-06,
+      "loss": 0.01,
+      "step": 4073
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5123743805901246e-06,
+      "loss": 0.0133,
+      "step": 4074
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5082847108410592e-06,
+      "loss": 0.0136,
+      "step": 4075
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.504197895035886e-06,
+      "loss": 0.0106,
+      "step": 4076
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5001139347314675e-06,
+      "loss": 0.0126,
+      "step": 4077
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4960328314835746e-06,
+      "loss": 0.0098,
+      "step": 4078
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4919545868468896e-06,
+      "loss": 0.0082,
+      "step": 4079
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.48787920237501e-06,
+      "loss": 0.0141,
+      "step": 4080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4838066796204318e-06,
+      "loss": 0.0126,
+      "step": 4081
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4797370201345785e-06,
+      "loss": 0.0091,
+      "step": 4082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4756702254677635e-06,
+      "loss": 0.0105,
+      "step": 4083
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4716062971692233e-06,
+      "loss": 0.0111,
+      "step": 4084
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4675452367871023e-06,
+      "loss": 0.0086,
+      "step": 4085
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.463487045868439e-06,
+      "loss": 0.0105,
+      "step": 4086
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4594317259591927e-06,
+      "loss": 0.0087,
+      "step": 4087
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.455379278604226e-06,
+      "loss": 0.0159,
+      "step": 4088
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4513297053472994e-06,
+      "loss": 0.0101,
+      "step": 4089
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.447283007731087e-06,
+      "loss": 0.0071,
+      "step": 4090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4432391872971694e-06,
+      "loss": 0.0093,
+      "step": 4091
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4391982455860186e-06,
+      "loss": 0.017,
+      "step": 4092
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.435160184137021e-06,
+      "loss": 0.0146,
+      "step": 4093
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.431125004488466e-06,
+      "loss": 0.0146,
+      "step": 4094
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4270927081775362e-06,
+      "loss": 0.019,
+      "step": 4095
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4230632967403246e-06,
+      "loss": 0.0106,
+      "step": 4096
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4190367717118233e-06,
+      "loss": 0.0125,
+      "step": 4097
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.41501313462592e-06,
+      "loss": 0.0114,
+      "step": 4098
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.410992387015406e-06,
+      "loss": 0.0147,
+      "step": 4099
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4069745304119775e-06,
+      "loss": 0.0151,
+      "step": 4100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4029595663462147e-06,
+      "loss": 0.015,
+      "step": 4101
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3989474963476134e-06,
+      "loss": 0.0156,
+      "step": 4102
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.394938321944551e-06,
+      "loss": 0.0186,
+      "step": 4103
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3909320446643115e-06,
+      "loss": 0.0202,
+      "step": 4104
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3869286660330737e-06,
+      "loss": 0.0203,
+      "step": 4105
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3829281875759146e-06,
+      "loss": 0.0129,
+      "step": 4106
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3789306108167965e-06,
+      "loss": 0.008,
+      "step": 4107
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3749359372785884e-06,
+      "loss": 0.0112,
+      "step": 4108
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3709441684830436e-06,
+      "loss": 0.0141,
+      "step": 4109
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3669553059508143e-06,
+      "loss": 0.0102,
+      "step": 4110
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.362969351201445e-06,
+      "loss": 0.018,
+      "step": 4111
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3589863057533748e-06,
+      "loss": 0.0108,
+      "step": 4112
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.355006171123928e-06,
+      "loss": 0.0142,
+      "step": 4113
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3510289488293223e-06,
+      "loss": 0.0117,
+      "step": 4114
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3470546403846683e-06,
+      "loss": 0.0073,
+      "step": 4115
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3430832473039666e-06,
+      "loss": 0.0104,
+      "step": 4116
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3391147711001063e-06,
+      "loss": 0.0151,
+      "step": 4117
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3351492132848665e-06,
+      "loss": 0.0102,
+      "step": 4118
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.331186575368911e-06,
+      "loss": 0.0127,
+      "step": 4119
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.327226858861792e-06,
+      "loss": 0.0203,
+      "step": 4120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3232700652719518e-06,
+      "loss": 0.0087,
+      "step": 4121
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.319316196106719e-06,
+      "loss": 0.0128,
+      "step": 4122
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3153652528723046e-06,
+      "loss": 0.0178,
+      "step": 4123
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3114172370738106e-06,
+      "loss": 0.0156,
+      "step": 4124
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3074721502152195e-06,
+      "loss": 0.012,
+      "step": 4125
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.303529993799395e-06,
+      "loss": 0.0123,
+      "step": 4126
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2995907693280904e-06,
+      "loss": 0.0151,
+      "step": 4127
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.295654478301942e-06,
+      "loss": 0.0129,
+      "step": 4128
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.291721122220464e-06,
+      "loss": 0.0122,
+      "step": 4129
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.28779070258206e-06,
+      "loss": 0.0182,
+      "step": 4130
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.283863220884007e-06,
+      "loss": 0.0124,
+      "step": 4131
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.279938678622464e-06,
+      "loss": 0.0099,
+      "step": 4132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.276017077292474e-06,
+      "loss": 0.0113,
+      "step": 4133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.272098418387958e-06,
+      "loss": 0.0133,
+      "step": 4134
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2681827034017157e-06,
+      "loss": 0.0151,
+      "step": 4135
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.264269933825429e-06,
+      "loss": 0.0166,
+      "step": 4136
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.260360111149652e-06,
+      "loss": 0.0099,
+      "step": 4137
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.256453236863815e-06,
+      "loss": 0.0149,
+      "step": 4138
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2525493124562325e-06,
+      "loss": 0.0103,
+      "step": 4139
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.24864833941409e-06,
+      "loss": 0.0127,
+      "step": 4140
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.244750319223453e-06,
+      "loss": 0.0169,
+      "step": 4141
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.240855253369256e-06,
+      "loss": 0.0108,
+      "step": 4142
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2369631433353147e-06,
+      "loss": 0.0108,
+      "step": 4143
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.23307399060431e-06,
+      "loss": 0.0136,
+      "step": 4144
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2291877966578045e-06,
+      "loss": 0.0111,
+      "step": 4145
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2253045629762317e-06,
+      "loss": 0.0129,
+      "step": 4146
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2214242910388993e-06,
+      "loss": 0.0123,
+      "step": 4147
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.217546982323977e-06,
+      "loss": 0.0122,
+      "step": 4148
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2136726383085183e-06,
+      "loss": 0.0086,
+      "step": 4149
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.209801260468437e-06,
+      "loss": 0.0125,
+      "step": 4150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.205932850278525e-06,
+      "loss": 0.0084,
+      "step": 4151
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2020674092124383e-06,
+      "loss": 0.0102,
+      "step": 4152
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.198204938742707e-06,
+      "loss": 0.0106,
+      "step": 4153
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1943454403407215e-06,
+      "loss": 0.0115,
+      "step": 4154
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1904889154767484e-06,
+      "loss": 0.0099,
+      "step": 4155
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.186635365619918e-06,
+      "loss": 0.0089,
+      "step": 4156
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1827847922382238e-06,
+      "loss": 0.0082,
+      "step": 4157
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.178937196798534e-06,
+      "loss": 0.0116,
+      "step": 4158
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1750925807665724e-06,
+      "loss": 0.0076,
+      "step": 4159
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.171250945606933e-06,
+      "loss": 0.0102,
+      "step": 4160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1674122927830764e-06,
+      "loss": 0.0102,
+      "step": 4161
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.163576623757326e-06,
+      "loss": 0.0092,
+      "step": 4162
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1597439399908626e-06,
+      "loss": 0.0196,
+      "step": 4163
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1559142429437385e-06,
+      "loss": 0.0155,
+      "step": 4164
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.152087534074859e-06,
+      "loss": 0.0136,
+      "step": 4165
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1482638148419987e-06,
+      "loss": 0.011,
+      "step": 4166
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1444430867017918e-06,
+      "loss": 0.0105,
+      "step": 4167
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.140625351109733e-06,
+      "loss": 0.02,
+      "step": 4168
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1368106095201747e-06,
+      "loss": 0.0159,
+      "step": 4169
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1329988633863253e-06,
+      "loss": 0.0158,
+      "step": 4170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.129190114160261e-06,
+      "loss": 0.0159,
+      "step": 4171
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1253843632929118e-06,
+      "loss": 0.0146,
+      "step": 4172
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.121581612234066e-06,
+      "loss": 0.0153,
+      "step": 4173
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.11778186243237e-06,
+      "loss": 0.008,
+      "step": 4174
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1139851153353254e-06,
+      "loss": 0.0099,
+      "step": 4175
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1101913723892864e-06,
+      "loss": 0.0107,
+      "step": 4176
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1064006350394693e-06,
+      "loss": 0.0127,
+      "step": 4177
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1026129047299436e-06,
+      "loss": 0.0124,
+      "step": 4178
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0988281829036318e-06,
+      "loss": 0.0084,
+      "step": 4179
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0950464710023132e-06,
+      "loss": 0.0101,
+      "step": 4180
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.091267770466616e-06,
+      "loss": 0.0137,
+      "step": 4181
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0874920827360223e-06,
+      "loss": 0.0129,
+      "step": 4182
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.083719409248869e-06,
+      "loss": 0.0127,
+      "step": 4183
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.079949751442344e-06,
+      "loss": 0.0174,
+      "step": 4184
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0761831107524842e-06,
+      "loss": 0.013,
+      "step": 4185
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0724194886141847e-06,
+      "loss": 0.0159,
+      "step": 4186
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.068658886461179e-06,
+      "loss": 0.0122,
+      "step": 4187
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.064901305726055e-06,
+      "loss": 0.0135,
+      "step": 4188
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0611467478402515e-06,
+      "loss": 0.0118,
+      "step": 4189
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.057395214234057e-06,
+      "loss": 0.0099,
+      "step": 4190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.053646706336604e-06,
+      "loss": 0.0121,
+      "step": 4191
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0499012255758775e-06,
+      "loss": 0.0056,
+      "step": 4192
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0461587733787036e-06,
+      "loss": 0.0121,
+      "step": 4193
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0424193511707522e-06,
+      "loss": 0.0107,
+      "step": 4194
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0386829603765503e-06,
+      "loss": 0.009,
+      "step": 4195
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.034949602419459e-06,
+      "loss": 0.0143,
+      "step": 4196
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0312192787216932e-06,
+      "loss": 0.0095,
+      "step": 4197
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0274919907043033e-06,
+      "loss": 0.0105,
+      "step": 4198
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.02376773978719e-06,
+      "loss": 0.0142,
+      "step": 4199
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0200465273890914e-06,
+      "loss": 0.0111,
+      "step": 4200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0163283549275915e-06,
+      "loss": 0.0145,
+      "step": 4201
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0126132238191167e-06,
+      "loss": 0.0076,
+      "step": 4202
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0089011354789357e-06,
+      "loss": 0.0145,
+      "step": 4203
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.005192091321152e-06,
+      "loss": 0.0122,
+      "step": 4204
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0014860927587177e-06,
+      "loss": 0.0053,
+      "step": 4205
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9977831412034165e-06,
+      "loss": 0.0112,
+      "step": 4206
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9940832380658772e-06,
+      "loss": 0.0092,
+      "step": 4207
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.990386384755565e-06,
+      "loss": 0.0148,
+      "step": 4208
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986692582680787e-06,
+      "loss": 0.0078,
+      "step": 4209
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.98300183324868e-06,
+      "loss": 0.0099,
+      "step": 4210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9793141378652283e-06,
+      "loss": 0.0068,
+      "step": 4211
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.97562949793524e-06,
+      "loss": 0.009,
+      "step": 4212
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.97194791486237e-06,
+      "loss": 0.0128,
+      "step": 4213
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.968269390049109e-06,
+      "loss": 0.0098,
+      "step": 4214
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9645939248967705e-06,
+      "loss": 0.0113,
+      "step": 4215
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9609215208055168e-06,
+      "loss": 0.0141,
+      "step": 4216
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.957252179174338e-06,
+      "loss": 0.0118,
+      "step": 4217
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9535859014010525e-06,
+      "loss": 0.008,
+      "step": 4218
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9499226888823206e-06,
+      "loss": 0.0085,
+      "step": 4219
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9462625430136326e-06,
+      "loss": 0.0065,
+      "step": 4220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9426054651893043e-06,
+      "loss": 0.01,
+      "step": 4221
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.938951456802488e-06,
+      "loss": 0.0078,
+      "step": 4222
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9353005192451714e-06,
+      "loss": 0.0079,
+      "step": 4223
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9316526539081592e-06,
+      "loss": 0.0087,
+      "step": 4224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9280078621810994e-06,
+      "loss": 0.0113,
+      "step": 4225
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.924366145452459e-06,
+      "loss": 0.0197,
+      "step": 4226
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.92072750510954e-06,
+      "loss": 0.009,
+      "step": 4227
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9170919425384695e-06,
+      "loss": 0.0153,
+      "step": 4228
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.913459459124203e-06,
+      "loss": 0.0163,
+      "step": 4229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.0103,
+      "step": 4230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9062037353000461e-06,
+      "loss": 0.0106,
+      "step": 4231
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9025804976541929e-06,
+      "loss": 0.0119,
+      "step": 4232
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8989603446932303e-06,
+      "loss": 0.011,
+      "step": 4233
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8953432777962445e-06,
+      "loss": 0.0132,
+      "step": 4234
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.891729298341145e-06,
+      "loss": 0.015,
+      "step": 4235
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8881184077046655e-06,
+      "loss": 0.0098,
+      "step": 4236
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.884510607262362e-06,
+      "loss": 0.0096,
+      "step": 4237
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.880905898388612e-06,
+      "loss": 0.0124,
+      "step": 4238
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.877304282456618e-06,
+      "loss": 0.0101,
+      "step": 4239
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8737057608384058e-06,
+      "loss": 0.0083,
+      "step": 4240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8701103349048176e-06,
+      "loss": 0.0066,
+      "step": 4241
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.866518006025524e-06,
+      "loss": 0.0089,
+      "step": 4242
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.862928775569005e-06,
+      "loss": 0.0107,
+      "step": 4243
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8593426449025676e-06,
+      "loss": 0.0122,
+      "step": 4244
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.855759615392335e-06,
+      "loss": 0.0171,
+      "step": 4245
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8521796884032517e-06,
+      "loss": 0.0074,
+      "step": 4246
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8486028652990796e-06,
+      "loss": 0.0099,
+      "step": 4247
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8450291474423999e-06,
+      "loss": 0.0111,
+      "step": 4248
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8414585361946048e-06,
+      "loss": 0.0084,
+      "step": 4249
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.837891032915905e-06,
+      "loss": 0.0085,
+      "step": 4250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8343266389653326e-06,
+      "loss": 0.0076,
+      "step": 4251
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8307653557007289e-06,
+      "loss": 0.0129,
+      "step": 4252
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8272071844787575e-06,
+      "loss": 0.0121,
+      "step": 4253
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8236521266548856e-06,
+      "loss": 0.008,
+      "step": 4254
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8201001835834065e-06,
+      "loss": 0.006,
+      "step": 4255
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8165513566174153e-06,
+      "loss": 0.0096,
+      "step": 4256
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8130056471088299e-06,
+      "loss": 0.0098,
+      "step": 4257
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8094630564083737e-06,
+      "loss": 0.013,
+      "step": 4258
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.80592358586559e-06,
+      "loss": 0.0056,
+      "step": 4259
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8023872368288209e-06,
+      "loss": 0.0152,
+      "step": 4260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7988540106452334e-06,
+      "loss": 0.009,
+      "step": 4261
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7953239086607932e-06,
+      "loss": 0.0099,
+      "step": 4262
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7917969322202832e-06,
+      "loss": 0.0133,
+      "step": 4263
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7882730826672922e-06,
+      "loss": 0.0059,
+      "step": 4264
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7847523613442218e-06,
+      "loss": 0.0118,
+      "step": 4265
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7812347695922749e-06,
+      "loss": 0.0121,
+      "step": 4266
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7777203087514717e-06,
+      "loss": 0.0131,
+      "step": 4267
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7742089801606278e-06,
+      "loss": 0.0137,
+      "step": 4268
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7707007851573743e-06,
+      "loss": 0.0085,
+      "step": 4269
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7671957250781503e-06,
+      "loss": 0.0083,
+      "step": 4270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7636938012581917e-06,
+      "loss": 0.008,
+      "step": 4271
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7601950150315462e-06,
+      "loss": 0.0102,
+      "step": 4272
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7566993677310684e-06,
+      "loss": 0.0068,
+      "step": 4273
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7532068606884067e-06,
+      "loss": 0.0114,
+      "step": 4274
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7497174952340246e-06,
+      "loss": 0.0141,
+      "step": 4275
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7462312726971842e-06,
+      "loss": 0.0073,
+      "step": 4276
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7427481944059477e-06,
+      "loss": 0.0094,
+      "step": 4277
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7392682616871836e-06,
+      "loss": 0.0074,
+      "step": 4278
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.735791475866564e-06,
+      "loss": 0.0089,
+      "step": 4279
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7323178382685512e-06,
+      "loss": 0.009,
+      "step": 4280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7288473502164217e-06,
+      "loss": 0.0212,
+      "step": 4281
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7253800130322472e-06,
+      "loss": 0.014,
+      "step": 4282
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.721915828036893e-06,
+      "loss": 0.0098,
+      "step": 4283
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7184547965500309e-06,
+      "loss": 0.0141,
+      "step": 4284
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.714996919890134e-06,
+      "loss": 0.012,
+      "step": 4285
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7115421993744607e-06,
+      "loss": 0.0102,
+      "step": 4286
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.708090636319083e-06,
+      "loss": 0.0074,
+      "step": 4287
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7046422320388556e-06,
+      "loss": 0.0122,
+      "step": 4288
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7011969878474398e-06,
+      "loss": 0.0053,
+      "step": 4289
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6977549050572906e-06,
+      "loss": 0.0067,
+      "step": 4290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6943159849796586e-06,
+      "loss": 0.0081,
+      "step": 4291
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6908802289245852e-06,
+      "loss": 0.0161,
+      "step": 4292
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6874476382009143e-06,
+      "loss": 0.0082,
+      "step": 4293
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6840182141162753e-06,
+      "loss": 0.0142,
+      "step": 4294
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.680591957977099e-06,
+      "loss": 0.0112,
+      "step": 4295
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6771688710886059e-06,
+      "loss": 0.011,
+      "step": 4296
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6737489547548091e-06,
+      "loss": 0.0041,
+      "step": 4297
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6703322102785168e-06,
+      "loss": 0.0113,
+      "step": 4298
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6669186389613235e-06,
+      "loss": 0.0143,
+      "step": 4299
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6635082421036163e-06,
+      "loss": 0.0139,
+      "step": 4300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6601010210045765e-06,
+      "loss": 0.0118,
+      "step": 4301
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6566969769621744e-06,
+      "loss": 0.0137,
+      "step": 4302
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6532961112731672e-06,
+      "loss": 0.0155,
+      "step": 4303
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6498984252331085e-06,
+      "loss": 0.0092,
+      "step": 4304
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6465039201363298e-06,
+      "loss": 0.01,
+      "step": 4305
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6431125972759566e-06,
+      "loss": 0.0081,
+      "step": 4306
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.639724457943903e-06,
+      "loss": 0.0089,
+      "step": 4307
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6363395034308704e-06,
+      "loss": 0.0103,
+      "step": 4308
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6329577350263459e-06,
+      "loss": 0.0104,
+      "step": 4309
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6295791540186034e-06,
+      "loss": 0.0106,
+      "step": 4310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6262037616947023e-06,
+      "loss": 0.0116,
+      "step": 4311
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.622831559340482e-06,
+      "loss": 0.0139,
+      "step": 4312
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6194625482405768e-06,
+      "loss": 0.0076,
+      "step": 4313
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.616096729678398e-06,
+      "loss": 0.0143,
+      "step": 4314
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6127341049361466e-06,
+      "loss": 0.0098,
+      "step": 4315
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.609374675294798e-06,
+      "loss": 0.0078,
+      "step": 4316
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6060184420341197e-06,
+      "loss": 0.0129,
+      "step": 4317
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6026654064326553e-06,
+      "loss": 0.0079,
+      "step": 4318
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5993155697677322e-06,
+      "loss": 0.011,
+      "step": 4319
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5959689333154593e-06,
+      "loss": 0.019,
+      "step": 4320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5926254983507317e-06,
+      "loss": 0.0107,
+      "step": 4321
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5892852661472136e-06,
+      "loss": 0.0076,
+      "step": 4322
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5859482379773584e-06,
+      "loss": 0.0112,
+      "step": 4323
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5826144151123934e-06,
+      "loss": 0.0095,
+      "step": 4324
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5792837988223285e-06,
+      "loss": 0.0193,
+      "step": 4325
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5759563903759545e-06,
+      "loss": 0.0092,
+      "step": 4326
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5726321910408304e-06,
+      "loss": 0.0104,
+      "step": 4327
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5693112020833012e-06,
+      "loss": 0.0097,
+      "step": 4328
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5659934247684893e-06,
+      "loss": 0.0089,
+      "step": 4329
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562678860360286e-06,
+      "loss": 0.0176,
+      "step": 4330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5593675101213656e-06,
+      "loss": 0.0186,
+      "step": 4331
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5560593753131802e-06,
+      "loss": 0.0099,
+      "step": 4332
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5527544571959452e-06,
+      "loss": 0.0114,
+      "step": 4333
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.549452757028662e-06,
+      "loss": 0.0132,
+      "step": 4334
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.546154276069105e-06,
+      "loss": 0.0128,
+      "step": 4335
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5428590155738133e-06,
+      "loss": 0.0094,
+      "step": 4336
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5395669767981093e-06,
+      "loss": 0.0155,
+      "step": 4337
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5362781609960853e-06,
+      "loss": 0.0122,
+      "step": 4338
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5329925694206006e-06,
+      "loss": 0.0111,
+      "step": 4339
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5297102033232924e-06,
+      "loss": 0.0189,
+      "step": 4340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.52643106395457e-06,
+      "loss": 0.0096,
+      "step": 4341
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5231551525636045e-06,
+      "loss": 0.01,
+      "step": 4342
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5198824703983494e-06,
+      "loss": 0.0117,
+      "step": 4343
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5166130187055173e-06,
+      "loss": 0.0124,
+      "step": 4344
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5133467987305972e-06,
+      "loss": 0.01,
+      "step": 4345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.510083811717844e-06,
+      "loss": 0.0088,
+      "step": 4346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5068240589102844e-06,
+      "loss": 0.0119,
+      "step": 4347
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5035675415497064e-06,
+      "loss": 0.0079,
+      "step": 4348
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.500314260876673e-06,
+      "loss": 0.0061,
+      "step": 4349
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4970642181305052e-06,
+      "loss": 0.0117,
+      "step": 4350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4938174145493e-06,
+      "loss": 0.0132,
+      "step": 4351
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4905738513699163e-06,
+      "loss": 0.0078,
+      "step": 4352
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4873335298279801e-06,
+      "loss": 0.0126,
+      "step": 4353
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4840964511578792e-06,
+      "loss": 0.0142,
+      "step": 4354
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4808626165927653e-06,
+      "loss": 0.0079,
+      "step": 4355
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4776320273645595e-06,
+      "loss": 0.0055,
+      "step": 4356
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.474404684703943e-06,
+      "loss": 0.009,
+      "step": 4357
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.471180589840363e-06,
+      "loss": 0.0103,
+      "step": 4358
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4679597440020276e-06,
+      "loss": 0.0108,
+      "step": 4359
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4647421484159075e-06,
+      "loss": 0.0095,
+      "step": 4360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4615278043077308e-06,
+      "loss": 0.011,
+      "step": 4361
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.458316712901995e-06,
+      "loss": 0.0138,
+      "step": 4362
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4551088754219522e-06,
+      "loss": 0.0094,
+      "step": 4363
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.45190429308962e-06,
+      "loss": 0.0188,
+      "step": 4364
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4487029671257747e-06,
+      "loss": 0.0097,
+      "step": 4365
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4455048987499453e-06,
+      "loss": 0.0089,
+      "step": 4366
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.442310089180431e-06,
+      "loss": 0.0113,
+      "step": 4367
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4391185396342787e-06,
+      "loss": 0.0074,
+      "step": 4368
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4359302513273e-06,
+      "loss": 0.0165,
+      "step": 4369
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4327452254740648e-06,
+      "loss": 0.0221,
+      "step": 4370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4295634632878986e-06,
+      "loss": 0.0113,
+      "step": 4371
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4263849659808781e-06,
+      "loss": 0.0092,
+      "step": 4372
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4232097347638485e-06,
+      "loss": 0.0112,
+      "step": 4373
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4200377708463953e-06,
+      "loss": 0.0102,
+      "step": 4374
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.416869075436873e-06,
+      "loss": 0.0106,
+      "step": 4375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4137036497423838e-06,
+      "loss": 0.0151,
+      "step": 4376
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4105414949687878e-06,
+      "loss": 0.0108,
+      "step": 4377
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4073826123206946e-06,
+      "loss": 0.0158,
+      "step": 4378
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4042270030014737e-06,
+      "loss": 0.0103,
+      "step": 4379
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4010746682132393e-06,
+      "loss": 0.0171,
+      "step": 4380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3979256091568638e-06,
+      "loss": 0.0112,
+      "step": 4381
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3947798270319757e-06,
+      "loss": 0.0106,
+      "step": 4382
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3916373230369428e-06,
+      "loss": 0.0104,
+      "step": 4383
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3884980983688966e-06,
+      "loss": 0.0105,
+      "step": 4384
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.385362154223714e-06,
+      "loss": 0.0184,
+      "step": 4385
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.382229491796019e-06,
+      "loss": 0.009,
+      "step": 4386
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3791001122791914e-06,
+      "loss": 0.0136,
+      "step": 4387
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.375974016865359e-06,
+      "loss": 0.0079,
+      "step": 4388
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3728512067453948e-06,
+      "loss": 0.0116,
+      "step": 4389
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3697316831089236e-06,
+      "loss": 0.0145,
+      "step": 4390
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3666154471443194e-06,
+      "loss": 0.0128,
+      "step": 4391
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3635025000386992e-06,
+      "loss": 0.011,
+      "step": 4392
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.360392842977931e-06,
+      "loss": 0.0138,
+      "step": 4393
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.35728647714663e-06,
+      "loss": 0.0188,
+      "step": 4394
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.354183403728152e-06,
+      "loss": 0.0192,
+      "step": 4395
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3510836239046044e-06,
+      "loss": 0.0078,
+      "step": 4396
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3479871388568399e-06,
+      "loss": 0.0084,
+      "step": 4397
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3448939497644508e-06,
+      "loss": 0.0165,
+      "step": 4398
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3418040578057812e-06,
+      "loss": 0.0129,
+      "step": 4399
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3387174641579105e-06,
+      "loss": 0.0144,
+      "step": 4400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.335634169996668e-06,
+      "loss": 0.0114,
+      "step": 4401
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3325541764966243e-06,
+      "loss": 0.0146,
+      "step": 4402
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3294774848310954e-06,
+      "loss": 0.0097,
+      "step": 4403
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3264040961721325e-06,
+      "loss": 0.0124,
+      "step": 4404
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3233340116905357e-06,
+      "loss": 0.0131,
+      "step": 4405
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3202672325558408e-06,
+      "loss": 0.0135,
+      "step": 4406
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3172037599363285e-06,
+      "loss": 0.0164,
+      "step": 4407
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3141435949990188e-06,
+      "loss": 0.0084,
+      "step": 4408
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3110867389096727e-06,
+      "loss": 0.0148,
+      "step": 4409
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3080331928327873e-06,
+      "loss": 0.0099,
+      "step": 4410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3049829579315987e-06,
+      "loss": 0.0127,
+      "step": 4411
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3019360353680866e-06,
+      "loss": 0.0113,
+      "step": 4412
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2988924263029657e-06,
+      "loss": 0.0116,
+      "step": 4413
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2958521318956874e-06,
+      "loss": 0.0088,
+      "step": 4414
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2928151533044452e-06,
+      "loss": 0.0153,
+      "step": 4415
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.289781491686164e-06,
+      "loss": 0.0085,
+      "step": 4416
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2867511481965046e-06,
+      "loss": 0.0114,
+      "step": 4417
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2837241239898669e-06,
+      "loss": 0.0159,
+      "step": 4418
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2807004202193874e-06,
+      "loss": 0.0134,
+      "step": 4419
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2776800380369347e-06,
+      "loss": 0.0116,
+      "step": 4420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2746629785931175e-06,
+      "loss": 0.0091,
+      "step": 4421
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2716492430372708e-06,
+      "loss": 0.0098,
+      "step": 4422
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2686388325174637e-06,
+      "loss": 0.0084,
+      "step": 4423
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.265631748180507e-06,
+      "loss": 0.0068,
+      "step": 4424
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2626279911719375e-06,
+      "loss": 0.0085,
+      "step": 4425
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2596275626360267e-06,
+      "loss": 0.0132,
+      "step": 4426
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2566304637157812e-06,
+      "loss": 0.0121,
+      "step": 4427
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.253636695552931e-06,
+      "loss": 0.0138,
+      "step": 4428
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.250646259287943e-06,
+      "loss": 0.0082,
+      "step": 4429
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2476591560600138e-06,
+      "loss": 0.0082,
+      "step": 4430
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2446753870070716e-06,
+      "loss": 0.0066,
+      "step": 4431
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.241694953265773e-06,
+      "loss": 0.0084,
+      "step": 4432
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2387178559715063e-06,
+      "loss": 0.0146,
+      "step": 4433
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2357440962583823e-06,
+      "loss": 0.0126,
+      "step": 4434
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.23277367525925e-06,
+      "loss": 0.0097,
+      "step": 4435
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2298065941056746e-06,
+      "loss": 0.0059,
+      "step": 4436
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2268428539279608e-06,
+      "loss": 0.007,
+      "step": 4437
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2238824558551365e-06,
+      "loss": 0.0095,
+      "step": 4438
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2209254010149507e-06,
+      "loss": 0.012,
+      "step": 4439
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2179716905338868e-06,
+      "loss": 0.0054,
+      "step": 4440
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2150213255371523e-06,
+      "loss": 0.0146,
+      "step": 4441
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.212074307148674e-06,
+      "loss": 0.0152,
+      "step": 4442
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2091306364911126e-06,
+      "loss": 0.0163,
+      "step": 4443
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.20619031468585e-06,
+      "loss": 0.0078,
+      "step": 4444
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.203253342852988e-06,
+      "loss": 0.0082,
+      "step": 4445
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2003197221113594e-06,
+      "loss": 0.0097,
+      "step": 4446
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1973894535785192e-06,
+      "loss": 0.0092,
+      "step": 4447
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1944625383707377e-06,
+      "loss": 0.0085,
+      "step": 4448
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.191538977603016e-06,
+      "loss": 0.0169,
+      "step": 4449
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1886187723890785e-06,
+      "loss": 0.0111,
+      "step": 4450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1857019238413615e-06,
+      "loss": 0.0101,
+      "step": 4451
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1827884330710326e-06,
+      "loss": 0.0106,
+      "step": 4452
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1798783011879766e-06,
+      "loss": 0.0075,
+      "step": 4453
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1769715293007955e-06,
+      "loss": 0.0071,
+      "step": 4454
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1740681185168179e-06,
+      "loss": 0.0128,
+      "step": 4455
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1711680699420857e-06,
+      "loss": 0.0158,
+      "step": 4456
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1682713846813631e-06,
+      "loss": 0.0121,
+      "step": 4457
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.165378063838133e-06,
+      "loss": 0.0106,
+      "step": 4458
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1624881085145979e-06,
+      "loss": 0.0099,
+      "step": 4459
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.159601519811674e-06,
+      "loss": 0.0121,
+      "step": 4460
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1567182988290015e-06,
+      "loss": 0.006,
+      "step": 4461
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.153838446664929e-06,
+      "loss": 0.011,
+      "step": 4462
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1509619644165282e-06,
+      "loss": 0.009,
+      "step": 4463
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1480888531795854e-06,
+      "loss": 0.0147,
+      "step": 4464
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1452191140486057e-06,
+      "loss": 0.0131,
+      "step": 4465
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1423527481168029e-06,
+      "loss": 0.0105,
+      "step": 4466
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.139489756476112e-06,
+      "loss": 0.0119,
+      "step": 4467
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1366301402171775e-06,
+      "loss": 0.01,
+      "step": 4468
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1337739004293612e-06,
+      "loss": 0.0124,
+      "step": 4469
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1309210382007395e-06,
+      "loss": 0.01,
+      "step": 4470
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1280715546181032e-06,
+      "loss": 0.0128,
+      "step": 4471
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1252254507669492e-06,
+      "loss": 0.0099,
+      "step": 4472
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1223827277314915e-06,
+      "loss": 0.0112,
+      "step": 4473
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.119543386594656e-06,
+      "loss": 0.0065,
+      "step": 4474
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.116707428438082e-06,
+      "loss": 0.0103,
+      "step": 4475
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1138748543421163e-06,
+      "loss": 0.0072,
+      "step": 4476
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1110456653858226e-06,
+      "loss": 0.0113,
+      "step": 4477
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1082198626469687e-06,
+      "loss": 0.0168,
+      "step": 4478
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1053974472020313e-06,
+      "loss": 0.0129,
+      "step": 4479
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1025784201262036e-06,
+      "loss": 0.0113,
+      "step": 4480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0997627824933833e-06,
+      "loss": 0.0144,
+      "step": 4481
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0969505353761788e-06,
+      "loss": 0.0087,
+      "step": 4482
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0941416798459082e-06,
+      "loss": 0.0108,
+      "step": 4483
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0913362169725938e-06,
+      "loss": 0.0106,
+      "step": 4484
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0885341478249645e-06,
+      "loss": 0.0054,
+      "step": 4485
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0857354734704627e-06,
+      "loss": 0.0146,
+      "step": 4486
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0829401949752328e-06,
+      "loss": 0.0104,
+      "step": 4487
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.080148313404127e-06,
+      "loss": 0.013,
+      "step": 4488
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0773598298207056e-06,
+      "loss": 0.0087,
+      "step": 4489
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0745747452872291e-06,
+      "loss": 0.0092,
+      "step": 4490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0717930608646653e-06,
+      "loss": 0.009,
+      "step": 4491
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0690147776126892e-06,
+      "loss": 0.0135,
+      "step": 4492
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0662398965896793e-06,
+      "loss": 0.0111,
+      "step": 4493
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0634684188527155e-06,
+      "loss": 0.0082,
+      "step": 4494
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.060700345457586e-06,
+      "loss": 0.0121,
+      "step": 4495
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0579356774587778e-06,
+      "loss": 0.0046,
+      "step": 4496
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0551744159094778e-06,
+      "loss": 0.0086,
+      "step": 4497
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0524165618615845e-06,
+      "loss": 0.0119,
+      "step": 4498
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0496621163656907e-06,
+      "loss": 0.0123,
+      "step": 4499
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0469110804710948e-06,
+      "loss": 0.0089,
+      "step": 4500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.044163455225793e-06,
+      "loss": 0.0147,
+      "step": 4501
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0414192416764868e-06,
+      "loss": 0.0135,
+      "step": 4502
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0386784408685713e-06,
+      "loss": 0.0171,
+      "step": 4503
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0359410538461468e-06,
+      "loss": 0.0093,
+      "step": 4504
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0332070816520124e-06,
+      "loss": 0.0079,
+      "step": 4505
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0304765253276673e-06,
+      "loss": 0.0047,
+      "step": 4506
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0277493859133036e-06,
+      "loss": 0.0099,
+      "step": 4507
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0250256644478196e-06,
+      "loss": 0.0103,
+      "step": 4508
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0223053619688073e-06,
+      "loss": 0.0156,
+      "step": 4509
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0195884795125544e-06,
+      "loss": 0.0071,
+      "step": 4510
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.016875018114053e-06,
+      "loss": 0.0132,
+      "step": 4511
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0141649788069808e-06,
+      "loss": 0.0101,
+      "step": 4512
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0114583626237217e-06,
+      "loss": 0.0137,
+      "step": 4513
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0087551705953523e-06,
+      "loss": 0.0132,
+      "step": 4514
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0060554037516468e-06,
+      "loss": 0.0096,
+      "step": 4515
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0033590631210665e-06,
+      "loss": 0.0089,
+      "step": 4516
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.000666149730779e-06,
+      "loss": 0.0117,
+      "step": 4517
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.979766646066368e-07,
+      "loss": 0.0117,
+      "step": 4518
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.952906087731916e-07,
+      "loss": 0.0128,
+      "step": 4519
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.926079832536873e-07,
+      "loss": 0.0061,
+      "step": 4520
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.899287890700627e-07,
+      "loss": 0.0105,
+      "step": 4521
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.872530272429447e-07,
+      "loss": 0.0118,
+      "step": 4522
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.845806987916596e-07,
+      "loss": 0.0083,
+      "step": 4523
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.81911804734218e-07,
+      "loss": 0.0094,
+      "step": 4524
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.792463460873292e-07,
+      "loss": 0.0108,
+      "step": 4525
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.765843238663886e-07,
+      "loss": 0.0081,
+      "step": 4526
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.739257390854872e-07,
+      "loss": 0.0152,
+      "step": 4527
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.71270592757404e-07,
+      "loss": 0.0076,
+      "step": 4528
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.68618885893604e-07,
+      "loss": 0.012,
+      "step": 4529
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.659706195042484e-07,
+      "loss": 0.0125,
+      "step": 4530
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.633257945981867e-07,
+      "loss": 0.0095,
+      "step": 4531
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.60684412182955e-07,
+      "loss": 0.0098,
+      "step": 4532
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.58046473264781e-07,
+      "loss": 0.007,
+      "step": 4533
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.55411978848576e-07,
+      "loss": 0.0106,
+      "step": 4534
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.52780929937942e-07,
+      "loss": 0.0097,
+      "step": 4535
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.501533275351693e-07,
+      "loss": 0.0041,
+      "step": 4536
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.475291726412339e-07,
+      "loss": 0.0137,
+      "step": 4537
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.449084662557984e-07,
+      "loss": 0.0124,
+      "step": 4538
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.422912093772152e-07,
+      "loss": 0.0119,
+      "step": 4539
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.396774030025169e-07,
+      "loss": 0.0119,
+      "step": 4540
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.370670481274224e-07,
+      "loss": 0.0123,
+      "step": 4541
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.344601457463409e-07,
+      "loss": 0.0098,
+      "step": 4542
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.318566968523613e-07,
+      "loss": 0.0151,
+      "step": 4543
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.292567024372589e-07,
+      "loss": 0.0166,
+      "step": 4544
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.266601634914962e-07,
+      "loss": 0.0075,
+      "step": 4545
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.240670810042129e-07,
+      "loss": 0.0127,
+      "step": 4546
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.214774559632333e-07,
+      "loss": 0.0141,
+      "step": 4547
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.188912893550694e-07,
+      "loss": 0.0076,
+      "step": 4548
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.163085821649109e-07,
+      "loss": 0.0091,
+      "step": 4549
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.13729335376633e-07,
+      "loss": 0.0123,
+      "step": 4550
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.11153549972793e-07,
+      "loss": 0.017,
+      "step": 4551
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.08581226934625e-07,
+      "loss": 0.0056,
+      "step": 4552
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.060123672420451e-07,
+      "loss": 0.0116,
+      "step": 4553
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.034469718736538e-07,
+      "loss": 0.0136,
+      "step": 4554
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.008850418067306e-07,
+      "loss": 0.0106,
+      "step": 4555
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.983265780172346e-07,
+      "loss": 0.0139,
+      "step": 4556
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.957715814798018e-07,
+      "loss": 0.0152,
+      "step": 4557
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.932200531677537e-07,
+      "loss": 0.0055,
+      "step": 4558
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.906719940530805e-07,
+      "loss": 0.0066,
+      "step": 4559
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.881274051064603e-07,
+      "loss": 0.0124,
+      "step": 4560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.855862872972454e-07,
+      "loss": 0.0114,
+      "step": 4561
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.83048641593468e-07,
+      "loss": 0.0064,
+      "step": 4562
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.805144689618317e-07,
+      "loss": 0.0078,
+      "step": 4563
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.779837703677252e-07,
+      "loss": 0.0084,
+      "step": 4564
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.754565467752063e-07,
+      "loss": 0.0108,
+      "step": 4565
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.729327991470138e-07,
+      "loss": 0.0121,
+      "step": 4566
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.704125284445619e-07,
+      "loss": 0.0109,
+      "step": 4567
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.678957356279372e-07,
+      "loss": 0.0111,
+      "step": 4568
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.653824216559048e-07,
+      "loss": 0.0092,
+      "step": 4569
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.628725874859045e-07,
+      "loss": 0.007,
+      "step": 4570
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.60366234074046e-07,
+      "loss": 0.0069,
+      "step": 4571
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.578633623751187e-07,
+      "loss": 0.0143,
+      "step": 4572
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.553639733425834e-07,
+      "loss": 0.0123,
+      "step": 4573
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.528680679285716e-07,
+      "loss": 0.0061,
+      "step": 4574
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.50375647083892e-07,
+      "loss": 0.0147,
+      "step": 4575
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.478867117580258e-07,
+      "loss": 0.0085,
+      "step": 4576
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.454012628991204e-07,
+      "loss": 0.0101,
+      "step": 4577
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.429193014540015e-07,
+      "loss": 0.0092,
+      "step": 4578
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.404408283681653e-07,
+      "loss": 0.0084,
+      "step": 4579
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.379658445857752e-07,
+      "loss": 0.0128,
+      "step": 4580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.354943510496682e-07,
+      "loss": 0.0116,
+      "step": 4581
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.330263487013534e-07,
+      "loss": 0.0115,
+      "step": 4582
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.305618384810088e-07,
+      "loss": 0.0149,
+      "step": 4583
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.281008213274788e-07,
+      "loss": 0.0083,
+      "step": 4584
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.256432981782802e-07,
+      "loss": 0.0129,
+      "step": 4585
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.231892699695976e-07,
+      "loss": 0.0093,
+      "step": 4586
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.207387376362852e-07,
+      "loss": 0.007,
+      "step": 4587
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.182917021118664e-07,
+      "loss": 0.0086,
+      "step": 4588
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.158481643285332e-07,
+      "loss": 0.0136,
+      "step": 4589
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.134081252171389e-07,
+      "loss": 0.0079,
+      "step": 4590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.109715857072087e-07,
+      "loss": 0.0095,
+      "step": 4591
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.085385467269346e-07,
+      "loss": 0.0068,
+      "step": 4592
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.061090092031765e-07,
+      "loss": 0.0069,
+      "step": 4593
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.036829740614571e-07,
+      "loss": 0.0097,
+      "step": 4594
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.012604422259674e-07,
+      "loss": 0.0065,
+      "step": 4595
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.988414146195633e-07,
+      "loss": 0.0159,
+      "step": 4596
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.964258921637613e-07,
+      "loss": 0.0071,
+      "step": 4597
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.940138757787507e-07,
+      "loss": 0.0064,
+      "step": 4598
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.916053663833789e-07,
+      "loss": 0.0063,
+      "step": 4599
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.892003648951607e-07,
+      "loss": 0.0098,
+      "step": 4600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.867988722302744e-07,
+      "loss": 0.0122,
+      "step": 4601
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.844008893035592e-07,
+      "loss": 0.011,
+      "step": 4602
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82006417028518e-07,
+      "loss": 0.0109,
+      "step": 4603
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.796154563173164e-07,
+      "loss": 0.0079,
+      "step": 4604
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.772280080807848e-07,
+      "loss": 0.0106,
+      "step": 4605
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.748440732284135e-07,
+      "loss": 0.0149,
+      "step": 4606
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72463652668356e-07,
+      "loss": 0.0124,
+      "step": 4607
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700867473074225e-07,
+      "loss": 0.0124,
+      "step": 4608
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.677133580510876e-07,
+      "loss": 0.0082,
+      "step": 4609
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.653434858034858e-07,
+      "loss": 0.0125,
+      "step": 4610
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.629771314674139e-07,
+      "loss": 0.0114,
+      "step": 4611
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.606142959443264e-07,
+      "loss": 0.0085,
+      "step": 4612
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.582549801343353e-07,
+      "loss": 0.0054,
+      "step": 4613
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.558991849362174e-07,
+      "loss": 0.0129,
+      "step": 4614
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.535469112474015e-07,
+      "loss": 0.0104,
+      "step": 4615
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.511981599639806e-07,
+      "loss": 0.0088,
+      "step": 4616
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.488529319807036e-07,
+      "loss": 0.0159,
+      "step": 4617
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.46511228190977e-07,
+      "loss": 0.0056,
+      "step": 4618
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.441730494868649e-07,
+      "loss": 0.0084,
+      "step": 4619
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.418383967590892e-07,
+      "loss": 0.0076,
+      "step": 4620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.395072708970263e-07,
+      "loss": 0.006,
+      "step": 4621
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.371796727887115e-07,
+      "loss": 0.0091,
+      "step": 4622
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.348556033208376e-07,
+      "loss": 0.0145,
+      "step": 4623
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.325350633787487e-07,
+      "loss": 0.0089,
+      "step": 4624
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.302180538464466e-07,
+      "loss": 0.0105,
+      "step": 4625
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.279045756065917e-07,
+      "loss": 0.0107,
+      "step": 4626
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.255946295404914e-07,
+      "loss": 0.013,
+      "step": 4627
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.232882165281141e-07,
+      "loss": 0.0123,
+      "step": 4628
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.209853374480813e-07,
+      "loss": 0.0112,
+      "step": 4629
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.186859931776635e-07,
+      "loss": 0.0115,
+      "step": 4630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.163901845927911e-07,
+      "loss": 0.0092,
+      "step": 4631
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.140979125680458e-07,
+      "loss": 0.0126,
+      "step": 4632
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.118091779766578e-07,
+      "loss": 0.0062,
+      "step": 4633
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.095239816905142e-07,
+      "loss": 0.0078,
+      "step": 4634
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.07242324580154e-07,
+      "loss": 0.0059,
+      "step": 4635
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.049642075147645e-07,
+      "loss": 0.0109,
+      "step": 4636
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.026896313621889e-07,
+      "loss": 0.0077,
+      "step": 4637
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.004185969889188e-07,
+      "loss": 0.01,
+      "step": 4638
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.981511052600964e-07,
+      "loss": 0.0143,
+      "step": 4639
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.958871570395165e-07,
+      "loss": 0.0117,
+      "step": 4640
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.936267531896212e-07,
+      "loss": 0.0116,
+      "step": 4641
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.913698945715031e-07,
+      "loss": 0.0115,
+      "step": 4642
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.891165820449075e-07,
+      "loss": 0.0159,
+      "step": 4643
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.868668164682257e-07,
+      "loss": 0.0079,
+      "step": 4644
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.846205986984955e-07,
+      "loss": 0.0099,
+      "step": 4645
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.823779295914112e-07,
+      "loss": 0.0077,
+      "step": 4646
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.80138810001304e-07,
+      "loss": 0.0109,
+      "step": 4647
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.779032407811636e-07,
+      "loss": 0.0081,
+      "step": 4648
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.756712227826201e-07,
+      "loss": 0.0094,
+      "step": 4649
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.734427568559543e-07,
+      "loss": 0.007,
+      "step": 4650
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.712178438500949e-07,
+      "loss": 0.0076,
+      "step": 4651
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.689964846126129e-07,
+      "loss": 0.0093,
+      "step": 4652
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.667786799897269e-07,
+      "loss": 0.0155,
+      "step": 4653
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.645644308263021e-07,
+      "loss": 0.0095,
+      "step": 4654
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.623537379658507e-07,
+      "loss": 0.0129,
+      "step": 4655
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.601466022505287e-07,
+      "loss": 0.0057,
+      "step": 4656
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.579430245211383e-07,
+      "loss": 0.0094,
+      "step": 4657
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.557430056171221e-07,
+      "loss": 0.0111,
+      "step": 4658
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.535465463765711e-07,
+      "loss": 0.006,
+      "step": 4659
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.513536476362193e-07,
+      "loss": 0.0124,
+      "step": 4660
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.491643102314438e-07,
+      "loss": 0.0123,
+      "step": 4661
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.469785349962654e-07,
+      "loss": 0.0136,
+      "step": 4662
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.447963227633503e-07,
+      "loss": 0.0142,
+      "step": 4663
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.426176743640034e-07,
+      "loss": 0.0073,
+      "step": 4664
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.404425906281708e-07,
+      "loss": 0.0125,
+      "step": 4665
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.382710723844454e-07,
+      "loss": 0.019,
+      "step": 4666
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.36103120460061e-07,
+      "loss": 0.0076,
+      "step": 4667
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.339387356808912e-07,
+      "loss": 0.0144,
+      "step": 4668
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.31777918871449e-07,
+      "loss": 0.0047,
+      "step": 4669
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.296206708548935e-07,
+      "loss": 0.0163,
+      "step": 4670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.274669924530186e-07,
+      "loss": 0.0071,
+      "step": 4671
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.253168844862612e-07,
+      "loss": 0.0099,
+      "step": 4672
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.231703477736994e-07,
+      "loss": 0.0095,
+      "step": 4673
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.210273831330482e-07,
+      "loss": 0.0104,
+      "step": 4674
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.188879913806623e-07,
+      "loss": 0.0131,
+      "step": 4675
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.167521733315374e-07,
+      "loss": 0.0116,
+      "step": 4676
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.146199297993039e-07,
+      "loss": 0.0105,
+      "step": 4677
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.124912615962341e-07,
+      "loss": 0.0072,
+      "step": 4678
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.103661695332364e-07,
+      "loss": 0.0089,
+      "step": 4679
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.082446544198606e-07,
+      "loss": 0.011,
+      "step": 4680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.061267170642859e-07,
+      "loss": 0.016,
+      "step": 4681
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.040123582733382e-07,
+      "loss": 0.0094,
+      "step": 4682
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.019015788524718e-07,
+      "loss": 0.0124,
+      "step": 4683
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.997943796057814e-07,
+      "loss": 0.0088,
+      "step": 4684
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.976907613360006e-07,
+      "loss": 0.0076,
+      "step": 4685
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.955907248444926e-07,
+      "loss": 0.0176,
+      "step": 4686
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.934942709312597e-07,
+      "loss": 0.0107,
+      "step": 4687
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.914014003949408e-07,
+      "loss": 0.017,
+      "step": 4688
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.893121140328062e-07,
+      "loss": 0.0095,
+      "step": 4689
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.872264126407623e-07,
+      "loss": 0.01,
+      "step": 4690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.851442970133525e-07,
+      "loss": 0.0112,
+      "step": 4691
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.830657679437491e-07,
+      "loss": 0.0107,
+      "step": 4692
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.809908262237617e-07,
+      "loss": 0.0108,
+      "step": 4693
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.789194726438341e-07,
+      "loss": 0.008,
+      "step": 4694
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.76851707993038e-07,
+      "loss": 0.0074,
+      "step": 4695
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.747875330590857e-07,
+      "loss": 0.006,
+      "step": 4696
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.727269486283138e-07,
+      "loss": 0.0091,
+      "step": 4697
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.706699554856965e-07,
+      "loss": 0.0086,
+      "step": 4698
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.686165544148392e-07,
+      "loss": 0.0114,
+      "step": 4699
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.665667461979785e-07,
+      "loss": 0.0109,
+      "step": 4700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.645205316159796e-07,
+      "loss": 0.0126,
+      "step": 4701
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.624779114483447e-07,
+      "loss": 0.0092,
+      "step": 4702
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.604388864732002e-07,
+      "loss": 0.0066,
+      "step": 4703
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.584034574673058e-07,
+      "loss": 0.0115,
+      "step": 4704
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.563716252060525e-07,
+      "loss": 0.0085,
+      "step": 4705
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.54343390463461e-07,
+      "loss": 0.0116,
+      "step": 4706
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.523187540121777e-07,
+      "loss": 0.0096,
+      "step": 4707
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.502977166234857e-07,
+      "loss": 0.0068,
+      "step": 4708
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.48280279067287e-07,
+      "loss": 0.0051,
+      "step": 4709
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.462664421121199e-07,
+      "loss": 0.0079,
+      "step": 4710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.442562065251489e-07,
+      "loss": 0.0091,
+      "step": 4711
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.422495730721688e-07,
+      "loss": 0.0098,
+      "step": 4712
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.402465425175984e-07,
+      "loss": 0.0121,
+      "step": 4713
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.382471156244829e-07,
+      "loss": 0.0077,
+      "step": 4714
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.362512931544994e-07,
+      "loss": 0.0075,
+      "step": 4715
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.3425907586795e-07,
+      "loss": 0.0068,
+      "step": 4716
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.322704645237631e-07,
+      "loss": 0.0128,
+      "step": 4717
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.302854598794938e-07,
+      "loss": 0.0127,
+      "step": 4718
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.283040626913228e-07,
+      "loss": 0.0094,
+      "step": 4719
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.263262737140562e-07,
+      "loss": 0.0115,
+      "step": 4720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.24352093701126e-07,
+      "loss": 0.0123,
+      "step": 4721
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.223815234045903e-07,
+      "loss": 0.0139,
+      "step": 4722
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.204145635751301e-07,
+      "loss": 0.0102,
+      "step": 4723
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.184512149620547e-07,
+      "loss": 0.0123,
+      "step": 4724
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.164914783132913e-07,
+      "loss": 0.0062,
+      "step": 4725
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.145353543753994e-07,
+      "loss": 0.0123,
+      "step": 4726
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.125828438935543e-07,
+      "loss": 0.0076,
+      "step": 4727
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.106339476115596e-07,
+      "loss": 0.0103,
+      "step": 4728
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.086886662718405e-07,
+      "loss": 0.0077,
+      "step": 4729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.067470006154473e-07,
+      "loss": 0.0109,
+      "step": 4730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.048089513820487e-07,
+      "loss": 0.0094,
+      "step": 4731
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.028745193099405e-07,
+      "loss": 0.0063,
+      "step": 4732
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.009437051360344e-07,
+      "loss": 0.0122,
+      "step": 4733
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.990165095958699e-07,
+      "loss": 0.0133,
+      "step": 4734
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.970929334236063e-07,
+      "loss": 0.0087,
+      "step": 4735
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.95172977352023e-07,
+      "loss": 0.0121,
+      "step": 4736
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.932566421125207e-07,
+      "loss": 0.0089,
+      "step": 4737
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.913439284351207e-07,
+      "loss": 0.0133,
+      "step": 4738
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.0055,
+      "step": 4739
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.875293686798144e-07,
+      "loss": 0.0106,
+      "step": 4740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.856275240550546e-07,
+      "loss": 0.0085,
+      "step": 4741
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.837293038986812e-07,
+      "loss": 0.0054,
+      "step": 4742
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.818347089338193e-07,
+      "loss": 0.0082,
+      "step": 4743
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.799437398822082e-07,
+      "loss": 0.0144,
+      "step": 4744
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.780563974642038e-07,
+      "loss": 0.0065,
+      "step": 4745
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7617268239878443e-07,
+      "loss": 0.0101,
+      "step": 4746
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.742925954035449e-07,
+      "loss": 0.0148,
+      "step": 4747
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7241613719469784e-07,
+      "loss": 0.0149,
+      "step": 4748
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7054330848707366e-07,
+      "loss": 0.0094,
+      "step": 4749
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6867410999412165e-07,
+      "loss": 0.0106,
+      "step": 4750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.668085424279045e-07,
+      "loss": 0.0143,
+      "step": 4751
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.649466064991059e-07,
+      "loss": 0.0043,
+      "step": 4752
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.63088302917023e-07,
+      "loss": 0.0053,
+      "step": 4753
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.612336323895694e-07,
+      "loss": 0.0058,
+      "step": 4754
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.593825956232767e-07,
+      "loss": 0.0107,
+      "step": 4755
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.57535193323293e-07,
+      "loss": 0.0079,
+      "step": 4756
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5569142619337647e-07,
+      "loss": 0.012,
+      "step": 4757
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.538512949359075e-07,
+      "loss": 0.0071,
+      "step": 4758
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.520148002518754e-07,
+      "loss": 0.0099,
+      "step": 4759
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.50181942840886e-07,
+      "loss": 0.0106,
+      "step": 4760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.483527234011631e-07,
+      "loss": 0.0045,
+      "step": 4761
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4652714262954143e-07,
+      "loss": 0.0046,
+      "step": 4762
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.44705201221467e-07,
+      "loss": 0.0101,
+      "step": 4763
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.428868998710067e-07,
+      "loss": 0.0121,
+      "step": 4764
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.410722392708311e-07,
+      "loss": 0.0093,
+      "step": 4765
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.392612201122326e-07,
+      "loss": 0.0066,
+      "step": 4766
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.374538430851116e-07,
+      "loss": 0.0087,
+      "step": 4767
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.356501088779841e-07,
+      "loss": 0.0139,
+      "step": 4768
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.338500181779759e-07,
+      "loss": 0.0078,
+      "step": 4769
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.32053571670823e-07,
+      "loss": 0.0083,
+      "step": 4770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3026077004087784e-07,
+      "loss": 0.0071,
+      "step": 4771
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2847161397110204e-07,
+      "loss": 0.0062,
+      "step": 4772
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.266861041430681e-07,
+      "loss": 0.0069,
+      "step": 4773
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2490424123696195e-07,
+      "loss": 0.0075,
+      "step": 4774
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2312602593157725e-07,
+      "loss": 0.0075,
+      "step": 4775
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.213514589043177e-07,
+      "loss": 0.0064,
+      "step": 4776
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1958054083119926e-07,
+      "loss": 0.0118,
+      "step": 4777
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1781327238684775e-07,
+      "loss": 0.013,
+      "step": 4778
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1604965424449917e-07,
+      "loss": 0.0189,
+      "step": 4779
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1428968707599713e-07,
+      "loss": 0.009,
+      "step": 4780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.125333715517954e-07,
+      "loss": 0.0079,
+      "step": 4781
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.107807083409554e-07,
+      "loss": 0.0088,
+      "step": 4782
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0903169811114864e-07,
+      "loss": 0.0103,
+      "step": 4783
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0728634152865545e-07,
+      "loss": 0.0108,
+      "step": 4784
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0554463925836285e-07,
+      "loss": 0.0084,
+      "step": 4785
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0380659196376794e-07,
+      "loss": 0.0125,
+      "step": 4786
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0207220030697436e-07,
+      "loss": 0.012,
+      "step": 4787
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.003414649486892e-07,
+      "loss": 0.008,
+      "step": 4788
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9861438654823283e-07,
+      "loss": 0.0126,
+      "step": 4789
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9689096576353e-07,
+      "loss": 0.0116,
+      "step": 4790
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9517120325111126e-07,
+      "loss": 0.0121,
+      "step": 4791
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9345509966611686e-07,
+      "loss": 0.0128,
+      "step": 4792
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.917426556622883e-07,
+      "loss": 0.0149,
+      "step": 4793
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9003387189197606e-07,
+      "loss": 0.007,
+      "step": 4794
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8832874900613495e-07,
+      "loss": 0.0067,
+      "step": 4795
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8662728765432646e-07,
+      "loss": 0.0083,
+      "step": 4796
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8492948848471655e-07,
+      "loss": 0.0094,
+      "step": 4797
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.832353521440768e-07,
+      "loss": 0.0081,
+      "step": 4798
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8154487927778094e-07,
+      "loss": 0.0131,
+      "step": 4799
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.798580705298116e-07,
+      "loss": 0.0073,
+      "step": 4800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.781749265427503e-07,
+      "loss": 0.0118,
+      "step": 4801
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.764954479577865e-07,
+      "loss": 0.0075,
+      "step": 4802
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.748196354147127e-07,
+      "loss": 0.0074,
+      "step": 4803
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.731474895519216e-07,
+      "loss": 0.0132,
+      "step": 4804
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.714790110064137e-07,
+      "loss": 0.0079,
+      "step": 4805
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6981420041379057e-07,
+      "loss": 0.0035,
+      "step": 4806
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.681530584082538e-07,
+      "loss": 0.0091,
+      "step": 4807
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6649558562261377e-07,
+      "loss": 0.0101,
+      "step": 4808
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.648417826882744e-07,
+      "loss": 0.0052,
+      "step": 4809
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.631916502352495e-07,
+      "loss": 0.0136,
+      "step": 4810
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.615451888921517e-07,
+      "loss": 0.0104,
+      "step": 4811
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5990239928619497e-07,
+      "loss": 0.0088,
+      "step": 4812
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5826328204319193e-07,
+      "loss": 0.0072,
+      "step": 4813
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5662783778756205e-07,
+      "loss": 0.0088,
+      "step": 4814
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.549960671423203e-07,
+      "loss": 0.01,
+      "step": 4815
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.533679707290838e-07,
+      "loss": 0.0123,
+      "step": 4816
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.517435491680721e-07,
+      "loss": 0.009,
+      "step": 4817
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5012280307810344e-07,
+      "loss": 0.0137,
+      "step": 4818
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.48505733076594e-07,
+      "loss": 0.0102,
+      "step": 4819
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4689233977956203e-07,
+      "loss": 0.0093,
+      "step": 4820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.452826238016238e-07,
+      "loss": 0.0053,
+      "step": 4821
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.436765857559965e-07,
+      "loss": 0.0116,
+      "step": 4822
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.420742262544929e-07,
+      "loss": 0.0079,
+      "step": 4823
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.404755459075293e-07,
+      "loss": 0.0091,
+      "step": 4824
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.388805453241162e-07,
+      "loss": 0.0099,
+      "step": 4825
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.37289225111862e-07,
+      "loss": 0.0085,
+      "step": 4826
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.357015858769763e-07,
+      "loss": 0.0055,
+      "step": 4827
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.341176282242653e-07,
+      "loss": 0.0127,
+      "step": 4828
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3253735275713185e-07,
+      "loss": 0.0126,
+      "step": 4829
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3096076007757773e-07,
+      "loss": 0.0087,
+      "step": 4830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2938785078620025e-07,
+      "loss": 0.0099,
+      "step": 4831
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.278186254821925e-07,
+      "loss": 0.0079,
+      "step": 4832
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2625308476334717e-07,
+      "loss": 0.0069,
+      "step": 4833
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.246912292260507e-07,
+      "loss": 0.0096,
+      "step": 4834
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2313305946528593e-07,
+      "loss": 0.019,
+      "step": 4835
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2157857607463596e-07,
+      "loss": 0.0064,
+      "step": 4836
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.200277796462736e-07,
+      "loss": 0.0109,
+      "step": 4837
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.184806707709698e-07,
+      "loss": 0.0115,
+      "step": 4838
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.169372500380885e-07,
+      "loss": 0.0099,
+      "step": 4839
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1539751803559394e-07,
+      "loss": 0.0089,
+      "step": 4840
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.138614753500402e-07,
+      "loss": 0.0065,
+      "step": 4841
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.123291225665792e-07,
+      "loss": 0.0093,
+      "step": 4842
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1080046026895406e-07,
+      "loss": 0.0174,
+      "step": 4843
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.092754890395033e-07,
+      "loss": 0.007,
+      "step": 4844
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.077542094591601e-07,
+      "loss": 0.0078,
+      "step": 4845
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.062366221074509e-07,
+      "loss": 0.0057,
+      "step": 4846
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0472272756249535e-07,
+      "loss": 0.01,
+      "step": 4847
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0321252640100883e-07,
+      "loss": 0.0074,
+      "step": 4848
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0170601919829457e-07,
+      "loss": 0.0103,
+      "step": 4849
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.002032065282512e-07,
+      "loss": 0.0067,
+      "step": 4850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9870408896337075e-07,
+      "loss": 0.0065,
+      "step": 4851
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.972086670747376e-07,
+      "loss": 0.0109,
+      "step": 4852
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9571694143202934e-07,
+      "loss": 0.0075,
+      "step": 4853
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.942289126035114e-07,
+      "loss": 0.0092,
+      "step": 4854
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9274458115604365e-07,
+      "loss": 0.0112,
+      "step": 4855
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9126394765507823e-07,
+      "loss": 0.009,
+      "step": 4856
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8978701266465624e-07,
+      "loss": 0.0102,
+      "step": 4857
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8831377674741203e-07,
+      "loss": 0.0065,
+      "step": 4858
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8684424046456996e-07,
+      "loss": 0.0152,
+      "step": 4859
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.853784043759444e-07,
+      "loss": 0.0109,
+      "step": 4860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.839162690399411e-07,
+      "loss": 0.0156,
+      "step": 4861
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.824578350135576e-07,
+      "loss": 0.0094,
+      "step": 4862
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.810031028523752e-07,
+      "loss": 0.0086,
+      "step": 4863
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7955207311057167e-07,
+      "loss": 0.0078,
+      "step": 4864
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7810474634091254e-07,
+      "loss": 0.0089,
+      "step": 4865
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7666112309475026e-07,
+      "loss": 0.009,
+      "step": 4866
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.752212039220303e-07,
+      "loss": 0.0066,
+      "step": 4867
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.737849893712841e-07,
+      "loss": 0.0081,
+      "step": 4868
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7235247998963155e-07,
+      "loss": 0.0118,
+      "step": 4869
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7092367632278513e-07,
+      "loss": 0.0096,
+      "step": 4870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.694985789150406e-07,
+      "loss": 0.0103,
+      "step": 4871
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6807718830928473e-07,
+      "loss": 0.0104,
+      "step": 4872
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.666595050469911e-07,
+      "loss": 0.0126,
+      "step": 4873
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6524552966822324e-07,
+      "loss": 0.0087,
+      "step": 4874
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.63835262711627e-07,
+      "loss": 0.0112,
+      "step": 4875
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.624287047144425e-07,
+      "loss": 0.0132,
+      "step": 4876
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.610258562124901e-07,
+      "loss": 0.0125,
+      "step": 4877
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5962671774018234e-07,
+      "loss": 0.0118,
+      "step": 4878
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5823128983051503e-07,
+      "loss": 0.0144,
+      "step": 4879
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.568395730150741e-07,
+      "loss": 0.0085,
+      "step": 4880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.554515678240266e-07,
+      "loss": 0.0099,
+      "step": 4881
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.540672747861295e-07,
+      "loss": 0.0082,
+      "step": 4882
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.526866944287254e-07,
+      "loss": 0.0091,
+      "step": 4883
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5130982727774036e-07,
+      "loss": 0.0102,
+      "step": 4884
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4993667385768807e-07,
+      "loss": 0.0079,
+      "step": 4885
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4856723469166787e-07,
+      "loss": 0.0127,
+      "step": 4886
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.472015103013614e-07,
+      "loss": 0.0083,
+      "step": 4887
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.458395012070369e-07,
+      "loss": 0.0143,
+      "step": 4888
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4448120792754826e-07,
+      "loss": 0.0056,
+      "step": 4889
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.431266309803315e-07,
+      "loss": 0.0119,
+      "step": 4890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4177577088140834e-07,
+      "loss": 0.0131,
+      "step": 4891
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4042862814538605e-07,
+      "loss": 0.0125,
+      "step": 4892
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3908520328545293e-07,
+      "loss": 0.0092,
+      "step": 4893
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3774549681338188e-07,
+      "loss": 0.0086,
+      "step": 4894
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.364095092395291e-07,
+      "loss": 0.0112,
+      "step": 4895
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3507724107283526e-07,
+      "loss": 0.0115,
+      "step": 4896
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3374869282082435e-07,
+      "loss": 0.0081,
+      "step": 4897
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3242386498960267e-07,
+      "loss": 0.0077,
+      "step": 4898
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3110275808385764e-07,
+      "loss": 0.0073,
+      "step": 4899
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2978537260685995e-07,
+      "loss": 0.0094,
+      "step": 4900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.284717090604638e-07,
+      "loss": 0.0084,
+      "step": 4901
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2716176794510547e-07,
+      "loss": 0.0138,
+      "step": 4902
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2585554975980252e-07,
+      "loss": 0.0138,
+      "step": 4903
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2455305500215575e-07,
+      "loss": 0.0132,
+      "step": 4904
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2325428416834382e-07,
+      "loss": 0.0098,
+      "step": 4905
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2195923775312878e-07,
+      "loss": 0.0099,
+      "step": 4906
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2066791624985594e-07,
+      "loss": 0.01,
+      "step": 4907
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1938032015044964e-07,
+      "loss": 0.0048,
+      "step": 4908
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1809644994541634e-07,
+      "loss": 0.0091,
+      "step": 4909
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.168163061238393e-07,
+      "loss": 0.0155,
+      "step": 4910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.155398891733873e-07,
+      "loss": 0.0144,
+      "step": 4911
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1426719958030585e-07,
+      "loss": 0.0042,
+      "step": 4912
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1299823782942263e-07,
+      "loss": 0.0087,
+      "step": 4913
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1173300440414324e-07,
+      "loss": 0.007,
+      "step": 4914
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1047149978645653e-07,
+      "loss": 0.0105,
+      "step": 4915
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.092137244569259e-07,
+      "loss": 0.0093,
+      "step": 4916
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0795967889469916e-07,
+      "loss": 0.0125,
+      "step": 4917
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.067093635774975e-07,
+      "loss": 0.0074,
+      "step": 4918
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.054627789816266e-07,
+      "loss": 0.0145,
+      "step": 4919
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.042199255819688e-07,
+      "loss": 0.0104,
+      "step": 4920
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0298080385198426e-07,
+      "loss": 0.0077,
+      "step": 4921
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.017454142637121e-07,
+      "loss": 0.0077,
+      "step": 4922
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0051375728777133e-07,
+      "loss": 0.0091,
+      "step": 4923
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.992858333933545e-07,
+      "loss": 0.0106,
+      "step": 4924
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.980616430482385e-07,
+      "loss": 0.0097,
+      "step": 4925
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9684118671877251e-07,
+      "loss": 0.018,
+      "step": 4926
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9562446486988573e-07,
+      "loss": 0.0116,
+      "step": 4927
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9441147796508408e-07,
+      "loss": 0.0091,
+      "step": 4928
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9320222646645126e-07,
+      "loss": 0.0092,
+      "step": 4929
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9199671083464766e-07,
+      "loss": 0.0162,
+      "step": 4930
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9079493152891037e-07,
+      "loss": 0.0082,
+      "step": 4931
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8959688900705319e-07,
+      "loss": 0.0079,
+      "step": 4932
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.884025837254655e-07,
+      "loss": 0.0047,
+      "step": 4933
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8721201613911555e-07,
+      "loss": 0.0108,
+      "step": 4934
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8602518670154502e-07,
+      "loss": 0.0097,
+      "step": 4935
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8484209586487335e-07,
+      "loss": 0.0064,
+      "step": 4936
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8366274407979444e-07,
+      "loss": 0.0143,
+      "step": 4937
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8248713179557788e-07,
+      "loss": 0.0101,
+      "step": 4938
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8131525946006978e-07,
+      "loss": 0.011,
+      "step": 4939
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8014712751969089e-07,
+      "loss": 0.0061,
+      "step": 4940
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7898273641943743e-07,
+      "loss": 0.0078,
+      "step": 4941
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7782208660288124e-07,
+      "loss": 0.0098,
+      "step": 4942
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.766651785121676e-07,
+      "loss": 0.0132,
+      "step": 4943
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7551201258801388e-07,
+      "loss": 0.0061,
+      "step": 4944
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.743625892697176e-07,
+      "loss": 0.0116,
+      "step": 4945
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7321690899514742e-07,
+      "loss": 0.01,
+      "step": 4946
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7207497220074643e-07,
+      "loss": 0.0122,
+      "step": 4947
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7093677932153218e-07,
+      "loss": 0.0101,
+      "step": 4948
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6980233079109453e-07,
+      "loss": 0.0078,
+      "step": 4949
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6867162704159666e-07,
+      "loss": 0.0133,
+      "step": 4950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6754466850377737e-07,
+      "loss": 0.0089,
+      "step": 4951
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6642145560694878e-07,
+      "loss": 0.0092,
+      "step": 4952
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6530198877899417e-07,
+      "loss": 0.0071,
+      "step": 4953
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6418626844637242e-07,
+      "loss": 0.008,
+      "step": 4954
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6307429503411132e-07,
+      "loss": 0.0081,
+      "step": 4955
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6196606896581314e-07,
+      "loss": 0.0114,
+      "step": 4956
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.608615906636535e-07,
+      "loss": 0.0073,
+      "step": 4957
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5976086054838024e-07,
+      "loss": 0.0092,
+      "step": 4958
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5866387903931247e-07,
+      "loss": 0.0099,
+      "step": 4959
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.575706465543436e-07,
+      "loss": 0.01,
+      "step": 4960
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5648116350993392e-07,
+      "loss": 0.0063,
+      "step": 4961
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5539543032111916e-07,
+      "loss": 0.0089,
+      "step": 4962
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5431344740150513e-07,
+      "loss": 0.0091,
+      "step": 4963
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.532352151632721e-07,
+      "loss": 0.0084,
+      "step": 4964
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5216073401716713e-07,
+      "loss": 0.01,
+      "step": 4965
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5109000437250942e-07,
+      "loss": 0.0123,
+      "step": 4966
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.500230266371927e-07,
+      "loss": 0.0117,
+      "step": 4967
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4895980121767627e-07,
+      "loss": 0.0036,
+      "step": 4968
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4790032851899172e-07,
+      "loss": 0.0062,
+      "step": 4969
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4684460894474394e-07,
+      "loss": 0.0129,
+      "step": 4970
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.457926428971046e-07,
+      "loss": 0.0072,
+      "step": 4971
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4474443077681644e-07,
+      "loss": 0.0061,
+      "step": 4972
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.436999729831934e-07,
+      "loss": 0.0111,
+      "step": 4973
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4265926991411605e-07,
+      "loss": 0.0126,
+      "step": 4974
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4162232196603843e-07,
+      "loss": 0.0107,
+      "step": 4975
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4058912953398228e-07,
+      "loss": 0.0113,
+      "step": 4976
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3955969301153949e-07,
+      "loss": 0.0118,
+      "step": 4977
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3853401279086853e-07,
+      "loss": 0.0136,
+      "step": 4978
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.375120892627002e-07,
+      "loss": 0.0065,
+      "step": 4979
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.36493922816332e-07,
+      "loss": 0.0094,
+      "step": 4980
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3547951383963141e-07,
+      "loss": 0.0108,
+      "step": 4981
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3446886271903493e-07,
+      "loss": 0.0159,
+      "step": 4982
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3346196983954674e-07,
+      "loss": 0.0179,
+      "step": 4983
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3245883558473672e-07,
+      "loss": 0.0113,
+      "step": 4984
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3145946033674916e-07,
+      "loss": 0.0074,
+      "step": 4985
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.304638444762907e-07,
+      "loss": 0.0087,
+      "step": 4986
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2947198838263786e-07,
+      "loss": 0.0131,
+      "step": 4987
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2848389243363514e-07,
+      "loss": 0.0095,
+      "step": 4988
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.274995570056947e-07,
+      "loss": 0.0122,
+      "step": 4989
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2651898247379445e-07,
+      "loss": 0.0109,
+      "step": 4990
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2554216921148444e-07,
+      "loss": 0.0103,
+      "step": 4991
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2456911759087476e-07,
+      "loss": 0.0084,
+      "step": 4992
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.235998279826478e-07,
+      "loss": 0.0132,
+      "step": 4993
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2263430075605043e-07,
+      "loss": 0.0101,
+      "step": 4994
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.216725362788973e-07,
+      "loss": 0.007,
+      "step": 4995
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2071453491756978e-07,
+      "loss": 0.012,
+      "step": 4996
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1976029703701598e-07,
+      "loss": 0.023,
+      "step": 4997
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1880982300074839e-07,
+      "loss": 0.0098,
+      "step": 4998
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1786311317084742e-07,
+      "loss": 0.0125,
+      "step": 4999
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1692016790795791e-07,
+      "loss": 0.0085,
+      "step": 5000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.159809875712925e-07,
+      "loss": 0.011,
+      "step": 5001
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1504557251862946e-07,
+      "loss": 0.0117,
+      "step": 5002
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1411392310631153e-07,
+      "loss": 0.0056,
+      "step": 5003
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1318603968924813e-07,
+      "loss": 0.0119,
+      "step": 5004
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1226192262091317e-07,
+      "loss": 0.0069,
+      "step": 5005
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1134157225334397e-07,
+      "loss": 0.0103,
+      "step": 5006
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.104249889371467e-07,
+      "loss": 0.0043,
+      "step": 5007
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0951217302148986e-07,
+      "loss": 0.0135,
+      "step": 5008
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0860312485410862e-07,
+      "loss": 0.0138,
+      "step": 5009
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0769784478130263e-07,
+      "loss": 0.012,
+      "step": 5010
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0679633314793269e-07,
+      "loss": 0.005,
+      "step": 5011
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0589859029742855e-07,
+      "loss": 0.0093,
+      "step": 5012
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0500461657178107e-07,
+      "loss": 0.0046,
+      "step": 5013
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0411441231154784e-07,
+      "loss": 0.0069,
+      "step": 5014
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.032279778558487e-07,
+      "loss": 0.0126,
+      "step": 5015
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0234531354236799e-07,
+      "loss": 0.0081,
+      "step": 5016
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0146641970735338e-07,
+      "loss": 0.0121,
+      "step": 5017
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0059129668561707e-07,
+      "loss": 0.0142,
+      "step": 5018
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.971994481053348e-08,
+      "loss": 0.0063,
+      "step": 5019
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.885236441404155e-08,
+      "loss": 0.0116,
+      "step": 5020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.798855582664357e-08,
+      "loss": 0.01,
+      "step": 5021
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.712851937740409e-08,
+      "loss": 0.0086,
+      "step": 5022
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.627225539395102e-08,
+      "loss": 0.0049,
+      "step": 5023
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.541976420247568e-08,
+      "loss": 0.008,
+      "step": 5024
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.45710461277316e-08,
+      "loss": 0.008,
+      "step": 5025
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.372610149303462e-08,
+      "loss": 0.0087,
+      "step": 5026
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.288493062026394e-08,
+      "loss": 0.0083,
+      "step": 5027
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.204753382986097e-08,
+      "loss": 0.0082,
+      "step": 5028
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.121391144082948e-08,
+      "loss": 0.0115,
+      "step": 5029
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.038406377073427e-08,
+      "loss": 0.0034,
+      "step": 5030
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.955799113570363e-08,
+      "loss": 0.0062,
+      "step": 5031
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.873569385042691e-08,
+      "loss": 0.0125,
+      "step": 5032
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.791717222815577e-08,
+      "loss": 0.0088,
+      "step": 5033
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.7102426580703e-08,
+      "loss": 0.008,
+      "step": 5034
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.629145721844367e-08,
+      "loss": 0.008,
+      "step": 5035
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.548426445031288e-08,
+      "loss": 0.0072,
+      "step": 5036
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.468084858380909e-08,
+      "loss": 0.0086,
+      "step": 5037
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.388120992499083e-08,
+      "loss": 0.0063,
+      "step": 5038
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.308534877847663e-08,
+      "loss": 0.0103,
+      "step": 5039
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.229326544744843e-08,
+      "loss": 0.0086,
+      "step": 5040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.150496023364707e-08,
+      "loss": 0.0097,
+      "step": 5041
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.072043343737457e-08,
+      "loss": 0.0134,
+      "step": 5042
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.993968535749519e-08,
+      "loss": 0.0087,
+      "step": 5043
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.916271629143102e-08,
+      "loss": 0.0139,
+      "step": 5044
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.838952653516529e-08,
+      "loss": 0.007,
+      "step": 5045
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.762011638324351e-08,
+      "loss": 0.0114,
+      "step": 5046
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.6854486128769e-08,
+      "loss": 0.0093,
+      "step": 5047
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.609263606340622e-08,
+      "loss": 0.0081,
+      "step": 5048
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.533456647737969e-08,
+      "loss": 0.0138,
+      "step": 5049
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.458027765947285e-08,
+      "loss": 0.0154,
+      "step": 5050
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.382976989703027e-08,
+      "loss": 0.0048,
+      "step": 5051
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.308304347595441e-08,
+      "loss": 0.0085,
+      "step": 5052
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.23400986807099e-08,
+      "loss": 0.015,
+      "step": 5053
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.160093579431704e-08,
+      "loss": 0.0099,
+      "step": 5054
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.086555509835946e-08,
+      "loss": 0.004,
+      "step": 5055
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.013395687297641e-08,
+      "loss": 0.0083,
+      "step": 5056
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.94061413968683e-08,
+      "loss": 0.0166,
+      "step": 5057
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.868210894729333e-08,
+      "loss": 0.0092,
+      "step": 5058
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.796185980007086e-08,
+      "loss": 0.0161,
+      "step": 5059
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.724539422957588e-08,
+      "loss": 0.0079,
+      "step": 5060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.653271250874448e-08,
+      "loss": 0.0121,
+      "step": 5061
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.582381490906953e-08,
+      "loss": 0.0125,
+      "step": 5062
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.511870170060386e-08,
+      "loss": 0.0095,
+      "step": 5063
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.441737315195818e-08,
+      "loss": 0.0116,
+      "step": 5064
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.371982953030098e-08,
+      "loss": 0.0087,
+      "step": 5065
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.302607110135973e-08,
+      "loss": 0.0069,
+      "step": 5066
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.233609812941743e-08,
+      "loss": 0.0118,
+      "step": 5067
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.164991087731831e-08,
+      "loss": 0.0052,
+      "step": 5068
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.096750960646325e-08,
+      "loss": 0.0107,
+      "step": 5069
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.028889457681098e-08,
+      "loss": 0.0099,
+      "step": 5070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.961406604687803e-08,
+      "loss": 0.0098,
+      "step": 5071
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.894302427373544e-08,
+      "loss": 0.0072,
+      "step": 5072
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.827576951301761e-08,
+      "loss": 0.0113,
+      "step": 5073
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.761230201891121e-08,
+      "loss": 0.0081,
+      "step": 5074
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.6952622044161855e-08,
+      "loss": 0.0089,
+      "step": 5075
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.629672984007295e-08,
+      "loss": 0.0126,
+      "step": 5076
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.5644625656504634e-08,
+      "loss": 0.0104,
+      "step": 5077
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.4996309741873755e-08,
+      "loss": 0.0096,
+      "step": 5078
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.435178234315386e-08,
+      "loss": 0.0084,
+      "step": 5079
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.371104370587521e-08,
+      "loss": 0.0132,
+      "step": 5080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.307409407412589e-08,
+      "loss": 0.006,
+      "step": 5081
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.244093369054848e-08,
+      "loss": 0.0139,
+      "step": 5082
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.181156279634447e-08,
+      "loss": 0.0119,
+      "step": 5083
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.118598163126986e-08,
+      "loss": 0.0041,
+      "step": 5084
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.0564190433638474e-08,
+      "loss": 0.0113,
+      "step": 5085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.99461894403197e-08,
+      "loss": 0.0095,
+      "step": 5086
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9331978886737465e-08,
+      "loss": 0.0102,
+      "step": 5087
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8721559006873473e-08,
+      "loss": 0.0073,
+      "step": 5088
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.811493003326728e-08,
+      "loss": 0.0073,
+      "step": 5089
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.7512092197009585e-08,
+      "loss": 0.0082,
+      "step": 5090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6913045727752235e-08,
+      "loss": 0.0083,
+      "step": 5091
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.631779085369714e-08,
+      "loss": 0.0082,
+      "step": 5092
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.572632780160513e-08,
+      "loss": 0.0142,
+      "step": 5093
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.513865679679374e-08,
+      "loss": 0.0098,
+      "step": 5094
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.45547780631328e-08,
+      "loss": 0.0104,
+      "step": 5095
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3974691823049917e-08,
+      "loss": 0.0087,
+      "step": 5096
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3398398297527235e-08,
+      "loss": 0.007,
+      "step": 5097
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2825897706100237e-08,
+      "loss": 0.0114,
+      "step": 5098
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2257190266861145e-08,
+      "loss": 0.0131,
+      "step": 5099
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.169227619645999e-08,
+      "loss": 0.0136,
+      "step": 5100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.113115571009574e-08,
+      "loss": 0.0058,
+      "step": 5101
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.057382902152629e-08,
+      "loss": 0.004,
+      "step": 5102
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0020296343065144e-08,
+      "loss": 0.007,
+      "step": 5103
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9470557885575856e-08,
+      "loss": 0.0063,
+      "step": 5104
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.89246138584809e-08,
+      "loss": 0.0106,
+      "step": 5105
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.838246446975724e-08,
+      "loss": 0.0084,
+      "step": 5106
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7844109925931905e-08,
+      "loss": 0.0092,
+      "step": 5107
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.730955043209083e-08,
+      "loss": 0.0055,
+      "step": 5108
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6778786191873363e-08,
+      "loss": 0.0099,
+      "step": 5109
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.625181740747108e-08,
+      "loss": 0.0094,
+      "step": 5110
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.57286442796323e-08,
+      "loss": 0.0088,
+      "step": 5111
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.520926700765648e-08,
+      "loss": 0.0136,
+      "step": 5112
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.46936857893998e-08,
+      "loss": 0.0101,
+      "step": 5113
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.418190082127071e-08,
+      "loss": 0.0135,
+      "step": 5114
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3673912298232134e-08,
+      "loss": 0.013,
+      "step": 5115
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.31697204138004e-08,
+      "loss": 0.0103,
+      "step": 5116
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.266932536004741e-08,
+      "loss": 0.0094,
+      "step": 5117
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2172727327594024e-08,
+      "loss": 0.007,
+      "step": 5118
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.167992650561891e-08,
+      "loss": 0.0104,
+      "step": 5119
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1190923081853005e-08,
+      "loss": 0.0115,
+      "step": 5120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0705717242581755e-08,
+      "loss": 0.0084,
+      "step": 5121
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.022430917264063e-08,
+      "loss": 0.0155,
+      "step": 5122
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9746699055421824e-08,
+      "loss": 0.0074,
+      "step": 5123
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9272887072868683e-08,
+      "loss": 0.0097,
+      "step": 5124
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8802873405477937e-08,
+      "loss": 0.0129,
+      "step": 5125
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8336658232300805e-08,
+      "loss": 0.0108,
+      "step": 5126
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7874241730939666e-08,
+      "loss": 0.0086,
+      "step": 5127
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7415624077551383e-08,
+      "loss": 0.0161,
+      "step": 5128
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.6960805446842874e-08,
+      "loss": 0.0079,
+      "step": 5129
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6509786012077765e-08,
+      "loss": 0.0056,
+      "step": 5130
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.606256594506862e-08,
+      "loss": 0.0097,
+      "step": 5131
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.561914541618471e-08,
+      "loss": 0.0104,
+      "step": 5132
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.517952459434425e-08,
+      "loss": 0.0072,
+      "step": 5133
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4743703647018837e-08,
+      "loss": 0.0123,
+      "step": 5134
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4311682740234544e-08,
+      "loss": 0.0092,
+      "step": 5135
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3883462038566397e-08,
+      "loss": 0.0124,
+      "step": 5136
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3459041705145014e-08,
+      "loss": 0.0157,
+      "step": 5137
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3038421901651064e-08,
+      "loss": 0.0085,
+      "step": 5138
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2621602788319708e-08,
+      "loss": 0.0076,
+      "step": 5139
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.220858452393504e-08,
+      "loss": 0.0066,
+      "step": 5140
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1799367265835646e-08,
+      "loss": 0.0142,
+      "step": 5141
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1393951169911276e-08,
+      "loss": 0.0164,
+      "step": 5142
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.099233639060394e-08,
+      "loss": 0.0086,
+      "step": 5143
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0594523080907926e-08,
+      "loss": 0.0084,
+      "step": 5144
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.020051139236978e-08,
+      "loss": 0.0094,
+      "step": 5145
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9810301475083892e-08,
+      "loss": 0.0043,
+      "step": 5146
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.942389347770246e-08,
+      "loss": 0.0078,
+      "step": 5147
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9041287547424404e-08,
+      "loss": 0.0136,
+      "step": 5148
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8662483830003132e-08,
+      "loss": 0.019,
+      "step": 5149
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8287482469743212e-08,
+      "loss": 0.0082,
+      "step": 5150
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7916283609498152e-08,
+      "loss": 0.0061,
+      "step": 5151
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7548887390675952e-08,
+      "loss": 0.0059,
+      "step": 5152
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.718529395323687e-08,
+      "loss": 0.0079,
+      "step": 5153
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6825503435686784e-08,
+      "loss": 0.0063,
+      "step": 5154
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.64695159750905e-08,
+      "loss": 0.0093,
+      "step": 5155
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6117331707057316e-08,
+      "loss": 0.0125,
+      "step": 5156
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5768950765752133e-08,
+      "loss": 0.0098,
+      "step": 5157
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5424373283889904e-08,
+      "loss": 0.0108,
+      "step": 5158
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5083599392734515e-08,
+      "loss": 0.008,
+      "step": 5159
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4746629222104348e-08,
+      "loss": 0.0135,
+      "step": 5160
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4413462900366714e-08,
+      "loss": 0.0161,
+      "step": 5161
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4084100554440094e-08,
+      "loss": 0.0075,
+      "step": 5162
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.375854230979301e-08,
+      "loss": 0.0121,
+      "step": 5163
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3436788290447366e-08,
+      "loss": 0.012,
+      "step": 5164
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3118838618972896e-08,
+      "loss": 0.0077,
+      "step": 5165
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2804693416491599e-08,
+      "loss": 0.0108,
+      "step": 5166
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2494352802677745e-08,
+      "loss": 0.0153,
+      "step": 5167
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2187816895752324e-08,
+      "loss": 0.0074,
+      "step": 5168
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1885085812490815e-08,
+      "loss": 0.007,
+      "step": 5169
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1586159668216523e-08,
+      "loss": 0.0073,
+      "step": 5170
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1291038576806134e-08,
+      "loss": 0.0135,
+      "step": 5171
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0999722650681944e-08,
+      "loss": 0.0124,
+      "step": 5172
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0712212000822952e-08,
+      "loss": 0.0069,
+      "step": 5173
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0428506736752664e-08,
+      "loss": 0.0102,
+      "step": 5174
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0148606966550178e-08,
+      "loss": 0.0105,
+      "step": 5175
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.872512796840205e-09,
+      "loss": 0.0075,
+      "step": 5176
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.600224332800833e-09,
+      "loss": 0.0053,
+      "step": 5177
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.3317416781602e-09,
+      "loss": 0.006,
+      "step": 5178
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.06706493519538e-09,
+      "loss": 0.0109,
+      "step": 5179
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.806194204732388e-09,
+      "loss": 0.0144,
+      "step": 5180
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.549129586151727e-09,
+      "loss": 0.0081,
+      "step": 5181
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.295871177379511e-09,
+      "loss": 0.0082,
+      "step": 5182
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.046419074895228e-09,
+      "loss": 0.0134,
+      "step": 5183
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.80077337372509e-09,
+      "loss": 0.0121,
+      "step": 5184
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.558934167447573e-09,
+      "loss": 0.0117,
+      "step": 5185
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.320901548191206e-09,
+      "loss": 0.013,
+      "step": 5186
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.0866756066323424e-09,
+      "loss": 0.0073,
+      "step": 5187
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.856256432000719e-09,
+      "loss": 0.0142,
+      "step": 5188
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.629644112071676e-09,
+      "loss": 0.0124,
+      "step": 5189
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.406838733173937e-09,
+      "loss": 0.0103,
+      "step": 5190
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.187840380182941e-09,
+      "loss": 0.0098,
+      "step": 5191
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.972649136527509e-09,
+      "loss": 0.0047,
+      "step": 5192
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.761265084180956e-09,
+      "loss": 0.0116,
+      "step": 5193
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5536883036722e-09,
+      "loss": 0.008,
+      "step": 5194
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.3499188740757656e-09,
+      "loss": 0.0104,
+      "step": 5195
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.149956873017336e-09,
+      "loss": 0.0127,
+      "step": 5196
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9538023766704246e-09,
+      "loss": 0.0092,
+      "step": 5197
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.761455459760811e-09,
+      "loss": 0.0092,
+      "step": 5198
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.5729161955621045e-09,
+      "loss": 0.007,
+      "step": 5199
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.388184655897964e-09,
+      "loss": 0.0149,
+      "step": 5200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.207260911140987e-09,
+      "loss": 0.0073,
+      "step": 5201
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.030145030212707e-09,
+      "loss": 0.0114,
+      "step": 5202
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.856837080585818e-09,
+      "loss": 0.0081,
+      "step": 5203
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.6873371282808433e-09,
+      "loss": 0.0086,
+      "step": 5204
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.5216452378694644e-09,
+      "loss": 0.0124,
+      "step": 5205
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.3597614724689696e-09,
+      "loss": 0.0107,
+      "step": 5206
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.2016858937511388e-09,
+      "loss": 0.0115,
+      "step": 5207
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0474185619333573e-09,
+      "loss": 0.0126,
+      "step": 5208
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.896959535781951e-09,
+      "loss": 0.0087,
+      "step": 5209
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.750308872615515e-09,
+      "loss": 0.0069,
+      "step": 5210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.60746662829936e-09,
+      "loss": 0.01,
+      "step": 5211
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.468432857248848e-09,
+      "loss": 0.0073,
+      "step": 5212
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.333207612428279e-09,
+      "loss": 0.009,
+      "step": 5213
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.201790945352e-09,
+      "loss": 0.0079,
+      "step": 5214
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.074182906081079e-09,
+      "loss": 0.0137,
+      "step": 5215
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9503835432299613e-09,
+      "loss": 0.0117,
+      "step": 5216
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8303929039564793e-09,
+      "loss": 0.0084,
+      "step": 5217
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7142110339740669e-09,
+      "loss": 0.0098,
+      "step": 5218
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.601837977538434e-09,
+      "loss": 0.0142,
+      "step": 5219
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4932737774608908e-09,
+      "loss": 0.0062,
+      "step": 5220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3885184750961345e-09,
+      "loss": 0.0107,
+      "step": 5221
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2875721103511318e-09,
+      "loss": 0.0073,
+      "step": 5222
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.190434721681788e-09,
+      "loss": 0.011,
+      "step": 5223
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.097106346090726e-09,
+      "loss": 0.013,
+      "step": 5224
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0075870191328384e-09,
+      "loss": 0.0051,
+      "step": 5225
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.218767749097357e-10,
+      "loss": 0.0131,
+      "step": 5226
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.399756460719666e-10,
+      "loss": 0.0093,
+      "step": 5227
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.618836638190186e-10,
+      "loss": 0.0101,
+      "step": 5228
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.876008579004278e-10,
+      "loss": 0.0157,
+      "step": 5229
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.171272566146691e-10,
+      "loss": 0.0083,
+      "step": 5230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.50462886806935e-10,
+      "loss": 0.009,
+      "step": 5231
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.876077738746876e-10,
+      "loss": 0.0139,
+      "step": 5232
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.285619417598863e-10,
+      "loss": 0.0084,
+      "step": 5233
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7332541295787004e-10,
+      "loss": 0.0095,
+      "step": 5234
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.218982085106959e-10,
+      "loss": 0.0066,
+      "step": 5235
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.74280348008249e-10,
+      "loss": 0.009,
+      "step": 5236
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3047184959157365e-10,
+      "loss": 0.0104,
+      "step": 5237
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.904727299473219e-10,
+      "loss": 0.0084,
+      "step": 5238
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5428300431552524e-10,
+      "loss": 0.0098,
+      "step": 5239
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.219026864818229e-10,
+      "loss": 0.0094,
+      "step": 5240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.333178878079274e-11,
+      "loss": 0.0094,
+      "step": 5241
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.857032209706128e-11,
+      "loss": 0.0128,
+      "step": 5242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7618295861973087e-11,
+      "loss": 0.007,
+      "step": 5243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.047571805914196e-11,
+      "loss": 0.0081,
+      "step": 5244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7142595217789492e-11,
+      "loss": 0.0075,
+      "step": 5245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.61893241718603e-12,
+      "loss": 0.0215,
+      "step": 5246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9047332855404168e-12,
+      "loss": 0.0147,
+      "step": 5247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.0069,
+      "step": 5248
+    },
+    {
+      "epoch": 2.0,
+      "step": 5248,
+      "total_flos": 267048097677312.0,
+      "train_loss": 0.026737265504934618,
+      "train_runtime": 71028.3987,
+      "train_samples_per_second": 8.272,
+      "train_steps_per_second": 0.074
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5248,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 188888,
+  "total_flos": 267048097677312.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}