diff --git "a/checkpoint-2195/trainer_state.json" "b/checkpoint-2195/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2195/trainer_state.json"
@@ -0,0 +1,13196 @@
+{
+  "best_metric": 0.4701635241508484,
+  "best_model_checkpoint": "english_accents_classification/checkpoint-2195",
+  "epoch": 1.0,
+  "global_step": 2195,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 0.2697,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 0.4396,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-08,
+      "loss": 0.5077,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 0.1614,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-07,
+      "loss": 1.2664,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-07,
+      "loss": 0.5497,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 0.5701,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 0.8319,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 0.0962,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-07,
+      "loss": 0.5706,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1999999999999998e-07,
+      "loss": 0.0664,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-07,
+      "loss": 0.5393,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6e-07,
+      "loss": 0.4106,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8e-07,
+      "loss": 0.1913,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-07,
+      "loss": 1.0994,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2e-07,
+      "loss": 0.9898,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 0.7844,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-07,
+      "loss": 0.157,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7999999999999996e-07,
+      "loss": 0.9783,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-07,
+      "loss": 0.5995,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1999999999999995e-07,
+      "loss": 0.6481,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3999999999999997e-07,
+      "loss": 0.9527,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6e-07,
+      "loss": 0.2092,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-07,
+      "loss": 0.8957,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-07,
+      "loss": 1.1157,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2e-07,
+      "loss": 0.7431,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 0.9357,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6e-07,
+      "loss": 1.2533,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8e-07,
+      "loss": 0.2219,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-07,
+      "loss": 0.0506,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2e-07,
+      "loss": 0.3284,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4e-07,
+      "loss": 0.3158,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6e-07,
+      "loss": 0.4881,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 0.8376,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7e-07,
+      "loss": 0.61,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-07,
+      "loss": 0.2311,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4e-07,
+      "loss": 1.3468,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.599999999999999e-07,
+      "loss": 0.2379,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 0.4719,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8e-07,
+      "loss": 0.2502,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.199999999999999e-07,
+      "loss": 0.208,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.399999999999999e-07,
+      "loss": 0.9285,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.599999999999999e-07,
+      "loss": 0.7614,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.799999999999999e-07,
+      "loss": 0.8994,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9e-07,
+      "loss": 1.0965,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.2e-07,
+      "loss": 0.0415,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.399999999999999e-07,
+      "loss": 0.5892,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.6e-07,
+      "loss": 0.5744,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.8e-07,
+      "loss": 0.4612,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-06,
+      "loss": 0.2253,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995337995337995e-07,
+      "loss": 0.7475,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990675990675991e-07,
+      "loss": 0.5024,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986013986013985e-07,
+      "loss": 0.1949,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.981351981351981e-07,
+      "loss": 0.3692,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976689976689975e-07,
+      "loss": 0.2752,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972027972027973e-07,
+      "loss": 0.2037,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.967365967365967e-07,
+      "loss": 0.416,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.962703962703963e-07,
+      "loss": 0.6794,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.958041958041957e-07,
+      "loss": 0.3761,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.953379953379953e-07,
+      "loss": 0.2819,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.948717948717949e-07,
+      "loss": 0.3834,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.944055944055944e-07,
+      "loss": 0.811,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.939393939393938e-07,
+      "loss": 0.1078,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.934731934731934e-07,
+      "loss": 0.2558,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.93006993006993e-07,
+      "loss": 0.6641,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.925407925407924e-07,
+      "loss": 0.2897,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92074592074592e-07,
+      "loss": 0.1813,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.916083916083916e-07,
+      "loss": 0.0429,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.911421911421912e-07,
+      "loss": 1.0502,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.906759906759906e-07,
+      "loss": 0.5908,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.902097902097902e-07,
+      "loss": 0.8051,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.897435897435898e-07,
+      "loss": 1.8102,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.892773892773891e-07,
+      "loss": 0.5972,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.888111888111887e-07,
+      "loss": 0.7426,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.883449883449883e-07,
+      "loss": 0.5157,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.87878787878788e-07,
+      "loss": 1.0276,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.874125874125873e-07,
+      "loss": 0.1409,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.869463869463869e-07,
+      "loss": 0.8096,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.864801864801865e-07,
+      "loss": 0.2062,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.86013986013986e-07,
+      "loss": 0.4916,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.855477855477855e-07,
+      "loss": 0.416,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.85081585081585e-07,
+      "loss": 0.7442,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.846153846153847e-07,
+      "loss": 0.5714,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.84149184149184e-07,
+      "loss": 0.4612,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.836829836829836e-07,
+      "loss": 0.7769,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.832167832167832e-07,
+      "loss": 1.0952,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.827505827505828e-07,
+      "loss": 0.2853,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.822843822843822e-07,
+      "loss": 0.1978,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.818181818181818e-07,
+      "loss": 0.3178,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.813519813519812e-07,
+      "loss": 1.019,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.80885780885781e-07,
+      "loss": 0.7215,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.804195804195804e-07,
+      "loss": 0.3357,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.7995337995338e-07,
+      "loss": 0.3128,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.794871794871793e-07,
+      "loss": 0.231,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.79020979020979e-07,
+      "loss": 0.697,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.785547785547785e-07,
+      "loss": 0.5477,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.780885780885781e-07,
+      "loss": 0.788,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.776223776223775e-07,
+      "loss": 0.505,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.77156177156177e-07,
+      "loss": 0.5423,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.766899766899767e-07,
+      "loss": 1.041,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.76223776223776e-07,
+      "loss": 0.7435,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.757575757575757e-07,
+      "loss": 1.2017,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.752913752913753e-07,
+      "loss": 0.3809,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.748251748251749e-07,
+      "loss": 0.1952,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.743589743589742e-07,
+      "loss": 0.5914,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.738927738927738e-07,
+      "loss": 0.1348,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.734265734265734e-07,
+      "loss": 0.7089,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.72960372960373e-07,
+      "loss": 0.5135,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.724941724941724e-07,
+      "loss": 0.6271,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.72027972027972e-07,
+      "loss": 0.121,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.715617715617716e-07,
+      "loss": 0.4388,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.71095571095571e-07,
+      "loss": 1.0235,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.706293706293706e-07,
+      "loss": 1.1125,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.701631701631702e-07,
+      "loss": 1.1468,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.696969696969698e-07,
+      "loss": 0.6173,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.692307692307691e-07,
+      "loss": 0.6534,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.687645687645687e-07,
+      "loss": 0.8306,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.682983682983683e-07,
+      "loss": 0.2338,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.678321678321677e-07,
+      "loss": 0.3129,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.673659673659673e-07,
+      "loss": 0.4549,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.66899766899767e-07,
+      "loss": 0.445,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.664335664335665e-07,
+      "loss": 0.3978,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.659673659673659e-07,
+      "loss": 0.3891,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.655011655011655e-07,
+      "loss": 0.2928,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.650349650349648e-07,
+      "loss": 0.803,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.645687645687647e-07,
+      "loss": 1.0638,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.64102564102564e-07,
+      "loss": 0.3401,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.636363636363636e-07,
+      "loss": 0.6608,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.63170163170163e-07,
+      "loss": 0.2608,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.627039627039626e-07,
+      "loss": 0.454,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.622377622377622e-07,
+      "loss": 0.2701,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.617715617715618e-07,
+      "loss": 0.8713,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.613053613053612e-07,
+      "loss": 0.2444,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.608391608391608e-07,
+      "loss": 0.214,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.603729603729604e-07,
+      "loss": 0.3274,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.599067599067597e-07,
+      "loss": 0.5751,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.594405594405593e-07,
+      "loss": 0.4116,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.58974358974359e-07,
+      "loss": 0.5622,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.585081585081585e-07,
+      "loss": 0.9027,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.58041958041958e-07,
+      "loss": 0.2139,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.575757575757575e-07,
+      "loss": 0.3488,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.57109557109557e-07,
+      "loss": 0.7612,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.566433566433567e-07,
+      "loss": 0.8103,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.56177156177156e-07,
+      "loss": 0.5668,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.557109557109557e-07,
+      "loss": 0.6247,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.552447552447553e-07,
+      "loss": 0.6235,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.547785547785546e-07,
+      "loss": 0.2373,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.543123543123542e-07,
+      "loss": 0.9669,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.538461538461538e-07,
+      "loss": 0.2583,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.533799533799533e-07,
+      "loss": 0.5178,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.529137529137529e-07,
+      "loss": 0.3697,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.524475524475524e-07,
+      "loss": 0.1705,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.51981351981352e-07,
+      "loss": 1.2947,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.515151515151515e-07,
+      "loss": 0.8781,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.51048951048951e-07,
+      "loss": 1.04,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.505827505827505e-07,
+      "loss": 0.1065,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.501165501165502e-07,
+      "loss": 0.4191,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.496503496503497e-07,
+      "loss": 1.357,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.491841491841491e-07,
+      "loss": 0.7947,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.487179487179486e-07,
+      "loss": 0.4605,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.482517482517482e-07,
+      "loss": 0.3693,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.477855477855477e-07,
+      "loss": 0.8224,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.473193473193473e-07,
+      "loss": 0.4088,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.468531468531468e-07,
+      "loss": 0.399,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.463869463869464e-07,
+      "loss": 0.1534,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.459207459207459e-07,
+      "loss": 0.0754,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.454545454545454e-07,
+      "loss": 1.6669,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.449883449883449e-07,
+      "loss": 0.8876,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.445221445221446e-07,
+      "loss": 0.4591,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.44055944055944e-07,
+      "loss": 0.1595,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.435897435897435e-07,
+      "loss": 0.4859,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.43123543123543e-07,
+      "loss": 0.7,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.426573426573426e-07,
+      "loss": 1.2747,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.421911421911422e-07,
+      "loss": 0.0871,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.417249417249417e-07,
+      "loss": 0.8679,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.412587412587412e-07,
+      "loss": 0.4158,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.407925407925408e-07,
+      "loss": 0.1375,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.403263403263403e-07,
+      "loss": 0.3705,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.398601398601398e-07,
+      "loss": 0.739,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.393939393939395e-07,
+      "loss": 1.0175,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.389277389277389e-07,
+      "loss": 0.2079,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.384615384615384e-07,
+      "loss": 0.6417,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.379953379953379e-07,
+      "loss": 0.1363,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.375291375291375e-07,
+      "loss": 1.0034,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.37062937062937e-07,
+      "loss": 0.2313,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.365967365967366e-07,
+      "loss": 0.3471,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.361305361305361e-07,
+      "loss": 0.8119,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.356643356643357e-07,
+      "loss": 0.7503,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.351981351981352e-07,
+      "loss": 0.442,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.347319347319347e-07,
+      "loss": 0.4107,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.342657342657341e-07,
+      "loss": 0.3514,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.337995337995338e-07,
+      "loss": 0.7022,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.333333333333333e-07,
+      "loss": 0.538,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.328671328671328e-07,
+      "loss": 1.0115,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.324009324009323e-07,
+      "loss": 0.2248,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.319347319347319e-07,
+      "loss": 1.277,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.314685314685314e-07,
+      "loss": 0.3915,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.31002331002331e-07,
+      "loss": 0.1461,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.305361305361305e-07,
+      "loss": 0.5608,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.300699300699301e-07,
+      "loss": 0.3596,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.296037296037295e-07,
+      "loss": 0.126,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.29137529137529e-07,
+      "loss": 0.4719,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.286713286713286e-07,
+      "loss": 0.5669,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.282051282051282e-07,
+      "loss": 0.506,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.277389277389277e-07,
+      "loss": 0.6553,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.272727272727272e-07,
+      "loss": 0.1117,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.268065268065267e-07,
+      "loss": 0.672,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.263403263403263e-07,
+      "loss": 0.971,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.258741258741259e-07,
+      "loss": 0.7821,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.254079254079254e-07,
+      "loss": 1.355,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.24941724941725e-07,
+      "loss": 0.9473,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.244755244755244e-07,
+      "loss": 0.6754,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.240093240093239e-07,
+      "loss": 0.8051,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.235431235431234e-07,
+      "loss": 0.7452,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.230769230769231e-07,
+      "loss": 0.1877,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.226107226107226e-07,
+      "loss": 1.1804,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.221445221445221e-07,
+      "loss": 1.1095,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.216783216783216e-07,
+      "loss": 0.5993,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.212121212121212e-07,
+      "loss": 0.8311,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.207459207459207e-07,
+      "loss": 0.6753,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.202797202797203e-07,
+      "loss": 0.7097,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.198135198135198e-07,
+      "loss": 1.3988,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.193473193473193e-07,
+      "loss": 0.2051,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.188811188811188e-07,
+      "loss": 0.4065,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.184149184149183e-07,
+      "loss": 0.4529,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.179487179487179e-07,
+      "loss": 1.0767,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.174825174825175e-07,
+      "loss": 0.6716,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.17016317016317e-07,
+      "loss": 0.0634,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.165501165501165e-07,
+      "loss": 0.2511,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.16083916083916e-07,
+      "loss": 1.2341,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.156177156177156e-07,
+      "loss": 0.0915,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.151515151515152e-07,
+      "loss": 0.8349,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.146853146853147e-07,
+      "loss": 0.2143,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.142191142191141e-07,
+      "loss": 0.2114,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.137529137529137e-07,
+      "loss": 0.9342,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.132867132867132e-07,
+      "loss": 0.0391,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.128205128205127e-07,
+      "loss": 1.5104,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.123543123543123e-07,
+      "loss": 0.3984,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.118881118881119e-07,
+      "loss": 0.3413,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.114219114219114e-07,
+      "loss": 0.6566,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.109557109557109e-07,
+      "loss": 0.5664,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.104895104895104e-07,
+      "loss": 0.371,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.1002331002331e-07,
+      "loss": 0.2574,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.095571095571096e-07,
+      "loss": 0.4314,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.09090909090909e-07,
+      "loss": 0.6212,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.086247086247086e-07,
+      "loss": 0.6465,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.081585081585081e-07,
+      "loss": 0.3893,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.076923076923076e-07,
+      "loss": 0.5573,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.072261072261071e-07,
+      "loss": 0.4188,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.067599067599068e-07,
+      "loss": 0.4821,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.062937062937063e-07,
+      "loss": 1.2224,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.058275058275058e-07,
+      "loss": 1.9969,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.053613053613053e-07,
+      "loss": 0.9049,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.048951048951049e-07,
+      "loss": 0.8997,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.044289044289045e-07,
+      "loss": 0.2375,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.039627039627039e-07,
+      "loss": 1.1948,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.034965034965034e-07,
+      "loss": 0.3155,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.03030303030303e-07,
+      "loss": 0.8782,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.025641025641025e-07,
+      "loss": 0.2583,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.02097902097902e-07,
+      "loss": 0.7527,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.016317016317016e-07,
+      "loss": 1.168,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.011655011655012e-07,
+      "loss": 0.7106,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.006993006993007e-07,
+      "loss": 0.7575,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.002331002331002e-07,
+      "loss": 0.4053,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.997668997668997e-07,
+      "loss": 0.5331,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.993006993006992e-07,
+      "loss": 0.3071,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.988344988344988e-07,
+      "loss": 0.6292,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.983682983682983e-07,
+      "loss": 1.2794,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.979020979020978e-07,
+      "loss": 0.4406,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.974358974358974e-07,
+      "loss": 0.3652,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.969696969696969e-07,
+      "loss": 0.1426,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.965034965034964e-07,
+      "loss": 0.342,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.96037296037296e-07,
+      "loss": 0.4244,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.955710955710956e-07,
+      "loss": 0.3634,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.951048951048951e-07,
+      "loss": 1.0175,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.946386946386945e-07,
+      "loss": 0.2506,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.941724941724941e-07,
+      "loss": 0.3747,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.937062937062937e-07,
+      "loss": 0.3832,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.932400932400932e-07,
+      "loss": 0.452,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.927738927738927e-07,
+      "loss": 0.4871,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.923076923076923e-07,
+      "loss": 0.1589,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.918414918414918e-07,
+      "loss": 0.0207,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.913752913752913e-07,
+      "loss": 0.7373,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.909090909090909e-07,
+      "loss": 1.0132,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.904428904428905e-07,
+      "loss": 0.3139,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.8997668997669e-07,
+      "loss": 0.4113,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.895104895104894e-07,
+      "loss": 0.8191,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.890442890442889e-07,
+      "loss": 0.4734,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.885780885780885e-07,
+      "loss": 0.4433,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.881118881118881e-07,
+      "loss": 0.3514,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.876456876456876e-07,
+      "loss": 0.7342,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.871794871794871e-07,
+      "loss": 0.469,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.867132867132867e-07,
+      "loss": 0.2603,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.862470862470862e-07,
+      "loss": 0.2582,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.857808857808857e-07,
+      "loss": 0.6944,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.853146853146853e-07,
+      "loss": 0.3737,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.848484848484849e-07,
+      "loss": 0.353,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.843822843822843e-07,
+      "loss": 0.9881,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.839160839160838e-07,
+      "loss": 0.561,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.834498834498833e-07,
+      "loss": 0.869,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.82983682983683e-07,
+      "loss": 1.0002,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.825174825174825e-07,
+      "loss": 0.6014,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.82051282051282e-07,
+      "loss": 0.8237,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.815850815850815e-07,
+      "loss": 0.5378,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.811188811188811e-07,
+      "loss": 0.3802,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.806526806526806e-07,
+      "loss": 0.4107,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.801864801864802e-07,
+      "loss": 0.4252,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.797202797202798e-07,
+      "loss": 0.4022,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.792540792540792e-07,
+      "loss": 0.1576,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.787878787878787e-07,
+      "loss": 0.4549,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.783216783216782e-07,
+      "loss": 0.7608,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.778554778554778e-07,
+      "loss": 0.3814,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.773892773892774e-07,
+      "loss": 0.4314,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.769230769230769e-07,
+      "loss": 0.5934,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.764568764568764e-07,
+      "loss": 1.0723,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.75990675990676e-07,
+      "loss": 0.0799,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.755244755244755e-07,
+      "loss": 0.176,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.75058275058275e-07,
+      "loss": 0.6018,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.745920745920746e-07,
+      "loss": 0.3065,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.741258741258741e-07,
+      "loss": 0.6418,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.736596736596736e-07,
+      "loss": 0.8705,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.731934731934731e-07,
+      "loss": 0.321,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.727272727272726e-07,
+      "loss": 0.9236,
+      "step": 323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.722610722610723e-07,
+      "loss": 0.1764,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.717948717948718e-07,
+      "loss": 0.1456,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.713286713286713e-07,
+      "loss": 0.3441,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.708624708624708e-07,
+      "loss": 0.741,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.703962703962704e-07,
+      "loss": 0.8835,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.699300699300699e-07,
+      "loss": 1.6253,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.694638694638695e-07,
+      "loss": 1.0812,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.689976689976689e-07,
+      "loss": 0.8224,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.685314685314685e-07,
+      "loss": 0.5518,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.68065268065268e-07,
+      "loss": 0.6023,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.675990675990675e-07,
+      "loss": 0.6288,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.67132867132867e-07,
+      "loss": 0.4006,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.666666666666667e-07,
+      "loss": 0.4891,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.662004662004662e-07,
+      "loss": 0.9205,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.657342657342657e-07,
+      "loss": 0.6212,
+      "step": 338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.652680652680653e-07,
+      "loss": 0.2526,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.648018648018648e-07,
+      "loss": 0.3449,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.643356643356642e-07,
+      "loss": 0.3619,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.638694638694638e-07,
+      "loss": 1.1025,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.634032634032634e-07,
+      "loss": 0.9264,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.629370629370629e-07,
+      "loss": 0.1858,
+      "step": 344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.624708624708624e-07,
+      "loss": 0.2834,
+      "step": 345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.620046620046619e-07,
+      "loss": 0.1827,
+      "step": 346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.615384615384616e-07,
+      "loss": 0.5381,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.610722610722611e-07,
+      "loss": 0.3102,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.606060606060606e-07,
+      "loss": 0.4673,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.601398601398601e-07,
+      "loss": 1.2077,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.596736596736597e-07,
+      "loss": 0.3813,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.592074592074591e-07,
+      "loss": 1.3662,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.587412587412587e-07,
+      "loss": 0.5387,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.582750582750582e-07,
+      "loss": 0.967,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.578088578088578e-07,
+      "loss": 0.1306,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.573426573426573e-07,
+      "loss": 1.3893,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.568764568764568e-07,
+      "loss": 0.4239,
+      "step": 357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.564102564102563e-07,
+      "loss": 0.7249,
+      "step": 358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.55944055944056e-07,
+      "loss": 0.4501,
+      "step": 359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.554778554778555e-07,
+      "loss": 0.2733,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.55011655011655e-07,
+      "loss": 0.603,
+      "step": 361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.545454545454544e-07,
+      "loss": 0.4453,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.54079254079254e-07,
+      "loss": 0.0618,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.536130536130535e-07,
+      "loss": 0.7462,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.531468531468531e-07,
+      "loss": 0.7528,
+      "step": 365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.526806526806526e-07,
+      "loss": 0.5247,
+      "step": 366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.522144522144522e-07,
+      "loss": 0.9413,
+      "step": 367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.517482517482517e-07,
+      "loss": 0.1754,
+      "step": 368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.512820512820512e-07,
+      "loss": 0.8789,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.508158508158507e-07,
+      "loss": 0.5791,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.503496503496504e-07,
+      "loss": 1.2487,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.498834498834499e-07,
+      "loss": 0.497,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.494172494172493e-07,
+      "loss": 0.5976,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.489510489510489e-07,
+      "loss": 0.3764,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.484848484848484e-07,
+      "loss": 0.1883,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.48018648018648e-07,
+      "loss": 0.1701,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.475524475524475e-07,
+      "loss": 0.4337,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.470862470862471e-07,
+      "loss": 0.4545,
+      "step": 378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.466200466200466e-07,
+      "loss": 0.8228,
+      "step": 379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.461538461538461e-07,
+      "loss": 0.3225,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.456876456876456e-07,
+      "loss": 1.4789,
+      "step": 381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.452214452214453e-07,
+      "loss": 0.5279,
+      "step": 382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.447552447552448e-07,
+      "loss": 0.5917,
+      "step": 383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 8.442890442890442e-07,
+      "loss": 0.8379,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.438228438228437e-07,
+      "loss": 0.1413,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.433566433566433e-07,
+      "loss": 0.8944,
+      "step": 386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.428904428904428e-07,
+      "loss": 0.4638,
+      "step": 387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.424242424242424e-07,
+      "loss": 0.2795,
+      "step": 388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.419580419580419e-07,
+      "loss": 0.6223,
+      "step": 389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.414918414918415e-07,
+      "loss": 0.5884,
+      "step": 390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.41025641025641e-07,
+      "loss": 0.5431,
+      "step": 391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.405594405594405e-07,
+      "loss": 0.2101,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.4009324009324e-07,
+      "loss": 0.1908,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.396270396270397e-07,
+      "loss": 0.5586,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.391608391608391e-07,
+      "loss": 0.6525,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.386946386946386e-07,
+      "loss": 0.4527,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.382284382284381e-07,
+      "loss": 0.1222,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.377622377622377e-07,
+      "loss": 0.0786,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.372960372960373e-07,
+      "loss": 0.4928,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.368298368298368e-07,
+      "loss": 0.3677,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.363636363636363e-07,
+      "loss": 0.9614,
+      "step": 401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.358974358974359e-07,
+      "loss": 0.3454,
+      "step": 402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.354312354312354e-07,
+      "loss": 0.4066,
+      "step": 403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.349650349650349e-07,
+      "loss": 0.7512,
+      "step": 404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.344988344988346e-07,
+      "loss": 0.3592,
+      "step": 405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.34032634032634e-07,
+      "loss": 0.7145,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.335664335664335e-07,
+      "loss": 0.6653,
+      "step": 407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.33100233100233e-07,
+      "loss": 1.4909,
+      "step": 408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.326340326340326e-07,
+      "loss": 0.0793,
+      "step": 409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.321678321678321e-07,
+      "loss": 0.2574,
+      "step": 410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.317016317016317e-07,
+      "loss": 0.4923,
+      "step": 411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.312354312354312e-07,
+      "loss": 1.1847,
+      "step": 412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.307692307692308e-07,
+      "loss": 0.2086,
+      "step": 413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.303030303030303e-07,
+      "loss": 0.4983,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.298368298368298e-07,
+      "loss": 1.1936,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.293706293706292e-07,
+      "loss": 0.2827,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.289044289044289e-07,
+      "loss": 0.3458,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.284382284382284e-07,
+      "loss": 0.6335,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.279720279720279e-07,
+      "loss": 0.5442,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.275058275058274e-07,
+      "loss": 0.0426,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.27039627039627e-07,
+      "loss": 0.116,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.265734265734266e-07,
+      "loss": 0.6586,
+      "step": 422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.261072261072261e-07,
+      "loss": 0.3375,
+      "step": 423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.256410256410256e-07,
+      "loss": 0.2099,
+      "step": 424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.251748251748252e-07,
+      "loss": 0.2468,
+      "step": 425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.247086247086247e-07,
+      "loss": 0.1405,
+      "step": 426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.242424242424241e-07,
+      "loss": 2.8083,
+      "step": 427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.237762237762237e-07,
+      "loss": 0.8359,
+      "step": 428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.233100233100233e-07,
+      "loss": 0.0882,
+      "step": 429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.228438228438228e-07,
+      "loss": 0.6334,
+      "step": 430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.223776223776223e-07,
+      "loss": 0.1033,
+      "step": 431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.219114219114218e-07,
+      "loss": 1.1003,
+      "step": 432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.214452214452214e-07,
+      "loss": 0.307,
+      "step": 433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.20979020979021e-07,
+      "loss": 0.4105,
+      "step": 434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.205128205128205e-07,
+      "loss": 1.0082,
+      "step": 435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.200466200466201e-07,
+      "loss": 0.122,
+      "step": 436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.195804195804196e-07,
+      "loss": 0.9319,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.19114219114219e-07,
+      "loss": 0.5291,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.186480186480185e-07,
+      "loss": 1.0978,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.181818181818182e-07,
+      "loss": 0.7561,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.177156177156177e-07,
+      "loss": 0.5643,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.172494172494172e-07,
+      "loss": 0.5011,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.167832167832167e-07,
+      "loss": 0.0722,
+      "step": 443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.163170163170163e-07,
+      "loss": 0.2172,
+      "step": 444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.158508158508159e-07,
+      "loss": 0.7358,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.153846153846154e-07,
+      "loss": 0.6059,
+      "step": 446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.149184149184149e-07,
+      "loss": 0.8357,
+      "step": 447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.144522144522145e-07,
+      "loss": 0.7287,
+      "step": 448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.139860139860139e-07,
+      "loss": 0.6888,
+      "step": 449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.135198135198134e-07,
+      "loss": 0.2778,
+      "step": 450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.13053613053613e-07,
+      "loss": 0.9167,
+      "step": 451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.125874125874126e-07,
+      "loss": 1.2744,
+      "step": 452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.121212121212121e-07,
+      "loss": 0.4655,
+      "step": 453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.116550116550116e-07,
+      "loss": 0.8254,
+      "step": 454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.111888111888111e-07,
+      "loss": 0.511,
+      "step": 455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.107226107226107e-07,
+      "loss": 0.6044,
+      "step": 456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.102564102564103e-07,
+      "loss": 0.296,
+      "step": 457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.097902097902098e-07,
+      "loss": 0.4576,
+      "step": 458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.093240093240092e-07,
+      "loss": 0.4951,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.088578088578088e-07,
+      "loss": 0.1649,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.083916083916083e-07,
+      "loss": 0.5389,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.079254079254078e-07,
+      "loss": 0.7125,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.074592074592074e-07,
+      "loss": 1.1012,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.06993006993007e-07,
+      "loss": 1.1142,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.065268065268065e-07,
+      "loss": 0.5619,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.06060606060606e-07,
+      "loss": 0.2885,
+      "step": 466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.055944055944056e-07,
+      "loss": 0.2114,
+      "step": 467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.051282051282052e-07,
+      "loss": 0.0783,
+      "step": 468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.046620046620047e-07,
+      "loss": 0.1381,
+      "step": 469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.041958041958041e-07,
+      "loss": 0.2473,
+      "step": 470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.037296037296037e-07,
+      "loss": 0.2033,
+      "step": 471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.032634032634032e-07,
+      "loss": 0.8793,
+      "step": 472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.027972027972027e-07,
+      "loss": 0.5735,
+      "step": 473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.023310023310023e-07,
+      "loss": 0.4858,
+      "step": 474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.018648018648019e-07,
+      "loss": 0.5786,
+      "step": 475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.013986013986014e-07,
+      "loss": 1.0541,
+      "step": 476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.009324009324009e-07,
+      "loss": 0.6782,
+      "step": 477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.004662004662004e-07,
+      "loss": 0.2468,
+      "step": 478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8e-07,
+      "loss": 0.4491,
+      "step": 479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.995337995337996e-07,
+      "loss": 0.591,
+      "step": 480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.99067599067599e-07,
+      "loss": 0.9483,
+      "step": 481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.986013986013985e-07,
+      "loss": 0.4905,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.981351981351981e-07,
+      "loss": 0.5913,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.976689976689976e-07,
+      "loss": 0.7051,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.972027972027971e-07,
+      "loss": 0.9754,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.967365967365967e-07,
+      "loss": 0.2501,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.962703962703963e-07,
+      "loss": 0.4256,
+      "step": 487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.958041958041958e-07,
+      "loss": 0.5701,
+      "step": 488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.953379953379953e-07,
+      "loss": 0.0877,
+      "step": 489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.948717948717948e-07,
+      "loss": 1.0885,
+      "step": 490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.944055944055944e-07,
+      "loss": 0.7774,
+      "step": 491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.939393939393939e-07,
+      "loss": 1.4065,
+      "step": 492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.934731934731934e-07,
+      "loss": 0.0685,
+      "step": 493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.930069930069929e-07,
+      "loss": 0.0919,
+      "step": 494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.925407925407925e-07,
+      "loss": 0.118,
+      "step": 495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.92074592074592e-07,
+      "loss": 1.224,
+      "step": 496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.916083916083916e-07,
+      "loss": 0.32,
+      "step": 497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.911421911421911e-07,
+      "loss": 0.762,
+      "step": 498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.906759906759907e-07,
+      "loss": 0.124,
+      "step": 499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.902097902097902e-07,
+      "loss": 0.5359,
+      "step": 500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.897435897435897e-07,
+      "loss": 0.8952,
+      "step": 501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.892773892773893e-07,
+      "loss": 0.3708,
+      "step": 502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.888111888111888e-07,
+      "loss": 0.9437,
+      "step": 503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.883449883449883e-07,
+      "loss": 0.1453,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.878787878787878e-07,
+      "loss": 0.4129,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.874125874125874e-07,
+      "loss": 0.6186,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.869463869463869e-07,
+      "loss": 0.3918,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.864801864801864e-07,
+      "loss": 0.4111,
+      "step": 508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.86013986013986e-07,
+      "loss": 0.2357,
+      "step": 509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.855477855477856e-07,
+      "loss": 0.3628,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.850815850815851e-07,
+      "loss": 0.7919,
+      "step": 511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.846153846153846e-07,
+      "loss": 0.9205,
+      "step": 512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.84149184149184e-07,
+      "loss": 0.4913,
+      "step": 513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.836829836829836e-07,
+      "loss": 0.6741,
+      "step": 514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.832167832167832e-07,
+      "loss": 0.5809,
+      "step": 515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.827505827505827e-07,
+      "loss": 0.5979,
+      "step": 516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.822843822843822e-07,
+      "loss": 1.0076,
+      "step": 517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.818181818181818e-07,
+      "loss": 0.2618,
+      "step": 518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.813519813519813e-07,
+      "loss": 0.5734,
+      "step": 519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.808857808857809e-07,
+      "loss": 0.1092,
+      "step": 520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.804195804195804e-07,
+      "loss": 0.1758,
+      "step": 521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.7995337995338e-07,
+      "loss": 0.7294,
+      "step": 522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.794871794871795e-07,
+      "loss": 0.4579,
+      "step": 523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.790209790209789e-07,
+      "loss": 0.9238,
+      "step": 524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.785547785547784e-07,
+      "loss": 0.4159,
+      "step": 525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.780885780885781e-07,
+      "loss": 0.0819,
+      "step": 526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.776223776223776e-07,
+      "loss": 0.4152,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.771561771561771e-07,
+      "loss": 0.2772,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.766899766899766e-07,
+      "loss": 0.5274,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.762237762237762e-07,
+      "loss": 0.6125,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.757575757575757e-07,
+      "loss": 0.4459,
+      "step": 531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.752913752913753e-07,
+      "loss": 0.4989,
+      "step": 532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.748251748251749e-07,
+      "loss": 0.4665,
+      "step": 533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.743589743589744e-07,
+      "loss": 0.6049,
+      "step": 534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.738927738927738e-07,
+      "loss": 1.096,
+      "step": 535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.734265734265733e-07,
+      "loss": 0.3898,
+      "step": 536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.729603729603729e-07,
+      "loss": 0.5501,
+      "step": 537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.724941724941725e-07,
+      "loss": 0.2218,
+      "step": 538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.72027972027972e-07,
+      "loss": 0.4602,
+      "step": 539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.715617715617715e-07,
+      "loss": 0.8771,
+      "step": 540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.710955710955711e-07,
+      "loss": 0.2172,
+      "step": 541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.706293706293706e-07,
+      "loss": 0.5483,
+      "step": 542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.701631701631701e-07,
+      "loss": 0.4111,
+      "step": 543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.696969696969697e-07,
+      "loss": 0.2698,
+      "step": 544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.692307692307693e-07,
+      "loss": 0.9074,
+      "step": 545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.687645687645687e-07,
+      "loss": 0.0843,
+      "step": 546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.682983682983682e-07,
+      "loss": 0.2039,
+      "step": 547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.678321678321677e-07,
+      "loss": 0.9356,
+      "step": 548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.673659673659674e-07,
+      "loss": 0.2067,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.668997668997669e-07,
+      "loss": 0.449,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.664335664335664e-07,
+      "loss": 0.8091,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.659673659673659e-07,
+      "loss": 0.9983,
+      "step": 552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.655011655011655e-07,
+      "loss": 0.5557,
+      "step": 553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.65034965034965e-07,
+      "loss": 0.4489,
+      "step": 554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.645687645687646e-07,
+      "loss": 1.5593,
+      "step": 555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.64102564102564e-07,
+      "loss": 0.4399,
+      "step": 556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.636363636363636e-07,
+      "loss": 0.4531,
+      "step": 557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.631701631701631e-07,
+      "loss": 0.4996,
+      "step": 558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 7.627039627039626e-07,
+      "loss": 0.9265,
+      "step": 559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.622377622377621e-07,
+      "loss": 0.2142,
+      "step": 560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.617715617715618e-07,
+      "loss": 0.4238,
+      "step": 561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.613053613053613e-07,
+      "loss": 0.5906,
+      "step": 562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.608391608391608e-07,
+      "loss": 0.2782,
+      "step": 563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.603729603729604e-07,
+      "loss": 0.4895,
+      "step": 564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.599067599067599e-07,
+      "loss": 1.154,
+      "step": 565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.594405594405594e-07,
+      "loss": 0.3406,
+      "step": 566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.589743589743589e-07,
+      "loss": 0.6152,
+      "step": 567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.585081585081585e-07,
+      "loss": 0.0553,
+      "step": 568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.58041958041958e-07,
+      "loss": 0.9163,
+      "step": 569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.575757575757575e-07,
+      "loss": 0.7248,
+      "step": 570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.57109557109557e-07,
+      "loss": 0.7236,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.566433566433567e-07,
+      "loss": 0.8627,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.561771561771562e-07,
+      "loss": 0.1891,
+      "step": 573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.557109557109557e-07,
+      "loss": 1.4896,
+      "step": 574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.552447552447552e-07,
+      "loss": 0.4418,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.547785547785548e-07,
+      "loss": 0.9025,
+      "step": 576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.543123543123543e-07,
+      "loss": 0.3289,
+      "step": 577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.538461538461538e-07,
+      "loss": 0.9482,
+      "step": 578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.533799533799533e-07,
+      "loss": 0.5447,
+      "step": 579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.529137529137529e-07,
+      "loss": 1.0946,
+      "step": 580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.524475524475524e-07,
+      "loss": 0.966,
+      "step": 581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.519813519813519e-07,
+      "loss": 0.1451,
+      "step": 582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.515151515151514e-07,
+      "loss": 0.5892,
+      "step": 583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.510489510489511e-07,
+      "loss": 0.6694,
+      "step": 584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.505827505827506e-07,
+      "loss": 0.4421,
+      "step": 585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.501165501165501e-07,
+      "loss": 0.4364,
+      "step": 586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.496503496503496e-07,
+      "loss": 1.5936,
+      "step": 587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.491841491841491e-07,
+      "loss": 0.9019,
+      "step": 588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.487179487179486e-07,
+      "loss": 0.5802,
+      "step": 589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.482517482517482e-07,
+      "loss": 0.2344,
+      "step": 590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.477855477855477e-07,
+      "loss": 0.4624,
+      "step": 591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.473193473193473e-07,
+      "loss": 0.9747,
+      "step": 592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.468531468531468e-07,
+      "loss": 0.192,
+      "step": 593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.463869463869463e-07,
+      "loss": 0.6448,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.45920745920746e-07,
+      "loss": 0.1995,
+      "step": 595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.454545454545455e-07,
+      "loss": 0.616,
+      "step": 596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.44988344988345e-07,
+      "loss": 0.3549,
+      "step": 597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.445221445221445e-07,
+      "loss": 0.2595,
+      "step": 598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.44055944055944e-07,
+      "loss": 0.7183,
+      "step": 599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.435897435897435e-07,
+      "loss": 0.48,
+      "step": 600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.431235431235431e-07,
+      "loss": 0.4473,
+      "step": 601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.426573426573426e-07,
+      "loss": 0.854,
+      "step": 602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.421911421911422e-07,
+      "loss": 0.2863,
+      "step": 603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.417249417249417e-07,
+      "loss": 0.5683,
+      "step": 604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.412587412587412e-07,
+      "loss": 0.6995,
+      "step": 605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.407925407925407e-07,
+      "loss": 0.3636,
+      "step": 606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.403263403263404e-07,
+      "loss": 0.1277,
+      "step": 607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.398601398601399e-07,
+      "loss": 0.3896,
+      "step": 608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.393939393939394e-07,
+      "loss": 1.193,
+      "step": 609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.389277389277388e-07,
+      "loss": 0.3392,
+      "step": 610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.384615384615384e-07,
+      "loss": 0.479,
+      "step": 611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.379953379953379e-07,
+      "loss": 0.2749,
+      "step": 612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.375291375291375e-07,
+      "loss": 0.7041,
+      "step": 613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.37062937062937e-07,
+      "loss": 0.549,
+      "step": 614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.365967365967366e-07,
+      "loss": 0.3097,
+      "step": 615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.361305361305361e-07,
+      "loss": 0.3603,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.356643356643356e-07,
+      "loss": 0.2611,
+      "step": 617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.351981351981351e-07,
+      "loss": 0.8296,
+      "step": 618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.347319347319348e-07,
+      "loss": 0.5213,
+      "step": 619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.342657342657343e-07,
+      "loss": 0.8234,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.337995337995337e-07,
+      "loss": 0.3815,
+      "step": 621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.333333333333332e-07,
+      "loss": 0.2186,
+      "step": 622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.328671328671328e-07,
+      "loss": 0.0456,
+      "step": 623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.324009324009324e-07,
+      "loss": 0.7225,
+      "step": 624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.319347319347319e-07,
+      "loss": 0.1286,
+      "step": 625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.314685314685314e-07,
+      "loss": 0.4527,
+      "step": 626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.31002331002331e-07,
+      "loss": 0.2511,
+      "step": 627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.305361305361305e-07,
+      "loss": 0.8114,
+      "step": 628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.3006993006993e-07,
+      "loss": 0.1985,
+      "step": 629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.296037296037297e-07,
+      "loss": 2.0004,
+      "step": 630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.291375291375292e-07,
+      "loss": 0.5199,
+      "step": 631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.286713286713286e-07,
+      "loss": 0.3489,
+      "step": 632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.282051282051281e-07,
+      "loss": 0.7127,
+      "step": 633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.277389277389277e-07,
+      "loss": 0.4068,
+      "step": 634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.272727272727272e-07,
+      "loss": 0.3731,
+      "step": 635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.268065268065268e-07,
+      "loss": 0.3247,
+      "step": 636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.263403263403263e-07,
+      "loss": 0.6279,
+      "step": 637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.258741258741259e-07,
+      "loss": 0.1738,
+      "step": 638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.254079254079254e-07,
+      "loss": 0.22,
+      "step": 639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.249417249417249e-07,
+      "loss": 0.7301,
+      "step": 640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.244755244755244e-07,
+      "loss": 0.6889,
+      "step": 641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.24009324009324e-07,
+      "loss": 0.469,
+      "step": 642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.235431235431235e-07,
+      "loss": 0.4373,
+      "step": 643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.23076923076923e-07,
+      "loss": 0.3698,
+      "step": 644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.226107226107225e-07,
+      "loss": 0.2808,
+      "step": 645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.221445221445221e-07,
+      "loss": 0.156,
+      "step": 646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.216783216783217e-07,
+      "loss": 1.0243,
+      "step": 647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.212121212121212e-07,
+      "loss": 0.8786,
+      "step": 648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.207459207459207e-07,
+      "loss": 0.4455,
+      "step": 649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.202797202797203e-07,
+      "loss": 0.2673,
+      "step": 650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.198135198135198e-07,
+      "loss": 0.2978,
+      "step": 651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.193473193473193e-07,
+      "loss": 0.8041,
+      "step": 652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.188811188811188e-07,
+      "loss": 0.3472,
+      "step": 653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.184149184149184e-07,
+      "loss": 1.0304,
+      "step": 654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.179487179487179e-07,
+      "loss": 0.5389,
+      "step": 655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.174825174825174e-07,
+      "loss": 0.7798,
+      "step": 656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.170163170163169e-07,
+      "loss": 0.6907,
+      "step": 657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.165501165501165e-07,
+      "loss": 0.4744,
+      "step": 658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.160839160839161e-07,
+      "loss": 0.4906,
+      "step": 659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.156177156177156e-07,
+      "loss": 1.2929,
+      "step": 660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.151515151515152e-07,
+      "loss": 0.7174,
+      "step": 661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.146853146853147e-07,
+      "loss": 0.9202,
+      "step": 662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.142191142191142e-07,
+      "loss": 0.2229,
+      "step": 663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.137529137529136e-07,
+      "loss": 0.9465,
+      "step": 664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.132867132867133e-07,
+      "loss": 0.8508,
+      "step": 665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.128205128205128e-07,
+      "loss": 0.1409,
+      "step": 666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.123543123543123e-07,
+      "loss": 0.2377,
+      "step": 667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.118881118881118e-07,
+      "loss": 0.2681,
+      "step": 668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.114219114219114e-07,
+      "loss": 0.7092,
+      "step": 669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.10955710955711e-07,
+      "loss": 0.2626,
+      "step": 670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.104895104895105e-07,
+      "loss": 1.7374,
+      "step": 671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.1002331002331e-07,
+      "loss": 0.8308,
+      "step": 672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.095571095571096e-07,
+      "loss": 1.0125,
+      "step": 673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.09090909090909e-07,
+      "loss": 0.3591,
+      "step": 674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.086247086247085e-07,
+      "loss": 0.3007,
+      "step": 675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.081585081585081e-07,
+      "loss": 0.0582,
+      "step": 676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.076923076923077e-07,
+      "loss": 0.2481,
+      "step": 677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.072261072261072e-07,
+      "loss": 0.9296,
+      "step": 678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.067599067599067e-07,
+      "loss": 0.5227,
+      "step": 679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.062937062937062e-07,
+      "loss": 0.1773,
+      "step": 680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.058275058275058e-07,
+      "loss": 0.4333,
+      "step": 681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.053613053613054e-07,
+      "loss": 0.2543,
+      "step": 682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.048951048951049e-07,
+      "loss": 0.1989,
+      "step": 683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.044289044289044e-07,
+      "loss": 0.7562,
+      "step": 684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.03962703962704e-07,
+      "loss": 1.1069,
+      "step": 685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.034965034965034e-07,
+      "loss": 1.0119,
+      "step": 686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.030303030303029e-07,
+      "loss": 0.3457,
+      "step": 687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.025641025641025e-07,
+      "loss": 0.7513,
+      "step": 688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.020979020979021e-07,
+      "loss": 0.346,
+      "step": 689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.016317016317016e-07,
+      "loss": 0.466,
+      "step": 690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.011655011655011e-07,
+      "loss": 0.7783,
+      "step": 691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.006993006993007e-07,
+      "loss": 0.7091,
+      "step": 692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.002331002331003e-07,
+      "loss": 0.1712,
+      "step": 693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.997668997668998e-07,
+      "loss": 0.8583,
+      "step": 694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.993006993006993e-07,
+      "loss": 0.6647,
+      "step": 695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.988344988344988e-07,
+      "loss": 0.4658,
+      "step": 696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.983682983682983e-07,
+      "loss": 0.2909,
+      "step": 697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.979020979020978e-07,
+      "loss": 0.4177,
+      "step": 698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.974358974358974e-07,
+      "loss": 0.3257,
+      "step": 699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.96969696969697e-07,
+      "loss": 1.0362,
+      "step": 700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.965034965034965e-07,
+      "loss": 0.4355,
+      "step": 701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.96037296037296e-07,
+      "loss": 0.5001,
+      "step": 702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.955710955710955e-07,
+      "loss": 1.9289,
+      "step": 703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.951048951048951e-07,
+      "loss": 0.5622,
+      "step": 704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.946386946386947e-07,
+      "loss": 0.1424,
+      "step": 705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.941724941724942e-07,
+      "loss": 0.4432,
+      "step": 706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.937062937062936e-07,
+      "loss": 0.8576,
+      "step": 707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.932400932400932e-07,
+      "loss": 0.1652,
+      "step": 708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.927738927738927e-07,
+      "loss": 0.6545,
+      "step": 709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.923076923076922e-07,
+      "loss": 0.426,
+      "step": 710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.918414918414918e-07,
+      "loss": 1.4261,
+      "step": 711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.913752913752914e-07,
+      "loss": 0.5645,
+      "step": 712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.909090909090909e-07,
+      "loss": 0.2521,
+      "step": 713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.904428904428904e-07,
+      "loss": 0.3289,
+      "step": 714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.899766899766899e-07,
+      "loss": 0.4881,
+      "step": 715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.895104895104896e-07,
+      "loss": 0.1005,
+      "step": 716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.89044289044289e-07,
+      "loss": 0.952,
+      "step": 717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.885780885780885e-07,
+      "loss": 0.799,
+      "step": 718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.88111888111888e-07,
+      "loss": 1.1688,
+      "step": 719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.876456876456876e-07,
+      "loss": 0.3663,
+      "step": 720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.871794871794871e-07,
+      "loss": 1.023,
+      "step": 721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.867132867132867e-07,
+      "loss": 0.4777,
+      "step": 722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.862470862470863e-07,
+      "loss": 0.7662,
+      "step": 723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.857808857808858e-07,
+      "loss": 0.5865,
+      "step": 724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.853146853146853e-07,
+      "loss": 0.626,
+      "step": 725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.848484848484848e-07,
+      "loss": 0.6988,
+      "step": 726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.843822843822844e-07,
+      "loss": 0.4424,
+      "step": 727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.83916083916084e-07,
+      "loss": 0.8099,
+      "step": 728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.834498834498834e-07,
+      "loss": 0.2409,
+      "step": 729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.829836829836829e-07,
+      "loss": 0.0832,
+      "step": 730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.825174825174825e-07,
+      "loss": 0.7694,
+      "step": 731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.82051282051282e-07,
+      "loss": 0.8596,
+      "step": 732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.815850815850815e-07,
+      "loss": 0.1651,
+      "step": 733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.811188811188811e-07,
+      "loss": 0.1754,
+      "step": 734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.806526806526807e-07,
+      "loss": 0.1363,
+      "step": 735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.801864801864802e-07,
+      "loss": 1.4782,
+      "step": 736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.797202797202797e-07,
+      "loss": 0.4921,
+      "step": 737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.792540792540792e-07,
+      "loss": 0.3358,
+      "step": 738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.787878787878789e-07,
+      "loss": 0.333,
+      "step": 739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.783216783216783e-07,
+      "loss": 1.2981,
+      "step": 740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.778554778554778e-07,
+      "loss": 0.7363,
+      "step": 741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.773892773892773e-07,
+      "loss": 0.322,
+      "step": 742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.769230769230769e-07,
+      "loss": 0.354,
+      "step": 743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.764568764568764e-07,
+      "loss": 0.3649,
+      "step": 744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.75990675990676e-07,
+      "loss": 0.177,
+      "step": 745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.755244755244755e-07,
+      "loss": 1.1313,
+      "step": 746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.750582750582751e-07,
+      "loss": 0.5067,
+      "step": 747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.745920745920746e-07,
+      "loss": 0.4973,
+      "step": 748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.74125874125874e-07,
+      "loss": 0.1478,
+      "step": 749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.736596736596735e-07,
+      "loss": 0.1934,
+      "step": 750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.731934731934732e-07,
+      "loss": 0.9429,
+      "step": 751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.727272727272727e-07,
+      "loss": 0.9369,
+      "step": 752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.722610722610722e-07,
+      "loss": 0.324,
+      "step": 753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.717948717948717e-07,
+      "loss": 0.1592,
+      "step": 754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.713286713286713e-07,
+      "loss": 0.4343,
+      "step": 755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.708624708624708e-07,
+      "loss": 0.3117,
+      "step": 756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.703962703962704e-07,
+      "loss": 0.3361,
+      "step": 757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.6993006993007e-07,
+      "loss": 0.4498,
+      "step": 758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.694638694638695e-07,
+      "loss": 0.4312,
+      "step": 759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.68997668997669e-07,
+      "loss": 0.1613,
+      "step": 760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.685314685314684e-07,
+      "loss": 0.4924,
+      "step": 761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.680652680652681e-07,
+      "loss": 0.8686,
+      "step": 762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.675990675990676e-07,
+      "loss": 0.4411,
+      "step": 763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.671328671328671e-07,
+      "loss": 1.3651,
+      "step": 764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.666666666666666e-07,
+      "loss": 0.125,
+      "step": 765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.662004662004662e-07,
+      "loss": 0.3629,
+      "step": 766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.657342657342657e-07,
+      "loss": 1.5701,
+      "step": 767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.652680652680653e-07,
+      "loss": 0.4364,
+      "step": 768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.648018648018648e-07,
+      "loss": 0.3138,
+      "step": 769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.643356643356644e-07,
+      "loss": 1.4343,
+      "step": 770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.638694638694638e-07,
+      "loss": 0.2085,
+      "step": 771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.634032634032633e-07,
+      "loss": 0.6832,
+      "step": 772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.629370629370628e-07,
+      "loss": 0.805,
+      "step": 773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.624708624708625e-07,
+      "loss": 0.8554,
+      "step": 774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.62004662004662e-07,
+      "loss": 0.7197,
+      "step": 775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.615384615384615e-07,
+      "loss": 0.7173,
+      "step": 776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.61072261072261e-07,
+      "loss": 0.6236,
+      "step": 777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.606060606060606e-07,
+      "loss": 1.375,
+      "step": 778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.601398601398601e-07,
+      "loss": 0.0678,
+      "step": 779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.596736596736597e-07,
+      "loss": 0.773,
+      "step": 780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.592074592074592e-07,
+      "loss": 1.3608,
+      "step": 781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.587412587412587e-07,
+      "loss": 1.148,
+      "step": 782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.582750582750582e-07,
+      "loss": 0.0592,
+      "step": 783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.578088578088577e-07,
+      "loss": 0.3076,
+      "step": 784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.573426573426572e-07,
+      "loss": 1.0073,
+      "step": 785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.568764568764569e-07,
+      "loss": 0.9216,
+      "step": 786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.564102564102564e-07,
+      "loss": 0.6902,
+      "step": 787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.559440559440559e-07,
+      "loss": 0.2872,
+      "step": 788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.554778554778555e-07,
+      "loss": 0.737,
+      "step": 789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.55011655011655e-07,
+      "loss": 0.4093,
+      "step": 790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.545454545454546e-07,
+      "loss": 0.7355,
+      "step": 791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.54079254079254e-07,
+      "loss": 0.7997,
+      "step": 792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.536130536130536e-07,
+      "loss": 0.2296,
+      "step": 793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.531468531468531e-07,
+      "loss": 0.7687,
+      "step": 794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.526806526806526e-07,
+      "loss": 0.7359,
+      "step": 795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.522144522144521e-07,
+      "loss": 0.5697,
+      "step": 796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.517482517482518e-07,
+      "loss": 0.5422,
+      "step": 797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.512820512820513e-07,
+      "loss": 0.4921,
+      "step": 798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.508158508158508e-07,
+      "loss": 0.1306,
+      "step": 799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.503496503496503e-07,
+      "loss": 0.4487,
+      "step": 800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.498834498834499e-07,
+      "loss": 1.0596,
+      "step": 801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.494172494172494e-07,
+      "loss": 0.4857,
+      "step": 802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.48951048951049e-07,
+      "loss": 0.2449,
+      "step": 803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.484848484848484e-07,
+      "loss": 0.2279,
+      "step": 804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.48018648018648e-07,
+      "loss": 1.1134,
+      "step": 805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.475524475524475e-07,
+      "loss": 0.4614,
+      "step": 806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.47086247086247e-07,
+      "loss": 0.5403,
+      "step": 807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.466200466200465e-07,
+      "loss": 0.6212,
+      "step": 808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.461538461538462e-07,
+      "loss": 0.3189,
+      "step": 809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.456876456876457e-07,
+      "loss": 0.8012,
+      "step": 810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.452214452214452e-07,
+      "loss": 0.2633,
+      "step": 811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.447552447552447e-07,
+      "loss": 0.7894,
+      "step": 812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.442890442890443e-07,
+      "loss": 0.4239,
+      "step": 813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.438228438228437e-07,
+      "loss": 0.2416,
+      "step": 814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.433566433566433e-07,
+      "loss": 1.3657,
+      "step": 815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.428904428904428e-07,
+      "loss": 0.6306,
+      "step": 816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.424242424242424e-07,
+      "loss": 0.1341,
+      "step": 817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.419580419580419e-07,
+      "loss": 0.2708,
+      "step": 818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.414918414918414e-07,
+      "loss": 1.6897,
+      "step": 819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.410256410256411e-07,
+      "loss": 0.4262,
+      "step": 820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.405594405594406e-07,
+      "loss": 0.3179,
+      "step": 821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.400932400932401e-07,
+      "loss": 0.0454,
+      "step": 822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.396270396270396e-07,
+      "loss": 0.5917,
+      "step": 823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.391608391608392e-07,
+      "loss": 0.4682,
+      "step": 824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.386946386946386e-07,
+      "loss": 0.1851,
+      "step": 825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.382284382284382e-07,
+      "loss": 0.2704,
+      "step": 826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.377622377622377e-07,
+      "loss": 0.4835,
+      "step": 827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.372960372960373e-07,
+      "loss": 0.8426,
+      "step": 828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.368298368298368e-07,
+      "loss": 0.5634,
+      "step": 829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.363636363636363e-07,
+      "loss": 0.7153,
+      "step": 830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.358974358974358e-07,
+      "loss": 0.0264,
+      "step": 831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.354312354312355e-07,
+      "loss": 0.1927,
+      "step": 832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.34965034965035e-07,
+      "loss": 0.1112,
+      "step": 833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.344988344988345e-07,
+      "loss": 0.5788,
+      "step": 834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.34032634032634e-07,
+      "loss": 0.587,
+      "step": 835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.335664335664335e-07,
+      "loss": 0.6168,
+      "step": 836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.33100233100233e-07,
+      "loss": 0.4908,
+      "step": 837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.326340326340326e-07,
+      "loss": 1.1949,
+      "step": 838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.321678321678321e-07,
+      "loss": 0.3669,
+      "step": 839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.317016317016317e-07,
+      "loss": 0.2089,
+      "step": 840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.312354312354312e-07,
+      "loss": 1.2992,
+      "step": 841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.307692307692307e-07,
+      "loss": 1.0708,
+      "step": 842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.303030303030303e-07,
+      "loss": 0.4692,
+      "step": 843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.298368298368299e-07,
+      "loss": 0.7796,
+      "step": 844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.293706293706294e-07,
+      "loss": 0.5851,
+      "step": 845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.289044289044288e-07,
+      "loss": 0.7419,
+      "step": 846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.284382284382283e-07,
+      "loss": 0.2276,
+      "step": 847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.279720279720279e-07,
+      "loss": 0.3461,
+      "step": 848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.275058275058275e-07,
+      "loss": 0.4584,
+      "step": 849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.27039627039627e-07,
+      "loss": 0.2265,
+      "step": 850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.265734265734266e-07,
+      "loss": 0.1214,
+      "step": 851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.261072261072261e-07,
+      "loss": 0.3078,
+      "step": 852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.256410256410256e-07,
+      "loss": 0.6016,
+      "step": 853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.251748251748251e-07,
+      "loss": 0.1066,
+      "step": 854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.247086247086248e-07,
+      "loss": 0.4685,
+      "step": 855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.242424242424243e-07,
+      "loss": 0.8281,
+      "step": 856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.237762237762237e-07,
+      "loss": 0.4948,
+      "step": 857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.233100233100232e-07,
+      "loss": 0.4544,
+      "step": 858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.228438228438228e-07,
+      "loss": 0.0572,
+      "step": 859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.223776223776223e-07,
+      "loss": 0.375,
+      "step": 860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.219114219114219e-07,
+      "loss": 0.3615,
+      "step": 861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.214452214452214e-07,
+      "loss": 0.3689,
+      "step": 862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.20979020979021e-07,
+      "loss": 0.0378,
+      "step": 863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.205128205128205e-07,
+      "loss": 0.5685,
+      "step": 864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.2004662004662e-07,
+      "loss": 0.4744,
+      "step": 865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.195804195804195e-07,
+      "loss": 0.8278,
+      "step": 866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.191142191142192e-07,
+      "loss": 0.769,
+      "step": 867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.186480186480186e-07,
+      "loss": 0.3657,
+      "step": 868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.181818181818181e-07,
+      "loss": 0.2125,
+      "step": 869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.177156177156176e-07,
+      "loss": 0.4491,
+      "step": 870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.172494172494172e-07,
+      "loss": 0.486,
+      "step": 871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.167832167832168e-07,
+      "loss": 0.6125,
+      "step": 872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.163170163170163e-07,
+      "loss": 0.5995,
+      "step": 873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.158508158508158e-07,
+      "loss": 0.3385,
+      "step": 874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.153846153846154e-07,
+      "loss": 0.6277,
+      "step": 875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.149184149184149e-07,
+      "loss": 0.0779,
+      "step": 876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.144522144522144e-07,
+      "loss": 0.4417,
+      "step": 877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.13986013986014e-07,
+      "loss": 0.4958,
+      "step": 878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.135198135198135e-07,
+      "loss": 0.6204,
+      "step": 879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.13053613053613e-07,
+      "loss": 0.4668,
+      "step": 880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.125874125874125e-07,
+      "loss": 0.337,
+      "step": 881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.12121212121212e-07,
+      "loss": 0.1201,
+      "step": 882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.116550116550116e-07,
+      "loss": 0.6472,
+      "step": 883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.111888111888112e-07,
+      "loss": 0.603,
+      "step": 884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.107226107226107e-07,
+      "loss": 0.4613,
+      "step": 885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.102564102564103e-07,
+      "loss": 0.3553,
+      "step": 886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.097902097902098e-07,
+      "loss": 0.8766,
+      "step": 887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.093240093240093e-07,
+      "loss": 0.4461,
+      "step": 888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.088578088578087e-07,
+      "loss": 1.4474,
+      "step": 889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.083916083916084e-07,
+      "loss": 1.1053,
+      "step": 890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.079254079254079e-07,
+      "loss": 0.9185,
+      "step": 891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.074592074592074e-07,
+      "loss": 0.1675,
+      "step": 892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.069930069930069e-07,
+      "loss": 0.5948,
+      "step": 893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.065268065268065e-07,
+      "loss": 0.8992,
+      "step": 894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 0.6338,
+      "step": 895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.055944055944056e-07,
+      "loss": 0.2656,
+      "step": 896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.051282051282051e-07,
+      "loss": 1.7608,
+      "step": 897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.046620046620047e-07,
+      "loss": 0.7667,
+      "step": 898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.041958041958042e-07,
+      "loss": 0.4371,
+      "step": 899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.037296037296036e-07,
+      "loss": 0.817,
+      "step": 900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.032634032634032e-07,
+      "loss": 0.9314,
+      "step": 901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.027972027972028e-07,
+      "loss": 0.2962,
+      "step": 902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.023310023310023e-07,
+      "loss": 0.2777,
+      "step": 903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.018648018648018e-07,
+      "loss": 0.1973,
+      "step": 904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.013986013986013e-07,
+      "loss": 0.0843,
+      "step": 905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.009324009324009e-07,
+      "loss": 1.0896,
+      "step": 906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.004662004662005e-07,
+      "loss": 0.1639,
+      "step": 907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6e-07,
+      "loss": 1.1745,
+      "step": 908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.995337995337995e-07,
+      "loss": 1.1945,
+      "step": 909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.990675990675991e-07,
+      "loss": 0.9512,
+      "step": 910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.986013986013985e-07,
+      "loss": 0.2707,
+      "step": 911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.98135198135198e-07,
+      "loss": 0.6397,
+      "step": 912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.976689976689976e-07,
+      "loss": 0.3658,
+      "step": 913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.972027972027972e-07,
+      "loss": 0.8314,
+      "step": 914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.967365967365967e-07,
+      "loss": 1.3255,
+      "step": 915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.962703962703962e-07,
+      "loss": 1.1513,
+      "step": 916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.958041958041958e-07,
+      "loss": 0.643,
+      "step": 917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.953379953379954e-07,
+      "loss": 0.322,
+      "step": 918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.948717948717949e-07,
+      "loss": 0.6063,
+      "step": 919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.944055944055944e-07,
+      "loss": 0.1805,
+      "step": 920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.93939393939394e-07,
+      "loss": 1.65,
+      "step": 921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.934731934731934e-07,
+      "loss": 0.1634,
+      "step": 922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.930069930069929e-07,
+      "loss": 0.6185,
+      "step": 923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.925407925407925e-07,
+      "loss": 0.2162,
+      "step": 924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.920745920745921e-07,
+      "loss": 0.7929,
+      "step": 925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.916083916083916e-07,
+      "loss": 0.1463,
+      "step": 926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.911421911421911e-07,
+      "loss": 0.5874,
+      "step": 927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.906759906759906e-07,
+      "loss": 1.5184,
+      "step": 928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.902097902097902e-07,
+      "loss": 1.3625,
+      "step": 929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.897435897435898e-07,
+      "loss": 0.6798,
+      "step": 930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.892773892773893e-07,
+      "loss": 1.0501,
+      "step": 931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.888111888111887e-07,
+      "loss": 0.0814,
+      "step": 932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.883449883449883e-07,
+      "loss": 0.6375,
+      "step": 933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.878787878787878e-07,
+      "loss": 0.5101,
+      "step": 934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.874125874125873e-07,
+      "loss": 0.3023,
+      "step": 935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.869463869463869e-07,
+      "loss": 0.5563,
+      "step": 936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.864801864801865e-07,
+      "loss": 0.4692,
+      "step": 937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.86013986013986e-07,
+      "loss": 0.1816,
+      "step": 938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.855477855477855e-07,
+      "loss": 0.3124,
+      "step": 939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.85081585081585e-07,
+      "loss": 0.2883,
+      "step": 940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.846153846153847e-07,
+      "loss": 0.1705,
+      "step": 941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.841491841491842e-07,
+      "loss": 0.0912,
+      "step": 942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.836829836829836e-07,
+      "loss": 0.4108,
+      "step": 943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832167832167831e-07,
+      "loss": 0.7952,
+      "step": 944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.827505827505827e-07,
+      "loss": 0.7365,
+      "step": 945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.822843822843822e-07,
+      "loss": 0.953,
+      "step": 946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.818181818181818e-07,
+      "loss": 0.2147,
+      "step": 947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.813519813519814e-07,
+      "loss": 0.3313,
+      "step": 948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.808857808857809e-07,
+      "loss": 0.2236,
+      "step": 949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.804195804195804e-07,
+      "loss": 1.3323,
+      "step": 950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.799533799533799e-07,
+      "loss": 0.3831,
+      "step": 951
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.794871794871795e-07,
+      "loss": 0.269,
+      "step": 952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.790209790209791e-07,
+      "loss": 1.2104,
+      "step": 953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.785547785547785e-07,
+      "loss": 0.3631,
+      "step": 954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.78088578088578e-07,
+      "loss": 0.5794,
+      "step": 955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.776223776223776e-07,
+      "loss": 0.5668,
+      "step": 956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.771561771561771e-07,
+      "loss": 0.7525,
+      "step": 957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.766899766899766e-07,
+      "loss": 0.358,
+      "step": 958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.762237762237762e-07,
+      "loss": 0.7374,
+      "step": 959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.757575757575758e-07,
+      "loss": 0.4397,
+      "step": 960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.752913752913753e-07,
+      "loss": 1.4356,
+      "step": 961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.748251748251748e-07,
+      "loss": 0.0815,
+      "step": 962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.743589743589743e-07,
+      "loss": 0.7777,
+      "step": 963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.73892773892774e-07,
+      "loss": 0.9275,
+      "step": 964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.734265734265734e-07,
+      "loss": 0.364,
+      "step": 965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.729603729603729e-07,
+      "loss": 0.4032,
+      "step": 966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.724941724941724e-07,
+      "loss": 0.517,
+      "step": 967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.72027972027972e-07,
+      "loss": 0.592,
+      "step": 968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.715617715617715e-07,
+      "loss": 0.5364,
+      "step": 969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.710955710955711e-07,
+      "loss": 0.6995,
+      "step": 970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.706293706293706e-07,
+      "loss": 0.704,
+      "step": 971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.701631701631702e-07,
+      "loss": 0.3601,
+      "step": 972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.696969696969697e-07,
+      "loss": 0.2056,
+      "step": 973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.692307692307692e-07,
+      "loss": 0.618,
+      "step": 974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.687645687645686e-07,
+      "loss": 0.0553,
+      "step": 975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.682983682983683e-07,
+      "loss": 0.9146,
+      "step": 976
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.678321678321678e-07,
+      "loss": 0.8259,
+      "step": 977
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.673659673659673e-07,
+      "loss": 0.2095,
+      "step": 978
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.668997668997669e-07,
+      "loss": 0.6528,
+      "step": 979
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.664335664335664e-07,
+      "loss": 0.6213,
+      "step": 980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.659673659673659e-07,
+      "loss": 0.8254,
+      "step": 981
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.655011655011655e-07,
+      "loss": 0.2939,
+      "step": 982
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.650349650349651e-07,
+      "loss": 0.4255,
+      "step": 983
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.645687645687646e-07,
+      "loss": 0.2874,
+      "step": 984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.641025641025641e-07,
+      "loss": 0.7963,
+      "step": 985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.636363636363635e-07,
+      "loss": 1.018,
+      "step": 986
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.631701631701632e-07,
+      "loss": 0.7825,
+      "step": 987
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.627039627039627e-07,
+      "loss": 0.1608,
+      "step": 988
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.622377622377622e-07,
+      "loss": 0.5718,
+      "step": 989
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.617715617715617e-07,
+      "loss": 0.6325,
+      "step": 990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.613053613053613e-07,
+      "loss": 0.5383,
+      "step": 991
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.608391608391608e-07,
+      "loss": 0.4132,
+      "step": 992
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.603729603729604e-07,
+      "loss": 0.4742,
+      "step": 993
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.599067599067599e-07,
+      "loss": 0.5048,
+      "step": 994
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.594405594405595e-07,
+      "loss": 0.126,
+      "step": 995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.58974358974359e-07,
+      "loss": 0.2455,
+      "step": 996
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.585081585081584e-07,
+      "loss": 0.2728,
+      "step": 997
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.580419580419579e-07,
+      "loss": 0.4211,
+      "step": 998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.575757575757576e-07,
+      "loss": 0.1564,
+      "step": 999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.571095571095571e-07,
+      "loss": 0.474,
+      "step": 1000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.566433566433566e-07,
+      "loss": 0.5076,
+      "step": 1001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.561771561771561e-07,
+      "loss": 0.1431,
+      "step": 1002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.557109557109557e-07,
+      "loss": 0.274,
+      "step": 1003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.552447552447552e-07,
+      "loss": 0.3462,
+      "step": 1004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.547785547785548e-07,
+      "loss": 1.5451,
+      "step": 1005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.543123543123543e-07,
+      "loss": 0.2969,
+      "step": 1006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.538461538461539e-07,
+      "loss": 0.7122,
+      "step": 1007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.533799533799533e-07,
+      "loss": 0.4436,
+      "step": 1008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.529137529137528e-07,
+      "loss": 0.5099,
+      "step": 1009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.524475524475523e-07,
+      "loss": 0.7548,
+      "step": 1010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.51981351981352e-07,
+      "loss": 1.1795,
+      "step": 1011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.515151515151515e-07,
+      "loss": 0.2794,
+      "step": 1012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.51048951048951e-07,
+      "loss": 0.964,
+      "step": 1013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.505827505827506e-07,
+      "loss": 1.7137,
+      "step": 1014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.501165501165501e-07,
+      "loss": 0.3439,
+      "step": 1015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.496503496503497e-07,
+      "loss": 1.2303,
+      "step": 1016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.491841491841492e-07,
+      "loss": 0.0976,
+      "step": 1017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.487179487179488e-07,
+      "loss": 0.3131,
+      "step": 1018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.482517482517482e-07,
+      "loss": 0.1535,
+      "step": 1019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.477855477855477e-07,
+      "loss": 0.3926,
+      "step": 1020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.473193473193472e-07,
+      "loss": 0.2184,
+      "step": 1021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.468531468531469e-07,
+      "loss": 0.7352,
+      "step": 1022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.463869463869464e-07,
+      "loss": 0.1709,
+      "step": 1023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.459207459207459e-07,
+      "loss": 0.5934,
+      "step": 1024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.454545454545454e-07,
+      "loss": 0.1457,
+      "step": 1025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.44988344988345e-07,
+      "loss": 0.6743,
+      "step": 1026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.445221445221445e-07,
+      "loss": 0.5353,
+      "step": 1027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.440559440559441e-07,
+      "loss": 0.1233,
+      "step": 1028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.435897435897435e-07,
+      "loss": 0.7855,
+      "step": 1029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.431235431235431e-07,
+      "loss": 0.5688,
+      "step": 1030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.426573426573426e-07,
+      "loss": 0.8311,
+      "step": 1031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.421911421911421e-07,
+      "loss": 1.1425,
+      "step": 1032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.417249417249416e-07,
+      "loss": 0.3825,
+      "step": 1033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.412587412587413e-07,
+      "loss": 0.1429,
+      "step": 1034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.407925407925408e-07,
+      "loss": 0.3464,
+      "step": 1035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.403263403263403e-07,
+      "loss": 0.143,
+      "step": 1036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.398601398601398e-07,
+      "loss": 0.5071,
+      "step": 1037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.393939393939394e-07,
+      "loss": 1.5319,
+      "step": 1038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.38927738927739e-07,
+      "loss": 0.8046,
+      "step": 1039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.384615384615384e-07,
+      "loss": 0.172,
+      "step": 1040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.379953379953379e-07,
+      "loss": 0.4592,
+      "step": 1041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.375291375291375e-07,
+      "loss": 1.1533,
+      "step": 1042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.37062937062937e-07,
+      "loss": 0.7928,
+      "step": 1043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.365967365967365e-07,
+      "loss": 0.1959,
+      "step": 1044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.361305361305362e-07,
+      "loss": 0.0528,
+      "step": 1045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.356643356643357e-07,
+      "loss": 0.515,
+      "step": 1046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.351981351981352e-07,
+      "loss": 0.2359,
+      "step": 1047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.347319347319347e-07,
+      "loss": 0.6963,
+      "step": 1048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.342657342657343e-07,
+      "loss": 0.3876,
+      "step": 1049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.337995337995338e-07,
+      "loss": 0.8602,
+      "step": 1050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.333333333333333e-07,
+      "loss": 0.8207,
+      "step": 1051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.328671328671328e-07,
+      "loss": 0.2873,
+      "step": 1052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.324009324009324e-07,
+      "loss": 0.7289,
+      "step": 1053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.319347319347319e-07,
+      "loss": 0.1876,
+      "step": 1054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.314685314685314e-07,
+      "loss": 0.6438,
+      "step": 1055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.310023310023309e-07,
+      "loss": 0.4586,
+      "step": 1056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.305361305361306e-07,
+      "loss": 0.6102,
+      "step": 1057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.300699300699301e-07,
+      "loss": 0.2966,
+      "step": 1058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.296037296037296e-07,
+      "loss": 0.524,
+      "step": 1059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.291375291375291e-07,
+      "loss": 1.5293,
+      "step": 1060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.286713286713286e-07,
+      "loss": 0.3648,
+      "step": 1061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.282051282051282e-07,
+      "loss": 0.9094,
+      "step": 1062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.277389277389277e-07,
+      "loss": 0.5102,
+      "step": 1063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.272727272727272e-07,
+      "loss": 0.2422,
+      "step": 1064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.268065268065268e-07,
+      "loss": 0.4257,
+      "step": 1065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.263403263403263e-07,
+      "loss": 0.056,
+      "step": 1066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.258741258741258e-07,
+      "loss": 0.5555,
+      "step": 1067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.254079254079254e-07,
+      "loss": 0.8631,
+      "step": 1068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.24941724941725e-07,
+      "loss": 1.1202,
+      "step": 1069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.244755244755245e-07,
+      "loss": 0.3184,
+      "step": 1070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.24009324009324e-07,
+      "loss": 1.539,
+      "step": 1071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.235431235431234e-07,
+      "loss": 0.4841,
+      "step": 1072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.23076923076923e-07,
+      "loss": 1.0317,
+      "step": 1073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.226107226107226e-07,
+      "loss": 0.4278,
+      "step": 1074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.221445221445221e-07,
+      "loss": 0.1625,
+      "step": 1075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.216783216783217e-07,
+      "loss": 1.2915,
+      "step": 1076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.212121212121212e-07,
+      "loss": 0.8626,
+      "step": 1077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.207459207459207e-07,
+      "loss": 0.3601,
+      "step": 1078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.202797202797202e-07,
+      "loss": 1.2753,
+      "step": 1079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.198135198135199e-07,
+      "loss": 0.3853,
+      "step": 1080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.193473193473194e-07,
+      "loss": 0.1902,
+      "step": 1081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.188811188811189e-07,
+      "loss": 0.1175,
+      "step": 1082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.184149184149183e-07,
+      "loss": 0.2824,
+      "step": 1083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.179487179487179e-07,
+      "loss": 0.0866,
+      "step": 1084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.174825174825175e-07,
+      "loss": 0.1514,
+      "step": 1085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.17016317016317e-07,
+      "loss": 1.516,
+      "step": 1086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.165501165501165e-07,
+      "loss": 0.5197,
+      "step": 1087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.160839160839161e-07,
+      "loss": 0.4297,
+      "step": 1088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.156177156177156e-07,
+      "loss": 0.1319,
+      "step": 1089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.151515151515151e-07,
+      "loss": 0.1275,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.146853146853147e-07,
+      "loss": 0.1008,
+      "step": 1091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.142191142191143e-07,
+      "loss": 0.128,
+      "step": 1092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.137529137529138e-07,
+      "loss": 0.3994,
+      "step": 1093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.132867132867132e-07,
+      "loss": 1.0346,
+      "step": 1094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.128205128205127e-07,
+      "loss": 0.7775,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.123543123543123e-07,
+      "loss": 0.1383,
+      "step": 1096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.118881118881119e-07,
+      "loss": 1.1149,
+      "step": 1097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.114219114219114e-07,
+      "loss": 0.7578,
+      "step": 1098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.109557109557109e-07,
+      "loss": 0.993,
+      "step": 1099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.104895104895105e-07,
+      "loss": 0.8719,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1002331002331e-07,
+      "loss": 0.5209,
+      "step": 1101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.095571095571095e-07,
+      "loss": 0.6478,
+      "step": 1102
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.090909090909091e-07,
+      "loss": 0.1583,
+      "step": 1103
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.086247086247087e-07,
+      "loss": 0.8762,
+      "step": 1104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.081585081585081e-07,
+      "loss": 0.5676,
+      "step": 1105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.076923076923076e-07,
+      "loss": 0.483,
+      "step": 1106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.072261072261072e-07,
+      "loss": 0.3564,
+      "step": 1107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.067599067599067e-07,
+      "loss": 0.3168,
+      "step": 1108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.062937062937063e-07,
+      "loss": 0.335,
+      "step": 1109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.058275058275058e-07,
+      "loss": 0.5659,
+      "step": 1110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.053613053613054e-07,
+      "loss": 0.8586,
+      "step": 1111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.048951048951049e-07,
+      "loss": 0.6073,
+      "step": 1112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.044289044289044e-07,
+      "loss": 0.3322,
+      "step": 1113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03962703962704e-07,
+      "loss": 0.1048,
+      "step": 1114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.034965034965036e-07,
+      "loss": 0.561,
+      "step": 1115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03030303030303e-07,
+      "loss": 0.1852,
+      "step": 1116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.025641025641025e-07,
+      "loss": 0.859,
+      "step": 1117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.02097902097902e-07,
+      "loss": 0.721,
+      "step": 1118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.016317016317016e-07,
+      "loss": 0.9208,
+      "step": 1119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.011655011655012e-07,
+      "loss": 0.4768,
+      "step": 1120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.006993006993007e-07,
+      "loss": 1.0492,
+      "step": 1121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.002331002331002e-07,
+      "loss": 1.2038,
+      "step": 1122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.997668997668998e-07,
+      "loss": 0.1267,
+      "step": 1123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.993006993006993e-07,
+      "loss": 0.5164,
+      "step": 1124
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.988344988344988e-07,
+      "loss": 0.3369,
+      "step": 1125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.983682983682983e-07,
+      "loss": 0.2715,
+      "step": 1126
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.979020979020978e-07,
+      "loss": 0.5548,
+      "step": 1127
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.974358974358974e-07,
+      "loss": 0.3373,
+      "step": 1128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.969696969696969e-07,
+      "loss": 0.1075,
+      "step": 1129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.965034965034965e-07,
+      "loss": 0.2583,
+      "step": 1130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.96037296037296e-07,
+      "loss": 0.196,
+      "step": 1131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.955710955710956e-07,
+      "loss": 0.6667,
+      "step": 1132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.951048951048951e-07,
+      "loss": 1.2072,
+      "step": 1133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.946386946386946e-07,
+      "loss": 0.7056,
+      "step": 1134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.941724941724942e-07,
+      "loss": 0.9923,
+      "step": 1135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.937062937062936e-07,
+      "loss": 0.8537,
+      "step": 1136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.932400932400932e-07,
+      "loss": 1.2011,
+      "step": 1137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927738927738927e-07,
+      "loss": 0.6933,
+      "step": 1138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.923076923076923e-07,
+      "loss": 0.0973,
+      "step": 1139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.918414918414918e-07,
+      "loss": 0.6508,
+      "step": 1140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913752913752914e-07,
+      "loss": 0.5744,
+      "step": 1141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.909090909090909e-07,
+      "loss": 0.681,
+      "step": 1142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.904428904428905e-07,
+      "loss": 1.1797,
+      "step": 1143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8997668997669e-07,
+      "loss": 0.8247,
+      "step": 1144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.895104895104895e-07,
+      "loss": 0.5362,
+      "step": 1145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.890442890442891e-07,
+      "loss": 0.2926,
+      "step": 1146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.885780885780885e-07,
+      "loss": 0.3157,
+      "step": 1147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.88111888111888e-07,
+      "loss": 0.2962,
+      "step": 1148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.876456876456876e-07,
+      "loss": 0.0931,
+      "step": 1149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.871794871794871e-07,
+      "loss": 0.4025,
+      "step": 1150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.867132867132867e-07,
+      "loss": 0.8939,
+      "step": 1151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.862470862470862e-07,
+      "loss": 0.4307,
+      "step": 1152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.857808857808858e-07,
+      "loss": 0.6902,
+      "step": 1153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.853146853146853e-07,
+      "loss": 0.5368,
+      "step": 1154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.848484848484849e-07,
+      "loss": 0.7278,
+      "step": 1155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.843822843822844e-07,
+      "loss": 0.8736,
+      "step": 1156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.839160839160839e-07,
+      "loss": 0.6205,
+      "step": 1157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.834498834498834e-07,
+      "loss": 0.2612,
+      "step": 1158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.829836829836829e-07,
+      "loss": 0.3925,
+      "step": 1159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.825174825174824e-07,
+      "loss": 0.6927,
+      "step": 1160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.82051282051282e-07,
+      "loss": 0.2137,
+      "step": 1161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.815850815850815e-07,
+      "loss": 0.864,
+      "step": 1162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.811188811188811e-07,
+      "loss": 0.4197,
+      "step": 1163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.806526806526806e-07,
+      "loss": 0.2005,
+      "step": 1164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.801864801864802e-07,
+      "loss": 0.7415,
+      "step": 1165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797202797202797e-07,
+      "loss": 0.996,
+      "step": 1166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.792540792540793e-07,
+      "loss": 0.4479,
+      "step": 1167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.787878787878788e-07,
+      "loss": 1.5082,
+      "step": 1168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.783216783216783e-07,
+      "loss": 0.5187,
+      "step": 1169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.778554778554778e-07,
+      "loss": 2.1292,
+      "step": 1170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.773892773892773e-07,
+      "loss": 0.8727,
+      "step": 1171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.769230769230769e-07,
+      "loss": 1.1335,
+      "step": 1172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7645687645687646e-07,
+      "loss": 0.6399,
+      "step": 1173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.75990675990676e-07,
+      "loss": 0.102,
+      "step": 1174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.755244755244755e-07,
+      "loss": 1.1355,
+      "step": 1175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.750582750582751e-07,
+      "loss": 0.411,
+      "step": 1176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7459207459207457e-07,
+      "loss": 0.1679,
+      "step": 1177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.741258741258741e-07,
+      "loss": 0.5226,
+      "step": 1178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7365967365967365e-07,
+      "loss": 0.2692,
+      "step": 1179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.731934731934732e-07,
+      "loss": 0.6003,
+      "step": 1180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.727272727272727e-07,
+      "loss": 0.0839,
+      "step": 1181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.722610722610723e-07,
+      "loss": 0.4415,
+      "step": 1182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7179487179487176e-07,
+      "loss": 0.7383,
+      "step": 1183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.713286713286713e-07,
+      "loss": 0.6899,
+      "step": 1184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7086247086247085e-07,
+      "loss": 0.7013,
+      "step": 1185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.703962703962704e-07,
+      "loss": 0.0892,
+      "step": 1186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.699300699300699e-07,
+      "loss": 0.184,
+      "step": 1187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6946386946386947e-07,
+      "loss": 0.6213,
+      "step": 1188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6899766899766896e-07,
+      "loss": 0.5995,
+      "step": 1189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.685314685314685e-07,
+      "loss": 0.3634,
+      "step": 1190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6806526806526804e-07,
+      "loss": 0.5093,
+      "step": 1191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.675990675990676e-07,
+      "loss": 0.0741,
+      "step": 1192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6713286713286707e-07,
+      "loss": 0.1066,
+      "step": 1193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6666666666666666e-07,
+      "loss": 0.3505,
+      "step": 1194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6620046620046615e-07,
+      "loss": 0.4639,
+      "step": 1195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.657342657342657e-07,
+      "loss": 1.0891,
+      "step": 1196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6526806526806523e-07,
+      "loss": 0.6283,
+      "step": 1197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.648018648018648e-07,
+      "loss": 0.6146,
+      "step": 1198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.643356643356643e-07,
+      "loss": 0.4059,
+      "step": 1199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6386946386946386e-07,
+      "loss": 0.1354,
+      "step": 1200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6340326340326335e-07,
+      "loss": 0.3268,
+      "step": 1201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6293706293706294e-07,
+      "loss": 0.2105,
+      "step": 1202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.624708624708625e-07,
+      "loss": 0.4102,
+      "step": 1203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6200466200466197e-07,
+      "loss": 0.713,
+      "step": 1204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6153846153846156e-07,
+      "loss": 0.4751,
+      "step": 1205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6107226107226105e-07,
+      "loss": 0.7327,
+      "step": 1206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.606060606060606e-07,
+      "loss": 0.1659,
+      "step": 1207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6013986013986013e-07,
+      "loss": 0.7258,
+      "step": 1208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.596736596736597e-07,
+      "loss": 0.523,
+      "step": 1209
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5920745920745916e-07,
+      "loss": 0.735,
+      "step": 1210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5874125874125876e-07,
+      "loss": 0.8384,
+      "step": 1211
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5827505827505824e-07,
+      "loss": 0.2091,
+      "step": 1212
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.578088578088578e-07,
+      "loss": 0.1582,
+      "step": 1213
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.573426573426573e-07,
+      "loss": 0.3897,
+      "step": 1214
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5687645687645687e-07,
+      "loss": 0.3738,
+      "step": 1215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5641025641025636e-07,
+      "loss": 0.0506,
+      "step": 1216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5594405594405595e-07,
+      "loss": 0.7403,
+      "step": 1217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5547785547785544e-07,
+      "loss": 0.8138,
+      "step": 1218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.55011655011655e-07,
+      "loss": 0.1979,
+      "step": 1219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.545454545454545e-07,
+      "loss": 1.0579,
+      "step": 1220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5407925407925406e-07,
+      "loss": 0.2262,
+      "step": 1221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5361305361305355e-07,
+      "loss": 0.4402,
+      "step": 1222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5314685314685314e-07,
+      "loss": 0.7862,
+      "step": 1223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5268065268065263e-07,
+      "loss": 0.4156,
+      "step": 1224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.522144522144522e-07,
+      "loss": 0.3819,
+      "step": 1225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.517482517482517e-07,
+      "loss": 0.5583,
+      "step": 1226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5128205128205125e-07,
+      "loss": 0.2731,
+      "step": 1227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.508158508158508e-07,
+      "loss": 0.2563,
+      "step": 1228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5034965034965034e-07,
+      "loss": 1.1903,
+      "step": 1229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.498834498834498e-07,
+      "loss": 0.475,
+      "step": 1230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.494172494172494e-07,
+      "loss": 0.645,
+      "step": 1231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.489510489510489e-07,
+      "loss": 0.1444,
+      "step": 1232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4848484848484845e-07,
+      "loss": 0.3612,
+      "step": 1233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.48018648018648e-07,
+      "loss": 0.2932,
+      "step": 1234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4755244755244753e-07,
+      "loss": 0.2459,
+      "step": 1235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4708624708624707e-07,
+      "loss": 0.2916,
+      "step": 1236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.466200466200466e-07,
+      "loss": 1.5194,
+      "step": 1237
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4615384615384615e-07,
+      "loss": 0.3185,
+      "step": 1238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4568764568764564e-07,
+      "loss": 0.991,
+      "step": 1239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4522144522144524e-07,
+      "loss": 0.4337,
+      "step": 1240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.447552447552447e-07,
+      "loss": 0.2362,
+      "step": 1241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4428904428904427e-07,
+      "loss": 0.3634,
+      "step": 1242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.438228438228438e-07,
+      "loss": 0.2726,
+      "step": 1243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4335664335664335e-07,
+      "loss": 0.4459,
+      "step": 1244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4289044289044284e-07,
+      "loss": 0.4892,
+      "step": 1245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4242424242424243e-07,
+      "loss": 0.1055,
+      "step": 1246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.419580419580419e-07,
+      "loss": 0.2721,
+      "step": 1247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.414918414918415e-07,
+      "loss": 0.7644,
+      "step": 1248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.41025641025641e-07,
+      "loss": 0.1927,
+      "step": 1249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4055944055944054e-07,
+      "loss": 0.2564,
+      "step": 1250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.400932400932401e-07,
+      "loss": 0.3293,
+      "step": 1251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.396270396270396e-07,
+      "loss": 0.3229,
+      "step": 1252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.391608391608391e-07,
+      "loss": 1.3036,
+      "step": 1253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.386946386946387e-07,
+      "loss": 1.8625,
+      "step": 1254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.382284382284382e-07,
+      "loss": 0.4797,
+      "step": 1255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3776223776223774e-07,
+      "loss": 0.4565,
+      "step": 1256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.372960372960373e-07,
+      "loss": 0.8831,
+      "step": 1257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.368298368298368e-07,
+      "loss": 0.3425,
+      "step": 1258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.363636363636363e-07,
+      "loss": 0.5193,
+      "step": 1259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.358974358974359e-07,
+      "loss": 0.3593,
+      "step": 1260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.354312354312354e-07,
+      "loss": 0.1399,
+      "step": 1261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3496503496503493e-07,
+      "loss": 0.2038,
+      "step": 1262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.3449883449883447e-07,
+      "loss": 0.4916,
+      "step": 1263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.34032634032634e-07,
+      "loss": 0.2766,
+      "step": 1264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.335664335664335e-07,
+      "loss": 0.2359,
+      "step": 1265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.331002331002331e-07,
+      "loss": 0.219,
+      "step": 1266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.3263403263403263e-07,
+      "loss": 1.4228,
+      "step": 1267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.321678321678321e-07,
+      "loss": 0.1485,
+      "step": 1268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.317016317016317e-07,
+      "loss": 0.4534,
+      "step": 1269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.312354312354312e-07,
+      "loss": 0.3072,
+      "step": 1270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.307692307692308e-07,
+      "loss": 0.6947,
+      "step": 1271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.303030303030303e-07,
+      "loss": 0.8795,
+      "step": 1272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2983682983682983e-07,
+      "loss": 0.5017,
+      "step": 1273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2937062937062937e-07,
+      "loss": 1.2819,
+      "step": 1274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.289044289044289e-07,
+      "loss": 0.851,
+      "step": 1275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.284382284382284e-07,
+      "loss": 0.098,
+      "step": 1276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.27972027972028e-07,
+      "loss": 0.7469,
+      "step": 1277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.275058275058275e-07,
+      "loss": 0.7321,
+      "step": 1278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.27039627039627e-07,
+      "loss": 1.2458,
+      "step": 1279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2657342657342656e-07,
+      "loss": 0.1579,
+      "step": 1280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.261072261072261e-07,
+      "loss": 1.5567,
+      "step": 1281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.256410256410256e-07,
+      "loss": 0.0259,
+      "step": 1282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.251748251748252e-07,
+      "loss": 0.1621,
+      "step": 1283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.247086247086247e-07,
+      "loss": 0.9221,
+      "step": 1284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.242424242424242e-07,
+      "loss": 0.5552,
+      "step": 1285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2377622377622376e-07,
+      "loss": 0.2544,
+      "step": 1286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.233100233100233e-07,
+      "loss": 1.3287,
+      "step": 1287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.228438228438228e-07,
+      "loss": 0.3917,
+      "step": 1288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.223776223776224e-07,
+      "loss": 0.4212,
+      "step": 1289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2191142191142187e-07,
+      "loss": 0.1529,
+      "step": 1290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.214452214452214e-07,
+      "loss": 0.4211,
+      "step": 1291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2097902097902095e-07,
+      "loss": 0.909,
+      "step": 1292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.205128205128205e-07,
+      "loss": 0.3658,
+      "step": 1293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2004662004662e-07,
+      "loss": 0.8337,
+      "step": 1294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1958041958041957e-07,
+      "loss": 0.4213,
+      "step": 1295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1911421911421906e-07,
+      "loss": 0.219,
+      "step": 1296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1864801864801866e-07,
+      "loss": 0.4667,
+      "step": 1297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1818181818181814e-07,
+      "loss": 0.353,
+      "step": 1298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.177156177156177e-07,
+      "loss": 0.736,
+      "step": 1299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.172494172494173e-07,
+      "loss": 0.6538,
+      "step": 1300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1678321678321677e-07,
+      "loss": 0.1388,
+      "step": 1301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.163170163170163e-07,
+      "loss": 0.5082,
+      "step": 1302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1585081585081585e-07,
+      "loss": 0.2311,
+      "step": 1303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.153846153846154e-07,
+      "loss": 1.0096,
+      "step": 1304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.149184149184149e-07,
+      "loss": 0.2214,
+      "step": 1305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1445221445221447e-07,
+      "loss": 1.2773,
+      "step": 1306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1398601398601396e-07,
+      "loss": 0.3612,
+      "step": 1307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.135198135198135e-07,
+      "loss": 0.6598,
+      "step": 1308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1305361305361304e-07,
+      "loss": 1.0211,
+      "step": 1309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.125874125874126e-07,
+      "loss": 0.1795,
+      "step": 1310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1212121212121207e-07,
+      "loss": 0.7285,
+      "step": 1311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1165501165501167e-07,
+      "loss": 0.1263,
+      "step": 1312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1118881118881115e-07,
+      "loss": 0.821,
+      "step": 1313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.107226107226107e-07,
+      "loss": 0.2866,
+      "step": 1314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1025641025641024e-07,
+      "loss": 0.0596,
+      "step": 1315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.097902097902098e-07,
+      "loss": 0.5687,
+      "step": 1316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0932400932400927e-07,
+      "loss": 0.6788,
+      "step": 1317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0885780885780886e-07,
+      "loss": 0.1845,
+      "step": 1318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0839160839160835e-07,
+      "loss": 0.489,
+      "step": 1319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0792540792540794e-07,
+      "loss": 0.0495,
+      "step": 1320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0745920745920743e-07,
+      "loss": 0.8414,
+      "step": 1321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0699300699300697e-07,
+      "loss": 0.3227,
+      "step": 1322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.065268065268065e-07,
+      "loss": 0.8349,
+      "step": 1323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0606060606060605e-07,
+      "loss": 1.1866,
+      "step": 1324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0559440559440554e-07,
+      "loss": 0.2182,
+      "step": 1325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0512820512820514e-07,
+      "loss": 0.1291,
+      "step": 1326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.046620046620046e-07,
+      "loss": 0.2604,
+      "step": 1327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0419580419580416e-07,
+      "loss": 0.4235,
+      "step": 1328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.037296037296037e-07,
+      "loss": 0.3586,
+      "step": 1329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0326340326340325e-07,
+      "loss": 0.6187,
+      "step": 1330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.027972027972028e-07,
+      "loss": 0.3303,
+      "step": 1331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0233100233100233e-07,
+      "loss": 0.3421,
+      "step": 1332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0186480186480187e-07,
+      "loss": 0.2695,
+      "step": 1333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0139860139860136e-07,
+      "loss": 0.3929,
+      "step": 1334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0093240093240095e-07,
+      "loss": 0.426,
+      "step": 1335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0046620046620044e-07,
+      "loss": 0.7118,
+      "step": 1336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4e-07,
+      "loss": 0.5733,
+      "step": 1337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.995337995337995e-07,
+      "loss": 0.1757,
+      "step": 1338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9906759906759906e-07,
+      "loss": 0.4542,
+      "step": 1339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9860139860139855e-07,
+      "loss": 0.7215,
+      "step": 1340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9813519813519815e-07,
+      "loss": 0.5505,
+      "step": 1341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9766899766899763e-07,
+      "loss": 0.7659,
+      "step": 1342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.972027972027972e-07,
+      "loss": 0.0711,
+      "step": 1343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.967365967365967e-07,
+      "loss": 0.3459,
+      "step": 1344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9627039627039626e-07,
+      "loss": 0.1586,
+      "step": 1345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.958041958041958e-07,
+      "loss": 0.3245,
+      "step": 1346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9533799533799534e-07,
+      "loss": 0.1906,
+      "step": 1347
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9487179487179483e-07,
+      "loss": 0.2402,
+      "step": 1348
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.944055944055944e-07,
+      "loss": 0.5882,
+      "step": 1349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.939393939393939e-07,
+      "loss": 0.2183,
+      "step": 1350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9347319347319345e-07,
+      "loss": 0.4524,
+      "step": 1351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.93006993006993e-07,
+      "loss": 0.3278,
+      "step": 1352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9254079254079253e-07,
+      "loss": 0.2408,
+      "step": 1353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.92074592074592e-07,
+      "loss": 0.2897,
+      "step": 1354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.916083916083916e-07,
+      "loss": 2.0289,
+      "step": 1355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.911421911421911e-07,
+      "loss": 1.544,
+      "step": 1356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9067599067599065e-07,
+      "loss": 0.1822,
+      "step": 1357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.902097902097902e-07,
+      "loss": 0.8613,
+      "step": 1358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8974358974358973e-07,
+      "loss": 1.3098,
+      "step": 1359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.892773892773892e-07,
+      "loss": 0.3362,
+      "step": 1360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.888111888111888e-07,
+      "loss": 0.6526,
+      "step": 1361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.883449883449883e-07,
+      "loss": 0.1702,
+      "step": 1362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8787878787878784e-07,
+      "loss": 0.309,
+      "step": 1363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8741258741258743e-07,
+      "loss": 0.1863,
+      "step": 1364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.869463869463869e-07,
+      "loss": 0.3104,
+      "step": 1365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8648018648018646e-07,
+      "loss": 0.21,
+      "step": 1366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.86013986013986e-07,
+      "loss": 0.4667,
+      "step": 1367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8554778554778554e-07,
+      "loss": 0.1641,
+      "step": 1368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8508158508158503e-07,
+      "loss": 0.1878,
+      "step": 1369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8461538461538463e-07,
+      "loss": 1.2571,
+      "step": 1370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.841491841491841e-07,
+      "loss": 0.5864,
+      "step": 1371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.836829836829837e-07,
+      "loss": 0.347,
+      "step": 1372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.832167832167832e-07,
+      "loss": 0.806,
+      "step": 1373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8275058275058274e-07,
+      "loss": 0.1996,
+      "step": 1374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.822843822843823e-07,
+      "loss": 0.4851,
+      "step": 1375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.818181818181818e-07,
+      "loss": 0.3881,
+      "step": 1376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.813519813519813e-07,
+      "loss": 1.6872,
+      "step": 1377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.808857808857809e-07,
+      "loss": 0.139,
+      "step": 1378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.804195804195804e-07,
+      "loss": 1.2867,
+      "step": 1379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7995337995337993e-07,
+      "loss": 0.5601,
+      "step": 1380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7948717948717947e-07,
+      "loss": 0.7435,
+      "step": 1381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.79020979020979e-07,
+      "loss": 0.1817,
+      "step": 1382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.785547785547785e-07,
+      "loss": 0.8116,
+      "step": 1383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.780885780885781e-07,
+      "loss": 0.7225,
+      "step": 1384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.776223776223776e-07,
+      "loss": 0.4472,
+      "step": 1385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.771561771561771e-07,
+      "loss": 0.1688,
+      "step": 1386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7668997668997667e-07,
+      "loss": 1.2835,
+      "step": 1387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.762237762237762e-07,
+      "loss": 0.0914,
+      "step": 1388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.757575757575757e-07,
+      "loss": 0.6415,
+      "step": 1389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.752913752913753e-07,
+      "loss": 0.3129,
+      "step": 1390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.748251748251748e-07,
+      "loss": 0.5049,
+      "step": 1391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.743589743589743e-07,
+      "loss": 0.4286,
+      "step": 1392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7389277389277386e-07,
+      "loss": 1.0051,
+      "step": 1393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.734265734265734e-07,
+      "loss": 0.7236,
+      "step": 1394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.72960372960373e-07,
+      "loss": 0.4819,
+      "step": 1395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.724941724941725e-07,
+      "loss": 0.3052,
+      "step": 1396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.72027972027972e-07,
+      "loss": 0.2693,
+      "step": 1397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.7156177156177157e-07,
+      "loss": 0.2321,
+      "step": 1398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.710955710955711e-07,
+      "loss": 0.8198,
+      "step": 1399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.706293706293706e-07,
+      "loss": 0.5529,
+      "step": 1400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.701631701631702e-07,
+      "loss": 0.584,
+      "step": 1401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.696969696969697e-07,
+      "loss": 0.1469,
+      "step": 1402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.692307692307692e-07,
+      "loss": 0.2186,
+      "step": 1403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6876456876456876e-07,
+      "loss": 0.7869,
+      "step": 1404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.682983682983683e-07,
+      "loss": 0.3615,
+      "step": 1405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.678321678321678e-07,
+      "loss": 0.5277,
+      "step": 1406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.673659673659674e-07,
+      "loss": 0.6714,
+      "step": 1407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6689976689976687e-07,
+      "loss": 0.2446,
+      "step": 1408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.664335664335664e-07,
+      "loss": 1.0722,
+      "step": 1409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6596736596736595e-07,
+      "loss": 0.2744,
+      "step": 1410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.655011655011655e-07,
+      "loss": 0.3506,
+      "step": 1411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.65034965034965e-07,
+      "loss": 0.4294,
+      "step": 1412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.645687645687646e-07,
+      "loss": 0.5776,
+      "step": 1413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.6410256410256406e-07,
+      "loss": 0.1175,
+      "step": 1414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.636363636363636e-07,
+      "loss": 1.2198,
+      "step": 1415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6317016317016315e-07,
+      "loss": 1.1838,
+      "step": 1416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.627039627039627e-07,
+      "loss": 0.8986,
+      "step": 1417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.622377622377622e-07,
+      "loss": 1.0012,
+      "step": 1418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6177156177156177e-07,
+      "loss": 0.0339,
+      "step": 1419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6130536130536126e-07,
+      "loss": 0.6918,
+      "step": 1420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6083916083916085e-07,
+      "loss": 0.6328,
+      "step": 1421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.6037296037296034e-07,
+      "loss": 1.7465,
+      "step": 1422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.599067599067599e-07,
+      "loss": 0.1275,
+      "step": 1423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.594405594405594e-07,
+      "loss": 1.5815,
+      "step": 1424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5897435897435896e-07,
+      "loss": 1.2902,
+      "step": 1425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5850815850815845e-07,
+      "loss": 0.8122,
+      "step": 1426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5804195804195805e-07,
+      "loss": 1.2009,
+      "step": 1427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.575757575757576e-07,
+      "loss": 0.3992,
+      "step": 1428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.571095571095571e-07,
+      "loss": 0.147,
+      "step": 1429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5664335664335667e-07,
+      "loss": 0.5454,
+      "step": 1430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5617715617715616e-07,
+      "loss": 0.7166,
+      "step": 1431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.557109557109557e-07,
+      "loss": 0.6006,
+      "step": 1432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5524475524475524e-07,
+      "loss": 0.6968,
+      "step": 1433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.547785547785548e-07,
+      "loss": 0.6101,
+      "step": 1434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5431235431235427e-07,
+      "loss": 0.2454,
+      "step": 1435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5384615384615386e-07,
+      "loss": 0.2373,
+      "step": 1436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.5337995337995335e-07,
+      "loss": 1.0522,
+      "step": 1437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.529137529137529e-07,
+      "loss": 0.64,
+      "step": 1438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.5244755244755243e-07,
+      "loss": 0.2423,
+      "step": 1439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.51981351981352e-07,
+      "loss": 0.0531,
+      "step": 1440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.5151515151515146e-07,
+      "loss": 0.5059,
+      "step": 1441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.5104895104895106e-07,
+      "loss": 0.1081,
+      "step": 1442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.5058275058275054e-07,
+      "loss": 1.2175,
+      "step": 1443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.5011655011655014e-07,
+      "loss": 0.6046,
+      "step": 1444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4965034965034963e-07,
+      "loss": 0.4606,
+      "step": 1445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4918414918414917e-07,
+      "loss": 0.1571,
+      "step": 1446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.487179487179487e-07,
+      "loss": 0.1047,
+      "step": 1447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4825174825174825e-07,
+      "loss": 0.557,
+      "step": 1448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4778554778554774e-07,
+      "loss": 0.2232,
+      "step": 1449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4731934731934733e-07,
+      "loss": 1.4656,
+      "step": 1450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.468531468531468e-07,
+      "loss": 1.6771,
+      "step": 1451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4638694638694636e-07,
+      "loss": 0.7617,
+      "step": 1452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.459207459207459e-07,
+      "loss": 0.3658,
+      "step": 1453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4545454545454544e-07,
+      "loss": 1.0289,
+      "step": 1454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4498834498834493e-07,
+      "loss": 0.7998,
+      "step": 1455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.445221445221445e-07,
+      "loss": 0.843,
+      "step": 1456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.44055944055944e-07,
+      "loss": 0.9358,
+      "step": 1457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4358974358974356e-07,
+      "loss": 0.8865,
+      "step": 1458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4312354312354315e-07,
+      "loss": 1.3271,
+      "step": 1459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.4265734265734264e-07,
+      "loss": 1.4355,
+      "step": 1460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.421911421911422e-07,
+      "loss": 0.7029,
+      "step": 1461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.417249417249417e-07,
+      "loss": 0.0583,
+      "step": 1462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.4125874125874126e-07,
+      "loss": 0.593,
+      "step": 1463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.4079254079254075e-07,
+      "loss": 0.437,
+      "step": 1464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.4032634032634034e-07,
+      "loss": 1.2481,
+      "step": 1465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3986013986013983e-07,
+      "loss": 0.709,
+      "step": 1466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.393939393939394e-07,
+      "loss": 1.5351,
+      "step": 1467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.389277389277389e-07,
+      "loss": 0.4167,
+      "step": 1468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3846153846153845e-07,
+      "loss": 0.5588,
+      "step": 1469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.37995337995338e-07,
+      "loss": 0.3102,
+      "step": 1470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3752913752913754e-07,
+      "loss": 0.2864,
+      "step": 1471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.37062937062937e-07,
+      "loss": 1.182,
+      "step": 1472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.365967365967366e-07,
+      "loss": 0.0754,
+      "step": 1473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.361305361305361e-07,
+      "loss": 0.6244,
+      "step": 1474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3566433566433565e-07,
+      "loss": 0.3345,
+      "step": 1475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.351981351981352e-07,
+      "loss": 0.3211,
+      "step": 1476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3473193473193473e-07,
+      "loss": 0.1351,
+      "step": 1477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.342657342657342e-07,
+      "loss": 0.5221,
+      "step": 1478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.337995337995338e-07,
+      "loss": 0.037,
+      "step": 1479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.333333333333333e-07,
+      "loss": 0.3932,
+      "step": 1480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3286713286713284e-07,
+      "loss": 0.601,
+      "step": 1481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.324009324009324e-07,
+      "loss": 0.7133,
+      "step": 1482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.319347319347319e-07,
+      "loss": 0.3957,
+      "step": 1483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.314685314685314e-07,
+      "loss": 0.3856,
+      "step": 1484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.31002331002331e-07,
+      "loss": 0.4021,
+      "step": 1485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.305361305361305e-07,
+      "loss": 0.0793,
+      "step": 1486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3006993006993004e-07,
+      "loss": 0.5343,
+      "step": 1487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.296037296037296e-07,
+      "loss": 0.933,
+      "step": 1488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.291375291375291e-07,
+      "loss": 0.9162,
+      "step": 1489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.286713286713286e-07,
+      "loss": 0.4474,
+      "step": 1490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.282051282051282e-07,
+      "loss": 0.5749,
+      "step": 1491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2773892773892774e-07,
+      "loss": 0.3872,
+      "step": 1492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.272727272727273e-07,
+      "loss": 0.3035,
+      "step": 1493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.268065268065268e-07,
+      "loss": 0.7938,
+      "step": 1494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.263403263403263e-07,
+      "loss": 0.4001,
+      "step": 1495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.258741258741259e-07,
+      "loss": 0.7983,
+      "step": 1496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.254079254079254e-07,
+      "loss": 0.6909,
+      "step": 1497
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.2494172494172493e-07,
+      "loss": 0.4042,
+      "step": 1498
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.244755244755245e-07,
+      "loss": 0.0988,
+      "step": 1499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.24009324009324e-07,
+      "loss": 0.2075,
+      "step": 1500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.235431235431235e-07,
+      "loss": 0.6584,
+      "step": 1501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.230769230769231e-07,
+      "loss": 0.2205,
+      "step": 1502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.226107226107226e-07,
+      "loss": 0.0599,
+      "step": 1503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2214452214452213e-07,
+      "loss": 0.576,
+      "step": 1504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2167832167832167e-07,
+      "loss": 0.615,
+      "step": 1505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.212121212121212e-07,
+      "loss": 0.3847,
+      "step": 1506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.207459207459207e-07,
+      "loss": 0.1775,
+      "step": 1507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.202797202797203e-07,
+      "loss": 1.7855,
+      "step": 1508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.198135198135198e-07,
+      "loss": 0.0631,
+      "step": 1509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.193473193473193e-07,
+      "loss": 0.6971,
+      "step": 1510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1888111888111886e-07,
+      "loss": 1.6215,
+      "step": 1511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.184149184149184e-07,
+      "loss": 0.7908,
+      "step": 1512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.179487179487179e-07,
+      "loss": 0.6709,
+      "step": 1513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.174825174825175e-07,
+      "loss": 0.4172,
+      "step": 1514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.17016317016317e-07,
+      "loss": 0.3634,
+      "step": 1515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.165501165501165e-07,
+      "loss": 0.6608,
+      "step": 1516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1608391608391606e-07,
+      "loss": 0.9899,
+      "step": 1517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.156177156177156e-07,
+      "loss": 0.0281,
+      "step": 1518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1515151515151514e-07,
+      "loss": 1.1283,
+      "step": 1519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.146853146853147e-07,
+      "loss": 0.5552,
+      "step": 1520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1421911421911417e-07,
+      "loss": 1.2349,
+      "step": 1521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.1375291375291376e-07,
+      "loss": 0.4198,
+      "step": 1522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.132867132867133e-07,
+      "loss": 0.7522,
+      "step": 1523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.128205128205128e-07,
+      "loss": 0.2119,
+      "step": 1524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.123543123543124e-07,
+      "loss": 0.3811,
+      "step": 1525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.118881118881119e-07,
+      "loss": 0.9058,
+      "step": 1526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.114219114219114e-07,
+      "loss": 0.6299,
+      "step": 1527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.1095571095571096e-07,
+      "loss": 0.6501,
+      "step": 1528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.104895104895105e-07,
+      "loss": 0.2915,
+      "step": 1529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.1002331002331e-07,
+      "loss": 0.0594,
+      "step": 1530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.095571095571096e-07,
+      "loss": 0.1813,
+      "step": 1531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0909090909090907e-07,
+      "loss": 0.1437,
+      "step": 1532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.086247086247086e-07,
+      "loss": 0.3951,
+      "step": 1533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0815850815850815e-07,
+      "loss": 0.3267,
+      "step": 1534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.076923076923077e-07,
+      "loss": 1.5226,
+      "step": 1535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.072261072261072e-07,
+      "loss": 0.2739,
+      "step": 1536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0675990675990677e-07,
+      "loss": 0.5863,
+      "step": 1537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0629370629370626e-07,
+      "loss": 0.2751,
+      "step": 1538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.058275058275058e-07,
+      "loss": 0.1985,
+      "step": 1539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0536130536130534e-07,
+      "loss": 0.6588,
+      "step": 1540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.048951048951049e-07,
+      "loss": 0.4095,
+      "step": 1541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0442890442890437e-07,
+      "loss": 0.4658,
+      "step": 1542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0396270396270397e-07,
+      "loss": 1.1994,
+      "step": 1543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0349650349650345e-07,
+      "loss": 1.2612,
+      "step": 1544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 0.0438,
+      "step": 1545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0256410256410254e-07,
+      "loss": 0.5883,
+      "step": 1546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.020979020979021e-07,
+      "loss": 0.831,
+      "step": 1547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.016317016317016e-07,
+      "loss": 0.2011,
+      "step": 1548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0116550116550116e-07,
+      "loss": 0.576,
+      "step": 1549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0069930069930065e-07,
+      "loss": 1.029,
+      "step": 1550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0023310023310024e-07,
+      "loss": 0.4671,
+      "step": 1551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9976689976689973e-07,
+      "loss": 0.1979,
+      "step": 1552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9930069930069927e-07,
+      "loss": 0.7685,
+      "step": 1553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.988344988344988e-07,
+      "loss": 0.5203,
+      "step": 1554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9836829836829835e-07,
+      "loss": 0.8891,
+      "step": 1555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.979020979020979e-07,
+      "loss": 0.6489,
+      "step": 1556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9743589743589744e-07,
+      "loss": 0.4612,
+      "step": 1557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.96969696969697e-07,
+      "loss": 0.7395,
+      "step": 1558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9650349650349647e-07,
+      "loss": 0.6285,
+      "step": 1559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9603729603729606e-07,
+      "loss": 0.1626,
+      "step": 1560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9557109557109555e-07,
+      "loss": 0.7646,
+      "step": 1561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.951048951048951e-07,
+      "loss": 0.4275,
+      "step": 1562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9463869463869463e-07,
+      "loss": 1.0024,
+      "step": 1563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9417249417249417e-07,
+      "loss": 0.2467,
+      "step": 1564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9370629370629366e-07,
+      "loss": 0.591,
+      "step": 1565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9324009324009325e-07,
+      "loss": 0.2874,
+      "step": 1566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9277389277389274e-07,
+      "loss": 0.21,
+      "step": 1567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9230769230769234e-07,
+      "loss": 0.788,
+      "step": 1568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.918414918414918e-07,
+      "loss": 0.9125,
+      "step": 1569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9137529137529136e-07,
+      "loss": 0.3494,
+      "step": 1570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.909090909090909e-07,
+      "loss": 0.3747,
+      "step": 1571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9044289044289045e-07,
+      "loss": 0.8857,
+      "step": 1572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8997668997668993e-07,
+      "loss": 0.7421,
+      "step": 1573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8951048951048953e-07,
+      "loss": 0.8125,
+      "step": 1574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.89044289044289e-07,
+      "loss": 0.9891,
+      "step": 1575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8857808857808856e-07,
+      "loss": 1.3214,
+      "step": 1576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.881118881118881e-07,
+      "loss": 0.446,
+      "step": 1577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8764568764568764e-07,
+      "loss": 1.0014,
+      "step": 1578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8717948717948713e-07,
+      "loss": 0.6438,
+      "step": 1579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.867132867132867e-07,
+      "loss": 0.7656,
+      "step": 1580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.862470862470862e-07,
+      "loss": 0.5182,
+      "step": 1581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8578088578088575e-07,
+      "loss": 0.7928,
+      "step": 1582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.853146853146853e-07,
+      "loss": 1.541,
+      "step": 1583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8484848484848483e-07,
+      "loss": 0.2998,
+      "step": 1584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.843822843822843e-07,
+      "loss": 0.479,
+      "step": 1585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.839160839160839e-07,
+      "loss": 0.5773,
+      "step": 1586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8344988344988346e-07,
+      "loss": 0.9397,
+      "step": 1587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8298368298368295e-07,
+      "loss": 0.4507,
+      "step": 1588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8251748251748254e-07,
+      "loss": 1.1734,
+      "step": 1589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8205128205128203e-07,
+      "loss": 0.2449,
+      "step": 1590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.815850815850816e-07,
+      "loss": 0.425,
+      "step": 1591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.811188811188811e-07,
+      "loss": 0.5738,
+      "step": 1592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8065268065268065e-07,
+      "loss": 0.1427,
+      "step": 1593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.801864801864802e-07,
+      "loss": 0.6067,
+      "step": 1594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7972027972027973e-07,
+      "loss": 1.0921,
+      "step": 1595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.792540792540792e-07,
+      "loss": 0.2345,
+      "step": 1596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.787878787878788e-07,
+      "loss": 0.8504,
+      "step": 1597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.783216783216783e-07,
+      "loss": 0.1967,
+      "step": 1598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7785547785547784e-07,
+      "loss": 0.5786,
+      "step": 1599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.773892773892774e-07,
+      "loss": 0.8167,
+      "step": 1600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7692307692307693e-07,
+      "loss": 0.772,
+      "step": 1601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.764568764568764e-07,
+      "loss": 0.7497,
+      "step": 1602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.75990675990676e-07,
+      "loss": 0.4227,
+      "step": 1603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.755244755244755e-07,
+      "loss": 0.0984,
+      "step": 1604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7505827505827504e-07,
+      "loss": 0.5225,
+      "step": 1605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.745920745920746e-07,
+      "loss": 0.4677,
+      "step": 1606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.741258741258741e-07,
+      "loss": 0.6035,
+      "step": 1607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.736596736596736e-07,
+      "loss": 0.0826,
+      "step": 1608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.731934731934732e-07,
+      "loss": 0.2909,
+      "step": 1609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.727272727272727e-07,
+      "loss": 0.3522,
+      "step": 1610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7226107226107223e-07,
+      "loss": 0.6949,
+      "step": 1611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.7179487179487177e-07,
+      "loss": 0.989,
+      "step": 1612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.713286713286713e-07,
+      "loss": 0.7244,
+      "step": 1613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.708624708624708e-07,
+      "loss": 0.3174,
+      "step": 1614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.703962703962704e-07,
+      "loss": 0.8447,
+      "step": 1615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.699300699300699e-07,
+      "loss": 1.5845,
+      "step": 1616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.694638694638695e-07,
+      "loss": 0.5163,
+      "step": 1617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6899766899766897e-07,
+      "loss": 0.489,
+      "step": 1618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.685314685314685e-07,
+      "loss": 0.5757,
+      "step": 1619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.680652680652681e-07,
+      "loss": 0.5097,
+      "step": 1620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.675990675990676e-07,
+      "loss": 0.3903,
+      "step": 1621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6713286713286713e-07,
+      "loss": 1.5265,
+      "step": 1622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6666666666666667e-07,
+      "loss": 0.1445,
+      "step": 1623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.662004662004662e-07,
+      "loss": 0.5257,
+      "step": 1624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.657342657342657e-07,
+      "loss": 0.0632,
+      "step": 1625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.652680652680653e-07,
+      "loss": 0.3332,
+      "step": 1626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.648018648018648e-07,
+      "loss": 0.218,
+      "step": 1627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.643356643356643e-07,
+      "loss": 0.9621,
+      "step": 1628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6386946386946387e-07,
+      "loss": 1.2902,
+      "step": 1629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.634032634032634e-07,
+      "loss": 0.4245,
+      "step": 1630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.629370629370629e-07,
+      "loss": 0.6604,
+      "step": 1631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.624708624708625e-07,
+      "loss": 0.6068,
+      "step": 1632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.62004662004662e-07,
+      "loss": 0.7032,
+      "step": 1633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.615384615384615e-07,
+      "loss": 0.6871,
+      "step": 1634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.6107226107226106e-07,
+      "loss": 0.0967,
+      "step": 1635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.606060606060606e-07,
+      "loss": 0.2784,
+      "step": 1636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.601398601398601e-07,
+      "loss": 0.7009,
+      "step": 1637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.596736596736597e-07,
+      "loss": 0.6564,
+      "step": 1638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5920745920745917e-07,
+      "loss": 0.125,
+      "step": 1639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5874125874125877e-07,
+      "loss": 0.3194,
+      "step": 1640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5827505827505825e-07,
+      "loss": 0.1009,
+      "step": 1641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.578088578088578e-07,
+      "loss": 0.2448,
+      "step": 1642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5734265734265734e-07,
+      "loss": 0.1767,
+      "step": 1643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.568764568764569e-07,
+      "loss": 0.1597,
+      "step": 1644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5641025641025636e-07,
+      "loss": 0.1461,
+      "step": 1645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5594405594405596e-07,
+      "loss": 1.5658,
+      "step": 1646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5547785547785545e-07,
+      "loss": 1.5189,
+      "step": 1647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.55011655011655e-07,
+      "loss": 0.1235,
+      "step": 1648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5454545454545453e-07,
+      "loss": 0.9222,
+      "step": 1649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5407925407925407e-07,
+      "loss": 0.31,
+      "step": 1650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.536130536130536e-07,
+      "loss": 0.1995,
+      "step": 1651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5314685314685315e-07,
+      "loss": 0.5268,
+      "step": 1652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.526806526806527e-07,
+      "loss": 0.266,
+      "step": 1653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.522144522144522e-07,
+      "loss": 1.2065,
+      "step": 1654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.517482517482518e-07,
+      "loss": 0.4241,
+      "step": 1655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5128205128205126e-07,
+      "loss": 0.1006,
+      "step": 1656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.508158508158508e-07,
+      "loss": 0.3112,
+      "step": 1657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.5034965034965035e-07,
+      "loss": 0.7931,
+      "step": 1658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.498834498834499e-07,
+      "loss": 0.3829,
+      "step": 1659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.494172494172494e-07,
+      "loss": 0.5949,
+      "step": 1660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.489510489510489e-07,
+      "loss": 0.5507,
+      "step": 1661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4848484848484846e-07,
+      "loss": 0.5213,
+      "step": 1662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.48018648018648e-07,
+      "loss": 0.6773,
+      "step": 1663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4755244755244754e-07,
+      "loss": 0.7077,
+      "step": 1664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.470862470862471e-07,
+      "loss": 0.1645,
+      "step": 1665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.466200466200466e-07,
+      "loss": 0.1907,
+      "step": 1666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4615384615384616e-07,
+      "loss": 0.7946,
+      "step": 1667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.456876456876457e-07,
+      "loss": 1.0268,
+      "step": 1668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4522144522144525e-07,
+      "loss": 0.7084,
+      "step": 1669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4475524475524473e-07,
+      "loss": 0.2162,
+      "step": 1670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.442890442890443e-07,
+      "loss": 0.761,
+      "step": 1671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.438228438228438e-07,
+      "loss": 0.9584,
+      "step": 1672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4335664335664336e-07,
+      "loss": 0.6289,
+      "step": 1673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.428904428904429e-07,
+      "loss": 0.6175,
+      "step": 1674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4242424242424244e-07,
+      "loss": 0.488,
+      "step": 1675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4195804195804193e-07,
+      "loss": 0.1891,
+      "step": 1676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4149184149184147e-07,
+      "loss": 0.2208,
+      "step": 1677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.41025641025641e-07,
+      "loss": 0.5303,
+      "step": 1678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4055944055944055e-07,
+      "loss": 0.1323,
+      "step": 1679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.400932400932401e-07,
+      "loss": 0.7057,
+      "step": 1680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3962703962703963e-07,
+      "loss": 0.5883,
+      "step": 1681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.391608391608392e-07,
+      "loss": 0.1379,
+      "step": 1682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3869463869463866e-07,
+      "loss": 1.5102,
+      "step": 1683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3822843822843823e-07,
+      "loss": 0.2628,
+      "step": 1684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3776223776223774e-07,
+      "loss": 0.3345,
+      "step": 1685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3729603729603729e-07,
+      "loss": 0.3618,
+      "step": 1686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3682983682983683e-07,
+      "loss": 0.0736,
+      "step": 1687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3636363636363634e-07,
+      "loss": 0.1508,
+      "step": 1688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3589743589743588e-07,
+      "loss": 0.5686,
+      "step": 1689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3543123543123542e-07,
+      "loss": 0.4445,
+      "step": 1690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3496503496503494e-07,
+      "loss": 0.4633,
+      "step": 1691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3449883449883448e-07,
+      "loss": 0.733,
+      "step": 1692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3403263403263402e-07,
+      "loss": 0.459,
+      "step": 1693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3356643356643353e-07,
+      "loss": 0.6983,
+      "step": 1694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3310023310023308e-07,
+      "loss": 0.7462,
+      "step": 1695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3263403263403262e-07,
+      "loss": 0.7754,
+      "step": 1696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3216783216783216e-07,
+      "loss": 0.6875,
+      "step": 1697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3170163170163167e-07,
+      "loss": 0.5242,
+      "step": 1698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3123543123543124e-07,
+      "loss": 0.1085,
+      "step": 1699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3076923076923078e-07,
+      "loss": 0.691,
+      "step": 1700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.303030303030303e-07,
+      "loss": 0.3883,
+      "step": 1701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2983682983682984e-07,
+      "loss": 0.1069,
+      "step": 1702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2937062937062938e-07,
+      "loss": 1.0295,
+      "step": 1703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.289044289044289e-07,
+      "loss": 0.6692,
+      "step": 1704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2843822843822843e-07,
+      "loss": 1.5654,
+      "step": 1705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2797202797202797e-07,
+      "loss": 0.2303,
+      "step": 1706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.275058275058275e-07,
+      "loss": 0.6799,
+      "step": 1707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2703962703962703e-07,
+      "loss": 0.2522,
+      "step": 1708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2657342657342657e-07,
+      "loss": 0.3149,
+      "step": 1709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.261072261072261e-07,
+      "loss": 0.957,
+      "step": 1710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2564102564102563e-07,
+      "loss": 0.7722,
+      "step": 1711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2517482517482517e-07,
+      "loss": 0.999,
+      "step": 1712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.247086247086247e-07,
+      "loss": 0.9721,
+      "step": 1713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2424242424242422e-07,
+      "loss": 0.3104,
+      "step": 1714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2377622377622377e-07,
+      "loss": 0.5159,
+      "step": 1715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.233100233100233e-07,
+      "loss": 0.06,
+      "step": 1716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2284382284382282e-07,
+      "loss": 0.1987,
+      "step": 1717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2237762237762236e-07,
+      "loss": 0.4115,
+      "step": 1718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.219114219114219e-07,
+      "loss": 0.1015,
+      "step": 1719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2144522144522142e-07,
+      "loss": 0.0442,
+      "step": 1720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2097902097902096e-07,
+      "loss": 0.5134,
+      "step": 1721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.205128205128205e-07,
+      "loss": 1.0708,
+      "step": 1722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2004662004662004e-07,
+      "loss": 0.483,
+      "step": 1723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1958041958041956e-07,
+      "loss": 1.0553,
+      "step": 1724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.191142191142191e-07,
+      "loss": 0.9492,
+      "step": 1725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1864801864801864e-07,
+      "loss": 0.6813,
+      "step": 1726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1818181818181815e-07,
+      "loss": 0.5283,
+      "step": 1727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.177156177156177e-07,
+      "loss": 0.7738,
+      "step": 1728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1724941724941723e-07,
+      "loss": 0.3289,
+      "step": 1729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1678321678321675e-07,
+      "loss": 0.5422,
+      "step": 1730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1631701631701632e-07,
+      "loss": 0.1193,
+      "step": 1731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1585081585081586e-07,
+      "loss": 0.6549,
+      "step": 1732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153846153846154e-07,
+      "loss": 0.5241,
+      "step": 1733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1491841491841491e-07,
+      "loss": 0.4139,
+      "step": 1734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1445221445221446e-07,
+      "loss": 0.3249,
+      "step": 1735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13986013986014e-07,
+      "loss": 0.9192,
+      "step": 1736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.135198135198135e-07,
+      "loss": 0.5683,
+      "step": 1737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1305361305361305e-07,
+      "loss": 0.4813,
+      "step": 1738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.125874125874126e-07,
+      "loss": 0.7281,
+      "step": 1739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.121212121212121e-07,
+      "loss": 0.8647,
+      "step": 1740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1165501165501165e-07,
+      "loss": 0.7861,
+      "step": 1741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.111888111888112e-07,
+      "loss": 0.2636,
+      "step": 1742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.107226107226107e-07,
+      "loss": 0.6368,
+      "step": 1743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1025641025641025e-07,
+      "loss": 1.34,
+      "step": 1744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0979020979020979e-07,
+      "loss": 0.1893,
+      "step": 1745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0932400932400933e-07,
+      "loss": 0.449,
+      "step": 1746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0885780885780884e-07,
+      "loss": 0.258,
+      "step": 1747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0839160839160838e-07,
+      "loss": 0.0948,
+      "step": 1748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0792540792540792e-07,
+      "loss": 0.4116,
+      "step": 1749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0745920745920744e-07,
+      "loss": 0.3487,
+      "step": 1750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0699300699300698e-07,
+      "loss": 0.3793,
+      "step": 1751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0652680652680652e-07,
+      "loss": 0.1081,
+      "step": 1752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0606060606060604e-07,
+      "loss": 0.782,
+      "step": 1753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0559440559440558e-07,
+      "loss": 0.4457,
+      "step": 1754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0512820512820512e-07,
+      "loss": 0.5373,
+      "step": 1755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0466200466200463e-07,
+      "loss": 0.9892,
+      "step": 1756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0419580419580417e-07,
+      "loss": 0.1907,
+      "step": 1757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0372960372960372e-07,
+      "loss": 0.3256,
+      "step": 1758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0326340326340326e-07,
+      "loss": 0.736,
+      "step": 1759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0279720279720277e-07,
+      "loss": 0.7725,
+      "step": 1760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023310023310023e-07,
+      "loss": 0.8172,
+      "step": 1761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0186480186480185e-07,
+      "loss": 0.3932,
+      "step": 1762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013986013986014e-07,
+      "loss": 0.4086,
+      "step": 1763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0093240093240094e-07,
+      "loss": 0.5208,
+      "step": 1764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0046620046620048e-07,
+      "loss": 0.274,
+      "step": 1765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-07,
+      "loss": 0.8061,
+      "step": 1766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9953379953379953e-07,
+      "loss": 0.2664,
+      "step": 1767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9906759906759907e-07,
+      "loss": 0.4867,
+      "step": 1768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.986013986013986e-07,
+      "loss": 0.4676,
+      "step": 1769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9813519813519813e-07,
+      "loss": 0.7368,
+      "step": 1770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9766899766899767e-07,
+      "loss": 0.923,
+      "step": 1771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.972027972027972e-07,
+      "loss": 0.5327,
+      "step": 1772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9673659673659673e-07,
+      "loss": 0.7288,
+      "step": 1773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9627039627039627e-07,
+      "loss": 0.8219,
+      "step": 1774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.958041958041958e-07,
+      "loss": 0.1029,
+      "step": 1775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9533799533799532e-07,
+      "loss": 0.8173,
+      "step": 1776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9487179487179486e-07,
+      "loss": 0.2941,
+      "step": 1777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.944055944055944e-07,
+      "loss": 1.1697,
+      "step": 1778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9393939393939392e-07,
+      "loss": 0.4623,
+      "step": 1779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9347319347319346e-07,
+      "loss": 0.3121,
+      "step": 1780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.93006993006993e-07,
+      "loss": 0.2865,
+      "step": 1781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9254079254079252e-07,
+      "loss": 1.0403,
+      "step": 1782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9207459207459206e-07,
+      "loss": 0.0787,
+      "step": 1783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916083916083916e-07,
+      "loss": 0.4815,
+      "step": 1784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9114219114219114e-07,
+      "loss": 0.5237,
+      "step": 1785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9067599067599065e-07,
+      "loss": 0.3401,
+      "step": 1786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.902097902097902e-07,
+      "loss": 0.5594,
+      "step": 1787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8974358974358974e-07,
+      "loss": 0.5127,
+      "step": 1788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8927738927738925e-07,
+      "loss": 0.6448,
+      "step": 1789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.888111888111888e-07,
+      "loss": 0.4002,
+      "step": 1790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8834498834498833e-07,
+      "loss": 0.5568,
+      "step": 1791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8787878787878785e-07,
+      "loss": 0.3772,
+      "step": 1792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.874125874125874e-07,
+      "loss": 1.2252,
+      "step": 1793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8694638694638693e-07,
+      "loss": 0.7885,
+      "step": 1794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.864801864801865e-07,
+      "loss": 0.5085,
+      "step": 1795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.86013986013986e-07,
+      "loss": 0.2229,
+      "step": 1796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8554778554778555e-07,
+      "loss": 0.8408,
+      "step": 1797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.850815850815851e-07,
+      "loss": 0.5788,
+      "step": 1798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.846153846153846e-07,
+      "loss": 0.3297,
+      "step": 1799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8414918414918415e-07,
+      "loss": 0.3722,
+      "step": 1800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.836829836829837e-07,
+      "loss": 0.895,
+      "step": 1801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.832167832167832e-07,
+      "loss": 0.1902,
+      "step": 1802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8275058275058275e-07,
+      "loss": 0.2921,
+      "step": 1803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.822843822843823e-07,
+      "loss": 1.4293,
+      "step": 1804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.818181818181818e-07,
+      "loss": 1.0306,
+      "step": 1805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8135198135198134e-07,
+      "loss": 1.2959,
+      "step": 1806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8088578088578088e-07,
+      "loss": 0.1917,
+      "step": 1807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8041958041958043e-07,
+      "loss": 0.4836,
+      "step": 1808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7995337995337994e-07,
+      "loss": 0.8307,
+      "step": 1809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7948717948717948e-07,
+      "loss": 0.6805,
+      "step": 1810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7902097902097902e-07,
+      "loss": 0.109,
+      "step": 1811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7855477855477854e-07,
+      "loss": 1.0053,
+      "step": 1812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7808857808857808e-07,
+      "loss": 0.7912,
+      "step": 1813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7762237762237762e-07,
+      "loss": 0.9319,
+      "step": 1814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7715617715617713e-07,
+      "loss": 0.6224,
+      "step": 1815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7668997668997668e-07,
+      "loss": 0.3392,
+      "step": 1816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7622377622377622e-07,
+      "loss": 0.2877,
+      "step": 1817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7575757575757573e-07,
+      "loss": 0.8381,
+      "step": 1818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7529137529137527e-07,
+      "loss": 0.1874,
+      "step": 1819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7482517482517481e-07,
+      "loss": 0.2552,
+      "step": 1820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7435897435897435e-07,
+      "loss": 0.985,
+      "step": 1821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7389277389277387e-07,
+      "loss": 0.9177,
+      "step": 1822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.734265734265734e-07,
+      "loss": 0.3246,
+      "step": 1823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7296037296037295e-07,
+      "loss": 0.1172,
+      "step": 1824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7249417249417247e-07,
+      "loss": 1.0465,
+      "step": 1825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.72027972027972e-07,
+      "loss": 0.8806,
+      "step": 1826
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7156177156177157e-07,
+      "loss": 1.0652,
+      "step": 1827
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.710955710955711e-07,
+      "loss": 0.9329,
+      "step": 1828
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7062937062937063e-07,
+      "loss": 0.2663,
+      "step": 1829
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7016317016317017e-07,
+      "loss": 0.1381,
+      "step": 1830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696969696969697e-07,
+      "loss": 0.714,
+      "step": 1831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6923076923076923e-07,
+      "loss": 0.6471,
+      "step": 1832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6876456876456877e-07,
+      "loss": 0.5948,
+      "step": 1833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.682983682983683e-07,
+      "loss": 0.2481,
+      "step": 1834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6783216783216782e-07,
+      "loss": 0.5776,
+      "step": 1835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6736596736596737e-07,
+      "loss": 0.3299,
+      "step": 1836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.668997668997669e-07,
+      "loss": 1.2736,
+      "step": 1837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6643356643356642e-07,
+      "loss": 0.3607,
+      "step": 1838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6596736596736596e-07,
+      "loss": 0.116,
+      "step": 1839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.655011655011655e-07,
+      "loss": 0.8938,
+      "step": 1840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6503496503496502e-07,
+      "loss": 0.9088,
+      "step": 1841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6456876456876456e-07,
+      "loss": 0.7343,
+      "step": 1842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.641025641025641e-07,
+      "loss": 0.5076,
+      "step": 1843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6363636363636364e-07,
+      "loss": 0.2789,
+      "step": 1844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6317016317016316e-07,
+      "loss": 0.355,
+      "step": 1845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.627039627039627e-07,
+      "loss": 0.3869,
+      "step": 1846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6223776223776224e-07,
+      "loss": 0.3441,
+      "step": 1847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6177156177156175e-07,
+      "loss": 0.258,
+      "step": 1848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.613053613053613e-07,
+      "loss": 0.5665,
+      "step": 1849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6083916083916083e-07,
+      "loss": 0.1177,
+      "step": 1850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6037296037296035e-07,
+      "loss": 0.8811,
+      "step": 1851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.599067599067599e-07,
+      "loss": 0.3936,
+      "step": 1852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5944055944055943e-07,
+      "loss": 0.6328,
+      "step": 1853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5897435897435895e-07,
+      "loss": 0.7479,
+      "step": 1854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.585081585081585e-07,
+      "loss": 1.1534,
+      "step": 1855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5804195804195803e-07,
+      "loss": 0.6522,
+      "step": 1856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5757575757575757e-07,
+      "loss": 0.4813,
+      "step": 1857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5710955710955708e-07,
+      "loss": 0.1439,
+      "step": 1858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5664335664335665e-07,
+      "loss": 0.7418,
+      "step": 1859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.561771561771562e-07,
+      "loss": 0.4764,
+      "step": 1860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.557109557109557e-07,
+      "loss": 0.4162,
+      "step": 1861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5524475524475525e-07,
+      "loss": 1.1426,
+      "step": 1862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.547785547785548e-07,
+      "loss": 0.5982,
+      "step": 1863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.543123543123543e-07,
+      "loss": 0.4001,
+      "step": 1864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5384615384615385e-07,
+      "loss": 0.669,
+      "step": 1865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5337995337995339e-07,
+      "loss": 0.8787,
+      "step": 1866
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.529137529137529e-07,
+      "loss": 0.5912,
+      "step": 1867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5244755244755244e-07,
+      "loss": 0.8552,
+      "step": 1868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5198135198135198e-07,
+      "loss": 0.6685,
+      "step": 1869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5151515151515152e-07,
+      "loss": 0.0635,
+      "step": 1870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5104895104895104e-07,
+      "loss": 0.3925,
+      "step": 1871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5058275058275058e-07,
+      "loss": 0.4146,
+      "step": 1872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.5011655011655012e-07,
+      "loss": 0.5817,
+      "step": 1873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4965034965034964e-07,
+      "loss": 0.1036,
+      "step": 1874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4918414918414918e-07,
+      "loss": 0.3317,
+      "step": 1875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4871794871794872e-07,
+      "loss": 0.9239,
+      "step": 1876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4825174825174823e-07,
+      "loss": 0.1761,
+      "step": 1877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4778554778554777e-07,
+      "loss": 0.427,
+      "step": 1878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4731934731934731e-07,
+      "loss": 0.4161,
+      "step": 1879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4685314685314683e-07,
+      "loss": 0.1125,
+      "step": 1880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4638694638694637e-07,
+      "loss": 0.9213,
+      "step": 1881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.459207459207459e-07,
+      "loss": 0.4962,
+      "step": 1882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4545454545454545e-07,
+      "loss": 1.5972,
+      "step": 1883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4498834498834497e-07,
+      "loss": 0.4891,
+      "step": 1884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.445221445221445e-07,
+      "loss": 0.0952,
+      "step": 1885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4405594405594405e-07,
+      "loss": 0.2664,
+      "step": 1886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4358974358974356e-07,
+      "loss": 0.1926,
+      "step": 1887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.431235431235431e-07,
+      "loss": 0.2777,
+      "step": 1888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4265734265734265e-07,
+      "loss": 0.6575,
+      "step": 1889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4219114219114216e-07,
+      "loss": 0.5084,
+      "step": 1890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4172494172494173e-07,
+      "loss": 0.9407,
+      "step": 1891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4125874125874127e-07,
+      "loss": 0.0799,
+      "step": 1892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.407925407925408e-07,
+      "loss": 0.5357,
+      "step": 1893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4032634032634033e-07,
+      "loss": 0.5279,
+      "step": 1894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3986013986013987e-07,
+      "loss": 0.6588,
+      "step": 1895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.393939393939394e-07,
+      "loss": 0.5373,
+      "step": 1896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3892773892773892e-07,
+      "loss": 0.659,
+      "step": 1897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.3846153846153846e-07,
+      "loss": 1.006,
+      "step": 1898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.37995337995338e-07,
+      "loss": 0.3832,
+      "step": 1899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3752913752913752e-07,
+      "loss": 0.2409,
+      "step": 1900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3706293706293706e-07,
+      "loss": 0.1567,
+      "step": 1901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.365967365967366e-07,
+      "loss": 0.3107,
+      "step": 1902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3613053613053612e-07,
+      "loss": 0.2395,
+      "step": 1903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3566433566433566e-07,
+      "loss": 0.4833,
+      "step": 1904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.351981351981352e-07,
+      "loss": 0.3769,
+      "step": 1905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3473193473193474e-07,
+      "loss": 1.022,
+      "step": 1906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3426573426573425e-07,
+      "loss": 0.4667,
+      "step": 1907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.337995337995338e-07,
+      "loss": 0.2727,
+      "step": 1908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3333333333333334e-07,
+      "loss": 0.0392,
+      "step": 1909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3286713286713285e-07,
+      "loss": 0.6575,
+      "step": 1910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.324009324009324e-07,
+      "loss": 1.2816,
+      "step": 1911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3193473193473193e-07,
+      "loss": 0.5399,
+      "step": 1912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3146853146853145e-07,
+      "loss": 0.6426,
+      "step": 1913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.31002331002331e-07,
+      "loss": 0.2385,
+      "step": 1914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3053613053613053e-07,
+      "loss": 0.3742,
+      "step": 1915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.3006993006993004e-07,
+      "loss": 0.3162,
+      "step": 1916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2960372960372959e-07,
+      "loss": 0.196,
+      "step": 1917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2913752913752913e-07,
+      "loss": 0.3469,
+      "step": 1918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2867132867132867e-07,
+      "loss": 0.2828,
+      "step": 1919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2820512820512818e-07,
+      "loss": 0.6131,
+      "step": 1920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2773892773892772e-07,
+      "loss": 0.2872,
+      "step": 1921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2727272727272726e-07,
+      "loss": 0.271,
+      "step": 1922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.268065268065268e-07,
+      "loss": 0.2391,
+      "step": 1923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2634032634032635e-07,
+      "loss": 0.1684,
+      "step": 1924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.258741258741259e-07,
+      "loss": 0.5997,
+      "step": 1925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.254079254079254e-07,
+      "loss": 0.4311,
+      "step": 1926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2494172494172494e-07,
+      "loss": 0.7218,
+      "step": 1927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2447552447552446e-07,
+      "loss": 0.4189,
+      "step": 1928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.24009324009324e-07,
+      "loss": 0.4674,
+      "step": 1929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2354312354312354e-07,
+      "loss": 1.2126,
+      "step": 1930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2307692307692308e-07,
+      "loss": 0.285,
+      "step": 1931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2261072261072262e-07,
+      "loss": 0.058,
+      "step": 1932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2214452214452214e-07,
+      "loss": 0.1183,
+      "step": 1933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2167832167832168e-07,
+      "loss": 0.2955,
+      "step": 1934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2121212121212122e-07,
+      "loss": 0.4498,
+      "step": 1935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2074592074592073e-07,
+      "loss": 1.0001,
+      "step": 1936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2027972027972028e-07,
+      "loss": 0.9384,
+      "step": 1937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1981351981351982e-07,
+      "loss": 0.2978,
+      "step": 1938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1934731934731933e-07,
+      "loss": 0.1867,
+      "step": 1939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1888111888111887e-07,
+      "loss": 0.7664,
+      "step": 1940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1841491841491841e-07,
+      "loss": 0.6014,
+      "step": 1941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1794871794871794e-07,
+      "loss": 0.1212,
+      "step": 1942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1748251748251747e-07,
+      "loss": 0.1994,
+      "step": 1943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1701631701631701e-07,
+      "loss": 1.0024,
+      "step": 1944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1655011655011654e-07,
+      "loss": 0.8309,
+      "step": 1945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1608391608391608e-07,
+      "loss": 0.5289,
+      "step": 1946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1561771561771562e-07,
+      "loss": 0.7891,
+      "step": 1947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1515151515151515e-07,
+      "loss": 0.6403,
+      "step": 1948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1468531468531469e-07,
+      "loss": 0.4047,
+      "step": 1949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1421911421911422e-07,
+      "loss": 0.137,
+      "step": 1950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1375291375291374e-07,
+      "loss": 0.3262,
+      "step": 1951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1328671328671329e-07,
+      "loss": 0.0669,
+      "step": 1952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1282051282051281e-07,
+      "loss": 0.2893,
+      "step": 1953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1235431235431235e-07,
+      "loss": 0.1837,
+      "step": 1954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1188811188811188e-07,
+      "loss": 0.2471,
+      "step": 1955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1142191142191141e-07,
+      "loss": 0.398,
+      "step": 1956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1095571095571095e-07,
+      "loss": 0.8484,
+      "step": 1957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1048951048951048e-07,
+      "loss": 0.3953,
+      "step": 1958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1002331002331002e-07,
+      "loss": 0.9679,
+      "step": 1959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0955710955710955e-07,
+      "loss": 0.5586,
+      "step": 1960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0909090909090908e-07,
+      "loss": 0.5779,
+      "step": 1961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0862470862470862e-07,
+      "loss": 0.0881,
+      "step": 1962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.0815850815850816e-07,
+      "loss": 0.4927,
+      "step": 1963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.076923076923077e-07,
+      "loss": 0.3755,
+      "step": 1964
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0722610722610723e-07,
+      "loss": 0.5031,
+      "step": 1965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0675990675990676e-07,
+      "loss": 0.3342,
+      "step": 1966
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.062937062937063e-07,
+      "loss": 0.8059,
+      "step": 1967
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0582750582750582e-07,
+      "loss": 0.2609,
+      "step": 1968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0536130536130535e-07,
+      "loss": 0.3843,
+      "step": 1969
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0489510489510489e-07,
+      "loss": 0.3957,
+      "step": 1970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0442890442890442e-07,
+      "loss": 0.1975,
+      "step": 1971
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0396270396270396e-07,
+      "loss": 0.1827,
+      "step": 1972
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0349650349650349e-07,
+      "loss": 0.5841,
+      "step": 1973
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0303030303030302e-07,
+      "loss": 0.727,
+      "step": 1974
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0256410256410256e-07,
+      "loss": 0.4419,
+      "step": 1975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0209790209790209e-07,
+      "loss": 0.3244,
+      "step": 1976
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0163170163170163e-07,
+      "loss": 0.521,
+      "step": 1977
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0116550116550116e-07,
+      "loss": 0.9415,
+      "step": 1978
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.006993006993007e-07,
+      "loss": 0.4638,
+      "step": 1979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0023310023310024e-07,
+      "loss": 0.5992,
+      "step": 1980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.976689976689977e-08,
+      "loss": 0.2705,
+      "step": 1981
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.93006993006993e-08,
+      "loss": 0.7123,
+      "step": 1982
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.883449883449883e-08,
+      "loss": 0.2317,
+      "step": 1983
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.836829836829836e-08,
+      "loss": 1.1987,
+      "step": 1984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.79020979020979e-08,
+      "loss": 1.566,
+      "step": 1985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.743589743589743e-08,
+      "loss": 1.0482,
+      "step": 1986
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.696969696969696e-08,
+      "loss": 0.5798,
+      "step": 1987
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.65034965034965e-08,
+      "loss": 0.5794,
+      "step": 1988
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.603729603729603e-08,
+      "loss": 0.7694,
+      "step": 1989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.557109557109557e-08,
+      "loss": 0.6602,
+      "step": 1990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.51048951048951e-08,
+      "loss": 0.3369,
+      "step": 1991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.463869463869463e-08,
+      "loss": 0.3462,
+      "step": 1992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.417249417249417e-08,
+      "loss": 0.214,
+      "step": 1993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.37062937062937e-08,
+      "loss": 0.3956,
+      "step": 1994
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.324009324009325e-08,
+      "loss": 0.3004,
+      "step": 1995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.277389277389278e-08,
+      "loss": 1.5116,
+      "step": 1996
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.23076923076923e-08,
+      "loss": 0.239,
+      "step": 1997
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.184149184149185e-08,
+      "loss": 1.2595,
+      "step": 1998
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.137529137529137e-08,
+      "loss": 1.0604,
+      "step": 1999
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.09090909090909e-08,
+      "loss": 0.5663,
+      "step": 2000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.044289044289044e-08,
+      "loss": 1.1512,
+      "step": 2001
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.997668997668997e-08,
+      "loss": 0.31,
+      "step": 2002
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.951048951048951e-08,
+      "loss": 0.5903,
+      "step": 2003
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.904428904428904e-08,
+      "loss": 0.4808,
+      "step": 2004
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.857808857808857e-08,
+      "loss": 0.5296,
+      "step": 2005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.811188811188811e-08,
+      "loss": 0.2065,
+      "step": 2006
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.764568764568764e-08,
+      "loss": 0.1513,
+      "step": 2007
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.717948717948718e-08,
+      "loss": 0.7535,
+      "step": 2008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.67132867132867e-08,
+      "loss": 0.0929,
+      "step": 2009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.624708624708623e-08,
+      "loss": 0.8117,
+      "step": 2010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.578088578088579e-08,
+      "loss": 0.3245,
+      "step": 2011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.531468531468532e-08,
+      "loss": 0.2128,
+      "step": 2012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.484848484848486e-08,
+      "loss": 0.2581,
+      "step": 2013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.438228438228438e-08,
+      "loss": 0.8501,
+      "step": 2014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.391608391608391e-08,
+      "loss": 0.6805,
+      "step": 2015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.344988344988345e-08,
+      "loss": 0.7879,
+      "step": 2016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.298368298368298e-08,
+      "loss": 0.3862,
+      "step": 2017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.251748251748251e-08,
+      "loss": 0.1935,
+      "step": 2018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.205128205128205e-08,
+      "loss": 0.0556,
+      "step": 2019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.158508158508158e-08,
+      "loss": 1.6391,
+      "step": 2020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.111888111888112e-08,
+      "loss": 1.1229,
+      "step": 2021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.065268065268065e-08,
+      "loss": 0.2082,
+      "step": 2022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.018648018648017e-08,
+      "loss": 0.5198,
+      "step": 2023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.972027972027972e-08,
+      "loss": 0.9059,
+      "step": 2024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.925407925407924e-08,
+      "loss": 1.1163,
+      "step": 2025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.878787878787878e-08,
+      "loss": 0.6353,
+      "step": 2026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.832167832167833e-08,
+      "loss": 0.2771,
+      "step": 2027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.785547785547785e-08,
+      "loss": 0.779,
+      "step": 2028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.73892773892774e-08,
+      "loss": 0.0607,
+      "step": 2029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.692307692307692e-08,
+      "loss": 0.2553,
+      "step": 2030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.645687645687645e-08,
+      "loss": 0.4716,
+      "step": 2031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.599067599067599e-08,
+      "loss": 1.0063,
+      "step": 2032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.552447552447552e-08,
+      "loss": 0.5961,
+      "step": 2033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.505827505827506e-08,
+      "loss": 0.7928,
+      "step": 2034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.459207459207459e-08,
+      "loss": 0.0632,
+      "step": 2035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.412587412587412e-08,
+      "loss": 0.7956,
+      "step": 2036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.365967365967366e-08,
+      "loss": 1.4114,
+      "step": 2037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.319347319347319e-08,
+      "loss": 0.3039,
+      "step": 2038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.272727272727273e-08,
+      "loss": 1.4308,
+      "step": 2039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.226107226107225e-08,
+      "loss": 0.6737,
+      "step": 2040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.179487179487178e-08,
+      "loss": 0.0647,
+      "step": 2041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.132867132867132e-08,
+      "loss": 0.702,
+      "step": 2042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.086247086247086e-08,
+      "loss": 0.3816,
+      "step": 2043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.03962703962704e-08,
+      "loss": 0.595,
+      "step": 2044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.993006993006993e-08,
+      "loss": 0.2251,
+      "step": 2045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.946386946386946e-08,
+      "loss": 0.2412,
+      "step": 2046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.8997668997669e-08,
+      "loss": 0.4079,
+      "step": 2047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.853146853146853e-08,
+      "loss": 0.2401,
+      "step": 2048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.806526806526806e-08,
+      "loss": 0.1781,
+      "step": 2049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.75990675990676e-08,
+      "loss": 0.5929,
+      "step": 2050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.713286713286713e-08,
+      "loss": 0.8613,
+      "step": 2051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.666666666666667e-08,
+      "loss": 0.5108,
+      "step": 2052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.62004662004662e-08,
+      "loss": 0.185,
+      "step": 2053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.573426573426572e-08,
+      "loss": 0.6171,
+      "step": 2054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.526806526806526e-08,
+      "loss": 0.3765,
+      "step": 2055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.480186480186479e-08,
+      "loss": 1.0989,
+      "step": 2056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.433566433566433e-08,
+      "loss": 0.5295,
+      "step": 2057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.386946386946386e-08,
+      "loss": 0.2579,
+      "step": 2058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.34032634032634e-08,
+      "loss": 0.0505,
+      "step": 2059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.293706293706294e-08,
+      "loss": 0.7094,
+      "step": 2060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.247086247086247e-08,
+      "loss": 0.7877,
+      "step": 2061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.2004662004662e-08,
+      "loss": 0.7335,
+      "step": 2062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.153846153846154e-08,
+      "loss": 0.0997,
+      "step": 2063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.107226107226107e-08,
+      "loss": 0.181,
+      "step": 2064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.060606060606061e-08,
+      "loss": 2.0511,
+      "step": 2065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.013986013986014e-08,
+      "loss": 0.5275,
+      "step": 2066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.967365967365967e-08,
+      "loss": 0.5595,
+      "step": 2067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.9207459207459207e-08,
+      "loss": 0.4227,
+      "step": 2068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.8741258741258734e-08,
+      "loss": 0.1635,
+      "step": 2069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.827505827505827e-08,
+      "loss": 1.0078,
+      "step": 2070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.780885780885781e-08,
+      "loss": 0.0794,
+      "step": 2071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.7342657342657345e-08,
+      "loss": 0.2281,
+      "step": 2072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.687645687645687e-08,
+      "loss": 0.7738,
+      "step": 2073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.641025641025641e-08,
+      "loss": 0.3932,
+      "step": 2074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.594405594405594e-08,
+      "loss": 0.4775,
+      "step": 2075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5477855477855476e-08,
+      "loss": 1.1381,
+      "step": 2076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.501165501165501e-08,
+      "loss": 0.3471,
+      "step": 2077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.454545454545454e-08,
+      "loss": 2.2582,
+      "step": 2078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.407925407925408e-08,
+      "loss": 0.2901,
+      "step": 2079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3613053613053614e-08,
+      "loss": 0.3157,
+      "step": 2080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.314685314685315e-08,
+      "loss": 0.5571,
+      "step": 2081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2680652680652676e-08,
+      "loss": 0.6485,
+      "step": 2082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.221445221445221e-08,
+      "loss": 0.6824,
+      "step": 2083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1748251748251745e-08,
+      "loss": 0.5785,
+      "step": 2084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.128205128205128e-08,
+      "loss": 0.4643,
+      "step": 2085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.0815850815850814e-08,
+      "loss": 0.2132,
+      "step": 2086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.034965034965035e-08,
+      "loss": 0.9546,
+      "step": 2087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.988344988344988e-08,
+      "loss": 0.6265,
+      "step": 2088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.941724941724942e-08,
+      "loss": 0.2556,
+      "step": 2089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.895104895104895e-08,
+      "loss": 1.1349,
+      "step": 2090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.848484848484848e-08,
+      "loss": 0.2507,
+      "step": 2091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8018648018648014e-08,
+      "loss": 0.5707,
+      "step": 2092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.755244755244755e-08,
+      "loss": 0.6083,
+      "step": 2093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.708624708624708e-08,
+      "loss": 0.8694,
+      "step": 2094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6620046620046624e-08,
+      "loss": 0.5173,
+      "step": 2095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.615384615384615e-08,
+      "loss": 0.3919,
+      "step": 2096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.568764568764569e-08,
+      "loss": 0.6374,
+      "step": 2097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.522144522144522e-08,
+      "loss": 0.3822,
+      "step": 2098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4755244755244756e-08,
+      "loss": 0.4583,
+      "step": 2099
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4289044289044284e-08,
+      "loss": 0.0696,
+      "step": 2100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.382284382284382e-08,
+      "loss": 0.1097,
+      "step": 2101
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.335664335664335e-08,
+      "loss": 1.0454,
+      "step": 2102
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2890442890442894e-08,
+      "loss": 0.7211,
+      "step": 2103
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.242424242424243e-08,
+      "loss": 0.6312,
+      "step": 2104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1958041958041956e-08,
+      "loss": 2.0241,
+      "step": 2105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.149184149184149e-08,
+      "loss": 0.1098,
+      "step": 2106
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1025641025641025e-08,
+      "loss": 0.3639,
+      "step": 2107
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.055944055944056e-08,
+      "loss": 0.0375,
+      "step": 2108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.009324009324009e-08,
+      "loss": 0.7562,
+      "step": 2109
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.962703962703962e-08,
+      "loss": 0.5027,
+      "step": 2110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.916083916083916e-08,
+      "loss": 0.4785,
+      "step": 2111
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.86946386946387e-08,
+      "loss": 0.0832,
+      "step": 2112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8228438228438225e-08,
+      "loss": 0.6726,
+      "step": 2113
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.776223776223776e-08,
+      "loss": 0.5361,
+      "step": 2114
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7296037296037294e-08,
+      "loss": 1.0671,
+      "step": 2115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.682983682983683e-08,
+      "loss": 0.4025,
+      "step": 2116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.636363636363636e-08,
+      "loss": 0.6215,
+      "step": 2117
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.589743589743589e-08,
+      "loss": 0.3615,
+      "step": 2118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.543123543123543e-08,
+      "loss": 1.641,
+      "step": 2119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4965034965034967e-08,
+      "loss": 1.4563,
+      "step": 2120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.44988344988345e-08,
+      "loss": 0.7555,
+      "step": 2121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.403263403263403e-08,
+      "loss": 0.3852,
+      "step": 2122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3566433566433563e-08,
+      "loss": 0.0973,
+      "step": 2123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.31002331002331e-08,
+      "loss": 0.6215,
+      "step": 2124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.263403263403263e-08,
+      "loss": 0.3477,
+      "step": 2125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.216783216783217e-08,
+      "loss": 0.5334,
+      "step": 2126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.17016317016317e-08,
+      "loss": 0.2752,
+      "step": 2127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1235431235431236e-08,
+      "loss": 0.8817,
+      "step": 2128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.076923076923077e-08,
+      "loss": 0.4071,
+      "step": 2129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0303030303030305e-08,
+      "loss": 0.7979,
+      "step": 2130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.983682983682983e-08,
+      "loss": 0.5382,
+      "step": 2131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9370629370629367e-08,
+      "loss": 0.5727,
+      "step": 2132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8904428904428905e-08,
+      "loss": 0.2413,
+      "step": 2133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8438228438228436e-08,
+      "loss": 0.518,
+      "step": 2134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.797202797202797e-08,
+      "loss": 0.667,
+      "step": 2135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7505827505827505e-08,
+      "loss": 0.3612,
+      "step": 2136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.703962703962704e-08,
+      "loss": 0.6382,
+      "step": 2137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6573426573426574e-08,
+      "loss": 0.4035,
+      "step": 2138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6107226107226105e-08,
+      "loss": 0.2911,
+      "step": 2139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.564102564102564e-08,
+      "loss": 0.9374,
+      "step": 2140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5174825174825174e-08,
+      "loss": 0.5175,
+      "step": 2141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.470862470862471e-08,
+      "loss": 0.3051,
+      "step": 2142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.424242424242424e-08,
+      "loss": 0.2176,
+      "step": 2143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3776223776223774e-08,
+      "loss": 0.3988,
+      "step": 2144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3310023310023312e-08,
+      "loss": 1.7052,
+      "step": 2145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2843822843822843e-08,
+      "loss": 0.4015,
+      "step": 2146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2377622377622378e-08,
+      "loss": 0.2003,
+      "step": 2147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.191142191142191e-08,
+      "loss": 0.1317,
+      "step": 2148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1445221445221447e-08,
+      "loss": 0.6199,
+      "step": 2149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0979020979020978e-08,
+      "loss": 0.6665,
+      "step": 2150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0512820512820512e-08,
+      "loss": 0.8923,
+      "step": 2151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0046620046620044e-08,
+      "loss": 0.205,
+      "step": 2152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.958041958041958e-08,
+      "loss": 1.1063,
+      "step": 2153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9114219114219113e-08,
+      "loss": 0.2124,
+      "step": 2154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8648018648018647e-08,
+      "loss": 0.251,
+      "step": 2155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.818181818181818e-08,
+      "loss": 0.6486,
+      "step": 2156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7715617715617716e-08,
+      "loss": 0.5205,
+      "step": 2157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.724941724941725e-08,
+      "loss": 0.3417,
+      "step": 2158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6783216783216782e-08,
+      "loss": 0.5467,
+      "step": 2159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6317016317016316e-08,
+      "loss": 0.1839,
+      "step": 2160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.585081585081585e-08,
+      "loss": 0.0964,
+      "step": 2161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5384615384615385e-08,
+      "loss": 0.4385,
+      "step": 2162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4918414918414916e-08,
+      "loss": 0.888,
+      "step": 2163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4452214452214453e-08,
+      "loss": 0.5441,
+      "step": 2164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3986013986013985e-08,
+      "loss": 0.526,
+      "step": 2165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.351981351981352e-08,
+      "loss": 0.9328,
+      "step": 2166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3053613053613053e-08,
+      "loss": 0.8792,
+      "step": 2167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2587412587412587e-08,
+      "loss": 0.3692,
+      "step": 2168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.212121212121212e-08,
+      "loss": 0.4355,
+      "step": 2169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1655011655011656e-08,
+      "loss": 0.1234,
+      "step": 2170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1188811188811189e-08,
+      "loss": 0.5219,
+      "step": 2171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0722610722610723e-08,
+      "loss": 0.2721,
+      "step": 2172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0256410256410256e-08,
+      "loss": 0.3693,
+      "step": 2173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.79020979020979e-09,
+      "loss": 0.5556,
+      "step": 2174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.324009324009324e-09,
+      "loss": 1.9111,
+      "step": 2175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.857808857808858e-09,
+      "loss": 0.0442,
+      "step": 2176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.391608391608391e-09,
+      "loss": 0.6286,
+      "step": 2177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.925407925407925e-09,
+      "loss": 0.3885,
+      "step": 2178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.459207459207458e-09,
+      "loss": 0.6613,
+      "step": 2179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.993006993006993e-09,
+      "loss": 0.7269,
+      "step": 2180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.526806526806526e-09,
+      "loss": 0.1497,
+      "step": 2181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.06060606060606e-09,
+      "loss": 0.3453,
+      "step": 2182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5944055944055945e-09,
+      "loss": 1.5299,
+      "step": 2183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.128205128205128e-09,
+      "loss": 0.6051,
+      "step": 2184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.662004662004662e-09,
+      "loss": 0.1374,
+      "step": 2185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1958041958041954e-09,
+      "loss": 0.5615,
+      "step": 2186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.729603729603729e-09,
+      "loss": 1.0941,
+      "step": 2187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.263403263403263e-09,
+      "loss": 0.7907,
+      "step": 2188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7972027972027972e-09,
+      "loss": 0.9296,
+      "step": 2189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.331002331002331e-09,
+      "loss": 0.9808,
+      "step": 2190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8648018648018645e-09,
+      "loss": 0.1613,
+      "step": 2191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3986013986013986e-09,
+      "loss": 0.5117,
+      "step": 2192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.324009324009323e-10,
+      "loss": 0.772,
+      "step": 2193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.662004662004661e-10,
+      "loss": 0.8227,
+      "step": 2194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.0727,
+      "step": 2195
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8426447975397232,
+      "eval_loss": 0.4701635241508484,
+      "eval_roc_auc": 0.9680887374943532,
+      "eval_runtime": 73.8486,
+      "eval_samples_per_second": 26.419,
+      "eval_steps_per_second": 3.304,
+      "step": 2195
+    }
+  ],
+  "max_steps": 2195,
+  "num_train_epochs": 1,
+  "total_flos": 3.9841904159856e+17,
+  "trial_name": null,
+  "trial_params": null
+}