diff --git "a/checkpoint-10610/trainer_state.json" "b/checkpoint-10610/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-10610/trainer_state.json"
@@ -0,0 +1,63685 @@
+{
+  "best_metric": 0.3358200788497925,
+  "best_model_checkpoint": "./results/checkpoint-10610",
+  "epoch": 1.0,
+  "global_step": 10610,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6953,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.6998,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.698,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.6948,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6984,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.6936,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.6819,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7026,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 0.7001,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.6904,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.6829,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6738,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.6922,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.6812,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.6764,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6786,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.6803,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6653,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-05,
+      "loss": 0.6484,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.6926,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 0.6938,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.6603,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.699,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 0.7119,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6531,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.6884,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.6828,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.6843,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9e-05,
+      "loss": 0.7051,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 0.6435,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1e-05,
+      "loss": 0.6623,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.6702,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3e-05,
+      "loss": 0.637,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.6668,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 0.7307,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 0.6765,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7e-05,
+      "loss": 0.6776,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-05,
+      "loss": 0.6645,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.723,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 0.6541,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1e-05,
+      "loss": 0.714,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2e-05,
+      "loss": 0.6747,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3e-05,
+      "loss": 0.6815,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.6572,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 0.6186,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.6372,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7e-05,
+      "loss": 0.6484,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 0.6316,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9e-05,
+      "loss": 0.6217,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 0.6436,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999526515151515e-05,
+      "loss": 0.6647,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99905303030303e-05,
+      "loss": 0.7472,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998579545454546e-05,
+      "loss": 0.6212,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998106060606061e-05,
+      "loss": 0.6476,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997632575757576e-05,
+      "loss": 0.6523,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997159090909091e-05,
+      "loss": 0.7048,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9966856060606065e-05,
+      "loss": 0.6536,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996212121212122e-05,
+      "loss": 0.6471,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9957386363636366e-05,
+      "loss": 0.6978,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995265151515151e-05,
+      "loss": 0.6946,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994791666666667e-05,
+      "loss": 0.6762,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994318181818182e-05,
+      "loss": 0.6426,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9938446969696974e-05,
+      "loss": 0.6636,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993371212121212e-05,
+      "loss": 0.6893,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9928977272727275e-05,
+      "loss": 0.6325,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992424242424243e-05,
+      "loss": 0.6489,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991950757575758e-05,
+      "loss": 0.6706,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991477272727273e-05,
+      "loss": 0.6979,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9910037878787876e-05,
+      "loss": 0.6549,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990530303030303e-05,
+      "loss": 0.6472,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9900568181818184e-05,
+      "loss": 0.7219,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989583333333334e-05,
+      "loss": 0.6364,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9891098484848485e-05,
+      "loss": 0.6462,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988636363636364e-05,
+      "loss": 0.6763,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988162878787879e-05,
+      "loss": 0.6948,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9876893939393946e-05,
+      "loss": 0.7292,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987215909090909e-05,
+      "loss": 0.6665,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.986742424242424e-05,
+      "loss": 0.6221,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98626893939394e-05,
+      "loss": 0.6462,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985795454545455e-05,
+      "loss": 0.6669,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98532196969697e-05,
+      "loss": 0.6756,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984848484848485e-05,
+      "loss": 0.6498,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984375e-05,
+      "loss": 0.6112,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9839015151515156e-05,
+      "loss": 0.6246,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983428030303031e-05,
+      "loss": 0.6369,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9829545454545456e-05,
+      "loss": 0.6428,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9824810606060603e-05,
+      "loss": 0.6333,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9820075757575764e-05,
+      "loss": 0.6311,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981534090909091e-05,
+      "loss": 0.6608,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9810606060606065e-05,
+      "loss": 0.6766,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980587121212121e-05,
+      "loss": 0.6251,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9801136363636366e-05,
+      "loss": 0.6409,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979640151515152e-05,
+      "loss": 0.6459,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.979166666666667e-05,
+      "loss": 0.6888,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978693181818182e-05,
+      "loss": 0.7297,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978219696969697e-05,
+      "loss": 0.6514,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.977746212121213e-05,
+      "loss": 0.6034,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9772727272727275e-05,
+      "loss": 0.5657,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.976799242424243e-05,
+      "loss": 0.6888,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9763257575757575e-05,
+      "loss": 0.6262,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.975852272727273e-05,
+      "loss": 0.6187,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.975378787878788e-05,
+      "loss": 0.6928,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9749053030303037e-05,
+      "loss": 0.6149,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9744318181818184e-05,
+      "loss": 0.6627,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973958333333333e-05,
+      "loss": 0.6729,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973484848484849e-05,
+      "loss": 0.6758,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973011363636364e-05,
+      "loss": 0.5983,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972537878787879e-05,
+      "loss": 0.6216,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972064393939394e-05,
+      "loss": 0.6355,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.971590909090909e-05,
+      "loss": 0.7157,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9711174242424246e-05,
+      "loss": 0.694,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.97064393939394e-05,
+      "loss": 0.7043,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.970170454545455e-05,
+      "loss": 0.6337,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9696969696969694e-05,
+      "loss": 0.6013,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9692234848484855e-05,
+      "loss": 0.6872,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96875e-05,
+      "loss": 0.6771,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9682765151515155e-05,
+      "loss": 0.6557,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96780303030303e-05,
+      "loss": 0.6838,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9673295454545456e-05,
+      "loss": 0.6947,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.966856060606061e-05,
+      "loss": 0.6188,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9663825757575764e-05,
+      "loss": 0.6028,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965909090909091e-05,
+      "loss": 0.6278,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965435606060606e-05,
+      "loss": 0.6219,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964962121212122e-05,
+      "loss": 0.6168,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9644886363636365e-05,
+      "loss": 0.6669,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964015151515152e-05,
+      "loss": 0.6708,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9635416666666666e-05,
+      "loss": 0.7165,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963068181818182e-05,
+      "loss": 0.6215,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962594696969697e-05,
+      "loss": 0.6428,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962121212121213e-05,
+      "loss": 0.5536,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9616477272727274e-05,
+      "loss": 0.6144,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.961174242424242e-05,
+      "loss": 0.6507,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960700757575758e-05,
+      "loss": 0.5787,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960227272727273e-05,
+      "loss": 0.6871,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.959753787878788e-05,
+      "loss": 0.6048,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.959280303030303e-05,
+      "loss": 0.602,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958806818181819e-05,
+      "loss": 0.6311,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958333333333334e-05,
+      "loss": 0.6222,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.957859848484849e-05,
+      "loss": 0.6548,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.957386363636364e-05,
+      "loss": 0.6729,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9569128787878785e-05,
+      "loss": 0.674,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9564393939393945e-05,
+      "loss": 0.5696,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.955965909090909e-05,
+      "loss": 0.6851,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9554924242424246e-05,
+      "loss": 0.6823,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.955018939393939e-05,
+      "loss": 0.5822,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9545454545454553e-05,
+      "loss": 0.6579,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95407196969697e-05,
+      "loss": 0.7256,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9535984848484854e-05,
+      "loss": 0.6337,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953125e-05,
+      "loss": 0.6584,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.952651515151515e-05,
+      "loss": 0.6757,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.952178030303031e-05,
+      "loss": 0.6504,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9517045454545456e-05,
+      "loss": 0.7171,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951231060606061e-05,
+      "loss": 0.6959,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9507575757575756e-05,
+      "loss": 0.6589,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.950284090909092e-05,
+      "loss": 0.588,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9498106060606064e-05,
+      "loss": 0.5475,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.949337121212122e-05,
+      "loss": 0.7713,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9488636363636365e-05,
+      "loss": 0.7027,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948390151515152e-05,
+      "loss": 0.6749,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.947916666666667e-05,
+      "loss": 0.7106,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.947443181818182e-05,
+      "loss": 0.6932,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.946969696969697e-05,
+      "loss": 0.6481,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.946496212121212e-05,
+      "loss": 0.5506,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.946022727272728e-05,
+      "loss": 0.6206,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.945549242424243e-05,
+      "loss": 0.6861,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.945075757575758e-05,
+      "loss": 0.7716,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.944602272727273e-05,
+      "loss": 0.6876,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.944128787878788e-05,
+      "loss": 0.6937,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9436553030303036e-05,
+      "loss": 0.6288,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.943181818181818e-05,
+      "loss": 0.5913,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9427083333333336e-05,
+      "loss": 0.6085,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9422348484848483e-05,
+      "loss": 0.6221,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9417613636363644e-05,
+      "loss": 0.6143,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.941287878787879e-05,
+      "loss": 0.7282,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9408143939393945e-05,
+      "loss": 0.6363,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.940340909090909e-05,
+      "loss": 0.6266,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9398674242424246e-05,
+      "loss": 0.6102,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.93939393939394e-05,
+      "loss": 0.6816,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9389204545454546e-05,
+      "loss": 0.6531,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.93844696969697e-05,
+      "loss": 0.563,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.937973484848485e-05,
+      "loss": 0.7157,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.937500000000001e-05,
+      "loss": 0.7516,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9370265151515155e-05,
+      "loss": 0.6569,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.936553030303031e-05,
+      "loss": 0.5964,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9360795454545455e-05,
+      "loss": 0.6526,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.935606060606061e-05,
+      "loss": 0.6466,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.935132575757576e-05,
+      "loss": 0.6468,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.934659090909091e-05,
+      "loss": 0.6595,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9341856060606064e-05,
+      "loss": 0.5991,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.933712121212121e-05,
+      "loss": 0.7172,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.933238636363637e-05,
+      "loss": 0.6664,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.932765151515152e-05,
+      "loss": 0.6353,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.932291666666667e-05,
+      "loss": 0.5891,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.931818181818182e-05,
+      "loss": 0.5476,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.931344696969697e-05,
+      "loss": 0.6089,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9308712121212126e-05,
+      "loss": 0.6838,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.930397727272727e-05,
+      "loss": 0.6175,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.929924242424243e-05,
+      "loss": 0.6914,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9294507575757574e-05,
+      "loss": 0.6134,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9289772727272735e-05,
+      "loss": 0.6232,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.928503787878788e-05,
+      "loss": 0.6006,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9280303030303035e-05,
+      "loss": 0.6487,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.927556818181818e-05,
+      "loss": 0.5887,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9270833333333336e-05,
+      "loss": 0.6822,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.926609848484849e-05,
+      "loss": 0.6374,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.926136363636364e-05,
+      "loss": 0.5804,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.925662878787879e-05,
+      "loss": 0.6389,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.925189393939394e-05,
+      "loss": 0.6519,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92471590909091e-05,
+      "loss": 0.6472,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9242424242424245e-05,
+      "loss": 0.6729,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92376893939394e-05,
+      "loss": 0.6356,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9232954545454546e-05,
+      "loss": 0.5787,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92282196969697e-05,
+      "loss": 0.6709,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922348484848485e-05,
+      "loss": 0.6277,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921875e-05,
+      "loss": 0.6431,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9214015151515154e-05,
+      "loss": 0.5757,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92092803030303e-05,
+      "loss": 0.6953,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.920454545454546e-05,
+      "loss": 0.67,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919981060606061e-05,
+      "loss": 0.5929,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919507575757576e-05,
+      "loss": 0.6594,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919034090909091e-05,
+      "loss": 0.6173,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918560606060606e-05,
+      "loss": 0.6094,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918087121212122e-05,
+      "loss": 0.6259,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9176136363636364e-05,
+      "loss": 0.6264,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917140151515152e-05,
+      "loss": 0.664,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9166666666666665e-05,
+      "loss": 0.6611,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9161931818181825e-05,
+      "loss": 0.57,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915719696969697e-05,
+      "loss": 0.6026,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9152462121212126e-05,
+      "loss": 0.6434,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914772727272727e-05,
+      "loss": 0.7026,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914299242424243e-05,
+      "loss": 0.6179,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.913825757575758e-05,
+      "loss": 0.6555,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.913352272727273e-05,
+      "loss": 0.6167,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.912878787878788e-05,
+      "loss": 0.6018,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.912405303030303e-05,
+      "loss": 0.616,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911931818181819e-05,
+      "loss": 0.6166,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9114583333333336e-05,
+      "loss": 0.688,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.910984848484849e-05,
+      "loss": 0.6465,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9105113636363636e-05,
+      "loss": 0.5601,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.910037878787879e-05,
+      "loss": 0.6715,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9095643939393944e-05,
+      "loss": 0.5983,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.909090909090909e-05,
+      "loss": 0.6392,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9086174242424245e-05,
+      "loss": 0.5891,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90814393939394e-05,
+      "loss": 0.5856,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.907670454545455e-05,
+      "loss": 0.683,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90719696969697e-05,
+      "loss": 0.6583,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906723484848485e-05,
+      "loss": 0.5748,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90625e-05,
+      "loss": 0.5795,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9057765151515154e-05,
+      "loss": 0.7861,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.905303030303031e-05,
+      "loss": 0.7702,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9048295454545454e-05,
+      "loss": 0.6889,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.904356060606061e-05,
+      "loss": 0.6983,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.903882575757576e-05,
+      "loss": 0.5783,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9034090909090916e-05,
+      "loss": 0.615,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.902935606060606e-05,
+      "loss": 0.6,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9024621212121216e-05,
+      "loss": 0.7096,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9019886363636363e-05,
+      "loss": 0.5864,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.901515151515152e-05,
+      "loss": 0.6594,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.901041666666667e-05,
+      "loss": 0.7045,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.900568181818182e-05,
+      "loss": 0.6113,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.900094696969697e-05,
+      "loss": 0.6263,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8996212121212125e-05,
+      "loss": 0.7391,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.899147727272728e-05,
+      "loss": 0.6518,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8986742424242426e-05,
+      "loss": 0.6454,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.898200757575758e-05,
+      "loss": 0.6028,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.897727272727273e-05,
+      "loss": 0.6183,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.897253787878788e-05,
+      "loss": 0.642,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8967803030303035e-05,
+      "loss": 0.6354,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.896306818181818e-05,
+      "loss": 0.4913,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8958333333333335e-05,
+      "loss": 0.6596,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.895359848484849e-05,
+      "loss": 0.5441,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.894886363636364e-05,
+      "loss": 0.7032,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.894412878787879e-05,
+      "loss": 0.6508,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8939393939393944e-05,
+      "loss": 0.7171,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.893465909090909e-05,
+      "loss": 0.6485,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8929924242424244e-05,
+      "loss": 0.7194,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.89251893939394e-05,
+      "loss": 0.6989,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8920454545454545e-05,
+      "loss": 0.6033,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.89157196969697e-05,
+      "loss": 0.5899,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.891098484848485e-05,
+      "loss": 0.6726,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8906250000000006e-05,
+      "loss": 0.6122,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.890151515151515e-05,
+      "loss": 0.5736,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.889678030303031e-05,
+      "loss": 0.6511,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8892045454545454e-05,
+      "loss": 0.6099,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.888731060606061e-05,
+      "loss": 0.6278,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.888257575757576e-05,
+      "loss": 0.586,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.887784090909091e-05,
+      "loss": 0.5746,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.887310606060606e-05,
+      "loss": 0.5859,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8868371212121216e-05,
+      "loss": 0.655,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.886363636363637e-05,
+      "loss": 0.6844,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.885890151515152e-05,
+      "loss": 0.652,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.885416666666667e-05,
+      "loss": 0.5275,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.884943181818182e-05,
+      "loss": 0.6358,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.884469696969697e-05,
+      "loss": 0.617,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8839962121212125e-05,
+      "loss": 0.653,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.883522727272727e-05,
+      "loss": 0.4992,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8830492424242426e-05,
+      "loss": 0.631,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.882575757575758e-05,
+      "loss": 0.7737,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.882102272727273e-05,
+      "loss": 0.667,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.881628787878788e-05,
+      "loss": 0.621,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8811553030303034e-05,
+      "loss": 0.5928,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.880681818181819e-05,
+      "loss": 0.5611,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8802083333333335e-05,
+      "loss": 0.5238,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.879734848484849e-05,
+      "loss": 0.6603,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8792613636363636e-05,
+      "loss": 0.6902,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.878787878787879e-05,
+      "loss": 0.6885,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.878314393939394e-05,
+      "loss": 0.6008,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.87784090909091e-05,
+      "loss": 0.6259,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8773674242424244e-05,
+      "loss": 0.6093,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.87689393939394e-05,
+      "loss": 0.5876,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.876420454545455e-05,
+      "loss": 0.6493,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.87594696969697e-05,
+      "loss": 0.6143,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.875473484848485e-05,
+      "loss": 0.5896,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.875e-05,
+      "loss": 0.5822,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.874526515151515e-05,
+      "loss": 0.5426,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.874053030303031e-05,
+      "loss": 0.6351,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.873579545454546e-05,
+      "loss": 0.6571,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.873106060606061e-05,
+      "loss": 0.5963,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.872632575757576e-05,
+      "loss": 0.5587,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8721590909090915e-05,
+      "loss": 0.5959,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.871685606060606e-05,
+      "loss": 0.5835,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8712121212121216e-05,
+      "loss": 0.6125,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.870738636363636e-05,
+      "loss": 0.6303,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8702651515151516e-05,
+      "loss": 0.692,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.869791666666667e-05,
+      "loss": 0.5843,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8693181818181824e-05,
+      "loss": 0.6221,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.868844696969697e-05,
+      "loss": 0.6486,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8683712121212125e-05,
+      "loss": 0.6025,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.867897727272728e-05,
+      "loss": 0.5634,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8674242424242425e-05,
+      "loss": 0.591,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.866950757575758e-05,
+      "loss": 0.7124,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8664772727272726e-05,
+      "loss": 0.6057,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.866003787878788e-05,
+      "loss": 0.6698,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8655303030303034e-05,
+      "loss": 0.5955,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.865056818181819e-05,
+      "loss": 0.6121,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8645833333333334e-05,
+      "loss": 0.7128,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.864109848484849e-05,
+      "loss": 0.5803,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.863636363636364e-05,
+      "loss": 0.5855,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.863162878787879e-05,
+      "loss": 0.6444,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.862689393939394e-05,
+      "loss": 0.6395,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.862215909090909e-05,
+      "loss": 0.6117,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8617424242424243e-05,
+      "loss": 0.5936,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86126893939394e-05,
+      "loss": 0.622,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.860795454545455e-05,
+      "loss": 0.575,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86032196969697e-05,
+      "loss": 0.5522,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.859848484848485e-05,
+      "loss": 0.6177,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8593750000000005e-05,
+      "loss": 0.6432,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.858901515151515e-05,
+      "loss": 0.609,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8584280303030306e-05,
+      "loss": 0.6013,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.857954545454545e-05,
+      "loss": 0.5647,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.857481060606061e-05,
+      "loss": 0.7009,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.857007575757576e-05,
+      "loss": 0.6209,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8565340909090915e-05,
+      "loss": 0.6521,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.856060606060606e-05,
+      "loss": 0.6613,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8555871212121215e-05,
+      "loss": 0.6575,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.855113636363637e-05,
+      "loss": 0.5677,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8546401515151516e-05,
+      "loss": 0.6761,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.854166666666667e-05,
+      "loss": 0.608,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.853693181818182e-05,
+      "loss": 0.593,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.853219696969698e-05,
+      "loss": 0.6122,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8527462121212124e-05,
+      "loss": 0.5661,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.852272727272728e-05,
+      "loss": 0.5643,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8517992424242425e-05,
+      "loss": 0.6389,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.851325757575758e-05,
+      "loss": 0.6129,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.850852272727273e-05,
+      "loss": 0.5903,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.850378787878788e-05,
+      "loss": 0.6363,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.849905303030303e-05,
+      "loss": 0.6092,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.849431818181818e-05,
+      "loss": 0.5977,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.848958333333334e-05,
+      "loss": 0.7092,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.848484848484849e-05,
+      "loss": 0.6614,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.848011363636364e-05,
+      "loss": 0.6964,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.847537878787879e-05,
+      "loss": 0.638,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.847064393939394e-05,
+      "loss": 0.656,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8465909090909096e-05,
+      "loss": 0.6079,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.846117424242424e-05,
+      "loss": 0.5877,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.84564393939394e-05,
+      "loss": 0.6317,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8451704545454544e-05,
+      "loss": 0.6132,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8446969696969704e-05,
+      "loss": 0.6785,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.844223484848485e-05,
+      "loss": 0.6514,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 0.6767,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.843276515151515e-05,
+      "loss": 0.6567,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8428030303030306e-05,
+      "loss": 0.5489,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.842329545454546e-05,
+      "loss": 0.6585,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8418560606060607e-05,
+      "loss": 0.6329,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.841382575757576e-05,
+      "loss": 0.6497,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.840909090909091e-05,
+      "loss": 0.6225,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.840435606060607e-05,
+      "loss": 0.6308,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8399621212121215e-05,
+      "loss": 0.6175,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.839488636363637e-05,
+      "loss": 0.6313,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8390151515151516e-05,
+      "loss": 0.6205,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.838541666666667e-05,
+      "loss": 0.567,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.838068181818182e-05,
+      "loss": 0.5687,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.837594696969697e-05,
+      "loss": 0.5412,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8371212121212124e-05,
+      "loss": 0.5712,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.836647727272727e-05,
+      "loss": 0.5937,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.836174242424243e-05,
+      "loss": 0.6568,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.835700757575758e-05,
+      "loss": 0.6387,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.835227272727273e-05,
+      "loss": 0.5937,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.834753787878788e-05,
+      "loss": 0.6046,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.834280303030303e-05,
+      "loss": 0.6034,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.833806818181819e-05,
+      "loss": 0.5913,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8333333333333334e-05,
+      "loss": 0.6486,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.832859848484849e-05,
+      "loss": 0.6425,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8323863636363634e-05,
+      "loss": 0.6431,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8319128787878795e-05,
+      "loss": 0.5545,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.831439393939394e-05,
+      "loss": 0.7255,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8309659090909096e-05,
+      "loss": 0.6329,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.830492424242424e-05,
+      "loss": 0.5741,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8300189393939396e-05,
+      "loss": 0.5892,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.829545454545455e-05,
+      "loss": 0.6675,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.82907196969697e-05,
+      "loss": 0.6801,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.828598484848485e-05,
+      "loss": 0.5681,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.828125e-05,
+      "loss": 0.5908,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.827651515151516e-05,
+      "loss": 0.6276,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8271780303030305e-05,
+      "loss": 0.6256,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.826704545454546e-05,
+      "loss": 0.6267,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8262310606060606e-05,
+      "loss": 0.6477,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.825757575757576e-05,
+      "loss": 0.5381,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8252840909090914e-05,
+      "loss": 0.597,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.824810606060606e-05,
+      "loss": 0.5588,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8243371212121214e-05,
+      "loss": 0.6253,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.823863636363636e-05,
+      "loss": 0.6225,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.823390151515152e-05,
+      "loss": 0.5549,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.822916666666667e-05,
+      "loss": 0.6337,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.822443181818182e-05,
+      "loss": 0.5783,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.821969696969697e-05,
+      "loss": 0.6431,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8214962121212123e-05,
+      "loss": 0.511,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.821022727272728e-05,
+      "loss": 0.6163,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8205492424242424e-05,
+      "loss": 0.6447,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.820075757575758e-05,
+      "loss": 0.5346,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8196022727272725e-05,
+      "loss": 0.5152,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8191287878787885e-05,
+      "loss": 0.5245,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.818655303030303e-05,
+      "loss": 0.5845,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8181818181818186e-05,
+      "loss": 0.6112,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.817708333333333e-05,
+      "loss": 0.533,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.817234848484849e-05,
+      "loss": 0.5889,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.816761363636364e-05,
+      "loss": 0.5621,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.816287878787879e-05,
+      "loss": 0.5669,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.815814393939394e-05,
+      "loss": 0.6208,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.815340909090909e-05,
+      "loss": 0.594,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.814867424242425e-05,
+      "loss": 0.6896,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8143939393939396e-05,
+      "loss": 0.5809,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.813920454545455e-05,
+      "loss": 0.6677,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.81344696969697e-05,
+      "loss": 0.575,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.812973484848485e-05,
+      "loss": 0.6098,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8125000000000004e-05,
+      "loss": 0.6391,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.812026515151515e-05,
+      "loss": 0.6506,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8115530303030305e-05,
+      "loss": 0.4846,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.811079545454545e-05,
+      "loss": 0.6854,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.810606060606061e-05,
+      "loss": 0.5708,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.810132575757576e-05,
+      "loss": 0.6617,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.809659090909091e-05,
+      "loss": 0.5919,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.809185606060606e-05,
+      "loss": 0.6832,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8087121212121214e-05,
+      "loss": 0.6243,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.808238636363637e-05,
+      "loss": 0.6002,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8077651515151515e-05,
+      "loss": 0.5849,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.807291666666667e-05,
+      "loss": 0.6747,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8068181818181816e-05,
+      "loss": 0.6096,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8063446969696976e-05,
+      "loss": 0.6697,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.805871212121212e-05,
+      "loss": 0.5189,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.805397727272728e-05,
+      "loss": 0.5437,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8049242424242424e-05,
+      "loss": 0.5637,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.804450757575758e-05,
+      "loss": 0.5525,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.803977272727273e-05,
+      "loss": 0.5292,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.803503787878788e-05,
+      "loss": 0.6462,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.803030303030303e-05,
+      "loss": 0.6467,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8025568181818186e-05,
+      "loss": 0.5809,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.802083333333334e-05,
+      "loss": 0.7527,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8016098484848487e-05,
+      "loss": 0.6733,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.801136363636364e-05,
+      "loss": 0.6202,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.800662878787879e-05,
+      "loss": 0.6495,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.800189393939394e-05,
+      "loss": 0.6264,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7997159090909095e-05,
+      "loss": 0.6751,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.799242424242424e-05,
+      "loss": 0.6053,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7987689393939396e-05,
+      "loss": 0.6382,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.798295454545455e-05,
+      "loss": 0.6323,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.79782196969697e-05,
+      "loss": 0.5989,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.797348484848485e-05,
+      "loss": 0.6286,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7968750000000004e-05,
+      "loss": 0.6089,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.796401515151515e-05,
+      "loss": 0.6342,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7959280303030305e-05,
+      "loss": 0.5811,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.795454545454546e-05,
+      "loss": 0.6504,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7949810606060605e-05,
+      "loss": 0.5914,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.794507575757576e-05,
+      "loss": 0.638,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.794034090909091e-05,
+      "loss": 0.6412,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.793560606060607e-05,
+      "loss": 0.6096,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7930871212121214e-05,
+      "loss": 0.5749,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.792613636363637e-05,
+      "loss": 0.5811,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7921401515151514e-05,
+      "loss": 0.5874,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 0.6172,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.791193181818182e-05,
+      "loss": 0.5474,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.790719696969697e-05,
+      "loss": 0.5381,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.790246212121212e-05,
+      "loss": 0.5597,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7897727272727276e-05,
+      "loss": 0.56,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.789299242424243e-05,
+      "loss": 0.5698,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.788825757575758e-05,
+      "loss": 0.6056,
+      "step": 496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.788352272727273e-05,
+      "loss": 0.6365,
+      "step": 497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.787878787878788e-05,
+      "loss": 0.6647,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.787405303030303e-05,
+      "loss": 0.6961,
+      "step": 499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7869318181818185e-05,
+      "loss": 0.5969,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.786458333333333e-05,
+      "loss": 0.5672,
+      "step": 501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7859848484848486e-05,
+      "loss": 0.5196,
+      "step": 502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.785511363636364e-05,
+      "loss": 0.649,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7850378787878794e-05,
+      "loss": 0.7037,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.784564393939394e-05,
+      "loss": 0.583,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7840909090909094e-05,
+      "loss": 0.5895,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.783617424242424e-05,
+      "loss": 0.7128,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7831439393939395e-05,
+      "loss": 0.5749,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.782670454545455e-05,
+      "loss": 0.5253,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7821969696969696e-05,
+      "loss": 0.6055,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.781723484848485e-05,
+      "loss": 0.6267,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7812500000000003e-05,
+      "loss": 0.5645,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.780776515151516e-05,
+      "loss": 0.5194,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7803030303030304e-05,
+      "loss": 0.6499,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.779829545454546e-05,
+      "loss": 0.6159,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7793560606060605e-05,
+      "loss": 0.596,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.778882575757576e-05,
+      "loss": 0.6323,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.778409090909091e-05,
+      "loss": 0.5345,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.777935606060606e-05,
+      "loss": 0.6609,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.777462121212121e-05,
+      "loss": 0.6423,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.776988636363637e-05,
+      "loss": 0.62,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.776515151515152e-05,
+      "loss": 0.5415,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.776041666666667e-05,
+      "loss": 0.6616,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.775568181818182e-05,
+      "loss": 0.532,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7750946969696975e-05,
+      "loss": 0.5554,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.774621212121212e-05,
+      "loss": 0.5265,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7741477272727276e-05,
+      "loss": 0.696,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.773674242424242e-05,
+      "loss": 0.6888,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.773200757575758e-05,
+      "loss": 0.6618,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.772727272727273e-05,
+      "loss": 0.6737,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7722537878787884e-05,
+      "loss": 0.6055,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.771780303030303e-05,
+      "loss": 0.5701,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7713068181818185e-05,
+      "loss": 0.5959,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.770833333333334e-05,
+      "loss": 0.5426,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7703598484848486e-05,
+      "loss": 0.6881,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.769886363636364e-05,
+      "loss": 0.5945,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7694128787878786e-05,
+      "loss": 0.5787,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.768939393939394e-05,
+      "loss": 0.5644,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7684659090909094e-05,
+      "loss": 0.5369,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.767992424242425e-05,
+      "loss": 0.6441,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7675189393939395e-05,
+      "loss": 0.6107,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.767045454545455e-05,
+      "loss": 0.6098,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76657196969697e-05,
+      "loss": 0.6048,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.766098484848485e-05,
+      "loss": 0.6085,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.765625e-05,
+      "loss": 0.5042,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.765151515151515e-05,
+      "loss": 0.6291,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7646780303030304e-05,
+      "loss": 0.6305,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.764204545454546e-05,
+      "loss": 0.6854,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.763731060606061e-05,
+      "loss": 0.6655,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.763257575757576e-05,
+      "loss": 0.6005,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.762784090909091e-05,
+      "loss": 0.5906,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7623106060606066e-05,
+      "loss": 0.619,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.761837121212121e-05,
+      "loss": 0.5969,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7613636363636367e-05,
+      "loss": 0.604,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7608901515151514e-05,
+      "loss": 0.6328,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.760416666666667e-05,
+      "loss": 0.5561,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.759943181818182e-05,
+      "loss": 0.58,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7594696969696975e-05,
+      "loss": 0.534,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.758996212121212e-05,
+      "loss": 0.5442,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7585227272727276e-05,
+      "loss": 0.5862,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.758049242424243e-05,
+      "loss": 0.7086,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7575757575757576e-05,
+      "loss": 0.5716,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.757102272727273e-05,
+      "loss": 0.6093,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.756628787878788e-05,
+      "loss": 0.5727,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.756155303030303e-05,
+      "loss": 0.6463,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7556818181818185e-05,
+      "loss": 0.5207,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.755208333333334e-05,
+      "loss": 0.5835,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7547348484848485e-05,
+      "loss": 0.6402,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.754261363636364e-05,
+      "loss": 0.5905,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.753787878787879e-05,
+      "loss": 0.5679,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.753314393939394e-05,
+      "loss": 0.5383,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7528409090909094e-05,
+      "loss": 0.5757,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.752367424242424e-05,
+      "loss": 0.5934,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.75189393939394e-05,
+      "loss": 0.5677,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.751420454545455e-05,
+      "loss": 0.5984,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.75094696969697e-05,
+      "loss": 0.6716,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.750473484848485e-05,
+      "loss": 0.6396,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.75e-05,
+      "loss": 0.6298,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7495265151515156e-05,
+      "loss": 0.6455,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74905303030303e-05,
+      "loss": 0.6591,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.748579545454546e-05,
+      "loss": 0.6619,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7481060606060604e-05,
+      "loss": 0.6394,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7476325757575765e-05,
+      "loss": 0.5989,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.747159090909091e-05,
+      "loss": 0.5841,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7466856060606065e-05,
+      "loss": 0.6427,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.746212121212121e-05,
+      "loss": 0.5445,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7457386363636366e-05,
+      "loss": 0.5866,
+      "step": 587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.745265151515152e-05,
+      "loss": 0.6262,
+      "step": 588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.744791666666667e-05,
+      "loss": 0.6227,
+      "step": 589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.744318181818182e-05,
+      "loss": 0.5867,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.743844696969697e-05,
+      "loss": 0.5892,
+      "step": 591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.743371212121213e-05,
+      "loss": 0.6145,
+      "step": 592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7428977272727275e-05,
+      "loss": 0.6387,
+      "step": 593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.742424242424243e-05,
+      "loss": 0.6458,
+      "step": 594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7419507575757576e-05,
+      "loss": 0.617,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.741477272727273e-05,
+      "loss": 0.7181,
+      "step": 596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7410037878787883e-05,
+      "loss": 0.6047,
+      "step": 597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.740530303030303e-05,
+      "loss": 0.5446,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7400568181818184e-05,
+      "loss": 0.6347,
+      "step": 599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.739583333333333e-05,
+      "loss": 0.5703,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.739109848484849e-05,
+      "loss": 0.5761,
+      "step": 601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.738636363636364e-05,
+      "loss": 0.5067,
+      "step": 602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.738162878787879e-05,
+      "loss": 0.6852,
+      "step": 603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.737689393939394e-05,
+      "loss": 0.5735,
+      "step": 604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.737215909090909e-05,
+      "loss": 0.637,
+      "step": 605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.736742424242425e-05,
+      "loss": 0.5388,
+      "step": 606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7362689393939394e-05,
+      "loss": 0.6011,
+      "step": 607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.735795454545455e-05,
+      "loss": 0.6132,
+      "step": 608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7353219696969695e-05,
+      "loss": 0.6198,
+      "step": 609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7348484848484855e-05,
+      "loss": 0.6093,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.734375e-05,
+      "loss": 0.669,
+      "step": 611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7339015151515156e-05,
+      "loss": 0.5918,
+      "step": 612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.73342803030303e-05,
+      "loss": 0.598,
+      "step": 613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.732954545454546e-05,
+      "loss": 0.6677,
+      "step": 614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.732481060606061e-05,
+      "loss": 0.5285,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.732007575757576e-05,
+      "loss": 0.6186,
+      "step": 616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.731534090909091e-05,
+      "loss": 0.5467,
+      "step": 617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.731060606060606e-05,
+      "loss": 0.5171,
+      "step": 618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.730587121212122e-05,
+      "loss": 0.6334,
+      "step": 619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7301136363636366e-05,
+      "loss": 0.6356,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.729640151515152e-05,
+      "loss": 0.5773,
+      "step": 621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7291666666666666e-05,
+      "loss": 0.6164,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.728693181818182e-05,
+      "loss": 0.5686,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7282196969696974e-05,
+      "loss": 0.5513,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.727746212121212e-05,
+      "loss": 0.7283,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7272727272727275e-05,
+      "loss": 0.5671,
+      "step": 626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.726799242424242e-05,
+      "loss": 0.5868,
+      "step": 627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.726325757575758e-05,
+      "loss": 0.4949,
+      "step": 628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.725852272727273e-05,
+      "loss": 0.5167,
+      "step": 629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.725378787878788e-05,
+      "loss": 0.5479,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.724905303030303e-05,
+      "loss": 0.6116,
+      "step": 631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7244318181818184e-05,
+      "loss": 0.5981,
+      "step": 632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.723958333333334e-05,
+      "loss": 0.5608,
+      "step": 633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7234848484848485e-05,
+      "loss": 0.5748,
+      "step": 634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.723011363636364e-05,
+      "loss": 0.5071,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7225378787878785e-05,
+      "loss": 0.5879,
+      "step": 636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7220643939393946e-05,
+      "loss": 0.6786,
+      "step": 637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.721590909090909e-05,
+      "loss": 0.6874,
+      "step": 638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7211174242424247e-05,
+      "loss": 0.6794,
+      "step": 639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7206439393939394e-05,
+      "loss": 0.5591,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.720170454545455e-05,
+      "loss": 0.6756,
+      "step": 641
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.71969696969697e-05,
+      "loss": 0.5671,
+      "step": 642
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.719223484848485e-05,
+      "loss": 0.6545,
+      "step": 643
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.71875e-05,
+      "loss": 0.7172,
+      "step": 644
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.718276515151515e-05,
+      "loss": 0.5703,
+      "step": 645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.717803030303031e-05,
+      "loss": 0.634,
+      "step": 646
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7173295454545456e-05,
+      "loss": 0.6136,
+      "step": 647
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.716856060606061e-05,
+      "loss": 0.5704,
+      "step": 648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.716382575757576e-05,
+      "loss": 0.599,
+      "step": 649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.715909090909091e-05,
+      "loss": 0.6231,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7154356060606065e-05,
+      "loss": 0.6108,
+      "step": 651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.714962121212121e-05,
+      "loss": 0.6252,
+      "step": 652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7144886363636365e-05,
+      "loss": 0.5065,
+      "step": 653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.714015151515151e-05,
+      "loss": 0.6508,
+      "step": 654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.713541666666667e-05,
+      "loss": 0.6222,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.713068181818182e-05,
+      "loss": 0.613,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7125946969696974e-05,
+      "loss": 0.5481,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.712121212121212e-05,
+      "loss": 0.6445,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7116477272727274e-05,
+      "loss": 0.6111,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.711174242424243e-05,
+      "loss": 0.5622,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7107007575757575e-05,
+      "loss": 0.5695,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.710227272727273e-05,
+      "loss": 0.5807,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7097537878787876e-05,
+      "loss": 0.6349,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7092803030303036e-05,
+      "loss": 0.6652,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.708806818181818e-05,
+      "loss": 0.7053,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.708333333333334e-05,
+      "loss": 0.6264,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7078598484848484e-05,
+      "loss": 0.5643,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.707386363636364e-05,
+      "loss": 0.5449,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.706912878787879e-05,
+      "loss": 0.6555,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.706439393939394e-05,
+      "loss": 0.5986,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.705965909090909e-05,
+      "loss": 0.7093,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.705492424242424e-05,
+      "loss": 0.6577,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.70501893939394e-05,
+      "loss": 0.6336,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.704545454545455e-05,
+      "loss": 0.6454,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.70407196969697e-05,
+      "loss": 0.5964,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.703598484848485e-05,
+      "loss": 0.6484,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.703125e-05,
+      "loss": 0.6989,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7026515151515155e-05,
+      "loss": 0.7406,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.70217803030303e-05,
+      "loss": 0.5868,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7017045454545456e-05,
+      "loss": 0.5786,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.70123106060606e-05,
+      "loss": 0.5992,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7007575757575763e-05,
+      "loss": 0.6123,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.700284090909091e-05,
+      "loss": 0.6068,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6998106060606064e-05,
+      "loss": 0.5318,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.699337121212121e-05,
+      "loss": 0.6288,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6988636363636365e-05,
+      "loss": 0.6536,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.698390151515152e-05,
+      "loss": 0.5559,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6979166666666666e-05,
+      "loss": 0.5937,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.697443181818182e-05,
+      "loss": 0.5317,
+      "step": 689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.696969696969697e-05,
+      "loss": 0.5899,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.696496212121213e-05,
+      "loss": 0.6022,
+      "step": 691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6960227272727274e-05,
+      "loss": 0.6475,
+      "step": 692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.695549242424243e-05,
+      "loss": 0.5865,
+      "step": 693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6950757575757575e-05,
+      "loss": 0.5978,
+      "step": 694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.694602272727273e-05,
+      "loss": 0.4722,
+      "step": 695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.694128787878788e-05,
+      "loss": 0.4544,
+      "step": 696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.693655303030303e-05,
+      "loss": 0.6176,
+      "step": 697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.693181818181818e-05,
+      "loss": 0.5959,
+      "step": 698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.692708333333334e-05,
+      "loss": 0.6378,
+      "step": 699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.692234848484849e-05,
+      "loss": 0.5853,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.691761363636364e-05,
+      "loss": 0.6752,
+      "step": 701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.691287878787879e-05,
+      "loss": 0.5911,
+      "step": 702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.690814393939394e-05,
+      "loss": 0.5123,
+      "step": 703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.690340909090909e-05,
+      "loss": 0.5055,
+      "step": 704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6898674242424246e-05,
+      "loss": 0.5582,
+      "step": 705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.689393939393939e-05,
+      "loss": 0.5784,
+      "step": 706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6889204545454546e-05,
+      "loss": 0.6017,
+      "step": 707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.68844696969697e-05,
+      "loss": 0.6213,
+      "step": 708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6879734848484854e-05,
+      "loss": 0.6031,
+      "step": 709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.5868,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6870265151515155e-05,
+      "loss": 0.6443,
+      "step": 711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.68655303030303e-05,
+      "loss": 0.5608,
+      "step": 712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6860795454545455e-05,
+      "loss": 0.4892,
+      "step": 713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.685606060606061e-05,
+      "loss": 0.5677,
+      "step": 714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6851325757575756e-05,
+      "loss": 0.6398,
+      "step": 715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.684659090909091e-05,
+      "loss": 0.4938,
+      "step": 716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6841856060606064e-05,
+      "loss": 0.6549,
+      "step": 717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.683712121212122e-05,
+      "loss": 0.6354,
+      "step": 718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6832386363636365e-05,
+      "loss": 0.6519,
+      "step": 719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.682765151515152e-05,
+      "loss": 0.581,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6822916666666665e-05,
+      "loss": 0.5419,
+      "step": 721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.681818181818182e-05,
+      "loss": 0.6283,
+      "step": 722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.681344696969697e-05,
+      "loss": 0.6411,
+      "step": 723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.680871212121212e-05,
+      "loss": 0.588,
+      "step": 724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6803977272727274e-05,
+      "loss": 0.6874,
+      "step": 725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.679924242424243e-05,
+      "loss": 0.506,
+      "step": 726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.679450757575758e-05,
+      "loss": 0.6079,
+      "step": 727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.678977272727273e-05,
+      "loss": 0.5651,
+      "step": 728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.678503787878788e-05,
+      "loss": 0.6625,
+      "step": 729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.678030303030303e-05,
+      "loss": 0.7183,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.677556818181818e-05,
+      "loss": 0.5677,
+      "step": 731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6770833333333336e-05,
+      "loss": 0.6139,
+      "step": 732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.676609848484848e-05,
+      "loss": 0.5477,
+      "step": 733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.676136363636364e-05,
+      "loss": 0.607,
+      "step": 734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.675662878787879e-05,
+      "loss": 0.6151,
+      "step": 735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6751893939393945e-05,
+      "loss": 0.646,
+      "step": 736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.674715909090909e-05,
+      "loss": 0.6489,
+      "step": 737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6742424242424245e-05,
+      "loss": 0.5543,
+      "step": 738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.67376893939394e-05,
+      "loss": 0.6294,
+      "step": 739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6732954545454546e-05,
+      "loss": 0.6314,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.67282196969697e-05,
+      "loss": 0.6003,
+      "step": 741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.672348484848485e-05,
+      "loss": 0.5865,
+      "step": 742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.671875e-05,
+      "loss": 0.6113,
+      "step": 743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6714015151515154e-05,
+      "loss": 0.6229,
+      "step": 744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.670928030303031e-05,
+      "loss": 0.654,
+      "step": 745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6704545454545455e-05,
+      "loss": 0.5318,
+      "step": 746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.669981060606061e-05,
+      "loss": 0.5429,
+      "step": 747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.669507575757576e-05,
+      "loss": 0.6436,
+      "step": 748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.669034090909091e-05,
+      "loss": 0.5792,
+      "step": 749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.668560606060606e-05,
+      "loss": 0.5268,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.668087121212121e-05,
+      "loss": 0.4896,
+      "step": 751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6676136363636364e-05,
+      "loss": 0.566,
+      "step": 752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.667140151515152e-05,
+      "loss": 0.6524,
+      "step": 753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.6215,
+      "step": 754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.666193181818182e-05,
+      "loss": 0.5857,
+      "step": 755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.665719696969697e-05,
+      "loss": 0.6681,
+      "step": 756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6652462121212126e-05,
+      "loss": 0.5682,
+      "step": 757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.664772727272727e-05,
+      "loss": 0.5294,
+      "step": 758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.664299242424243e-05,
+      "loss": 0.5527,
+      "step": 759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6638257575757574e-05,
+      "loss": 0.5289,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.663352272727273e-05,
+      "loss": 0.6182,
+      "step": 761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.662878787878788e-05,
+      "loss": 0.5591,
+      "step": 762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6624053030303035e-05,
+      "loss": 0.6248,
+      "step": 763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.661931818181818e-05,
+      "loss": 0.6615,
+      "step": 764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6614583333333336e-05,
+      "loss": 0.5966,
+      "step": 765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.660984848484849e-05,
+      "loss": 0.5488,
+      "step": 766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.660511363636364e-05,
+      "loss": 0.503,
+      "step": 767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.660037878787879e-05,
+      "loss": 0.5918,
+      "step": 768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.659564393939394e-05,
+      "loss": 0.6493,
+      "step": 769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.659090909090909e-05,
+      "loss": 0.5192,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6586174242424245e-05,
+      "loss": 0.555,
+      "step": 771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.65814393939394e-05,
+      "loss": 0.5143,
+      "step": 772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6576704545454546e-05,
+      "loss": 0.5402,
+      "step": 773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.65719696969697e-05,
+      "loss": 0.5659,
+      "step": 774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.656723484848485e-05,
+      "loss": 0.5863,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.65625e-05,
+      "loss": 0.5419,
+      "step": 776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6557765151515154e-05,
+      "loss": 0.4835,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.65530303030303e-05,
+      "loss": 0.6202,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6548295454545455e-05,
+      "loss": 0.5449,
+      "step": 779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.654356060606061e-05,
+      "loss": 0.593,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.653882575757576e-05,
+      "loss": 0.6221,
+      "step": 781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.653409090909091e-05,
+      "loss": 0.6024,
+      "step": 782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.652935606060606e-05,
+      "loss": 0.6652,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.652462121212122e-05,
+      "loss": 0.5503,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6519886363636364e-05,
+      "loss": 0.6684,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.651515151515152e-05,
+      "loss": 0.6089,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6510416666666664e-05,
+      "loss": 0.5444,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.650568181818182e-05,
+      "loss": 0.5605,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.650094696969697e-05,
+      "loss": 0.5266,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6496212121212126e-05,
+      "loss": 0.5863,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.649147727272727e-05,
+      "loss": 0.6724,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6486742424242426e-05,
+      "loss": 0.6454,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.648200757575758e-05,
+      "loss": 0.76,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.647727272727273e-05,
+      "loss": 0.5533,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.647253787878788e-05,
+      "loss": 0.5844,
+      "step": 795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.646780303030303e-05,
+      "loss": 0.6969,
+      "step": 796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.646306818181819e-05,
+      "loss": 0.6469,
+      "step": 797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6458333333333335e-05,
+      "loss": 0.6722,
+      "step": 798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.645359848484849e-05,
+      "loss": 0.5831,
+      "step": 799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6448863636363636e-05,
+      "loss": 0.5789,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.644412878787879e-05,
+      "loss": 0.5978,
+      "step": 801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6439393939393944e-05,
+      "loss": 0.6692,
+      "step": 802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.643465909090909e-05,
+      "loss": 0.5779,
+      "step": 803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6429924242424244e-05,
+      "loss": 0.5598,
+      "step": 804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.642518939393939e-05,
+      "loss": 0.5803,
+      "step": 805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.642045454545455e-05,
+      "loss": 0.5464,
+      "step": 806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.64157196969697e-05,
+      "loss": 0.5887,
+      "step": 807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.641098484848485e-05,
+      "loss": 0.6601,
+      "step": 808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.640625e-05,
+      "loss": 0.6014,
+      "step": 809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6401515151515154e-05,
+      "loss": 0.5914,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.639678030303031e-05,
+      "loss": 0.6059,
+      "step": 811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6392045454545454e-05,
+      "loss": 0.6689,
+      "step": 812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.638731060606061e-05,
+      "loss": 0.6089,
+      "step": 813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6382575757575755e-05,
+      "loss": 0.552,
+      "step": 814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6377840909090916e-05,
+      "loss": 0.5645,
+      "step": 815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.637310606060606e-05,
+      "loss": 0.5975,
+      "step": 816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6368371212121216e-05,
+      "loss": 0.6222,
+      "step": 817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.636363636363636e-05,
+      "loss": 0.6356,
+      "step": 818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.635890151515152e-05,
+      "loss": 0.7099,
+      "step": 819
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.635416666666667e-05,
+      "loss": 0.6591,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.634943181818182e-05,
+      "loss": 0.5786,
+      "step": 821
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.634469696969697e-05,
+      "loss": 0.5502,
+      "step": 822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.633996212121212e-05,
+      "loss": 0.6667,
+      "step": 823
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.633522727272728e-05,
+      "loss": 0.5643,
+      "step": 824
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6330492424242426e-05,
+      "loss": 0.5736,
+      "step": 825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.632575757575758e-05,
+      "loss": 0.6399,
+      "step": 826
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.632102272727273e-05,
+      "loss": 0.612,
+      "step": 827
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.631628787878788e-05,
+      "loss": 0.6002,
+      "step": 828
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6311553030303034e-05,
+      "loss": 0.5999,
+      "step": 829
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.630681818181818e-05,
+      "loss": 0.6148,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6302083333333335e-05,
+      "loss": 0.6518,
+      "step": 831
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.629734848484848e-05,
+      "loss": 0.508,
+      "step": 832
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.629261363636364e-05,
+      "loss": 0.47,
+      "step": 833
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.628787878787879e-05,
+      "loss": 0.5742,
+      "step": 834
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.628314393939394e-05,
+      "loss": 0.6466,
+      "step": 835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.627840909090909e-05,
+      "loss": 0.4955,
+      "step": 836
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6273674242424244e-05,
+      "loss": 0.5839,
+      "step": 837
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.62689393939394e-05,
+      "loss": 0.4875,
+      "step": 838
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6264204545454545e-05,
+      "loss": 0.6474,
+      "step": 839
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.62594696969697e-05,
+      "loss": 0.5737,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6254734848484846e-05,
+      "loss": 0.5869,
+      "step": 841
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 0.6428,
+      "step": 842
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.624526515151515e-05,
+      "loss": 0.4451,
+      "step": 843
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.624053030303031e-05,
+      "loss": 0.6737,
+      "step": 844
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6235795454545454e-05,
+      "loss": 0.5942,
+      "step": 845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.623106060606061e-05,
+      "loss": 0.6052,
+      "step": 846
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.622632575757576e-05,
+      "loss": 0.5801,
+      "step": 847
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.622159090909091e-05,
+      "loss": 0.616,
+      "step": 848
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.621685606060606e-05,
+      "loss": 0.6283,
+      "step": 849
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.621212121212121e-05,
+      "loss": 0.6738,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.620738636363637e-05,
+      "loss": 0.6185,
+      "step": 851
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.620265151515152e-05,
+      "loss": 0.5686,
+      "step": 852
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.619791666666667e-05,
+      "loss": 0.6651,
+      "step": 853
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.619318181818182e-05,
+      "loss": 0.573,
+      "step": 854
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.618844696969698e-05,
+      "loss": 0.6118,
+      "step": 855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6183712121212125e-05,
+      "loss": 0.5414,
+      "step": 856
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.617897727272727e-05,
+      "loss": 0.6373,
+      "step": 857
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6174242424242426e-05,
+      "loss": 0.5738,
+      "step": 858
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.616950757575757e-05,
+      "loss": 0.5701,
+      "step": 859
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.616477272727273e-05,
+      "loss": 0.6503,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.616003787878788e-05,
+      "loss": 0.697,
+      "step": 861
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6155303030303034e-05,
+      "loss": 0.6254,
+      "step": 862
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.615056818181818e-05,
+      "loss": 0.632,
+      "step": 863
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.614583333333334e-05,
+      "loss": 0.5834,
+      "step": 864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.614109848484849e-05,
+      "loss": 0.5386,
+      "step": 865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6136363636363635e-05,
+      "loss": 0.5359,
+      "step": 866
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.613162878787879e-05,
+      "loss": 0.5723,
+      "step": 867
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6126893939393936e-05,
+      "loss": 0.6917,
+      "step": 868
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.61221590909091e-05,
+      "loss": 0.6313,
+      "step": 869
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6117424242424244e-05,
+      "loss": 0.6023,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.61126893939394e-05,
+      "loss": 0.6114,
+      "step": 871
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6107954545454544e-05,
+      "loss": 0.6773,
+      "step": 872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6103219696969705e-05,
+      "loss": 0.6628,
+      "step": 873
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.609848484848485e-05,
+      "loss": 0.5087,
+      "step": 874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.609375e-05,
+      "loss": 0.5578,
+      "step": 875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.608901515151515e-05,
+      "loss": 0.6474,
+      "step": 876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.60842803030303e-05,
+      "loss": 0.7431,
+      "step": 877
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.607954545454546e-05,
+      "loss": 0.6757,
+      "step": 878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.607481060606061e-05,
+      "loss": 0.7532,
+      "step": 879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.607007575757576e-05,
+      "loss": 0.5965,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.606534090909091e-05,
+      "loss": 0.6238,
+      "step": 881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.606060606060607e-05,
+      "loss": 0.601,
+      "step": 882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6055871212121215e-05,
+      "loss": 0.5669,
+      "step": 883
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.605113636363636e-05,
+      "loss": 0.5915,
+      "step": 884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6046401515151516e-05,
+      "loss": 0.5956,
+      "step": 885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.604166666666666e-05,
+      "loss": 0.5636,
+      "step": 886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6036931818181824e-05,
+      "loss": 0.5833,
+      "step": 887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.603219696969697e-05,
+      "loss": 0.662,
+      "step": 888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6027462121212124e-05,
+      "loss": 0.5696,
+      "step": 889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.602272727272727e-05,
+      "loss": 0.5747,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.601799242424243e-05,
+      "loss": 0.57,
+      "step": 891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.601325757575758e-05,
+      "loss": 0.7179,
+      "step": 892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6008522727272726e-05,
+      "loss": 0.5687,
+      "step": 893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.600378787878788e-05,
+      "loss": 0.5695,
+      "step": 894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.599905303030303e-05,
+      "loss": 0.5696,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.599431818181819e-05,
+      "loss": 0.5649,
+      "step": 896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5989583333333334e-05,
+      "loss": 0.5556,
+      "step": 897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.598484848484849e-05,
+      "loss": 0.5324,
+      "step": 898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5980113636363635e-05,
+      "loss": 0.6139,
+      "step": 899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5975378787878796e-05,
+      "loss": 0.5597,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.597064393939394e-05,
+      "loss": 0.538,
+      "step": 901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.596590909090909e-05,
+      "loss": 0.5632,
+      "step": 902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.596117424242424e-05,
+      "loss": 0.634,
+      "step": 903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.59564393939394e-05,
+      "loss": 0.6318,
+      "step": 904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.595170454545455e-05,
+      "loss": 0.7037,
+      "step": 905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.59469696969697e-05,
+      "loss": 0.563,
+      "step": 906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.594223484848485e-05,
+      "loss": 0.6078,
+      "step": 907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.59375e-05,
+      "loss": 0.5511,
+      "step": 908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.593276515151516e-05,
+      "loss": 0.5644,
+      "step": 909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5928030303030306e-05,
+      "loss": 0.5819,
+      "step": 910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.592329545454545e-05,
+      "loss": 0.5497,
+      "step": 911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.591856060606061e-05,
+      "loss": 0.5527,
+      "step": 912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.591382575757576e-05,
+      "loss": 0.5302,
+      "step": 913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5909090909090914e-05,
+      "loss": 0.6419,
+      "step": 914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.590435606060606e-05,
+      "loss": 0.6073,
+      "step": 915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5899621212121215e-05,
+      "loss": 0.5613,
+      "step": 916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.589488636363636e-05,
+      "loss": 0.5914,
+      "step": 917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.589015151515152e-05,
+      "loss": 0.6444,
+      "step": 918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.588541666666667e-05,
+      "loss": 0.6455,
+      "step": 919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5880681818181817e-05,
+      "loss": 0.6097,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.587594696969697e-05,
+      "loss": 0.6331,
+      "step": 921
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5871212121212124e-05,
+      "loss": 0.6026,
+      "step": 922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.586647727272728e-05,
+      "loss": 0.5533,
+      "step": 923
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5861742424242425e-05,
+      "loss": 0.5875,
+      "step": 924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.585700757575758e-05,
+      "loss": 0.6044,
+      "step": 925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5852272727272726e-05,
+      "loss": 0.5884,
+      "step": 926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5847537878787886e-05,
+      "loss": 0.5631,
+      "step": 927
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.584280303030303e-05,
+      "loss": 0.5407,
+      "step": 928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.583806818181818e-05,
+      "loss": 0.5658,
+      "step": 929
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.6207,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.582859848484849e-05,
+      "loss": 0.6612,
+      "step": 931
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.582386363636364e-05,
+      "loss": 0.53,
+      "step": 932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.581912878787879e-05,
+      "loss": 0.639,
+      "step": 933
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.581439393939394e-05,
+      "loss": 0.5156,
+      "step": 934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.580965909090909e-05,
+      "loss": 0.5752,
+      "step": 935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.580492424242425e-05,
+      "loss": 0.6297,
+      "step": 936
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.58001893939394e-05,
+      "loss": 0.5146,
+      "step": 937
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5795454545454544e-05,
+      "loss": 0.6576,
+      "step": 938
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.57907196969697e-05,
+      "loss": 0.6014,
+      "step": 939
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.578598484848485e-05,
+      "loss": 0.5646,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5781250000000005e-05,
+      "loss": 0.6535,
+      "step": 941
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.577651515151515e-05,
+      "loss": 0.6164,
+      "step": 942
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5771780303030306e-05,
+      "loss": 0.5818,
+      "step": 943
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.576704545454545e-05,
+      "loss": 0.5965,
+      "step": 944
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.576231060606061e-05,
+      "loss": 0.5891,
+      "step": 945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.575757575757576e-05,
+      "loss": 0.5202,
+      "step": 946
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.575284090909091e-05,
+      "loss": 0.5937,
+      "step": 947
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.574810606060606e-05,
+      "loss": 0.5796,
+      "step": 948
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5743371212121215e-05,
+      "loss": 0.6262,
+      "step": 949
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.573863636363637e-05,
+      "loss": 0.5628,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5733901515151515e-05,
+      "loss": 0.6207,
+      "step": 951
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.572916666666667e-05,
+      "loss": 0.5136,
+      "step": 952
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5724431818181816e-05,
+      "loss": 0.5962,
+      "step": 953
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.571969696969698e-05,
+      "loss": 0.609,
+      "step": 954
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5714962121212124e-05,
+      "loss": 0.551,
+      "step": 955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.571022727272727e-05,
+      "loss": 0.5844,
+      "step": 956
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5705492424242424e-05,
+      "loss": 0.607,
+      "step": 957
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.570075757575758e-05,
+      "loss": 0.6317,
+      "step": 958
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.569602272727273e-05,
+      "loss": 0.5418,
+      "step": 959
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.569128787878788e-05,
+      "loss": 0.6074,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.568655303030303e-05,
+      "loss": 0.5645,
+      "step": 961
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5681818181818186e-05,
+      "loss": 0.5673,
+      "step": 962
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.567708333333334e-05,
+      "loss": 0.5657,
+      "step": 963
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.567234848484849e-05,
+      "loss": 0.5611,
+      "step": 964
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5667613636363634e-05,
+      "loss": 0.6151,
+      "step": 965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.566287878787879e-05,
+      "loss": 0.6631,
+      "step": 966
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.565814393939394e-05,
+      "loss": 0.624,
+      "step": 967
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5653409090909095e-05,
+      "loss": 0.5296,
+      "step": 968
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.564867424242424e-05,
+      "loss": 0.6445,
+      "step": 969
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5643939393939396e-05,
+      "loss": 0.6395,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.563920454545455e-05,
+      "loss": 0.4712,
+      "step": 971
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5634469696969704e-05,
+      "loss": 0.5338,
+      "step": 972
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.562973484848485e-05,
+      "loss": 0.5547,
+      "step": 973
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5625e-05,
+      "loss": 0.5857,
+      "step": 974
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.562026515151515e-05,
+      "loss": 0.5428,
+      "step": 975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5615530303030305e-05,
+      "loss": 0.5602,
+      "step": 976
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.561079545454546e-05,
+      "loss": 0.5895,
+      "step": 977
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5606060606060606e-05,
+      "loss": 0.5773,
+      "step": 978
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.560132575757576e-05,
+      "loss": 0.623,
+      "step": 979
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5596590909090913e-05,
+      "loss": 0.6109,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.559185606060607e-05,
+      "loss": 0.5819,
+      "step": 981
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5587121212121214e-05,
+      "loss": 0.5774,
+      "step": 982
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.558238636363636e-05,
+      "loss": 0.7441,
+      "step": 983
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5577651515151515e-05,
+      "loss": 0.5117,
+      "step": 984
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.557291666666667e-05,
+      "loss": 0.5731,
+      "step": 985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.556818181818182e-05,
+      "loss": 0.5415,
+      "step": 986
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.556344696969697e-05,
+      "loss": 0.661,
+      "step": 987
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.555871212121212e-05,
+      "loss": 0.5849,
+      "step": 988
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.555397727272728e-05,
+      "loss": 0.608,
+      "step": 989
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.554924242424243e-05,
+      "loss": 0.5316,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.554450757575758e-05,
+      "loss": 0.4829,
+      "step": 991
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.553977272727273e-05,
+      "loss": 0.5512,
+      "step": 992
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.553503787878788e-05,
+      "loss": 0.5577,
+      "step": 993
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.553030303030303e-05,
+      "loss": 0.6497,
+      "step": 994
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5525568181818186e-05,
+      "loss": 0.5578,
+      "step": 995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.552083333333333e-05,
+      "loss": 0.5286,
+      "step": 996
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.551609848484849e-05,
+      "loss": 0.6254,
+      "step": 997
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.551136363636364e-05,
+      "loss": 0.5578,
+      "step": 998
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5506628787878794e-05,
+      "loss": 0.7206,
+      "step": 999
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.550189393939394e-05,
+      "loss": 0.508,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5497159090909095e-05,
+      "loss": 0.6546,
+      "step": 1001
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.549242424242424e-05,
+      "loss": 0.6318,
+      "step": 1002
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5487689393939396e-05,
+      "loss": 0.5483,
+      "step": 1003
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.548295454545455e-05,
+      "loss": 0.6291,
+      "step": 1004
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5478219696969697e-05,
+      "loss": 0.4841,
+      "step": 1005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.547348484848485e-05,
+      "loss": 0.6018,
+      "step": 1006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5468750000000004e-05,
+      "loss": 0.6202,
+      "step": 1007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.546401515151516e-05,
+      "loss": 0.5906,
+      "step": 1008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5459280303030305e-05,
+      "loss": 0.5638,
+      "step": 1009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.5707,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5449810606060606e-05,
+      "loss": 0.5565,
+      "step": 1011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.544507575757576e-05,
+      "loss": 0.6041,
+      "step": 1012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.544034090909091e-05,
+      "loss": 0.5406,
+      "step": 1013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.543560606060606e-05,
+      "loss": 0.5205,
+      "step": 1014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5430871212121214e-05,
+      "loss": 0.5633,
+      "step": 1015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.542613636363637e-05,
+      "loss": 0.4676,
+      "step": 1016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.542140151515152e-05,
+      "loss": 0.5633,
+      "step": 1017
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.541666666666667e-05,
+      "loss": 0.6432,
+      "step": 1018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.541193181818182e-05,
+      "loss": 0.4552,
+      "step": 1019
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5407196969696976e-05,
+      "loss": 0.5068,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.540246212121212e-05,
+      "loss": 0.5697,
+      "step": 1021
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5397727272727277e-05,
+      "loss": 0.5241,
+      "step": 1022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5392992424242424e-05,
+      "loss": 0.4953,
+      "step": 1023
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.538825757575758e-05,
+      "loss": 0.5732,
+      "step": 1024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.538352272727273e-05,
+      "loss": 0.6171,
+      "step": 1025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5378787878787885e-05,
+      "loss": 0.5368,
+      "step": 1026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.537405303030303e-05,
+      "loss": 0.504,
+      "step": 1027
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5369318181818186e-05,
+      "loss": 0.5261,
+      "step": 1028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.536458333333334e-05,
+      "loss": 0.6843,
+      "step": 1029
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5359848484848486e-05,
+      "loss": 0.4361,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.535511363636364e-05,
+      "loss": 0.6128,
+      "step": 1031
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.535037878787879e-05,
+      "loss": 0.6283,
+      "step": 1032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.534564393939394e-05,
+      "loss": 0.5693,
+      "step": 1033
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5340909090909095e-05,
+      "loss": 0.6608,
+      "step": 1034
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.533617424242425e-05,
+      "loss": 0.4977,
+      "step": 1035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5331439393939395e-05,
+      "loss": 0.5329,
+      "step": 1036
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.532670454545455e-05,
+      "loss": 0.6501,
+      "step": 1037
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.53219696969697e-05,
+      "loss": 0.5201,
+      "step": 1038
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.531723484848485e-05,
+      "loss": 0.5447,
+      "step": 1039
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 0.5592,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.530776515151515e-05,
+      "loss": 0.6081,
+      "step": 1041
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5303030303030304e-05,
+      "loss": 0.64,
+      "step": 1042
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.529829545454546e-05,
+      "loss": 0.6117,
+      "step": 1043
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.529356060606061e-05,
+      "loss": 0.5191,
+      "step": 1044
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.528882575757576e-05,
+      "loss": 0.6299,
+      "step": 1045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.528409090909091e-05,
+      "loss": 0.4775,
+      "step": 1046
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5279356060606066e-05,
+      "loss": 0.4875,
+      "step": 1047
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5274621212121213e-05,
+      "loss": 0.534,
+      "step": 1048
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.526988636363637e-05,
+      "loss": 0.5264,
+      "step": 1049
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5265151515151514e-05,
+      "loss": 0.5613,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.526041666666667e-05,
+      "loss": 0.6107,
+      "step": 1051
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.525568181818182e-05,
+      "loss": 0.5721,
+      "step": 1052
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5250946969696975e-05,
+      "loss": 0.5773,
+      "step": 1053
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.524621212121212e-05,
+      "loss": 0.5902,
+      "step": 1054
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5241477272727276e-05,
+      "loss": 0.4975,
+      "step": 1055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.523674242424243e-05,
+      "loss": 0.5937,
+      "step": 1056
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.523200757575758e-05,
+      "loss": 0.6926,
+      "step": 1057
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.522727272727273e-05,
+      "loss": 0.5379,
+      "step": 1058
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.522253787878788e-05,
+      "loss": 0.5729,
+      "step": 1059
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.521780303030303e-05,
+      "loss": 0.5521,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5213068181818185e-05,
+      "loss": 0.5843,
+      "step": 1061
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.520833333333334e-05,
+      "loss": 0.4288,
+      "step": 1062
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5203598484848486e-05,
+      "loss": 0.6305,
+      "step": 1063
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.519886363636364e-05,
+      "loss": 0.5738,
+      "step": 1064
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5194128787878793e-05,
+      "loss": 0.5978,
+      "step": 1065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.518939393939394e-05,
+      "loss": 0.4735,
+      "step": 1066
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5184659090909094e-05,
+      "loss": 0.5682,
+      "step": 1067
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.517992424242424e-05,
+      "loss": 0.6939,
+      "step": 1068
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.51751893939394e-05,
+      "loss": 0.5539,
+      "step": 1069
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.517045454545455e-05,
+      "loss": 0.6572,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.51657196969697e-05,
+      "loss": 0.6604,
+      "step": 1071
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.516098484848485e-05,
+      "loss": 0.498,
+      "step": 1072
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.515625e-05,
+      "loss": 0.5551,
+      "step": 1073
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.515151515151516e-05,
+      "loss": 0.5666,
+      "step": 1074
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5146780303030304e-05,
+      "loss": 0.5764,
+      "step": 1075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.514204545454546e-05,
+      "loss": 0.525,
+      "step": 1076
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5137310606060605e-05,
+      "loss": 0.6096,
+      "step": 1077
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5132575757575765e-05,
+      "loss": 0.5835,
+      "step": 1078
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.512784090909091e-05,
+      "loss": 0.6774,
+      "step": 1079
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5123106060606066e-05,
+      "loss": 0.6108,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.511837121212121e-05,
+      "loss": 0.5086,
+      "step": 1081
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.511363636363637e-05,
+      "loss": 0.6266,
+      "step": 1082
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.510890151515152e-05,
+      "loss": 0.5611,
+      "step": 1083
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.510416666666667e-05,
+      "loss": 0.5826,
+      "step": 1084
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.509943181818182e-05,
+      "loss": 0.6528,
+      "step": 1085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.509469696969697e-05,
+      "loss": 0.5306,
+      "step": 1086
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.508996212121213e-05,
+      "loss": 0.6083,
+      "step": 1087
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5085227272727276e-05,
+      "loss": 0.5638,
+      "step": 1088
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.508049242424243e-05,
+      "loss": 0.5907,
+      "step": 1089
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5075757575757577e-05,
+      "loss": 0.6167,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.507102272727273e-05,
+      "loss": 0.6419,
+      "step": 1091
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5066287878787884e-05,
+      "loss": 0.5687,
+      "step": 1092
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.506155303030303e-05,
+      "loss": 0.5213,
+      "step": 1093
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5056818181818185e-05,
+      "loss": 0.527,
+      "step": 1094
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.505208333333333e-05,
+      "loss": 0.5708,
+      "step": 1095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.504734848484849e-05,
+      "loss": 0.5119,
+      "step": 1096
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.504261363636364e-05,
+      "loss": 0.5443,
+      "step": 1097
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.503787878787879e-05,
+      "loss": 0.6015,
+      "step": 1098
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.503314393939394e-05,
+      "loss": 0.5865,
+      "step": 1099
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5028409090909094e-05,
+      "loss": 0.5432,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.502367424242425e-05,
+      "loss": 0.5883,
+      "step": 1101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5018939393939395e-05,
+      "loss": 0.5497,
+      "step": 1102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.501420454545455e-05,
+      "loss": 0.5062,
+      "step": 1103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5009469696969695e-05,
+      "loss": 0.5294,
+      "step": 1104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5004734848484856e-05,
+      "loss": 0.5733,
+      "step": 1105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5e-05,
+      "loss": 0.6189,
+      "step": 1106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4995265151515157e-05,
+      "loss": 0.6278,
+      "step": 1107
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4990530303030304e-05,
+      "loss": 0.6448,
+      "step": 1108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.498579545454546e-05,
+      "loss": 0.5659,
+      "step": 1109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.498106060606061e-05,
+      "loss": 0.5406,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.497632575757576e-05,
+      "loss": 0.4508,
+      "step": 1111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.497159090909091e-05,
+      "loss": 0.5303,
+      "step": 1112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.496685606060606e-05,
+      "loss": 0.5441,
+      "step": 1113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.496212121212122e-05,
+      "loss": 0.5183,
+      "step": 1114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4957386363636366e-05,
+      "loss": 0.5264,
+      "step": 1115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.495265151515152e-05,
+      "loss": 0.5419,
+      "step": 1116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.494791666666667e-05,
+      "loss": 0.7084,
+      "step": 1117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.494318181818182e-05,
+      "loss": 0.611,
+      "step": 1118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4938446969696975e-05,
+      "loss": 0.5842,
+      "step": 1119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.493371212121212e-05,
+      "loss": 0.5404,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4928977272727275e-05,
+      "loss": 0.5301,
+      "step": 1121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.492424242424242e-05,
+      "loss": 0.6065,
+      "step": 1122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.491950757575758e-05,
+      "loss": 0.5571,
+      "step": 1123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.491477272727273e-05,
+      "loss": 0.6046,
+      "step": 1124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4910037878787884e-05,
+      "loss": 0.5718,
+      "step": 1125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.490530303030303e-05,
+      "loss": 0.5292,
+      "step": 1126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4900568181818184e-05,
+      "loss": 0.6113,
+      "step": 1127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.489583333333334e-05,
+      "loss": 0.6248,
+      "step": 1128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4891098484848485e-05,
+      "loss": 0.636,
+      "step": 1129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.488636363636364e-05,
+      "loss": 0.5178,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4881628787878786e-05,
+      "loss": 0.4868,
+      "step": 1131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4876893939393946e-05,
+      "loss": 0.6569,
+      "step": 1132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4872159090909093e-05,
+      "loss": 0.5743,
+      "step": 1133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.486742424242425e-05,
+      "loss": 0.5155,
+      "step": 1134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4862689393939394e-05,
+      "loss": 0.6119,
+      "step": 1135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.485795454545455e-05,
+      "loss": 0.6249,
+      "step": 1136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.48532196969697e-05,
+      "loss": 0.5699,
+      "step": 1137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.484848484848485e-05,
+      "loss": 0.5327,
+      "step": 1138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.484375e-05,
+      "loss": 0.4796,
+      "step": 1139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.483901515151515e-05,
+      "loss": 0.5981,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.483428030303031e-05,
+      "loss": 0.6487,
+      "step": 1141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.482954545454546e-05,
+      "loss": 0.5515,
+      "step": 1142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.482481060606061e-05,
+      "loss": 0.5549,
+      "step": 1143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.482007575757576e-05,
+      "loss": 0.6698,
+      "step": 1144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.481534090909091e-05,
+      "loss": 0.5624,
+      "step": 1145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4810606060606065e-05,
+      "loss": 0.518,
+      "step": 1146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.480587121212121e-05,
+      "loss": 0.6271,
+      "step": 1147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4801136363636366e-05,
+      "loss": 0.5488,
+      "step": 1148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.479640151515151e-05,
+      "loss": 0.5519,
+      "step": 1149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4791666666666673e-05,
+      "loss": 0.6022,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.478693181818182e-05,
+      "loss": 0.508,
+      "step": 1151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4782196969696974e-05,
+      "loss": 0.5208,
+      "step": 1152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.477746212121212e-05,
+      "loss": 0.5568,
+      "step": 1153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4772727272727275e-05,
+      "loss": 0.6055,
+      "step": 1154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.476799242424243e-05,
+      "loss": 0.6075,
+      "step": 1155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4763257575757576e-05,
+      "loss": 0.6619,
+      "step": 1156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.475852272727273e-05,
+      "loss": 0.5264,
+      "step": 1157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4753787878787876e-05,
+      "loss": 0.4361,
+      "step": 1158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.474905303030304e-05,
+      "loss": 0.6245,
+      "step": 1159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4744318181818184e-05,
+      "loss": 0.5957,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.473958333333334e-05,
+      "loss": 0.558,
+      "step": 1161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4734848484848485e-05,
+      "loss": 0.5177,
+      "step": 1162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.473011363636364e-05,
+      "loss": 0.5653,
+      "step": 1163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.472537878787879e-05,
+      "loss": 0.5151,
+      "step": 1164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.472064393939394e-05,
+      "loss": 0.5692,
+      "step": 1165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.471590909090909e-05,
+      "loss": 0.4926,
+      "step": 1166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.471117424242424e-05,
+      "loss": 0.5883,
+      "step": 1167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.47064393939394e-05,
+      "loss": 0.6566,
+      "step": 1168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.470170454545455e-05,
+      "loss": 0.6757,
+      "step": 1169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.46969696969697e-05,
+      "loss": 0.6958,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.469223484848485e-05,
+      "loss": 0.5424,
+      "step": 1171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.46875e-05,
+      "loss": 0.5435,
+      "step": 1172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4682765151515156e-05,
+      "loss": 0.5376,
+      "step": 1173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.46780303030303e-05,
+      "loss": 0.5401,
+      "step": 1174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4673295454545457e-05,
+      "loss": 0.6433,
+      "step": 1175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4668560606060604e-05,
+      "loss": 0.6282,
+      "step": 1176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4663825757575764e-05,
+      "loss": 0.5788,
+      "step": 1177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.465909090909091e-05,
+      "loss": 0.6169,
+      "step": 1178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4654356060606065e-05,
+      "loss": 0.6038,
+      "step": 1179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.464962121212121e-05,
+      "loss": 0.6431,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4644886363636366e-05,
+      "loss": 0.5885,
+      "step": 1181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.464015151515152e-05,
+      "loss": 0.5581,
+      "step": 1182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4635416666666666e-05,
+      "loss": 0.6356,
+      "step": 1183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.463068181818182e-05,
+      "loss": 0.4829,
+      "step": 1184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4625946969696974e-05,
+      "loss": 0.6252,
+      "step": 1185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.462121212121213e-05,
+      "loss": 0.5218,
+      "step": 1186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4616477272727275e-05,
+      "loss": 0.5772,
+      "step": 1187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.461174242424243e-05,
+      "loss": 0.6406,
+      "step": 1188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4607007575757575e-05,
+      "loss": 0.5553,
+      "step": 1189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.460227272727273e-05,
+      "loss": 0.5726,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.459753787878788e-05,
+      "loss": 0.5433,
+      "step": 1191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.459280303030303e-05,
+      "loss": 0.6535,
+      "step": 1192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4588068181818184e-05,
+      "loss": 0.6082,
+      "step": 1193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.458333333333334e-05,
+      "loss": 0.5709,
+      "step": 1194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.457859848484849e-05,
+      "loss": 0.6571,
+      "step": 1195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.457386363636364e-05,
+      "loss": 0.5166,
+      "step": 1196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.456912878787879e-05,
+      "loss": 0.6348,
+      "step": 1197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.456439393939394e-05,
+      "loss": 0.52,
+      "step": 1198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.455965909090909e-05,
+      "loss": 0.5622,
+      "step": 1199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4554924242424246e-05,
+      "loss": 0.5154,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.455018939393939e-05,
+      "loss": 0.58,
+      "step": 1201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.454545454545455e-05,
+      "loss": 0.5215,
+      "step": 1202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.45407196969697e-05,
+      "loss": 0.4663,
+      "step": 1203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4535984848484855e-05,
+      "loss": 0.6025,
+      "step": 1204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.453125e-05,
+      "loss": 0.5313,
+      "step": 1205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4526515151515155e-05,
+      "loss": 0.4875,
+      "step": 1206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.45217803030303e-05,
+      "loss": 0.6244,
+      "step": 1207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4517045454545456e-05,
+      "loss": 0.5552,
+      "step": 1208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.451231060606061e-05,
+      "loss": 0.6388,
+      "step": 1209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.450757575757576e-05,
+      "loss": 0.6006,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.450284090909091e-05,
+      "loss": 0.5881,
+      "step": 1211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4498106060606064e-05,
+      "loss": 0.6046,
+      "step": 1212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.449337121212122e-05,
+      "loss": 0.6942,
+      "step": 1213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4488636363636365e-05,
+      "loss": 0.6046,
+      "step": 1214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.448390151515152e-05,
+      "loss": 0.6354,
+      "step": 1215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4479166666666666e-05,
+      "loss": 0.572,
+      "step": 1216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.447443181818182e-05,
+      "loss": 0.5208,
+      "step": 1217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4469696969696973e-05,
+      "loss": 0.5483,
+      "step": 1218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.446496212121212e-05,
+      "loss": 0.5452,
+      "step": 1219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4460227272727274e-05,
+      "loss": 0.5352,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.445549242424243e-05,
+      "loss": 0.5355,
+      "step": 1221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.445075757575758e-05,
+      "loss": 0.6803,
+      "step": 1222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.444602272727273e-05,
+      "loss": 0.6005,
+      "step": 1223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.444128787878788e-05,
+      "loss": 0.5751,
+      "step": 1224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.443655303030303e-05,
+      "loss": 0.5913,
+      "step": 1225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.443181818181818e-05,
+      "loss": 0.5405,
+      "step": 1226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.442708333333334e-05,
+      "loss": 0.5664,
+      "step": 1227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4422348484848484e-05,
+      "loss": 0.4843,
+      "step": 1228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.441761363636364e-05,
+      "loss": 0.616,
+      "step": 1229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.441287878787879e-05,
+      "loss": 0.5322,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4408143939393945e-05,
+      "loss": 0.5667,
+      "step": 1231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.440340909090909e-05,
+      "loss": 0.6757,
+      "step": 1232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4398674242424246e-05,
+      "loss": 0.4545,
+      "step": 1233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.43939393939394e-05,
+      "loss": 0.4157,
+      "step": 1234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.438920454545455e-05,
+      "loss": 0.5383,
+      "step": 1235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.43844696969697e-05,
+      "loss": 0.5653,
+      "step": 1236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.437973484848485e-05,
+      "loss": 0.5281,
+      "step": 1237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4375e-05,
+      "loss": 0.6162,
+      "step": 1238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4370265151515155e-05,
+      "loss": 0.501,
+      "step": 1239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.436553030303031e-05,
+      "loss": 0.5013,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4360795454545456e-05,
+      "loss": 0.5247,
+      "step": 1241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.435606060606061e-05,
+      "loss": 0.7073,
+      "step": 1242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.435132575757576e-05,
+      "loss": 0.6197,
+      "step": 1243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.434659090909091e-05,
+      "loss": 0.6123,
+      "step": 1244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4341856060606064e-05,
+      "loss": 0.5655,
+      "step": 1245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.433712121212121e-05,
+      "loss": 0.5947,
+      "step": 1246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4332386363636365e-05,
+      "loss": 0.5733,
+      "step": 1247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.432765151515152e-05,
+      "loss": 0.5636,
+      "step": 1248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.432291666666667e-05,
+      "loss": 0.5839,
+      "step": 1249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.431818181818182e-05,
+      "loss": 0.5512,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.431344696969697e-05,
+      "loss": 0.5385,
+      "step": 1251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.430871212121213e-05,
+      "loss": 0.5719,
+      "step": 1252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4303977272727274e-05,
+      "loss": 0.5389,
+      "step": 1253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.429924242424243e-05,
+      "loss": 0.6521,
+      "step": 1254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4294507575757574e-05,
+      "loss": 0.6298,
+      "step": 1255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.428977272727273e-05,
+      "loss": 0.5272,
+      "step": 1256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.428503787878788e-05,
+      "loss": 0.443,
+      "step": 1257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4280303030303036e-05,
+      "loss": 0.5191,
+      "step": 1258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.427556818181818e-05,
+      "loss": 0.6443,
+      "step": 1259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4270833333333337e-05,
+      "loss": 0.5283,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.426609848484849e-05,
+      "loss": 0.5769,
+      "step": 1261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.426136363636364e-05,
+      "loss": 0.673,
+      "step": 1262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.425662878787879e-05,
+      "loss": 0.5559,
+      "step": 1263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.425189393939394e-05,
+      "loss": 0.5855,
+      "step": 1264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.424715909090909e-05,
+      "loss": 0.6189,
+      "step": 1265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4242424242424246e-05,
+      "loss": 0.5814,
+      "step": 1266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.42376893939394e-05,
+      "loss": 0.5372,
+      "step": 1267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4232954545454546e-05,
+      "loss": 0.5575,
+      "step": 1268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.42282196969697e-05,
+      "loss": 0.6492,
+      "step": 1269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4223484848484854e-05,
+      "loss": 0.5527,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.421875e-05,
+      "loss": 0.5558,
+      "step": 1271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4214015151515155e-05,
+      "loss": 0.6167,
+      "step": 1272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.42092803030303e-05,
+      "loss": 0.6661,
+      "step": 1273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4204545454545455e-05,
+      "loss": 0.5525,
+      "step": 1274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.419981060606061e-05,
+      "loss": 0.5487,
+      "step": 1275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.419507575757576e-05,
+      "loss": 0.5698,
+      "step": 1276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.419034090909091e-05,
+      "loss": 0.5286,
+      "step": 1277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4185606060606064e-05,
+      "loss": 0.4545,
+      "step": 1278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.418087121212122e-05,
+      "loss": 0.6051,
+      "step": 1279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4176136363636364e-05,
+      "loss": 0.5271,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.417140151515152e-05,
+      "loss": 0.5447,
+      "step": 1281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4166666666666665e-05,
+      "loss": 0.7695,
+      "step": 1282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.416193181818182e-05,
+      "loss": 0.6419,
+      "step": 1283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.415719696969697e-05,
+      "loss": 0.7686,
+      "step": 1284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4152462121212126e-05,
+      "loss": 0.7109,
+      "step": 1285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.414772727272727e-05,
+      "loss": 0.6428,
+      "step": 1286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.414299242424243e-05,
+      "loss": 0.5762,
+      "step": 1287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.413825757575758e-05,
+      "loss": 0.4572,
+      "step": 1288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.413352272727273e-05,
+      "loss": 0.6513,
+      "step": 1289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.412878787878788e-05,
+      "loss": 0.5745,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.412405303030303e-05,
+      "loss": 0.5628,
+      "step": 1291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.411931818181819e-05,
+      "loss": 0.5182,
+      "step": 1292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4114583333333336e-05,
+      "loss": 0.6568,
+      "step": 1293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.410984848484849e-05,
+      "loss": 0.6131,
+      "step": 1294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.410511363636364e-05,
+      "loss": 0.5119,
+      "step": 1295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.410037878787879e-05,
+      "loss": 0.5524,
+      "step": 1296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4095643939393944e-05,
+      "loss": 0.5861,
+      "step": 1297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.409090909090909e-05,
+      "loss": 0.5285,
+      "step": 1298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4086174242424245e-05,
+      "loss": 0.6194,
+      "step": 1299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.408143939393939e-05,
+      "loss": 0.5762,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.407670454545455e-05,
+      "loss": 0.5427,
+      "step": 1301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.40719696969697e-05,
+      "loss": 0.618,
+      "step": 1302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4067234848484853e-05,
+      "loss": 0.5826,
+      "step": 1303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.40625e-05,
+      "loss": 0.5057,
+      "step": 1304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4057765151515154e-05,
+      "loss": 0.5569,
+      "step": 1305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.405303030303031e-05,
+      "loss": 0.5599,
+      "step": 1306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4048295454545455e-05,
+      "loss": 0.5007,
+      "step": 1307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.404356060606061e-05,
+      "loss": 0.678,
+      "step": 1308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4038825757575756e-05,
+      "loss": 0.5845,
+      "step": 1309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4034090909090916e-05,
+      "loss": 0.5073,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.402935606060606e-05,
+      "loss": 0.6031,
+      "step": 1311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.402462121212122e-05,
+      "loss": 0.5609,
+      "step": 1312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4019886363636364e-05,
+      "loss": 0.4858,
+      "step": 1313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.401515151515152e-05,
+      "loss": 0.5447,
+      "step": 1314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.401041666666667e-05,
+      "loss": 0.4791,
+      "step": 1315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.400568181818182e-05,
+      "loss": 0.5959,
+      "step": 1316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.400094696969697e-05,
+      "loss": 0.569,
+      "step": 1317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.399621212121212e-05,
+      "loss": 0.5982,
+      "step": 1318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.399147727272728e-05,
+      "loss": 0.4157,
+      "step": 1319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.398674242424243e-05,
+      "loss": 0.6388,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.398200757575758e-05,
+      "loss": 0.6015,
+      "step": 1321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.397727272727273e-05,
+      "loss": 0.5399,
+      "step": 1322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.397253787878788e-05,
+      "loss": 0.4794,
+      "step": 1323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3967803030303035e-05,
+      "loss": 0.602,
+      "step": 1324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.396306818181818e-05,
+      "loss": 0.5286,
+      "step": 1325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3958333333333336e-05,
+      "loss": 0.569,
+      "step": 1326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.395359848484848e-05,
+      "loss": 0.5033,
+      "step": 1327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.394886363636364e-05,
+      "loss": 0.5029,
+      "step": 1328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.394412878787879e-05,
+      "loss": 0.5582,
+      "step": 1329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3939393939393944e-05,
+      "loss": 0.4521,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.393465909090909e-05,
+      "loss": 0.649,
+      "step": 1331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3929924242424245e-05,
+      "loss": 0.6214,
+      "step": 1332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.39251893939394e-05,
+      "loss": 0.6075,
+      "step": 1333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3920454545454545e-05,
+      "loss": 0.5797,
+      "step": 1334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.39157196969697e-05,
+      "loss": 0.6014,
+      "step": 1335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3910984848484846e-05,
+      "loss": 0.5694,
+      "step": 1336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.390625000000001e-05,
+      "loss": 0.5732,
+      "step": 1337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3901515151515154e-05,
+      "loss": 0.571,
+      "step": 1338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.389678030303031e-05,
+      "loss": 0.5678,
+      "step": 1339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3892045454545454e-05,
+      "loss": 0.6017,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.388731060606061e-05,
+      "loss": 0.5002,
+      "step": 1341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.388257575757576e-05,
+      "loss": 0.5245,
+      "step": 1342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.387784090909091e-05,
+      "loss": 0.5396,
+      "step": 1343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.387310606060606e-05,
+      "loss": 0.5302,
+      "step": 1344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.386837121212121e-05,
+      "loss": 0.5595,
+      "step": 1345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.386363636363637e-05,
+      "loss": 0.6392,
+      "step": 1346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.385890151515152e-05,
+      "loss": 0.5831,
+      "step": 1347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.385416666666667e-05,
+      "loss": 0.5741,
+      "step": 1348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.384943181818182e-05,
+      "loss": 0.5354,
+      "step": 1349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.384469696969697e-05,
+      "loss": 0.7344,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3839962121212126e-05,
+      "loss": 0.6803,
+      "step": 1351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.383522727272727e-05,
+      "loss": 0.6497,
+      "step": 1352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3830492424242426e-05,
+      "loss": 0.5032,
+      "step": 1353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.382575757575757e-05,
+      "loss": 0.575,
+      "step": 1354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3821022727272734e-05,
+      "loss": 0.5818,
+      "step": 1355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.381628787878788e-05,
+      "loss": 0.5344,
+      "step": 1356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3811553030303035e-05,
+      "loss": 0.5171,
+      "step": 1357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.380681818181818e-05,
+      "loss": 0.6272,
+      "step": 1358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3802083333333335e-05,
+      "loss": 0.7019,
+      "step": 1359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.379734848484849e-05,
+      "loss": 0.5734,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3792613636363636e-05,
+      "loss": 0.5803,
+      "step": 1361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.378787878787879e-05,
+      "loss": 0.5686,
+      "step": 1362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.378314393939394e-05,
+      "loss": 0.6536,
+      "step": 1363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.37784090909091e-05,
+      "loss": 0.5813,
+      "step": 1364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3773674242424244e-05,
+      "loss": 0.6074,
+      "step": 1365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.37689393939394e-05,
+      "loss": 0.6012,
+      "step": 1366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3764204545454545e-05,
+      "loss": 0.5282,
+      "step": 1367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.37594696969697e-05,
+      "loss": 0.6557,
+      "step": 1368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.375473484848485e-05,
+      "loss": 0.5963,
+      "step": 1369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.375e-05,
+      "loss": 0.5108,
+      "step": 1370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.374526515151515e-05,
+      "loss": 0.5617,
+      "step": 1371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.37405303030303e-05,
+      "loss": 0.518,
+      "step": 1372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.373579545454546e-05,
+      "loss": 0.6599,
+      "step": 1373
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.373106060606061e-05,
+      "loss": 0.5928,
+      "step": 1374
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.372632575757576e-05,
+      "loss": 0.507,
+      "step": 1375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.372159090909091e-05,
+      "loss": 0.5628,
+      "step": 1376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.371685606060606e-05,
+      "loss": 0.5462,
+      "step": 1377
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3712121212121216e-05,
+      "loss": 0.5441,
+      "step": 1378
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.370738636363636e-05,
+      "loss": 0.6549,
+      "step": 1379
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.370265151515152e-05,
+      "loss": 0.5753,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3697916666666664e-05,
+      "loss": 0.574,
+      "step": 1381
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3693181818181824e-05,
+      "loss": 0.565,
+      "step": 1382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.368844696969697e-05,
+      "loss": 0.4187,
+      "step": 1383
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3683712121212125e-05,
+      "loss": 0.4847,
+      "step": 1384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.367897727272727e-05,
+      "loss": 0.6015,
+      "step": 1385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3674242424242426e-05,
+      "loss": 0.5967,
+      "step": 1386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.366950757575758e-05,
+      "loss": 0.6385,
+      "step": 1387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3664772727272727e-05,
+      "loss": 0.644,
+      "step": 1388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.366003787878788e-05,
+      "loss": 0.4617,
+      "step": 1389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.365530303030303e-05,
+      "loss": 0.509,
+      "step": 1390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.365056818181819e-05,
+      "loss": 0.6292,
+      "step": 1391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3645833333333335e-05,
+      "loss": 0.5685,
+      "step": 1392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.364109848484849e-05,
+      "loss": 0.6151,
+      "step": 1393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3636363636363636e-05,
+      "loss": 0.6437,
+      "step": 1394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.363162878787879e-05,
+      "loss": 0.5359,
+      "step": 1395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.362689393939394e-05,
+      "loss": 0.5446,
+      "step": 1396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.362215909090909e-05,
+      "loss": 0.5955,
+      "step": 1397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3617424242424244e-05,
+      "loss": 0.5356,
+      "step": 1398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.36126893939394e-05,
+      "loss": 0.6679,
+      "step": 1399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.360795454545455e-05,
+      "loss": 0.4952,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.36032196969697e-05,
+      "loss": 0.5029,
+      "step": 1401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.359848484848485e-05,
+      "loss": 0.5885,
+      "step": 1402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.359375e-05,
+      "loss": 0.4884,
+      "step": 1403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.358901515151515e-05,
+      "loss": 0.432,
+      "step": 1404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.358428030303031e-05,
+      "loss": 0.6394,
+      "step": 1405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3579545454545454e-05,
+      "loss": 0.549,
+      "step": 1406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.357481060606061e-05,
+      "loss": 0.5144,
+      "step": 1407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.357007575757576e-05,
+      "loss": 0.5576,
+      "step": 1408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3565340909090915e-05,
+      "loss": 0.6546,
+      "step": 1409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.356060606060606e-05,
+      "loss": 0.7001,
+      "step": 1410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3555871212121216e-05,
+      "loss": 0.6487,
+      "step": 1411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.355113636363636e-05,
+      "loss": 0.5599,
+      "step": 1412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3546401515151516e-05,
+      "loss": 0.5025,
+      "step": 1413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.354166666666667e-05,
+      "loss": 0.5429,
+      "step": 1414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.353693181818182e-05,
+      "loss": 0.5741,
+      "step": 1415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.353219696969697e-05,
+      "loss": 0.5751,
+      "step": 1416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3527462121212125e-05,
+      "loss": 0.7024,
+      "step": 1417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.352272727272728e-05,
+      "loss": 0.7152,
+      "step": 1418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3517992424242425e-05,
+      "loss": 0.7138,
+      "step": 1419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.351325757575758e-05,
+      "loss": 0.5611,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3508522727272726e-05,
+      "loss": 0.5248,
+      "step": 1421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.350378787878788e-05,
+      "loss": 0.6108,
+      "step": 1422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3499053030303034e-05,
+      "loss": 0.5464,
+      "step": 1423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.349431818181818e-05,
+      "loss": 0.6156,
+      "step": 1424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3489583333333334e-05,
+      "loss": 0.5844,
+      "step": 1425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.348484848484849e-05,
+      "loss": 0.5505,
+      "step": 1426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.348011363636364e-05,
+      "loss": 0.5403,
+      "step": 1427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.347537878787879e-05,
+      "loss": 0.6028,
+      "step": 1428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.347064393939394e-05,
+      "loss": 0.5957,
+      "step": 1429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.346590909090909e-05,
+      "loss": 0.6113,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3461174242424243e-05,
+      "loss": 0.6402,
+      "step": 1431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.34564393939394e-05,
+      "loss": 0.5769,
+      "step": 1432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3451704545454544e-05,
+      "loss": 0.5769,
+      "step": 1433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.34469696969697e-05,
+      "loss": 0.6094,
+      "step": 1434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.344223484848485e-05,
+      "loss": 0.5446,
+      "step": 1435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3437500000000006e-05,
+      "loss": 0.6287,
+      "step": 1436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.343276515151515e-05,
+      "loss": 0.5164,
+      "step": 1437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3428030303030306e-05,
+      "loss": 0.5075,
+      "step": 1438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.342329545454545e-05,
+      "loss": 0.5406,
+      "step": 1439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.341856060606061e-05,
+      "loss": 0.6532,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.341382575757576e-05,
+      "loss": 0.5209,
+      "step": 1441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.340909090909091e-05,
+      "loss": 0.6635,
+      "step": 1442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.340435606060606e-05,
+      "loss": 0.5337,
+      "step": 1443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3399621212121215e-05,
+      "loss": 0.5075,
+      "step": 1444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.339488636363637e-05,
+      "loss": 0.5229,
+      "step": 1445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3390151515151516e-05,
+      "loss": 0.6216,
+      "step": 1446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.338541666666667e-05,
+      "loss": 0.5191,
+      "step": 1447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.338068181818182e-05,
+      "loss": 0.5113,
+      "step": 1448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.337594696969697e-05,
+      "loss": 0.6608,
+      "step": 1449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3371212121212124e-05,
+      "loss": 0.5515,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.336647727272727e-05,
+      "loss": 0.6748,
+      "step": 1451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3361742424242425e-05,
+      "loss": 0.5212,
+      "step": 1452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.335700757575758e-05,
+      "loss": 0.5124,
+      "step": 1453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.335227272727273e-05,
+      "loss": 0.5889,
+      "step": 1454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.334753787878788e-05,
+      "loss": 0.5746,
+      "step": 1455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.334280303030303e-05,
+      "loss": 0.49,
+      "step": 1456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.333806818181819e-05,
+      "loss": 0.6842,
+      "step": 1457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.5962,
+      "step": 1458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.332859848484849e-05,
+      "loss": 0.5227,
+      "step": 1459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3323863636363635e-05,
+      "loss": 0.5534,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.331912878787879e-05,
+      "loss": 0.4822,
+      "step": 1461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.331439393939394e-05,
+      "loss": 0.5438,
+      "step": 1462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3309659090909096e-05,
+      "loss": 0.5208,
+      "step": 1463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.330492424242424e-05,
+      "loss": 0.5962,
+      "step": 1464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.33001893939394e-05,
+      "loss": 0.5499,
+      "step": 1465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.329545454545455e-05,
+      "loss": 0.6234,
+      "step": 1466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.32907196969697e-05,
+      "loss": 0.6328,
+      "step": 1467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.328598484848485e-05,
+      "loss": 0.6322,
+      "step": 1468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.328125e-05,
+      "loss": 0.5369,
+      "step": 1469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.327651515151515e-05,
+      "loss": 0.6462,
+      "step": 1470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3271780303030306e-05,
+      "loss": 0.4792,
+      "step": 1471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.326704545454546e-05,
+      "loss": 0.4349,
+      "step": 1472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3262310606060607e-05,
+      "loss": 0.51,
+      "step": 1473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.325757575757576e-05,
+      "loss": 0.5468,
+      "step": 1474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3252840909090914e-05,
+      "loss": 0.6496,
+      "step": 1475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.324810606060606e-05,
+      "loss": 0.4688,
+      "step": 1476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3243371212121215e-05,
+      "loss": 0.6248,
+      "step": 1477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.323863636363636e-05,
+      "loss": 0.5915,
+      "step": 1478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3233901515151516e-05,
+      "loss": 0.6888,
+      "step": 1479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.322916666666667e-05,
+      "loss": 0.6298,
+      "step": 1480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.322443181818182e-05,
+      "loss": 0.5307,
+      "step": 1481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.321969696969697e-05,
+      "loss": 0.4664,
+      "step": 1482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3214962121212124e-05,
+      "loss": 0.7105,
+      "step": 1483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.321022727272728e-05,
+      "loss": 0.5544,
+      "step": 1484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3205492424242425e-05,
+      "loss": 0.5826,
+      "step": 1485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.320075757575758e-05,
+      "loss": 0.5447,
+      "step": 1486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3196022727272725e-05,
+      "loss": 0.5534,
+      "step": 1487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.319128787878788e-05,
+      "loss": 0.532,
+      "step": 1488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.318655303030303e-05,
+      "loss": 0.6171,
+      "step": 1489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.318181818181819e-05,
+      "loss": 0.56,
+      "step": 1490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3177083333333334e-05,
+      "loss": 0.5782,
+      "step": 1491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.317234848484849e-05,
+      "loss": 0.5624,
+      "step": 1492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.316761363636364e-05,
+      "loss": 0.5688,
+      "step": 1493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.316287878787879e-05,
+      "loss": 0.6191,
+      "step": 1494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.315814393939394e-05,
+      "loss": 0.6089,
+      "step": 1495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.315340909090909e-05,
+      "loss": 0.5558,
+      "step": 1496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.314867424242424e-05,
+      "loss": 0.582,
+      "step": 1497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3143939393939396e-05,
+      "loss": 0.5757,
+      "step": 1498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.313920454545455e-05,
+      "loss": 0.5151,
+      "step": 1499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.31344696969697e-05,
+      "loss": 0.6297,
+      "step": 1500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.312973484848485e-05,
+      "loss": 0.5882,
+      "step": 1501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3125000000000005e-05,
+      "loss": 0.567,
+      "step": 1502
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.312026515151515e-05,
+      "loss": 0.5551,
+      "step": 1503
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3115530303030305e-05,
+      "loss": 0.577,
+      "step": 1504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.311079545454545e-05,
+      "loss": 0.5635,
+      "step": 1505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3106060606060606e-05,
+      "loss": 0.5946,
+      "step": 1506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.310132575757576e-05,
+      "loss": 0.5938,
+      "step": 1507
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3096590909090914e-05,
+      "loss": 0.5751,
+      "step": 1508
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.309185606060606e-05,
+      "loss": 0.616,
+      "step": 1509
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3087121212121214e-05,
+      "loss": 0.562,
+      "step": 1510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.308238636363637e-05,
+      "loss": 0.5365,
+      "step": 1511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3077651515151515e-05,
+      "loss": 0.5913,
+      "step": 1512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.307291666666667e-05,
+      "loss": 0.4174,
+      "step": 1513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3068181818181816e-05,
+      "loss": 0.6125,
+      "step": 1514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3063446969696976e-05,
+      "loss": 0.597,
+      "step": 1515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3058712121212123e-05,
+      "loss": 0.6179,
+      "step": 1516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.305397727272728e-05,
+      "loss": 0.5842,
+      "step": 1517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3049242424242424e-05,
+      "loss": 0.4926,
+      "step": 1518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.304450757575758e-05,
+      "loss": 0.525,
+      "step": 1519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.303977272727273e-05,
+      "loss": 0.5504,
+      "step": 1520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.303503787878788e-05,
+      "loss": 0.4703,
+      "step": 1521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.303030303030303e-05,
+      "loss": 0.635,
+      "step": 1522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.302556818181818e-05,
+      "loss": 0.5773,
+      "step": 1523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.302083333333334e-05,
+      "loss": 0.531,
+      "step": 1524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.301609848484849e-05,
+      "loss": 0.5993,
+      "step": 1525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.301136363636364e-05,
+      "loss": 0.5628,
+      "step": 1526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.300662878787879e-05,
+      "loss": 0.514,
+      "step": 1527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.300189393939394e-05,
+      "loss": 0.6501,
+      "step": 1528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2997159090909095e-05,
+      "loss": 0.5829,
+      "step": 1529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.299242424242424e-05,
+      "loss": 0.4673,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2987689393939396e-05,
+      "loss": 0.5284,
+      "step": 1531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.298295454545454e-05,
+      "loss": 0.5264,
+      "step": 1532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2978219696969704e-05,
+      "loss": 0.5688,
+      "step": 1533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.297348484848485e-05,
+      "loss": 0.6764,
+      "step": 1534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2968750000000004e-05,
+      "loss": 0.5585,
+      "step": 1535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.296401515151515e-05,
+      "loss": 0.5829,
+      "step": 1536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2959280303030305e-05,
+      "loss": 0.6634,
+      "step": 1537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.295454545454546e-05,
+      "loss": 0.6212,
+      "step": 1538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2949810606060606e-05,
+      "loss": 0.6083,
+      "step": 1539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.294507575757576e-05,
+      "loss": 0.559,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2940340909090907e-05,
+      "loss": 0.67,
+      "step": 1541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.293560606060607e-05,
+      "loss": 0.5413,
+      "step": 1542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2930871212121214e-05,
+      "loss": 0.5607,
+      "step": 1543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.292613636363637e-05,
+      "loss": 0.5005,
+      "step": 1544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2921401515151515e-05,
+      "loss": 0.5606,
+      "step": 1545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.291666666666667e-05,
+      "loss": 0.4985,
+      "step": 1546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.291193181818182e-05,
+      "loss": 0.4355,
+      "step": 1547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.290719696969697e-05,
+      "loss": 0.5363,
+      "step": 1548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.290246212121212e-05,
+      "loss": 0.6084,
+      "step": 1549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.289772727272727e-05,
+      "loss": 0.5605,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.289299242424243e-05,
+      "loss": 0.4382,
+      "step": 1551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.288825757575758e-05,
+      "loss": 0.5898,
+      "step": 1552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.288352272727273e-05,
+      "loss": 0.5152,
+      "step": 1553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.287878787878788e-05,
+      "loss": 0.5118,
+      "step": 1554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.287405303030303e-05,
+      "loss": 0.631,
+      "step": 1555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2869318181818186e-05,
+      "loss": 0.4977,
+      "step": 1556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.286458333333333e-05,
+      "loss": 0.5888,
+      "step": 1557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2859848484848487e-05,
+      "loss": 0.491,
+      "step": 1558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2855113636363634e-05,
+      "loss": 0.4959,
+      "step": 1559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2850378787878794e-05,
+      "loss": 0.4871,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.284564393939394e-05,
+      "loss": 0.5886,
+      "step": 1561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2840909090909095e-05,
+      "loss": 0.6043,
+      "step": 1562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.283617424242424e-05,
+      "loss": 0.5849,
+      "step": 1563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2831439393939396e-05,
+      "loss": 0.5591,
+      "step": 1564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.282670454545455e-05,
+      "loss": 0.4518,
+      "step": 1565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2821969696969696e-05,
+      "loss": 0.5374,
+      "step": 1566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.281723484848485e-05,
+      "loss": 0.5599,
+      "step": 1567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.28125e-05,
+      "loss": 0.6428,
+      "step": 1568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.280776515151516e-05,
+      "loss": 0.5405,
+      "step": 1569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2803030303030305e-05,
+      "loss": 0.5393,
+      "step": 1570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.279829545454546e-05,
+      "loss": 0.6661,
+      "step": 1571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2793560606060605e-05,
+      "loss": 0.5757,
+      "step": 1572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.278882575757576e-05,
+      "loss": 0.5869,
+      "step": 1573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.278409090909091e-05,
+      "loss": 0.5894,
+      "step": 1574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.277935606060606e-05,
+      "loss": 0.4975,
+      "step": 1575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2774621212121214e-05,
+      "loss": 0.5921,
+      "step": 1576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.276988636363636e-05,
+      "loss": 0.5989,
+      "step": 1577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.276515151515152e-05,
+      "loss": 0.6422,
+      "step": 1578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.276041666666667e-05,
+      "loss": 0.5771,
+      "step": 1579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.275568181818182e-05,
+      "loss": 0.5637,
+      "step": 1580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.275094696969697e-05,
+      "loss": 0.6896,
+      "step": 1581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.274621212121212e-05,
+      "loss": 0.5739,
+      "step": 1582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2741477272727276e-05,
+      "loss": 0.5547,
+      "step": 1583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2736742424242423e-05,
+      "loss": 0.643,
+      "step": 1584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.273200757575758e-05,
+      "loss": 0.5256,
+      "step": 1585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2727272727272724e-05,
+      "loss": 0.4635,
+      "step": 1586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2722537878787885e-05,
+      "loss": 0.6064,
+      "step": 1587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.271780303030303e-05,
+      "loss": 0.5184,
+      "step": 1588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2713068181818185e-05,
+      "loss": 0.6444,
+      "step": 1589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.270833333333333e-05,
+      "loss": 0.4993,
+      "step": 1590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2703598484848486e-05,
+      "loss": 0.5492,
+      "step": 1591
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.269886363636364e-05,
+      "loss": 0.4821,
+      "step": 1592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.269412878787879e-05,
+      "loss": 0.6285,
+      "step": 1593
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.268939393939394e-05,
+      "loss": 0.5047,
+      "step": 1594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.268465909090909e-05,
+      "loss": 0.5249,
+      "step": 1595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.267992424242425e-05,
+      "loss": 0.5236,
+      "step": 1596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2675189393939395e-05,
+      "loss": 0.4953,
+      "step": 1597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.267045454545455e-05,
+      "loss": 0.52,
+      "step": 1598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2665719696969696e-05,
+      "loss": 0.6038,
+      "step": 1599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.266098484848485e-05,
+      "loss": 0.6558,
+      "step": 1600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2656250000000003e-05,
+      "loss": 0.587,
+      "step": 1601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.265151515151515e-05,
+      "loss": 0.5972,
+      "step": 1602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2646780303030304e-05,
+      "loss": 0.486,
+      "step": 1603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.264204545454545e-05,
+      "loss": 0.7204,
+      "step": 1604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.263731060606061e-05,
+      "loss": 0.6251,
+      "step": 1605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.263257575757576e-05,
+      "loss": 0.5902,
+      "step": 1606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.262784090909091e-05,
+      "loss": 0.6065,
+      "step": 1607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.262310606060606e-05,
+      "loss": 0.6623,
+      "step": 1608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.261837121212121e-05,
+      "loss": 0.5578,
+      "step": 1609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.261363636363637e-05,
+      "loss": 0.6431,
+      "step": 1610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2608901515151514e-05,
+      "loss": 0.6638,
+      "step": 1611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.260416666666667e-05,
+      "loss": 0.5902,
+      "step": 1612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2599431818181815e-05,
+      "loss": 0.5279,
+      "step": 1613
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2594696969696975e-05,
+      "loss": 0.5828,
+      "step": 1614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.258996212121212e-05,
+      "loss": 0.5427,
+      "step": 1615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2585227272727276e-05,
+      "loss": 0.6122,
+      "step": 1616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.258049242424242e-05,
+      "loss": 0.5412,
+      "step": 1617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.257575757575758e-05,
+      "loss": 0.561,
+      "step": 1618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.257102272727273e-05,
+      "loss": 0.6203,
+      "step": 1619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.256628787878788e-05,
+      "loss": 0.556,
+      "step": 1620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.256155303030303e-05,
+      "loss": 0.5996,
+      "step": 1621
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2556818181818185e-05,
+      "loss": 0.5526,
+      "step": 1622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.255208333333334e-05,
+      "loss": 0.5924,
+      "step": 1623
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2547348484848486e-05,
+      "loss": 0.5638,
+      "step": 1624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.254261363636364e-05,
+      "loss": 0.4911,
+      "step": 1625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2537878787878787e-05,
+      "loss": 0.4627,
+      "step": 1626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.253314393939394e-05,
+      "loss": 0.5747,
+      "step": 1627
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2528409090909094e-05,
+      "loss": 0.5268,
+      "step": 1628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.252367424242424e-05,
+      "loss": 0.5983,
+      "step": 1629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2518939393939395e-05,
+      "loss": 0.5737,
+      "step": 1630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.251420454545455e-05,
+      "loss": 0.5527,
+      "step": 1631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.25094696969697e-05,
+      "loss": 0.4922,
+      "step": 1632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.250473484848485e-05,
+      "loss": 0.5067,
+      "step": 1633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.25e-05,
+      "loss": 0.4963,
+      "step": 1634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.249526515151515e-05,
+      "loss": 0.5261,
+      "step": 1635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2490530303030304e-05,
+      "loss": 0.6377,
+      "step": 1636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.248579545454546e-05,
+      "loss": 0.628,
+      "step": 1637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2481060606060605e-05,
+      "loss": 0.6338,
+      "step": 1638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.247632575757576e-05,
+      "loss": 0.5038,
+      "step": 1639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.247159090909091e-05,
+      "loss": 0.5529,
+      "step": 1640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2466856060606066e-05,
+      "loss": 0.4659,
+      "step": 1641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.246212121212121e-05,
+      "loss": 0.5684,
+      "step": 1642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2457386363636367e-05,
+      "loss": 0.57,
+      "step": 1643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2452651515151514e-05,
+      "loss": 0.5413,
+      "step": 1644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.244791666666667e-05,
+      "loss": 0.6584,
+      "step": 1645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.244318181818182e-05,
+      "loss": 0.6206,
+      "step": 1646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.243844696969697e-05,
+      "loss": 0.5788,
+      "step": 1647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.243371212121212e-05,
+      "loss": 0.6193,
+      "step": 1648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2428977272727276e-05,
+      "loss": 0.5474,
+      "step": 1649
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.242424242424243e-05,
+      "loss": 0.585,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2419507575757576e-05,
+      "loss": 0.5152,
+      "step": 1651
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.241477272727273e-05,
+      "loss": 0.5422,
+      "step": 1652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.241003787878788e-05,
+      "loss": 0.4612,
+      "step": 1653
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.240530303030303e-05,
+      "loss": 0.5931,
+      "step": 1654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2400568181818185e-05,
+      "loss": 0.4753,
+      "step": 1655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.239583333333333e-05,
+      "loss": 0.5803,
+      "step": 1656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2391098484848485e-05,
+      "loss": 0.5688,
+      "step": 1657
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.238636363636364e-05,
+      "loss": 0.587,
+      "step": 1658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.238162878787879e-05,
+      "loss": 0.5923,
+      "step": 1659
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.237689393939394e-05,
+      "loss": 0.501,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2372159090909094e-05,
+      "loss": 0.5851,
+      "step": 1661
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.236742424242424e-05,
+      "loss": 0.5323,
+      "step": 1662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2362689393939394e-05,
+      "loss": 0.5465,
+      "step": 1663
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.235795454545455e-05,
+      "loss": 0.4994,
+      "step": 1664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2353219696969695e-05,
+      "loss": 0.4611,
+      "step": 1665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.234848484848485e-05,
+      "loss": 0.3853,
+      "step": 1666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.234375e-05,
+      "loss": 0.4817,
+      "step": 1667
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2339015151515156e-05,
+      "loss": 0.614,
+      "step": 1668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2334280303030303e-05,
+      "loss": 0.4771,
+      "step": 1669
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.232954545454546e-05,
+      "loss": 0.6564,
+      "step": 1670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2324810606060604e-05,
+      "loss": 0.5961,
+      "step": 1671
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.232007575757576e-05,
+      "loss": 0.5305,
+      "step": 1672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.231534090909091e-05,
+      "loss": 0.5963,
+      "step": 1673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.231060606060606e-05,
+      "loss": 0.5938,
+      "step": 1674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.230587121212121e-05,
+      "loss": 0.5072,
+      "step": 1675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2301136363636366e-05,
+      "loss": 0.5725,
+      "step": 1676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.229640151515152e-05,
+      "loss": 0.5716,
+      "step": 1677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.229166666666667e-05,
+      "loss": 0.5172,
+      "step": 1678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.228693181818182e-05,
+      "loss": 0.49,
+      "step": 1679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2282196969696974e-05,
+      "loss": 0.5897,
+      "step": 1680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.227746212121212e-05,
+      "loss": 0.5627,
+      "step": 1681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2272727272727275e-05,
+      "loss": 0.6644,
+      "step": 1682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.226799242424242e-05,
+      "loss": 0.6197,
+      "step": 1683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2263257575757576e-05,
+      "loss": 0.5109,
+      "step": 1684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.225852272727273e-05,
+      "loss": 0.6412,
+      "step": 1685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2253787878787883e-05,
+      "loss": 0.5535,
+      "step": 1686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.224905303030303e-05,
+      "loss": 0.6136,
+      "step": 1687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2244318181818184e-05,
+      "loss": 0.5557,
+      "step": 1688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.223958333333334e-05,
+      "loss": 0.4964,
+      "step": 1689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2234848484848485e-05,
+      "loss": 0.6067,
+      "step": 1690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.223011363636364e-05,
+      "loss": 0.5625,
+      "step": 1691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2225378787878786e-05,
+      "loss": 0.525,
+      "step": 1692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.222064393939394e-05,
+      "loss": 0.5856,
+      "step": 1693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.221590909090909e-05,
+      "loss": 0.5179,
+      "step": 1694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.221117424242425e-05,
+      "loss": 0.6403,
+      "step": 1695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2206439393939394e-05,
+      "loss": 0.6088,
+      "step": 1696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.220170454545455e-05,
+      "loss": 0.5906,
+      "step": 1697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.21969696969697e-05,
+      "loss": 0.6119,
+      "step": 1698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.219223484848485e-05,
+      "loss": 0.4996,
+      "step": 1699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.21875e-05,
+      "loss": 0.5408,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.218276515151515e-05,
+      "loss": 0.5947,
+      "step": 1701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.21780303030303e-05,
+      "loss": 0.5938,
+      "step": 1702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.217329545454546e-05,
+      "loss": 0.523,
+      "step": 1703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.216856060606061e-05,
+      "loss": 0.5149,
+      "step": 1704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.216382575757576e-05,
+      "loss": 0.502,
+      "step": 1705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.215909090909091e-05,
+      "loss": 0.5128,
+      "step": 1706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2154356060606065e-05,
+      "loss": 0.5997,
+      "step": 1707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.214962121212121e-05,
+      "loss": 0.5578,
+      "step": 1708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2144886363636366e-05,
+      "loss": 0.6359,
+      "step": 1709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.214015151515151e-05,
+      "loss": 0.6155,
+      "step": 1710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2135416666666667e-05,
+      "loss": 0.6772,
+      "step": 1711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.213068181818182e-05,
+      "loss": 0.5723,
+      "step": 1712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2125946969696974e-05,
+      "loss": 0.57,
+      "step": 1713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.212121212121212e-05,
+      "loss": 0.5889,
+      "step": 1714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2116477272727275e-05,
+      "loss": 0.6338,
+      "step": 1715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.211174242424243e-05,
+      "loss": 0.5726,
+      "step": 1716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2107007575757576e-05,
+      "loss": 0.4822,
+      "step": 1717
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.210227272727273e-05,
+      "loss": 0.5895,
+      "step": 1718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2097537878787876e-05,
+      "loss": 0.6314,
+      "step": 1719
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.209280303030303e-05,
+      "loss": 0.5456,
+      "step": 1720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2088068181818184e-05,
+      "loss": 0.6221,
+      "step": 1721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.208333333333334e-05,
+      "loss": 0.5414,
+      "step": 1722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2078598484848485e-05,
+      "loss": 0.6204,
+      "step": 1723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.207386363636364e-05,
+      "loss": 0.6595,
+      "step": 1724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.206912878787879e-05,
+      "loss": 0.5971,
+      "step": 1725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.206439393939394e-05,
+      "loss": 0.5969,
+      "step": 1726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.205965909090909e-05,
+      "loss": 0.5464,
+      "step": 1727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.205492424242424e-05,
+      "loss": 0.5637,
+      "step": 1728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.20501893939394e-05,
+      "loss": 0.5632,
+      "step": 1729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.204545454545455e-05,
+      "loss": 0.5709,
+      "step": 1730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.20407196969697e-05,
+      "loss": 0.5873,
+      "step": 1731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.203598484848485e-05,
+      "loss": 0.5938,
+      "step": 1732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.203125e-05,
+      "loss": 0.4857,
+      "step": 1733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2026515151515156e-05,
+      "loss": 0.5821,
+      "step": 1734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.20217803030303e-05,
+      "loss": 0.5175,
+      "step": 1735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2017045454545456e-05,
+      "loss": 0.6038,
+      "step": 1736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.20123106060606e-05,
+      "loss": 0.6116,
+      "step": 1737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2007575757575764e-05,
+      "loss": 0.4852,
+      "step": 1738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.200284090909091e-05,
+      "loss": 0.5151,
+      "step": 1739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1998106060606065e-05,
+      "loss": 0.579,
+      "step": 1740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.199337121212121e-05,
+      "loss": 0.6417,
+      "step": 1741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1988636363636365e-05,
+      "loss": 0.4962,
+      "step": 1742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.198390151515152e-05,
+      "loss": 0.5329,
+      "step": 1743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1979166666666666e-05,
+      "loss": 0.5332,
+      "step": 1744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.197443181818182e-05,
+      "loss": 0.5053,
+      "step": 1745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.196969696969697e-05,
+      "loss": 0.6538,
+      "step": 1746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.196496212121213e-05,
+      "loss": 0.568,
+      "step": 1747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1960227272727274e-05,
+      "loss": 0.5188,
+      "step": 1748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.195549242424243e-05,
+      "loss": 0.5566,
+      "step": 1749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1950757575757575e-05,
+      "loss": 0.5252,
+      "step": 1750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.194602272727273e-05,
+      "loss": 0.5973,
+      "step": 1751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.194128787878788e-05,
+      "loss": 0.494,
+      "step": 1752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.193655303030303e-05,
+      "loss": 0.6233,
+      "step": 1753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.193181818181818e-05,
+      "loss": 0.456,
+      "step": 1754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.192708333333333e-05,
+      "loss": 0.6228,
+      "step": 1755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.192234848484849e-05,
+      "loss": 0.5536,
+      "step": 1756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.191761363636364e-05,
+      "loss": 0.5968,
+      "step": 1757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.191287878787879e-05,
+      "loss": 0.5267,
+      "step": 1758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.190814393939394e-05,
+      "loss": 0.6518,
+      "step": 1759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.190340909090909e-05,
+      "loss": 0.5822,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1898674242424246e-05,
+      "loss": 0.4832,
+      "step": 1761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.189393939393939e-05,
+      "loss": 0.6069,
+      "step": 1762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.188920454545455e-05,
+      "loss": 0.5827,
+      "step": 1763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1884469696969694e-05,
+      "loss": 0.462,
+      "step": 1764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1879734848484854e-05,
+      "loss": 0.4887,
+      "step": 1765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1875e-05,
+      "loss": 0.6274,
+      "step": 1766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1870265151515155e-05,
+      "loss": 0.6092,
+      "step": 1767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.18655303030303e-05,
+      "loss": 0.6042,
+      "step": 1768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1860795454545456e-05,
+      "loss": 0.6042,
+      "step": 1769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.185606060606061e-05,
+      "loss": 0.5804,
+      "step": 1770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.185132575757576e-05,
+      "loss": 0.6,
+      "step": 1771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.184659090909091e-05,
+      "loss": 0.6276,
+      "step": 1772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.184185606060606e-05,
+      "loss": 0.4753,
+      "step": 1773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.183712121212122e-05,
+      "loss": 0.4679,
+      "step": 1774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1832386363636365e-05,
+      "loss": 0.4684,
+      "step": 1775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.182765151515152e-05,
+      "loss": 0.5444,
+      "step": 1776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1822916666666666e-05,
+      "loss": 0.5965,
+      "step": 1777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.181818181818182e-05,
+      "loss": 0.5604,
+      "step": 1778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.181344696969697e-05,
+      "loss": 0.5591,
+      "step": 1779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.180871212121212e-05,
+      "loss": 0.5438,
+      "step": 1780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1803977272727274e-05,
+      "loss": 0.5292,
+      "step": 1781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.179924242424242e-05,
+      "loss": 0.5766,
+      "step": 1782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.179450757575758e-05,
+      "loss": 0.5524,
+      "step": 1783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.178977272727273e-05,
+      "loss": 0.602,
+      "step": 1784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.178503787878788e-05,
+      "loss": 0.5463,
+      "step": 1785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.178030303030303e-05,
+      "loss": 0.4938,
+      "step": 1786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.177556818181819e-05,
+      "loss": 0.5995,
+      "step": 1787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.177083333333334e-05,
+      "loss": 0.5952,
+      "step": 1788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1766098484848484e-05,
+      "loss": 0.6133,
+      "step": 1789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.176136363636364e-05,
+      "loss": 0.555,
+      "step": 1790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1756628787878784e-05,
+      "loss": 0.5167,
+      "step": 1791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1751893939393945e-05,
+      "loss": 0.6183,
+      "step": 1792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.174715909090909e-05,
+      "loss": 0.4637,
+      "step": 1793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1742424242424246e-05,
+      "loss": 0.5629,
+      "step": 1794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.173768939393939e-05,
+      "loss": 0.5833,
+      "step": 1795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.173295454545455e-05,
+      "loss": 0.5102,
+      "step": 1796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.17282196969697e-05,
+      "loss": 0.5718,
+      "step": 1797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.172348484848485e-05,
+      "loss": 0.4932,
+      "step": 1798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.171875e-05,
+      "loss": 0.6199,
+      "step": 1799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.171401515151515e-05,
+      "loss": 0.5367,
+      "step": 1800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.170928030303031e-05,
+      "loss": 0.5864,
+      "step": 1801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1704545454545456e-05,
+      "loss": 0.5115,
+      "step": 1802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.169981060606061e-05,
+      "loss": 0.5234,
+      "step": 1803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1695075757575756e-05,
+      "loss": 0.5991,
+      "step": 1804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.169034090909092e-05,
+      "loss": 0.51,
+      "step": 1805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1685606060606064e-05,
+      "loss": 0.471,
+      "step": 1806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.168087121212121e-05,
+      "loss": 0.5062,
+      "step": 1807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1676136363636365e-05,
+      "loss": 0.5842,
+      "step": 1808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.167140151515151e-05,
+      "loss": 0.5397,
+      "step": 1809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.5642,
+      "step": 1810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.166193181818182e-05,
+      "loss": 0.5983,
+      "step": 1811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.165719696969697e-05,
+      "loss": 0.5119,
+      "step": 1812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.165246212121212e-05,
+      "loss": 0.6212,
+      "step": 1813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.164772727272728e-05,
+      "loss": 0.5181,
+      "step": 1814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.164299242424243e-05,
+      "loss": 0.5554,
+      "step": 1815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.163825757575758e-05,
+      "loss": 0.5568,
+      "step": 1816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.163352272727273e-05,
+      "loss": 0.5455,
+      "step": 1817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1628787878787875e-05,
+      "loss": 0.5162,
+      "step": 1818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1624053030303036e-05,
+      "loss": 0.5736,
+      "step": 1819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.161931818181818e-05,
+      "loss": 0.5892,
+      "step": 1820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1614583333333336e-05,
+      "loss": 0.5907,
+      "step": 1821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.160984848484848e-05,
+      "loss": 0.4991,
+      "step": 1822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1605113636363644e-05,
+      "loss": 0.4981,
+      "step": 1823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.160037878787879e-05,
+      "loss": 0.5146,
+      "step": 1824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1595643939393945e-05,
+      "loss": 0.4472,
+      "step": 1825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.159090909090909e-05,
+      "loss": 0.4542,
+      "step": 1826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.158617424242424e-05,
+      "loss": 0.5047,
+      "step": 1827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.15814393939394e-05,
+      "loss": 0.6175,
+      "step": 1828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1576704545454546e-05,
+      "loss": 0.5176,
+      "step": 1829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.15719696969697e-05,
+      "loss": 0.6237,
+      "step": 1830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.156723484848485e-05,
+      "loss": 0.6386,
+      "step": 1831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.156250000000001e-05,
+      "loss": 0.5771,
+      "step": 1832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1557765151515154e-05,
+      "loss": 0.5993,
+      "step": 1833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.155303030303031e-05,
+      "loss": 0.5204,
+      "step": 1834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1548295454545455e-05,
+      "loss": 0.5881,
+      "step": 1835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.15435606060606e-05,
+      "loss": 0.649,
+      "step": 1836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.153882575757576e-05,
+      "loss": 0.489,
+      "step": 1837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.153409090909091e-05,
+      "loss": 0.5475,
+      "step": 1838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.152935606060606e-05,
+      "loss": 0.5226,
+      "step": 1839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.152462121212121e-05,
+      "loss": 0.5258,
+      "step": 1840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.151988636363637e-05,
+      "loss": 0.5125,
+      "step": 1841
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.151515151515152e-05,
+      "loss": 0.6106,
+      "step": 1842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.151041666666667e-05,
+      "loss": 0.5838,
+      "step": 1843
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.150568181818182e-05,
+      "loss": 0.6977,
+      "step": 1844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.150094696969697e-05,
+      "loss": 0.5398,
+      "step": 1845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1496212121212126e-05,
+      "loss": 0.4691,
+      "step": 1846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.149147727272727e-05,
+      "loss": 0.4682,
+      "step": 1847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.148674242424243e-05,
+      "loss": 0.5478,
+      "step": 1848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1482007575757574e-05,
+      "loss": 0.5057,
+      "step": 1849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1477272727272734e-05,
+      "loss": 0.6524,
+      "step": 1850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.147253787878788e-05,
+      "loss": 0.5777,
+      "step": 1851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1467803030303035e-05,
+      "loss": 0.5331,
+      "step": 1852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.146306818181818e-05,
+      "loss": 0.5678,
+      "step": 1853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1458333333333336e-05,
+      "loss": 0.5812,
+      "step": 1854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.145359848484849e-05,
+      "loss": 0.4496,
+      "step": 1855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.144886363636364e-05,
+      "loss": 0.5447,
+      "step": 1856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.144412878787879e-05,
+      "loss": 0.6214,
+      "step": 1857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.143939393939394e-05,
+      "loss": 0.4777,
+      "step": 1858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.14346590909091e-05,
+      "loss": 0.5588,
+      "step": 1859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1429924242424245e-05,
+      "loss": 0.5883,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.14251893939394e-05,
+      "loss": 0.6027,
+      "step": 1861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1420454545454546e-05,
+      "loss": 0.5964,
+      "step": 1862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.14157196969697e-05,
+      "loss": 0.6187,
+      "step": 1863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.141098484848485e-05,
+      "loss": 0.4767,
+      "step": 1864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.140625e-05,
+      "loss": 0.5667,
+      "step": 1865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1401515151515154e-05,
+      "loss": 0.4687,
+      "step": 1866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.13967803030303e-05,
+      "loss": 0.5368,
+      "step": 1867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.139204545454546e-05,
+      "loss": 0.6477,
+      "step": 1868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.138731060606061e-05,
+      "loss": 0.5102,
+      "step": 1869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.138257575757576e-05,
+      "loss": 0.5998,
+      "step": 1870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.137784090909091e-05,
+      "loss": 0.5372,
+      "step": 1871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.137310606060606e-05,
+      "loss": 0.5352,
+      "step": 1872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.136837121212122e-05,
+      "loss": 0.5736,
+      "step": 1873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1363636363636364e-05,
+      "loss": 0.5851,
+      "step": 1874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.135890151515152e-05,
+      "loss": 0.4798,
+      "step": 1875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1354166666666664e-05,
+      "loss": 0.5601,
+      "step": 1876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1349431818181825e-05,
+      "loss": 0.4867,
+      "step": 1877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.134469696969697e-05,
+      "loss": 0.6694,
+      "step": 1878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1339962121212126e-05,
+      "loss": 0.6443,
+      "step": 1879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.133522727272727e-05,
+      "loss": 0.4656,
+      "step": 1880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1330492424242426e-05,
+      "loss": 0.446,
+      "step": 1881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.132575757575758e-05,
+      "loss": 0.5576,
+      "step": 1882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.132102272727273e-05,
+      "loss": 0.6392,
+      "step": 1883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.131628787878788e-05,
+      "loss": 0.5551,
+      "step": 1884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.131155303030303e-05,
+      "loss": 0.5136,
+      "step": 1885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.130681818181819e-05,
+      "loss": 0.5041,
+      "step": 1886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1302083333333336e-05,
+      "loss": 0.5576,
+      "step": 1887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.129734848484849e-05,
+      "loss": 0.5283,
+      "step": 1888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1292613636363636e-05,
+      "loss": 0.5616,
+      "step": 1889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.128787878787879e-05,
+      "loss": 0.4739,
+      "step": 1890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1283143939393944e-05,
+      "loss": 0.5335,
+      "step": 1891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.127840909090909e-05,
+      "loss": 0.5466,
+      "step": 1892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1273674242424245e-05,
+      "loss": 0.6208,
+      "step": 1893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.12689393939394e-05,
+      "loss": 0.517,
+      "step": 1894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.126420454545455e-05,
+      "loss": 0.5671,
+      "step": 1895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.12594696969697e-05,
+      "loss": 0.5823,
+      "step": 1896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.125473484848485e-05,
+      "loss": 0.6541,
+      "step": 1897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.125e-05,
+      "loss": 0.755,
+      "step": 1898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1245265151515154e-05,
+      "loss": 0.5348,
+      "step": 1899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.124053030303031e-05,
+      "loss": 0.5186,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1235795454545454e-05,
+      "loss": 0.5155,
+      "step": 1901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.123106060606061e-05,
+      "loss": 0.5375,
+      "step": 1902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.122632575757576e-05,
+      "loss": 0.5296,
+      "step": 1903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1221590909090916e-05,
+      "loss": 0.5234,
+      "step": 1904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.121685606060606e-05,
+      "loss": 0.5357,
+      "step": 1905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1212121212121216e-05,
+      "loss": 0.4985,
+      "step": 1906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.120738636363636e-05,
+      "loss": 0.4862,
+      "step": 1907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.120265151515152e-05,
+      "loss": 0.6104,
+      "step": 1908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.119791666666667e-05,
+      "loss": 0.5125,
+      "step": 1909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.119318181818182e-05,
+      "loss": 0.5316,
+      "step": 1910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.118844696969697e-05,
+      "loss": 0.4453,
+      "step": 1911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1183712121212125e-05,
+      "loss": 0.705,
+      "step": 1912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.117897727272728e-05,
+      "loss": 0.5573,
+      "step": 1913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1174242424242426e-05,
+      "loss": 0.5054,
+      "step": 1914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.116950757575758e-05,
+      "loss": 0.4923,
+      "step": 1915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.116477272727273e-05,
+      "loss": 0.5454,
+      "step": 1916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.116003787878788e-05,
+      "loss": 0.569,
+      "step": 1917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1155303030303034e-05,
+      "loss": 0.4733,
+      "step": 1918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.115056818181818e-05,
+      "loss": 0.5145,
+      "step": 1919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1145833333333335e-05,
+      "loss": 0.5276,
+      "step": 1920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.114109848484849e-05,
+      "loss": 0.519,
+      "step": 1921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.113636363636364e-05,
+      "loss": 0.5692,
+      "step": 1922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.113162878787879e-05,
+      "loss": 0.5154,
+      "step": 1923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.112689393939394e-05,
+      "loss": 0.5788,
+      "step": 1924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.112215909090909e-05,
+      "loss": 0.5089,
+      "step": 1925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1117424242424244e-05,
+      "loss": 0.4864,
+      "step": 1926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.11126893939394e-05,
+      "loss": 0.6791,
+      "step": 1927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1107954545454545e-05,
+      "loss": 0.5435,
+      "step": 1928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.11032196969697e-05,
+      "loss": 0.5553,
+      "step": 1929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.109848484848485e-05,
+      "loss": 0.6881,
+      "step": 1930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1093750000000006e-05,
+      "loss": 0.4741,
+      "step": 1931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.108901515151515e-05,
+      "loss": 0.5086,
+      "step": 1932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.108428030303031e-05,
+      "loss": 0.5952,
+      "step": 1933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1079545454545454e-05,
+      "loss": 0.5563,
+      "step": 1934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.107481060606061e-05,
+      "loss": 0.5325,
+      "step": 1935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.107007575757576e-05,
+      "loss": 0.664,
+      "step": 1936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.106534090909091e-05,
+      "loss": 0.5105,
+      "step": 1937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.106060606060606e-05,
+      "loss": 0.5845,
+      "step": 1938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1055871212121216e-05,
+      "loss": 0.5262,
+      "step": 1939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.105113636363637e-05,
+      "loss": 0.6457,
+      "step": 1940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.104640151515152e-05,
+      "loss": 0.4384,
+      "step": 1941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.104166666666667e-05,
+      "loss": 0.595,
+      "step": 1942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.103693181818182e-05,
+      "loss": 0.6041,
+      "step": 1943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.103219696969697e-05,
+      "loss": 0.5971,
+      "step": 1944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1027462121212125e-05,
+      "loss": 0.5651,
+      "step": 1945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.102272727272727e-05,
+      "loss": 0.5283,
+      "step": 1946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1017992424242426e-05,
+      "loss": 0.5084,
+      "step": 1947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.101325757575758e-05,
+      "loss": 0.5941,
+      "step": 1948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.100852272727273e-05,
+      "loss": 0.5487,
+      "step": 1949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.100378787878788e-05,
+      "loss": 0.6342,
+      "step": 1950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0999053030303034e-05,
+      "loss": 0.624,
+      "step": 1951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.099431818181819e-05,
+      "loss": 0.5832,
+      "step": 1952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0989583333333335e-05,
+      "loss": 0.5318,
+      "step": 1953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.098484848484849e-05,
+      "loss": 0.5704,
+      "step": 1954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0980113636363635e-05,
+      "loss": 0.6097,
+      "step": 1955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.097537878787879e-05,
+      "loss": 0.5576,
+      "step": 1956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.097064393939394e-05,
+      "loss": 0.5738,
+      "step": 1957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.09659090909091e-05,
+      "loss": 0.5972,
+      "step": 1958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0961174242424244e-05,
+      "loss": 0.5961,
+      "step": 1959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.09564393939394e-05,
+      "loss": 0.5092,
+      "step": 1960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.095170454545455e-05,
+      "loss": 0.5702,
+      "step": 1961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.09469696969697e-05,
+      "loss": 0.5214,
+      "step": 1962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.094223484848485e-05,
+      "loss": 0.6196,
+      "step": 1963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.09375e-05,
+      "loss": 0.6776,
+      "step": 1964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.093276515151515e-05,
+      "loss": 0.6429,
+      "step": 1965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0928030303030306e-05,
+      "loss": 0.5591,
+      "step": 1966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.092329545454546e-05,
+      "loss": 0.5806,
+      "step": 1967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.091856060606061e-05,
+      "loss": 0.5016,
+      "step": 1968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.091382575757576e-05,
+      "loss": 0.6098,
+      "step": 1969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 0.5424,
+      "step": 1970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.090435606060606e-05,
+      "loss": 0.6134,
+      "step": 1971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0899621212121215e-05,
+      "loss": 0.5679,
+      "step": 1972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.089488636363636e-05,
+      "loss": 0.5557,
+      "step": 1973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0890151515151516e-05,
+      "loss": 0.5724,
+      "step": 1974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.088541666666667e-05,
+      "loss": 0.594,
+      "step": 1975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0880681818181824e-05,
+      "loss": 0.5314,
+      "step": 1976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.087594696969697e-05,
+      "loss": 0.6037,
+      "step": 1977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0871212121212125e-05,
+      "loss": 0.658,
+      "step": 1978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.086647727272728e-05,
+      "loss": 0.5442,
+      "step": 1979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0861742424242425e-05,
+      "loss": 0.5902,
+      "step": 1980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.085700757575758e-05,
+      "loss": 0.5787,
+      "step": 1981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0852272727272726e-05,
+      "loss": 0.6407,
+      "step": 1982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.084753787878788e-05,
+      "loss": 0.5525,
+      "step": 1983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0842803030303034e-05,
+      "loss": 0.5601,
+      "step": 1984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.083806818181819e-05,
+      "loss": 0.5459,
+      "step": 1985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0833333333333334e-05,
+      "loss": 0.5096,
+      "step": 1986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.082859848484849e-05,
+      "loss": 0.5072,
+      "step": 1987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.082386363636364e-05,
+      "loss": 0.5803,
+      "step": 1988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.081912878787879e-05,
+      "loss": 0.5569,
+      "step": 1989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.081439393939394e-05,
+      "loss": 0.5636,
+      "step": 1990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.080965909090909e-05,
+      "loss": 0.5854,
+      "step": 1991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.080492424242424e-05,
+      "loss": 0.5578,
+      "step": 1992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.08001893939394e-05,
+      "loss": 0.5318,
+      "step": 1993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.079545454545455e-05,
+      "loss": 0.5287,
+      "step": 1994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.07907196969697e-05,
+      "loss": 0.5081,
+      "step": 1995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.078598484848485e-05,
+      "loss": 0.5392,
+      "step": 1996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0781250000000005e-05,
+      "loss": 0.6157,
+      "step": 1997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.077651515151515e-05,
+      "loss": 0.5829,
+      "step": 1998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0771780303030306e-05,
+      "loss": 0.5131,
+      "step": 1999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.076704545454545e-05,
+      "loss": 0.599,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.076231060606061e-05,
+      "loss": 0.5121,
+      "step": 2001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.075757575757576e-05,
+      "loss": 0.5394,
+      "step": 2002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0752840909090914e-05,
+      "loss": 0.6057,
+      "step": 2003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.074810606060606e-05,
+      "loss": 0.6317,
+      "step": 2004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0743371212121215e-05,
+      "loss": 0.5231,
+      "step": 2005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.073863636363637e-05,
+      "loss": 0.5357,
+      "step": 2006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0733901515151516e-05,
+      "loss": 0.6883,
+      "step": 2007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.072916666666667e-05,
+      "loss": 0.6384,
+      "step": 2008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0724431818181817e-05,
+      "loss": 0.4806,
+      "step": 2009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.071969696969698e-05,
+      "loss": 0.5333,
+      "step": 2010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0714962121212124e-05,
+      "loss": 0.5827,
+      "step": 2011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.071022727272728e-05,
+      "loss": 0.6479,
+      "step": 2012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0705492424242425e-05,
+      "loss": 0.4916,
+      "step": 2013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.070075757575758e-05,
+      "loss": 0.4911,
+      "step": 2014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.069602272727273e-05,
+      "loss": 0.6121,
+      "step": 2015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.069128787878788e-05,
+      "loss": 0.6559,
+      "step": 2016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.068655303030303e-05,
+      "loss": 0.6091,
+      "step": 2017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.068181818181818e-05,
+      "loss": 0.6064,
+      "step": 2018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.067708333333334e-05,
+      "loss": 0.5074,
+      "step": 2019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.067234848484849e-05,
+      "loss": 0.5662,
+      "step": 2020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.066761363636364e-05,
+      "loss": 0.6186,
+      "step": 2021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.066287878787879e-05,
+      "loss": 0.5782,
+      "step": 2022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.065814393939394e-05,
+      "loss": 0.5291,
+      "step": 2023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0653409090909096e-05,
+      "loss": 0.5606,
+      "step": 2024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.064867424242424e-05,
+      "loss": 0.5291,
+      "step": 2025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.06439393939394e-05,
+      "loss": 0.5803,
+      "step": 2026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0639204545454544e-05,
+      "loss": 0.5498,
+      "step": 2027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0634469696969704e-05,
+      "loss": 0.5629,
+      "step": 2028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.062973484848485e-05,
+      "loss": 0.5049,
+      "step": 2029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 0.538,
+      "step": 2030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.062026515151515e-05,
+      "loss": 0.5416,
+      "step": 2031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0615530303030306e-05,
+      "loss": 0.6018,
+      "step": 2032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.061079545454546e-05,
+      "loss": 0.5578,
+      "step": 2033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0606060606060606e-05,
+      "loss": 0.6518,
+      "step": 2034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.060132575757576e-05,
+      "loss": 0.5722,
+      "step": 2035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.059659090909091e-05,
+      "loss": 0.627,
+      "step": 2036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.059185606060607e-05,
+      "loss": 0.6125,
+      "step": 2037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0587121212121215e-05,
+      "loss": 0.5308,
+      "step": 2038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.058238636363637e-05,
+      "loss": 0.4898,
+      "step": 2039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0577651515151515e-05,
+      "loss": 0.5638,
+      "step": 2040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.057291666666667e-05,
+      "loss": 0.5202,
+      "step": 2041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.056818181818182e-05,
+      "loss": 0.5256,
+      "step": 2042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.056344696969697e-05,
+      "loss": 0.6061,
+      "step": 2043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0558712121212124e-05,
+      "loss": 0.5656,
+      "step": 2044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.055397727272727e-05,
+      "loss": 0.5818,
+      "step": 2045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.054924242424243e-05,
+      "loss": 0.5296,
+      "step": 2046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.054450757575758e-05,
+      "loss": 0.6364,
+      "step": 2047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.053977272727273e-05,
+      "loss": 0.5162,
+      "step": 2048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.053503787878788e-05,
+      "loss": 0.5272,
+      "step": 2049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.053030303030303e-05,
+      "loss": 0.5444,
+      "step": 2050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0525568181818186e-05,
+      "loss": 0.5642,
+      "step": 2051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0520833333333333e-05,
+      "loss": 0.5408,
+      "step": 2052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.051609848484849e-05,
+      "loss": 0.5236,
+      "step": 2053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0511363636363634e-05,
+      "loss": 0.5735,
+      "step": 2054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0506628787878795e-05,
+      "loss": 0.5389,
+      "step": 2055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.050189393939394e-05,
+      "loss": 0.5611,
+      "step": 2056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0497159090909095e-05,
+      "loss": 0.6381,
+      "step": 2057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.049242424242424e-05,
+      "loss": 0.5487,
+      "step": 2058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0487689393939396e-05,
+      "loss": 0.5709,
+      "step": 2059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.048295454545455e-05,
+      "loss": 0.5406,
+      "step": 2060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.04782196969697e-05,
+      "loss": 0.4991,
+      "step": 2061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.047348484848485e-05,
+      "loss": 0.508,
+      "step": 2062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.046875e-05,
+      "loss": 0.514,
+      "step": 2063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.046401515151516e-05,
+      "loss": 0.598,
+      "step": 2064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0459280303030305e-05,
+      "loss": 0.5925,
+      "step": 2065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.045454545454546e-05,
+      "loss": 0.5801,
+      "step": 2066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0449810606060606e-05,
+      "loss": 0.5398,
+      "step": 2067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.044507575757576e-05,
+      "loss": 0.5983,
+      "step": 2068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0440340909090914e-05,
+      "loss": 0.5578,
+      "step": 2069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.043560606060606e-05,
+      "loss": 0.5013,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0430871212121214e-05,
+      "loss": 0.6377,
+      "step": 2071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.042613636363636e-05,
+      "loss": 0.5351,
+      "step": 2072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.042140151515152e-05,
+      "loss": 0.5055,
+      "step": 2073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.041666666666667e-05,
+      "loss": 0.5017,
+      "step": 2074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.041193181818182e-05,
+      "loss": 0.696,
+      "step": 2075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.040719696969697e-05,
+      "loss": 0.499,
+      "step": 2076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.040246212121212e-05,
+      "loss": 0.5267,
+      "step": 2077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.039772727272728e-05,
+      "loss": 0.5222,
+      "step": 2078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0392992424242424e-05,
+      "loss": 0.6195,
+      "step": 2079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.038825757575758e-05,
+      "loss": 0.6389,
+      "step": 2080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0383522727272725e-05,
+      "loss": 0.6037,
+      "step": 2081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0378787878787885e-05,
+      "loss": 0.48,
+      "step": 2082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.037405303030303e-05,
+      "loss": 0.6359,
+      "step": 2083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0369318181818186e-05,
+      "loss": 0.5734,
+      "step": 2084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.036458333333333e-05,
+      "loss": 0.6437,
+      "step": 2085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.035984848484849e-05,
+      "loss": 0.5003,
+      "step": 2086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.035511363636364e-05,
+      "loss": 0.5647,
+      "step": 2087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.035037878787879e-05,
+      "loss": 0.6492,
+      "step": 2088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.034564393939394e-05,
+      "loss": 0.4711,
+      "step": 2089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.034090909090909e-05,
+      "loss": 0.4858,
+      "step": 2090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.033617424242425e-05,
+      "loss": 0.5032,
+      "step": 2091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0331439393939396e-05,
+      "loss": 0.5458,
+      "step": 2092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.032670454545455e-05,
+      "loss": 0.6076,
+      "step": 2093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0321969696969697e-05,
+      "loss": 0.6278,
+      "step": 2094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.031723484848485e-05,
+      "loss": 0.5859,
+      "step": 2095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0312500000000004e-05,
+      "loss": 0.4844,
+      "step": 2096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.030776515151515e-05,
+      "loss": 0.5354,
+      "step": 2097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0303030303030305e-05,
+      "loss": 0.6032,
+      "step": 2098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.029829545454545e-05,
+      "loss": 0.5035,
+      "step": 2099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.029356060606061e-05,
+      "loss": 0.5642,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.028882575757576e-05,
+      "loss": 0.5565,
+      "step": 2101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.028409090909091e-05,
+      "loss": 0.4446,
+      "step": 2102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.027935606060606e-05,
+      "loss": 0.6973,
+      "step": 2103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0274621212121214e-05,
+      "loss": 0.5272,
+      "step": 2104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.026988636363637e-05,
+      "loss": 0.6028,
+      "step": 2105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0265151515151515e-05,
+      "loss": 0.4689,
+      "step": 2106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.026041666666667e-05,
+      "loss": 0.6314,
+      "step": 2107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0255681818181815e-05,
+      "loss": 0.4624,
+      "step": 2108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0250946969696976e-05,
+      "loss": 0.6454,
+      "step": 2109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.024621212121212e-05,
+      "loss": 0.5413,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.024147727272728e-05,
+      "loss": 0.5293,
+      "step": 2111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0236742424242424e-05,
+      "loss": 0.5057,
+      "step": 2112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.023200757575758e-05,
+      "loss": 0.5629,
+      "step": 2113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.022727272727273e-05,
+      "loss": 0.4767,
+      "step": 2114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.022253787878788e-05,
+      "loss": 0.6165,
+      "step": 2115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.021780303030303e-05,
+      "loss": 0.5087,
+      "step": 2116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0213068181818186e-05,
+      "loss": 0.6225,
+      "step": 2117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.020833333333334e-05,
+      "loss": 0.4462,
+      "step": 2118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0203598484848486e-05,
+      "loss": 0.5561,
+      "step": 2119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.019886363636364e-05,
+      "loss": 0.6745,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.019412878787879e-05,
+      "loss": 0.5654,
+      "step": 2121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.018939393939394e-05,
+      "loss": 0.5133,
+      "step": 2122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0184659090909095e-05,
+      "loss": 0.5727,
+      "step": 2123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.017992424242424e-05,
+      "loss": 0.6057,
+      "step": 2124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0175189393939395e-05,
+      "loss": 0.5337,
+      "step": 2125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.017045454545455e-05,
+      "loss": 0.5664,
+      "step": 2126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.01657196969697e-05,
+      "loss": 0.5855,
+      "step": 2127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.016098484848485e-05,
+      "loss": 0.5557,
+      "step": 2128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0156250000000004e-05,
+      "loss": 0.5036,
+      "step": 2129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.015151515151515e-05,
+      "loss": 0.6551,
+      "step": 2130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0146780303030304e-05,
+      "loss": 0.5806,
+      "step": 2131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.014204545454546e-05,
+      "loss": 0.47,
+      "step": 2132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0137310606060605e-05,
+      "loss": 0.4974,
+      "step": 2133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.013257575757576e-05,
+      "loss": 0.4604,
+      "step": 2134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.012784090909091e-05,
+      "loss": 0.5326,
+      "step": 2135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0123106060606066e-05,
+      "loss": 0.6257,
+      "step": 2136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0118371212121213e-05,
+      "loss": 0.5332,
+      "step": 2137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.011363636363637e-05,
+      "loss": 0.6137,
+      "step": 2138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0108901515151514e-05,
+      "loss": 0.5042,
+      "step": 2139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.010416666666667e-05,
+      "loss": 0.6794,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.009943181818182e-05,
+      "loss": 0.4901,
+      "step": 2141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.009469696969697e-05,
+      "loss": 0.5538,
+      "step": 2142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.008996212121212e-05,
+      "loss": 0.6845,
+      "step": 2143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0085227272727276e-05,
+      "loss": 0.6952,
+      "step": 2144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.008049242424243e-05,
+      "loss": 0.5626,
+      "step": 2145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.007575757575758e-05,
+      "loss": 0.5259,
+      "step": 2146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.007102272727273e-05,
+      "loss": 0.5245,
+      "step": 2147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.006628787878788e-05,
+      "loss": 0.5755,
+      "step": 2148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.006155303030303e-05,
+      "loss": 0.4718,
+      "step": 2149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0056818181818185e-05,
+      "loss": 0.5336,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.005208333333333e-05,
+      "loss": 0.549,
+      "step": 2151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0047348484848486e-05,
+      "loss": 0.549,
+      "step": 2152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.004261363636364e-05,
+      "loss": 0.5041,
+      "step": 2153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0037878787878794e-05,
+      "loss": 0.5307,
+      "step": 2154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.003314393939394e-05,
+      "loss": 0.4783,
+      "step": 2155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0028409090909094e-05,
+      "loss": 0.3918,
+      "step": 2156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.002367424242424e-05,
+      "loss": 0.6136,
+      "step": 2157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0018939393939395e-05,
+      "loss": 0.5161,
+      "step": 2158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.001420454545455e-05,
+      "loss": 0.6066,
+      "step": 2159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0009469696969696e-05,
+      "loss": 0.5719,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.000473484848485e-05,
+      "loss": 0.4926,
+      "step": 2161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4e-05,
+      "loss": 0.4784,
+      "step": 2162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.999526515151516e-05,
+      "loss": 0.4849,
+      "step": 2163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9990530303030304e-05,
+      "loss": 0.5387,
+      "step": 2164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.998579545454546e-05,
+      "loss": 0.4685,
+      "step": 2165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9981060606060605e-05,
+      "loss": 0.5313,
+      "step": 2166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.997632575757576e-05,
+      "loss": 0.659,
+      "step": 2167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.997159090909091e-05,
+      "loss": 0.7156,
+      "step": 2168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.996685606060606e-05,
+      "loss": 0.6544,
+      "step": 2169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.996212121212121e-05,
+      "loss": 0.6231,
+      "step": 2170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.995738636363637e-05,
+      "loss": 0.5518,
+      "step": 2171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.995265151515152e-05,
+      "loss": 0.5616,
+      "step": 2172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.994791666666667e-05,
+      "loss": 0.5533,
+      "step": 2173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.994318181818182e-05,
+      "loss": 0.5306,
+      "step": 2174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9938446969696975e-05,
+      "loss": 0.6052,
+      "step": 2175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.993371212121212e-05,
+      "loss": 0.533,
+      "step": 2176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9928977272727276e-05,
+      "loss": 0.6013,
+      "step": 2177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.992424242424242e-05,
+      "loss": 0.5443,
+      "step": 2178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9919507575757577e-05,
+      "loss": 0.5272,
+      "step": 2179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.991477272727273e-05,
+      "loss": 0.5753,
+      "step": 2180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9910037878787884e-05,
+      "loss": 0.5391,
+      "step": 2181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.990530303030303e-05,
+      "loss": 0.4849,
+      "step": 2182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9900568181818185e-05,
+      "loss": 0.5948,
+      "step": 2183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.989583333333334e-05,
+      "loss": 0.6097,
+      "step": 2184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9891098484848486e-05,
+      "loss": 0.615,
+      "step": 2185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.988636363636364e-05,
+      "loss": 0.628,
+      "step": 2186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9881628787878786e-05,
+      "loss": 0.5737,
+      "step": 2187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.987689393939394e-05,
+      "loss": 0.5522,
+      "step": 2188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9872159090909094e-05,
+      "loss": 0.5533,
+      "step": 2189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.986742424242425e-05,
+      "loss": 0.586,
+      "step": 2190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9862689393939395e-05,
+      "loss": 0.556,
+      "step": 2191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.985795454545455e-05,
+      "loss": 0.5832,
+      "step": 2192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.98532196969697e-05,
+      "loss": 0.5562,
+      "step": 2193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.984848484848485e-05,
+      "loss": 0.5566,
+      "step": 2194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.984375e-05,
+      "loss": 0.5676,
+      "step": 2195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.983901515151515e-05,
+      "loss": 0.6112,
+      "step": 2196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9834280303030304e-05,
+      "loss": 0.5666,
+      "step": 2197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.982954545454546e-05,
+      "loss": 0.5109,
+      "step": 2198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.982481060606061e-05,
+      "loss": 0.5148,
+      "step": 2199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.982007575757576e-05,
+      "loss": 0.5496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.981534090909091e-05,
+      "loss": 0.5101,
+      "step": 2201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9810606060606066e-05,
+      "loss": 0.582,
+      "step": 2202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.980587121212121e-05,
+      "loss": 0.6022,
+      "step": 2203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9801136363636366e-05,
+      "loss": 0.5876,
+      "step": 2204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.979640151515151e-05,
+      "loss": 0.5245,
+      "step": 2205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.979166666666667e-05,
+      "loss": 0.6432,
+      "step": 2206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.978693181818182e-05,
+      "loss": 0.5421,
+      "step": 2207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9782196969696975e-05,
+      "loss": 0.5733,
+      "step": 2208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.977746212121212e-05,
+      "loss": 0.5492,
+      "step": 2209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9772727272727275e-05,
+      "loss": 0.6253,
+      "step": 2210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.976799242424243e-05,
+      "loss": 0.5266,
+      "step": 2211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9763257575757576e-05,
+      "loss": 0.54,
+      "step": 2212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.975852272727273e-05,
+      "loss": 0.5248,
+      "step": 2213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.975378787878788e-05,
+      "loss": 0.5421,
+      "step": 2214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.974905303030303e-05,
+      "loss": 0.5294,
+      "step": 2215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9744318181818184e-05,
+      "loss": 0.5237,
+      "step": 2216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.973958333333334e-05,
+      "loss": 0.6232,
+      "step": 2217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9734848484848485e-05,
+      "loss": 0.4984,
+      "step": 2218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.973011363636364e-05,
+      "loss": 0.4545,
+      "step": 2219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.972537878787879e-05,
+      "loss": 0.4971,
+      "step": 2220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.972064393939394e-05,
+      "loss": 0.5247,
+      "step": 2221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9715909090909093e-05,
+      "loss": 0.4763,
+      "step": 2222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.971117424242424e-05,
+      "loss": 0.5107,
+      "step": 2223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.97064393939394e-05,
+      "loss": 0.6081,
+      "step": 2224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.970170454545455e-05,
+      "loss": 0.5178,
+      "step": 2225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.96969696969697e-05,
+      "loss": 0.512,
+      "step": 2226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.969223484848485e-05,
+      "loss": 0.4872,
+      "step": 2227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.96875e-05,
+      "loss": 0.6013,
+      "step": 2228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9682765151515156e-05,
+      "loss": 0.5867,
+      "step": 2229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.96780303030303e-05,
+      "loss": 0.5331,
+      "step": 2230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.967329545454546e-05,
+      "loss": 0.4142,
+      "step": 2231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9668560606060604e-05,
+      "loss": 0.6688,
+      "step": 2232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9663825757575764e-05,
+      "loss": 0.5145,
+      "step": 2233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.965909090909091e-05,
+      "loss": 0.4454,
+      "step": 2234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9654356060606065e-05,
+      "loss": 0.6271,
+      "step": 2235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.964962121212121e-05,
+      "loss": 0.5171,
+      "step": 2236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9644886363636366e-05,
+      "loss": 0.605,
+      "step": 2237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.964015151515152e-05,
+      "loss": 0.4903,
+      "step": 2238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.963541666666667e-05,
+      "loss": 0.6363,
+      "step": 2239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.963068181818182e-05,
+      "loss": 0.5746,
+      "step": 2240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.962594696969697e-05,
+      "loss": 0.6117,
+      "step": 2241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.962121212121213e-05,
+      "loss": 0.5561,
+      "step": 2242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9616477272727275e-05,
+      "loss": 0.5831,
+      "step": 2243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.961174242424243e-05,
+      "loss": 0.5497,
+      "step": 2244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9607007575757576e-05,
+      "loss": 0.569,
+      "step": 2245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.960227272727273e-05,
+      "loss": 0.5343,
+      "step": 2246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.959753787878788e-05,
+      "loss": 0.4558,
+      "step": 2247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.959280303030303e-05,
+      "loss": 0.5348,
+      "step": 2248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9588068181818184e-05,
+      "loss": 0.4686,
+      "step": 2249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.958333333333333e-05,
+      "loss": 0.5357,
+      "step": 2250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.957859848484849e-05,
+      "loss": 0.5845,
+      "step": 2251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.957386363636364e-05,
+      "loss": 0.5366,
+      "step": 2252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.956912878787879e-05,
+      "loss": 0.4455,
+      "step": 2253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.956439393939394e-05,
+      "loss": 0.5663,
+      "step": 2254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.955965909090909e-05,
+      "loss": 0.5203,
+      "step": 2255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.955492424242425e-05,
+      "loss": 0.5803,
+      "step": 2256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9550189393939394e-05,
+      "loss": 0.4672,
+      "step": 2257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.954545454545455e-05,
+      "loss": 0.5705,
+      "step": 2258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9540719696969695e-05,
+      "loss": 0.5635,
+      "step": 2259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9535984848484855e-05,
+      "loss": 0.3868,
+      "step": 2260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.953125e-05,
+      "loss": 0.6081,
+      "step": 2261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9526515151515156e-05,
+      "loss": 0.6016,
+      "step": 2262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.95217803030303e-05,
+      "loss": 0.4302,
+      "step": 2263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9517045454545457e-05,
+      "loss": 0.5663,
+      "step": 2264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.951231060606061e-05,
+      "loss": 0.5024,
+      "step": 2265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.950757575757576e-05,
+      "loss": 0.6207,
+      "step": 2266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.950284090909091e-05,
+      "loss": 0.4441,
+      "step": 2267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.949810606060606e-05,
+      "loss": 0.542,
+      "step": 2268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.949337121212122e-05,
+      "loss": 0.4711,
+      "step": 2269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9488636363636366e-05,
+      "loss": 0.5323,
+      "step": 2270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.948390151515152e-05,
+      "loss": 0.5563,
+      "step": 2271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9479166666666666e-05,
+      "loss": 0.4573,
+      "step": 2272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.947443181818182e-05,
+      "loss": 0.4346,
+      "step": 2273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9469696969696974e-05,
+      "loss": 0.6298,
+      "step": 2274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.946496212121212e-05,
+      "loss": 0.5915,
+      "step": 2275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9460227272727275e-05,
+      "loss": 0.5282,
+      "step": 2276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.945549242424242e-05,
+      "loss": 0.5585,
+      "step": 2277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.945075757575758e-05,
+      "loss": 0.4736,
+      "step": 2278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.944602272727273e-05,
+      "loss": 0.4414,
+      "step": 2279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.944128787878788e-05,
+      "loss": 0.5332,
+      "step": 2280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.943655303030303e-05,
+      "loss": 0.5722,
+      "step": 2281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9431818181818184e-05,
+      "loss": 0.4802,
+      "step": 2282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.942708333333334e-05,
+      "loss": 0.4357,
+      "step": 2283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9422348484848484e-05,
+      "loss": 0.591,
+      "step": 2284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.941761363636364e-05,
+      "loss": 0.554,
+      "step": 2285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9412878787878785e-05,
+      "loss": 0.562,
+      "step": 2286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9408143939393946e-05,
+      "loss": 0.5467,
+      "step": 2287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.940340909090909e-05,
+      "loss": 0.5537,
+      "step": 2288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9398674242424246e-05,
+      "loss": 0.6561,
+      "step": 2289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.939393939393939e-05,
+      "loss": 0.6219,
+      "step": 2290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.938920454545455e-05,
+      "loss": 0.4916,
+      "step": 2291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.93844696969697e-05,
+      "loss": 0.5731,
+      "step": 2292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.937973484848485e-05,
+      "loss": 0.5209,
+      "step": 2293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9375e-05,
+      "loss": 0.5084,
+      "step": 2294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.937026515151515e-05,
+      "loss": 0.591,
+      "step": 2295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.936553030303031e-05,
+      "loss": 0.541,
+      "step": 2296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9360795454545456e-05,
+      "loss": 0.4472,
+      "step": 2297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.935606060606061e-05,
+      "loss": 0.4782,
+      "step": 2298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.935132575757576e-05,
+      "loss": 0.5863,
+      "step": 2299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.934659090909091e-05,
+      "loss": 0.4668,
+      "step": 2300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9341856060606064e-05,
+      "loss": 0.6036,
+      "step": 2301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.933712121212121e-05,
+      "loss": 0.5265,
+      "step": 2302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9332386363636365e-05,
+      "loss": 0.695,
+      "step": 2303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.932765151515151e-05,
+      "loss": 0.5758,
+      "step": 2304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.932291666666667e-05,
+      "loss": 0.4659,
+      "step": 2305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.931818181818182e-05,
+      "loss": 0.5422,
+      "step": 2306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9313446969696973e-05,
+      "loss": 0.4998,
+      "step": 2307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.930871212121212e-05,
+      "loss": 0.4926,
+      "step": 2308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9303977272727274e-05,
+      "loss": 0.5777,
+      "step": 2309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.929924242424243e-05,
+      "loss": 0.557,
+      "step": 2310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9294507575757575e-05,
+      "loss": 0.5244,
+      "step": 2311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.928977272727273e-05,
+      "loss": 0.4971,
+      "step": 2312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9285037878787876e-05,
+      "loss": 0.4973,
+      "step": 2313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9280303030303036e-05,
+      "loss": 0.5369,
+      "step": 2314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.927556818181818e-05,
+      "loss": 0.5348,
+      "step": 2315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.927083333333334e-05,
+      "loss": 0.5512,
+      "step": 2316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9266098484848484e-05,
+      "loss": 0.5275,
+      "step": 2317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.926136363636364e-05,
+      "loss": 0.4216,
+      "step": 2318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.925662878787879e-05,
+      "loss": 0.4928,
+      "step": 2319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.925189393939394e-05,
+      "loss": 0.5889,
+      "step": 2320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.924715909090909e-05,
+      "loss": 0.5352,
+      "step": 2321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.924242424242424e-05,
+      "loss": 0.5127,
+      "step": 2322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.92376893939394e-05,
+      "loss": 0.4578,
+      "step": 2323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.923295454545455e-05,
+      "loss": 0.4497,
+      "step": 2324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.92282196969697e-05,
+      "loss": 0.4519,
+      "step": 2325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.922348484848485e-05,
+      "loss": 0.5528,
+      "step": 2326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.921875e-05,
+      "loss": 0.4882,
+      "step": 2327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9214015151515155e-05,
+      "loss": 0.4911,
+      "step": 2328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.92092803030303e-05,
+      "loss": 0.4948,
+      "step": 2329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9204545454545456e-05,
+      "loss": 0.5638,
+      "step": 2330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.91998106060606e-05,
+      "loss": 0.5748,
+      "step": 2331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.919507575757576e-05,
+      "loss": 0.5284,
+      "step": 2332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.919034090909091e-05,
+      "loss": 0.5455,
+      "step": 2333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9185606060606064e-05,
+      "loss": 0.5907,
+      "step": 2334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.918087121212121e-05,
+      "loss": 0.5066,
+      "step": 2335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9176136363636365e-05,
+      "loss": 0.5737,
+      "step": 2336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.917140151515152e-05,
+      "loss": 0.5747,
+      "step": 2337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9166666666666665e-05,
+      "loss": 0.5029,
+      "step": 2338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.916193181818182e-05,
+      "loss": 0.5187,
+      "step": 2339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.915719696969697e-05,
+      "loss": 0.59,
+      "step": 2340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.915246212121213e-05,
+      "loss": 0.6326,
+      "step": 2341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9147727272727274e-05,
+      "loss": 0.4666,
+      "step": 2342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.914299242424243e-05,
+      "loss": 0.5393,
+      "step": 2343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9138257575757575e-05,
+      "loss": 0.4844,
+      "step": 2344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.913352272727273e-05,
+      "loss": 0.4274,
+      "step": 2345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.912878787878788e-05,
+      "loss": 0.6222,
+      "step": 2346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.912405303030303e-05,
+      "loss": 0.4791,
+      "step": 2347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.911931818181818e-05,
+      "loss": 0.4775,
+      "step": 2348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9114583333333337e-05,
+      "loss": 0.5672,
+      "step": 2349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.910984848484849e-05,
+      "loss": 0.5489,
+      "step": 2350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.910511363636364e-05,
+      "loss": 0.4975,
+      "step": 2351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.910037878787879e-05,
+      "loss": 0.5084,
+      "step": 2352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.909564393939394e-05,
+      "loss": 0.5118,
+      "step": 2353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.909090909090909e-05,
+      "loss": 0.5833,
+      "step": 2354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9086174242424246e-05,
+      "loss": 0.4907,
+      "step": 2355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.908143939393939e-05,
+      "loss": 0.5399,
+      "step": 2356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9076704545454546e-05,
+      "loss": 0.595,
+      "step": 2357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.90719696969697e-05,
+      "loss": 0.5698,
+      "step": 2358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9067234848484854e-05,
+      "loss": 0.6112,
+      "step": 2359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.90625e-05,
+      "loss": 0.5634,
+      "step": 2360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9057765151515155e-05,
+      "loss": 0.5993,
+      "step": 2361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.90530303030303e-05,
+      "loss": 0.5903,
+      "step": 2362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9048295454545455e-05,
+      "loss": 0.4878,
+      "step": 2363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.904356060606061e-05,
+      "loss": 0.6492,
+      "step": 2364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9038825757575756e-05,
+      "loss": 0.5058,
+      "step": 2365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.903409090909091e-05,
+      "loss": 0.5176,
+      "step": 2366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9029356060606064e-05,
+      "loss": 0.5385,
+      "step": 2367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.902462121212122e-05,
+      "loss": 0.4071,
+      "step": 2368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9019886363636364e-05,
+      "loss": 0.5954,
+      "step": 2369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.901515151515152e-05,
+      "loss": 0.5727,
+      "step": 2370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9010416666666665e-05,
+      "loss": 0.6527,
+      "step": 2371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.900568181818182e-05,
+      "loss": 0.5197,
+      "step": 2372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.900094696969697e-05,
+      "loss": 0.5649,
+      "step": 2373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.899621212121212e-05,
+      "loss": 0.4832,
+      "step": 2374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.899147727272727e-05,
+      "loss": 0.5019,
+      "step": 2375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.898674242424243e-05,
+      "loss": 0.5101,
+      "step": 2376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.898200757575758e-05,
+      "loss": 0.5682,
+      "step": 2377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.897727272727273e-05,
+      "loss": 0.5469,
+      "step": 2378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.897253787878788e-05,
+      "loss": 0.4923,
+      "step": 2379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.896780303030303e-05,
+      "loss": 0.51,
+      "step": 2380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.896306818181818e-05,
+      "loss": 0.5658,
+      "step": 2381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8958333333333336e-05,
+      "loss": 0.6555,
+      "step": 2382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.895359848484848e-05,
+      "loss": 0.485,
+      "step": 2383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.894886363636364e-05,
+      "loss": 0.474,
+      "step": 2384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.894412878787879e-05,
+      "loss": 0.4767,
+      "step": 2385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8939393939393944e-05,
+      "loss": 0.6868,
+      "step": 2386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.893465909090909e-05,
+      "loss": 0.5219,
+      "step": 2387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8929924242424245e-05,
+      "loss": 0.6113,
+      "step": 2388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.89251893939394e-05,
+      "loss": 0.5726,
+      "step": 2389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8920454545454546e-05,
+      "loss": 0.6199,
+      "step": 2390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.89157196969697e-05,
+      "loss": 0.5559,
+      "step": 2391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.891098484848485e-05,
+      "loss": 0.6204,
+      "step": 2392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.890625e-05,
+      "loss": 0.4929,
+      "step": 2393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8901515151515154e-05,
+      "loss": 0.5762,
+      "step": 2394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.889678030303031e-05,
+      "loss": 0.5004,
+      "step": 2395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8892045454545455e-05,
+      "loss": 0.5509,
+      "step": 2396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.888731060606061e-05,
+      "loss": 0.5711,
+      "step": 2397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.888257575757576e-05,
+      "loss": 0.59,
+      "step": 2398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.887784090909091e-05,
+      "loss": 0.6026,
+      "step": 2399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.887310606060606e-05,
+      "loss": 0.5649,
+      "step": 2400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.886837121212121e-05,
+      "loss": 0.5037,
+      "step": 2401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8863636363636364e-05,
+      "loss": 0.5516,
+      "step": 2402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.885890151515152e-05,
+      "loss": 0.6487,
+      "step": 2403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.885416666666667e-05,
+      "loss": 0.5474,
+      "step": 2404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.884943181818182e-05,
+      "loss": 0.5687,
+      "step": 2405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.884469696969697e-05,
+      "loss": 0.5643,
+      "step": 2406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8839962121212126e-05,
+      "loss": 0.525,
+      "step": 2407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.883522727272727e-05,
+      "loss": 0.5585,
+      "step": 2408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.883049242424243e-05,
+      "loss": 0.4936,
+      "step": 2409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8825757575757574e-05,
+      "loss": 0.428,
+      "step": 2410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.882102272727273e-05,
+      "loss": 0.4862,
+      "step": 2411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.881628787878788e-05,
+      "loss": 0.5497,
+      "step": 2412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8811553030303035e-05,
+      "loss": 0.5383,
+      "step": 2413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.880681818181818e-05,
+      "loss": 0.5052,
+      "step": 2414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8802083333333336e-05,
+      "loss": 0.5631,
+      "step": 2415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.879734848484849e-05,
+      "loss": 0.615,
+      "step": 2416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8792613636363636e-05,
+      "loss": 0.3965,
+      "step": 2417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.878787878787879e-05,
+      "loss": 0.5624,
+      "step": 2418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.878314393939394e-05,
+      "loss": 0.483,
+      "step": 2419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.877840909090909e-05,
+      "loss": 0.5447,
+      "step": 2420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8773674242424245e-05,
+      "loss": 0.5868,
+      "step": 2421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.87689393939394e-05,
+      "loss": 0.4053,
+      "step": 2422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8764204545454545e-05,
+      "loss": 0.4924,
+      "step": 2423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.87594696969697e-05,
+      "loss": 0.4502,
+      "step": 2424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.875473484848485e-05,
+      "loss": 0.5101,
+      "step": 2425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.875e-05,
+      "loss": 0.5755,
+      "step": 2426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8745265151515154e-05,
+      "loss": 0.6383,
+      "step": 2427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.87405303030303e-05,
+      "loss": 0.5832,
+      "step": 2428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8735795454545455e-05,
+      "loss": 0.5611,
+      "step": 2429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.873106060606061e-05,
+      "loss": 0.556,
+      "step": 2430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.872632575757576e-05,
+      "loss": 0.5109,
+      "step": 2431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.872159090909091e-05,
+      "loss": 0.4426,
+      "step": 2432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.871685606060606e-05,
+      "loss": 0.6463,
+      "step": 2433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8712121212121217e-05,
+      "loss": 0.5186,
+      "step": 2434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8707386363636364e-05,
+      "loss": 0.5842,
+      "step": 2435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.870265151515152e-05,
+      "loss": 0.6011,
+      "step": 2436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8697916666666664e-05,
+      "loss": 0.5453,
+      "step": 2437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.869318181818182e-05,
+      "loss": 0.5475,
+      "step": 2438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.868844696969697e-05,
+      "loss": 0.5259,
+      "step": 2439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8683712121212126e-05,
+      "loss": 0.5509,
+      "step": 2440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.867897727272727e-05,
+      "loss": 0.541,
+      "step": 2441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8674242424242426e-05,
+      "loss": 0.5192,
+      "step": 2442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.866950757575758e-05,
+      "loss": 0.6057,
+      "step": 2443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.866477272727273e-05,
+      "loss": 0.5697,
+      "step": 2444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.866003787878788e-05,
+      "loss": 0.589,
+      "step": 2445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.865530303030303e-05,
+      "loss": 0.5179,
+      "step": 2446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.865056818181819e-05,
+      "loss": 0.5228,
+      "step": 2447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8645833333333335e-05,
+      "loss": 0.473,
+      "step": 2448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.864109848484849e-05,
+      "loss": 0.4969,
+      "step": 2449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 0.4803,
+      "step": 2450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.863162878787879e-05,
+      "loss": 0.5842,
+      "step": 2451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8626893939393944e-05,
+      "loss": 0.5702,
+      "step": 2452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.862215909090909e-05,
+      "loss": 0.5146,
+      "step": 2453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8617424242424244e-05,
+      "loss": 0.4626,
+      "step": 2454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.861268939393939e-05,
+      "loss": 0.4576,
+      "step": 2455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.860795454545455e-05,
+      "loss": 0.5821,
+      "step": 2456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.86032196969697e-05,
+      "loss": 0.6081,
+      "step": 2457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.859848484848485e-05,
+      "loss": 0.4675,
+      "step": 2458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.859375e-05,
+      "loss": 0.4634,
+      "step": 2459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.858901515151515e-05,
+      "loss": 0.5685,
+      "step": 2460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.858428030303031e-05,
+      "loss": 0.4445,
+      "step": 2461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8579545454545454e-05,
+      "loss": 0.458,
+      "step": 2462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.857481060606061e-05,
+      "loss": 0.5379,
+      "step": 2463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8570075757575755e-05,
+      "loss": 0.6604,
+      "step": 2464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8565340909090915e-05,
+      "loss": 0.4683,
+      "step": 2465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.856060606060606e-05,
+      "loss": 0.5585,
+      "step": 2466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8555871212121216e-05,
+      "loss": 0.4991,
+      "step": 2467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.855113636363636e-05,
+      "loss": 0.6198,
+      "step": 2468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.854640151515152e-05,
+      "loss": 0.5561,
+      "step": 2469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.854166666666667e-05,
+      "loss": 0.4996,
+      "step": 2470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.853693181818182e-05,
+      "loss": 0.5113,
+      "step": 2471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.853219696969697e-05,
+      "loss": 0.5648,
+      "step": 2472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.852746212121212e-05,
+      "loss": 0.6059,
+      "step": 2473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.852272727272728e-05,
+      "loss": 0.4665,
+      "step": 2474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8517992424242426e-05,
+      "loss": 0.5171,
+      "step": 2475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.851325757575758e-05,
+      "loss": 0.5301,
+      "step": 2476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.850852272727273e-05,
+      "loss": 0.678,
+      "step": 2477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.850378787878788e-05,
+      "loss": 0.4955,
+      "step": 2478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8499053030303034e-05,
+      "loss": 0.4432,
+      "step": 2479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.849431818181818e-05,
+      "loss": 0.5933,
+      "step": 2480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8489583333333335e-05,
+      "loss": 0.477,
+      "step": 2481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.848484848484848e-05,
+      "loss": 0.3797,
+      "step": 2482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.848011363636364e-05,
+      "loss": 0.626,
+      "step": 2483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.847537878787879e-05,
+      "loss": 0.5458,
+      "step": 2484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.847064393939394e-05,
+      "loss": 0.4282,
+      "step": 2485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.846590909090909e-05,
+      "loss": 0.5384,
+      "step": 2486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8461174242424244e-05,
+      "loss": 0.5897,
+      "step": 2487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.84564393939394e-05,
+      "loss": 0.4726,
+      "step": 2488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8451704545454545e-05,
+      "loss": 0.5334,
+      "step": 2489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.84469696969697e-05,
+      "loss": 0.6163,
+      "step": 2490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8442234848484845e-05,
+      "loss": 0.6266,
+      "step": 2491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8437500000000006e-05,
+      "loss": 0.6515,
+      "step": 2492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.843276515151515e-05,
+      "loss": 0.5828,
+      "step": 2493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.842803030303031e-05,
+      "loss": 0.474,
+      "step": 2494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8423295454545454e-05,
+      "loss": 0.4347,
+      "step": 2495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.841856060606061e-05,
+      "loss": 0.6628,
+      "step": 2496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.841382575757576e-05,
+      "loss": 0.4176,
+      "step": 2497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.840909090909091e-05,
+      "loss": 0.499,
+      "step": 2498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.840435606060606e-05,
+      "loss": 0.537,
+      "step": 2499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.839962121212121e-05,
+      "loss": 0.639,
+      "step": 2500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.839488636363637e-05,
+      "loss": 0.5039,
+      "step": 2501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8390151515151516e-05,
+      "loss": 0.4362,
+      "step": 2502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.838541666666667e-05,
+      "loss": 0.5279,
+      "step": 2503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.838068181818182e-05,
+      "loss": 0.5207,
+      "step": 2504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.837594696969697e-05,
+      "loss": 0.4486,
+      "step": 2505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8371212121212125e-05,
+      "loss": 0.6511,
+      "step": 2506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.836647727272727e-05,
+      "loss": 0.5016,
+      "step": 2507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8361742424242425e-05,
+      "loss": 0.4482,
+      "step": 2508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.835700757575757e-05,
+      "loss": 0.5626,
+      "step": 2509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.835227272727273e-05,
+      "loss": 0.4527,
+      "step": 2510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.834753787878788e-05,
+      "loss": 0.5155,
+      "step": 2511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8342803030303034e-05,
+      "loss": 0.6061,
+      "step": 2512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.833806818181818e-05,
+      "loss": 0.5136,
+      "step": 2513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8333333333333334e-05,
+      "loss": 0.5191,
+      "step": 2514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.832859848484849e-05,
+      "loss": 0.4925,
+      "step": 2515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8323863636363635e-05,
+      "loss": 0.528,
+      "step": 2516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.831912878787879e-05,
+      "loss": 0.7103,
+      "step": 2517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8314393939393936e-05,
+      "loss": 0.5619,
+      "step": 2518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8309659090909097e-05,
+      "loss": 0.4884,
+      "step": 2519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8304924242424244e-05,
+      "loss": 0.4801,
+      "step": 2520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.83001893939394e-05,
+      "loss": 0.4692,
+      "step": 2521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8295454545454544e-05,
+      "loss": 0.5052,
+      "step": 2522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.82907196969697e-05,
+      "loss": 0.4954,
+      "step": 2523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.828598484848485e-05,
+      "loss": 0.4851,
+      "step": 2524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.828125e-05,
+      "loss": 0.5737,
+      "step": 2525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.827651515151515e-05,
+      "loss": 0.5621,
+      "step": 2526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.82717803030303e-05,
+      "loss": 0.6202,
+      "step": 2527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.826704545454546e-05,
+      "loss": 0.5496,
+      "step": 2528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.826231060606061e-05,
+      "loss": 0.4726,
+      "step": 2529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.825757575757576e-05,
+      "loss": 0.5652,
+      "step": 2530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.825284090909091e-05,
+      "loss": 0.6046,
+      "step": 2531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.824810606060606e-05,
+      "loss": 0.5785,
+      "step": 2532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8243371212121215e-05,
+      "loss": 0.6214,
+      "step": 2533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.823863636363636e-05,
+      "loss": 0.648,
+      "step": 2534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8233901515151516e-05,
+      "loss": 0.5096,
+      "step": 2535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.822916666666666e-05,
+      "loss": 0.5492,
+      "step": 2536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8224431818181824e-05,
+      "loss": 0.5569,
+      "step": 2537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.821969696969697e-05,
+      "loss": 0.5689,
+      "step": 2538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8214962121212124e-05,
+      "loss": 0.521,
+      "step": 2539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.821022727272727e-05,
+      "loss": 0.615,
+      "step": 2540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8205492424242425e-05,
+      "loss": 0.6129,
+      "step": 2541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.820075757575758e-05,
+      "loss": 0.5934,
+      "step": 2542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8196022727272726e-05,
+      "loss": 0.4913,
+      "step": 2543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.819128787878788e-05,
+      "loss": 0.5152,
+      "step": 2544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8186553030303027e-05,
+      "loss": 0.5529,
+      "step": 2545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.818181818181819e-05,
+      "loss": 0.5066,
+      "step": 2546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8177083333333334e-05,
+      "loss": 0.474,
+      "step": 2547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.817234848484849e-05,
+      "loss": 0.5866,
+      "step": 2548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8167613636363635e-05,
+      "loss": 0.4994,
+      "step": 2549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.816287878787879e-05,
+      "loss": 0.4816,
+      "step": 2550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.815814393939394e-05,
+      "loss": 0.5231,
+      "step": 2551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.815340909090909e-05,
+      "loss": 0.5001,
+      "step": 2552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.814867424242424e-05,
+      "loss": 0.6136,
+      "step": 2553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.81439393939394e-05,
+      "loss": 0.6396,
+      "step": 2554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.813920454545455e-05,
+      "loss": 0.6568,
+      "step": 2555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.81344696969697e-05,
+      "loss": 0.6653,
+      "step": 2556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.812973484848485e-05,
+      "loss": 0.4358,
+      "step": 2557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8125e-05,
+      "loss": 0.5813,
+      "step": 2558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.812026515151515e-05,
+      "loss": 0.4732,
+      "step": 2559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8115530303030306e-05,
+      "loss": 0.7682,
+      "step": 2560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.811079545454545e-05,
+      "loss": 0.765,
+      "step": 2561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.810606060606061e-05,
+      "loss": 0.6199,
+      "step": 2562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.810132575757576e-05,
+      "loss": 0.5252,
+      "step": 2563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8096590909090914e-05,
+      "loss": 0.5682,
+      "step": 2564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.809185606060606e-05,
+      "loss": 0.518,
+      "step": 2565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8087121212121215e-05,
+      "loss": 0.6206,
+      "step": 2566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.808238636363636e-05,
+      "loss": 0.5386,
+      "step": 2567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8077651515151516e-05,
+      "loss": 0.4978,
+      "step": 2568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.807291666666667e-05,
+      "loss": 0.5066,
+      "step": 2569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8068181818181816e-05,
+      "loss": 0.5252,
+      "step": 2570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.806344696969697e-05,
+      "loss": 0.4877,
+      "step": 2571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8058712121212124e-05,
+      "loss": 0.5722,
+      "step": 2572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.805397727272728e-05,
+      "loss": 0.6259,
+      "step": 2573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8049242424242425e-05,
+      "loss": 0.4851,
+      "step": 2574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.804450757575758e-05,
+      "loss": 0.5581,
+      "step": 2575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8039772727272725e-05,
+      "loss": 0.5246,
+      "step": 2576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.803503787878788e-05,
+      "loss": 0.4836,
+      "step": 2577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.803030303030303e-05,
+      "loss": 0.5619,
+      "step": 2578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.802556818181818e-05,
+      "loss": 0.5338,
+      "step": 2579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8020833333333334e-05,
+      "loss": 0.5227,
+      "step": 2580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.801609848484849e-05,
+      "loss": 0.5766,
+      "step": 2581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.801136363636364e-05,
+      "loss": 0.5516,
+      "step": 2582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.800662878787879e-05,
+      "loss": 0.464,
+      "step": 2583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.800189393939394e-05,
+      "loss": 0.519,
+      "step": 2584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.799715909090909e-05,
+      "loss": 0.5904,
+      "step": 2585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.799242424242424e-05,
+      "loss": 0.4225,
+      "step": 2586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7987689393939396e-05,
+      "loss": 0.5361,
+      "step": 2587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7982954545454543e-05,
+      "loss": 0.462,
+      "step": 2588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.79782196969697e-05,
+      "loss": 0.4105,
+      "step": 2589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.797348484848485e-05,
+      "loss": 0.5294,
+      "step": 2590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7968750000000005e-05,
+      "loss": 0.6691,
+      "step": 2591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.796401515151515e-05,
+      "loss": 0.5422,
+      "step": 2592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7959280303030305e-05,
+      "loss": 0.4969,
+      "step": 2593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.795454545454545e-05,
+      "loss": 0.5638,
+      "step": 2594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7949810606060606e-05,
+      "loss": 0.4208,
+      "step": 2595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.794507575757576e-05,
+      "loss": 0.4878,
+      "step": 2596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.794034090909091e-05,
+      "loss": 0.4663,
+      "step": 2597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.793560606060606e-05,
+      "loss": 0.4117,
+      "step": 2598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7930871212121214e-05,
+      "loss": 0.4561,
+      "step": 2599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.792613636363637e-05,
+      "loss": 0.5207,
+      "step": 2600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7921401515151515e-05,
+      "loss": 0.5823,
+      "step": 2601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.791666666666667e-05,
+      "loss": 0.5813,
+      "step": 2602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7911931818181816e-05,
+      "loss": 0.4872,
+      "step": 2603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.790719696969697e-05,
+      "loss": 0.6201,
+      "step": 2604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7902462121212124e-05,
+      "loss": 0.5627,
+      "step": 2605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.789772727272727e-05,
+      "loss": 0.6707,
+      "step": 2606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7892992424242424e-05,
+      "loss": 0.5035,
+      "step": 2607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.788825757575758e-05,
+      "loss": 0.6009,
+      "step": 2608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.788352272727273e-05,
+      "loss": 0.5978,
+      "step": 2609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.787878787878788e-05,
+      "loss": 0.5301,
+      "step": 2610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.787405303030303e-05,
+      "loss": 0.5596,
+      "step": 2611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7869318181818186e-05,
+      "loss": 0.4823,
+      "step": 2612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.786458333333333e-05,
+      "loss": 0.5263,
+      "step": 2613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.785984848484849e-05,
+      "loss": 0.5361,
+      "step": 2614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7855113636363634e-05,
+      "loss": 0.5627,
+      "step": 2615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.785037878787879e-05,
+      "loss": 0.5708,
+      "step": 2616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.784564393939394e-05,
+      "loss": 0.4236,
+      "step": 2617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7840909090909095e-05,
+      "loss": 0.4924,
+      "step": 2618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.783617424242424e-05,
+      "loss": 0.4725,
+      "step": 2619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7831439393939396e-05,
+      "loss": 0.5467,
+      "step": 2620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.782670454545455e-05,
+      "loss": 0.464,
+      "step": 2621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.78219696969697e-05,
+      "loss": 0.5685,
+      "step": 2622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.781723484848485e-05,
+      "loss": 0.5156,
+      "step": 2623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.78125e-05,
+      "loss": 0.6091,
+      "step": 2624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.780776515151515e-05,
+      "loss": 0.4736,
+      "step": 2625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7803030303030305e-05,
+      "loss": 0.5126,
+      "step": 2626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.779829545454546e-05,
+      "loss": 0.535,
+      "step": 2627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7793560606060606e-05,
+      "loss": 0.6887,
+      "step": 2628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.778882575757576e-05,
+      "loss": 0.6345,
+      "step": 2629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.778409090909091e-05,
+      "loss": 0.4308,
+      "step": 2630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.777935606060606e-05,
+      "loss": 0.5456,
+      "step": 2631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7774621212121214e-05,
+      "loss": 0.4891,
+      "step": 2632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.776988636363636e-05,
+      "loss": 0.5827,
+      "step": 2633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7765151515151515e-05,
+      "loss": 0.5368,
+      "step": 2634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.776041666666667e-05,
+      "loss": 0.5406,
+      "step": 2635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.775568181818182e-05,
+      "loss": 0.5764,
+      "step": 2636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.775094696969697e-05,
+      "loss": 0.5866,
+      "step": 2637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.774621212121212e-05,
+      "loss": 0.4973,
+      "step": 2638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.774147727272728e-05,
+      "loss": 0.5462,
+      "step": 2639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.773674242424243e-05,
+      "loss": 0.5414,
+      "step": 2640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.773200757575758e-05,
+      "loss": 0.5209,
+      "step": 2641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7727272727272725e-05,
+      "loss": 0.5622,
+      "step": 2642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.772253787878788e-05,
+      "loss": 0.5289,
+      "step": 2643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.771780303030303e-05,
+      "loss": 0.5678,
+      "step": 2644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7713068181818186e-05,
+      "loss": 0.5475,
+      "step": 2645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.770833333333333e-05,
+      "loss": 0.5796,
+      "step": 2646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.770359848484849e-05,
+      "loss": 0.5458,
+      "step": 2647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.769886363636364e-05,
+      "loss": 0.5389,
+      "step": 2648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7694128787878794e-05,
+      "loss": 0.5465,
+      "step": 2649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.768939393939394e-05,
+      "loss": 0.5587,
+      "step": 2650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.768465909090909e-05,
+      "loss": 0.5637,
+      "step": 2651
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.767992424242424e-05,
+      "loss": 0.5753,
+      "step": 2652
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7675189393939396e-05,
+      "loss": 0.5789,
+      "step": 2653
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.767045454545455e-05,
+      "loss": 0.583,
+      "step": 2654
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7665719696969696e-05,
+      "loss": 0.4978,
+      "step": 2655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.766098484848485e-05,
+      "loss": 0.5176,
+      "step": 2656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7656250000000004e-05,
+      "loss": 0.5342,
+      "step": 2657
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.765151515151516e-05,
+      "loss": 0.4713,
+      "step": 2658
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7646780303030305e-05,
+      "loss": 0.5045,
+      "step": 2659
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.764204545454545e-05,
+      "loss": 0.6296,
+      "step": 2660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7637310606060605e-05,
+      "loss": 0.4991,
+      "step": 2661
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.763257575757576e-05,
+      "loss": 0.5279,
+      "step": 2662
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.762784090909091e-05,
+      "loss": 0.4665,
+      "step": 2663
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.762310606060606e-05,
+      "loss": 0.5377,
+      "step": 2664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7618371212121214e-05,
+      "loss": 0.5603,
+      "step": 2665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.761363636363637e-05,
+      "loss": 0.5081,
+      "step": 2666
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.760890151515152e-05,
+      "loss": 0.5869,
+      "step": 2667
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.760416666666667e-05,
+      "loss": 0.4758,
+      "step": 2668
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7599431818181815e-05,
+      "loss": 0.7498,
+      "step": 2669
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7594696969696976e-05,
+      "loss": 0.511,
+      "step": 2670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.758996212121212e-05,
+      "loss": 0.5335,
+      "step": 2671
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7585227272727276e-05,
+      "loss": 0.5356,
+      "step": 2672
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7580492424242423e-05,
+      "loss": 0.5342,
+      "step": 2673
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.757575757575758e-05,
+      "loss": 0.4983,
+      "step": 2674
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.757102272727273e-05,
+      "loss": 0.5914,
+      "step": 2675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7566287878787885e-05,
+      "loss": 0.5905,
+      "step": 2676
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.756155303030303e-05,
+      "loss": 0.5068,
+      "step": 2677
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.755681818181818e-05,
+      "loss": 0.5406,
+      "step": 2678
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.755208333333334e-05,
+      "loss": 0.6351,
+      "step": 2679
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7547348484848486e-05,
+      "loss": 0.6412,
+      "step": 2680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.754261363636364e-05,
+      "loss": 0.5301,
+      "step": 2681
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.753787878787879e-05,
+      "loss": 0.5,
+      "step": 2682
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.753314393939394e-05,
+      "loss": 0.5953,
+      "step": 2683
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7528409090909094e-05,
+      "loss": 0.532,
+      "step": 2684
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.752367424242425e-05,
+      "loss": 0.4022,
+      "step": 2685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7518939393939395e-05,
+      "loss": 0.5634,
+      "step": 2686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.751420454545454e-05,
+      "loss": 0.6332,
+      "step": 2687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.75094696969697e-05,
+      "loss": 0.5798,
+      "step": 2688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.750473484848485e-05,
+      "loss": 0.5082,
+      "step": 2689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.5473,
+      "step": 2690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.749526515151515e-05,
+      "loss": 0.45,
+      "step": 2691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7490530303030304e-05,
+      "loss": 0.4614,
+      "step": 2692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.748579545454546e-05,
+      "loss": 0.5971,
+      "step": 2693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.748106060606061e-05,
+      "loss": 0.5672,
+      "step": 2694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.747632575757576e-05,
+      "loss": 0.537,
+      "step": 2695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7471590909090906e-05,
+      "loss": 0.4809,
+      "step": 2696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7466856060606066e-05,
+      "loss": 0.4192,
+      "step": 2697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.746212121212121e-05,
+      "loss": 0.5702,
+      "step": 2698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.745738636363637e-05,
+      "loss": 0.4544,
+      "step": 2699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7452651515151514e-05,
+      "loss": 0.4826,
+      "step": 2700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.744791666666667e-05,
+      "loss": 0.4754,
+      "step": 2701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.744318181818182e-05,
+      "loss": 0.5911,
+      "step": 2702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7438446969696975e-05,
+      "loss": 0.3857,
+      "step": 2703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.743371212121212e-05,
+      "loss": 0.5624,
+      "step": 2704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.742897727272727e-05,
+      "loss": 0.5656,
+      "step": 2705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.742424242424243e-05,
+      "loss": 0.4929,
+      "step": 2706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.741950757575758e-05,
+      "loss": 0.4694,
+      "step": 2707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.741477272727273e-05,
+      "loss": 0.4584,
+      "step": 2708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.741003787878788e-05,
+      "loss": 0.5521,
+      "step": 2709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.740530303030303e-05,
+      "loss": 0.4496,
+      "step": 2710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7400568181818185e-05,
+      "loss": 0.4465,
+      "step": 2711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.739583333333334e-05,
+      "loss": 0.5005,
+      "step": 2712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7391098484848486e-05,
+      "loss": 0.4368,
+      "step": 2713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.738636363636363e-05,
+      "loss": 0.4503,
+      "step": 2714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.738162878787879e-05,
+      "loss": 0.6163,
+      "step": 2715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.737689393939394e-05,
+      "loss": 0.6418,
+      "step": 2716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7372159090909094e-05,
+      "loss": 0.6662,
+      "step": 2717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.736742424242424e-05,
+      "loss": 0.5543,
+      "step": 2718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.73626893939394e-05,
+      "loss": 0.5319,
+      "step": 2719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.735795454545455e-05,
+      "loss": 0.4488,
+      "step": 2720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.73532196969697e-05,
+      "loss": 0.5334,
+      "step": 2721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.734848484848485e-05,
+      "loss": 0.4881,
+      "step": 2722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7343749999999996e-05,
+      "loss": 0.5869,
+      "step": 2723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.733901515151516e-05,
+      "loss": 0.553,
+      "step": 2724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7334280303030304e-05,
+      "loss": 0.433,
+      "step": 2725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.732954545454546e-05,
+      "loss": 0.5721,
+      "step": 2726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7324810606060605e-05,
+      "loss": 0.5272,
+      "step": 2727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7320075757575765e-05,
+      "loss": 0.5889,
+      "step": 2728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.731534090909091e-05,
+      "loss": 0.4878,
+      "step": 2729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7310606060606066e-05,
+      "loss": 0.5003,
+      "step": 2730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.730587121212121e-05,
+      "loss": 0.6053,
+      "step": 2731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.730113636363636e-05,
+      "loss": 0.5951,
+      "step": 2732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.729640151515152e-05,
+      "loss": 0.5251,
+      "step": 2733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.729166666666667e-05,
+      "loss": 0.5378,
+      "step": 2734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.728693181818182e-05,
+      "loss": 0.5019,
+      "step": 2735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.728219696969697e-05,
+      "loss": 0.4457,
+      "step": 2736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.727746212121213e-05,
+      "loss": 0.5232,
+      "step": 2737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7272727272727276e-05,
+      "loss": 0.6027,
+      "step": 2738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.726799242424243e-05,
+      "loss": 0.5996,
+      "step": 2739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7263257575757576e-05,
+      "loss": 0.5581,
+      "step": 2740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.725852272727272e-05,
+      "loss": 0.4791,
+      "step": 2741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7253787878787884e-05,
+      "loss": 0.4554,
+      "step": 2742
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.724905303030303e-05,
+      "loss": 0.5538,
+      "step": 2743
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7244318181818185e-05,
+      "loss": 0.5267,
+      "step": 2744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.723958333333333e-05,
+      "loss": 0.5624,
+      "step": 2745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.723484848484849e-05,
+      "loss": 0.5324,
+      "step": 2746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.723011363636364e-05,
+      "loss": 0.5,
+      "step": 2747
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.722537878787879e-05,
+      "loss": 0.6508,
+      "step": 2748
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.722064393939394e-05,
+      "loss": 0.4415,
+      "step": 2749
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.721590909090909e-05,
+      "loss": 0.5141,
+      "step": 2750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.721117424242425e-05,
+      "loss": 0.5807,
+      "step": 2751
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7206439393939394e-05,
+      "loss": 0.528,
+      "step": 2752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.720170454545455e-05,
+      "loss": 0.543,
+      "step": 2753
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7196969696969695e-05,
+      "loss": 0.496,
+      "step": 2754
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7192234848484856e-05,
+      "loss": 0.4753,
+      "step": 2755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.71875e-05,
+      "loss": 0.4823,
+      "step": 2756
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7182765151515156e-05,
+      "loss": 0.4753,
+      "step": 2757
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7178030303030303e-05,
+      "loss": 0.4737,
+      "step": 2758
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.717329545454545e-05,
+      "loss": 0.5396,
+      "step": 2759
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.716856060606061e-05,
+      "loss": 0.5727,
+      "step": 2760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.716382575757576e-05,
+      "loss": 0.5873,
+      "step": 2761
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.715909090909091e-05,
+      "loss": 0.5516,
+      "step": 2762
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.715435606060606e-05,
+      "loss": 0.6588,
+      "step": 2763
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.714962121212122e-05,
+      "loss": 0.5176,
+      "step": 2764
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7144886363636366e-05,
+      "loss": 0.4954,
+      "step": 2765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.714015151515152e-05,
+      "loss": 0.5579,
+      "step": 2766
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.713541666666667e-05,
+      "loss": 0.4901,
+      "step": 2767
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7130681818181814e-05,
+      "loss": 0.4779,
+      "step": 2768
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7125946969696974e-05,
+      "loss": 0.5452,
+      "step": 2769
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.712121212121212e-05,
+      "loss": 0.5566,
+      "step": 2770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7116477272727275e-05,
+      "loss": 0.4898,
+      "step": 2771
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.711174242424242e-05,
+      "loss": 0.5363,
+      "step": 2772
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.710700757575758e-05,
+      "loss": 0.5368,
+      "step": 2773
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.710227272727273e-05,
+      "loss": 0.5952,
+      "step": 2774
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7097537878787883e-05,
+      "loss": 0.4933,
+      "step": 2775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.709280303030303e-05,
+      "loss": 0.6423,
+      "step": 2776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7088068181818184e-05,
+      "loss": 0.5451,
+      "step": 2777
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.708333333333334e-05,
+      "loss": 0.5247,
+      "step": 2778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7078598484848485e-05,
+      "loss": 0.4458,
+      "step": 2779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.707386363636364e-05,
+      "loss": 0.4714,
+      "step": 2780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7069128787878786e-05,
+      "loss": 0.4781,
+      "step": 2781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7064393939393946e-05,
+      "loss": 0.5175,
+      "step": 2782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.705965909090909e-05,
+      "loss": 0.5753,
+      "step": 2783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.705492424242425e-05,
+      "loss": 0.5422,
+      "step": 2784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7050189393939394e-05,
+      "loss": 0.5255,
+      "step": 2785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.704545454545455e-05,
+      "loss": 0.5224,
+      "step": 2786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.70407196969697e-05,
+      "loss": 0.5831,
+      "step": 2787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.703598484848485e-05,
+      "loss": 0.4915,
+      "step": 2788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.703125e-05,
+      "loss": 0.4228,
+      "step": 2789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.702651515151515e-05,
+      "loss": 0.489,
+      "step": 2790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.702178030303031e-05,
+      "loss": 0.6123,
+      "step": 2791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.701704545454546e-05,
+      "loss": 0.4762,
+      "step": 2792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.701231060606061e-05,
+      "loss": 0.5951,
+      "step": 2793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.700757575757576e-05,
+      "loss": 0.4958,
+      "step": 2794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.700284090909091e-05,
+      "loss": 0.5298,
+      "step": 2795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6998106060606065e-05,
+      "loss": 0.43,
+      "step": 2796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.699337121212121e-05,
+      "loss": 0.4981,
+      "step": 2797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6988636363636366e-05,
+      "loss": 0.5523,
+      "step": 2798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.698390151515151e-05,
+      "loss": 0.4732,
+      "step": 2799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.697916666666667e-05,
+      "loss": 0.496,
+      "step": 2800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.697443181818182e-05,
+      "loss": 0.4519,
+      "step": 2801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6969696969696974e-05,
+      "loss": 0.4818,
+      "step": 2802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.696496212121212e-05,
+      "loss": 0.5713,
+      "step": 2803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6960227272727275e-05,
+      "loss": 0.5675,
+      "step": 2804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.695549242424243e-05,
+      "loss": 0.627,
+      "step": 2805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6950757575757576e-05,
+      "loss": 0.3809,
+      "step": 2806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.694602272727273e-05,
+      "loss": 0.5177,
+      "step": 2807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6941287878787876e-05,
+      "loss": 0.5323,
+      "step": 2808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.693655303030304e-05,
+      "loss": 0.4553,
+      "step": 2809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6931818181818184e-05,
+      "loss": 0.599,
+      "step": 2810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.692708333333334e-05,
+      "loss": 0.5925,
+      "step": 2811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6922348484848485e-05,
+      "loss": 0.5618,
+      "step": 2812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.691761363636364e-05,
+      "loss": 0.494,
+      "step": 2813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.691287878787879e-05,
+      "loss": 0.4854,
+      "step": 2814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.690814393939394e-05,
+      "loss": 0.6813,
+      "step": 2815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.690340909090909e-05,
+      "loss": 0.527,
+      "step": 2816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.689867424242424e-05,
+      "loss": 0.5785,
+      "step": 2817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.68939393939394e-05,
+      "loss": 0.5802,
+      "step": 2818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.688920454545455e-05,
+      "loss": 0.6141,
+      "step": 2819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.68844696969697e-05,
+      "loss": 0.5312,
+      "step": 2820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.687973484848485e-05,
+      "loss": 0.5041,
+      "step": 2821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6875e-05,
+      "loss": 0.4576,
+      "step": 2822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6870265151515156e-05,
+      "loss": 0.5343,
+      "step": 2823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.68655303030303e-05,
+      "loss": 0.5071,
+      "step": 2824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6860795454545456e-05,
+      "loss": 0.5045,
+      "step": 2825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.68560606060606e-05,
+      "loss": 0.5841,
+      "step": 2826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6851325757575764e-05,
+      "loss": 0.5418,
+      "step": 2827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.684659090909091e-05,
+      "loss": 0.617,
+      "step": 2828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6841856060606065e-05,
+      "loss": 0.514,
+      "step": 2829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.683712121212121e-05,
+      "loss": 0.5209,
+      "step": 2830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6832386363636365e-05,
+      "loss": 0.5945,
+      "step": 2831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.682765151515152e-05,
+      "loss": 0.4347,
+      "step": 2832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6822916666666666e-05,
+      "loss": 0.5568,
+      "step": 2833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.681818181818182e-05,
+      "loss": 0.5183,
+      "step": 2834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6813446969696974e-05,
+      "loss": 0.5817,
+      "step": 2835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.680871212121213e-05,
+      "loss": 0.5239,
+      "step": 2836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6803977272727274e-05,
+      "loss": 0.484,
+      "step": 2837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.679924242424243e-05,
+      "loss": 0.515,
+      "step": 2838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6794507575757575e-05,
+      "loss": 0.5108,
+      "step": 2839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.678977272727273e-05,
+      "loss": 0.5144,
+      "step": 2840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.678503787878788e-05,
+      "loss": 0.5071,
+      "step": 2841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.678030303030303e-05,
+      "loss": 0.4623,
+      "step": 2842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6775568181818183e-05,
+      "loss": 0.573,
+      "step": 2843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.677083333333334e-05,
+      "loss": 0.4961,
+      "step": 2844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.676609848484849e-05,
+      "loss": 0.561,
+      "step": 2845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.676136363636364e-05,
+      "loss": 0.5106,
+      "step": 2846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.675662878787879e-05,
+      "loss": 0.5709,
+      "step": 2847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.675189393939394e-05,
+      "loss": 0.4054,
+      "step": 2848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.674715909090909e-05,
+      "loss": 0.4638,
+      "step": 2849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6742424242424246e-05,
+      "loss": 0.4551,
+      "step": 2850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.673768939393939e-05,
+      "loss": 0.5415,
+      "step": 2851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.673295454545455e-05,
+      "loss": 0.4316,
+      "step": 2852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.67282196969697e-05,
+      "loss": 0.5742,
+      "step": 2853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6723484848484854e-05,
+      "loss": 0.3545,
+      "step": 2854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.671875e-05,
+      "loss": 0.4717,
+      "step": 2855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6714015151515155e-05,
+      "loss": 0.441,
+      "step": 2856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.67092803030303e-05,
+      "loss": 0.5163,
+      "step": 2857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6704545454545456e-05,
+      "loss": 0.5808,
+      "step": 2858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.669981060606061e-05,
+      "loss": 0.4387,
+      "step": 2859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.669507575757576e-05,
+      "loss": 0.5514,
+      "step": 2860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.669034090909091e-05,
+      "loss": 0.5672,
+      "step": 2861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6685606060606064e-05,
+      "loss": 0.5492,
+      "step": 2862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.668087121212122e-05,
+      "loss": 0.4769,
+      "step": 2863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6676136363636365e-05,
+      "loss": 0.5636,
+      "step": 2864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.667140151515152e-05,
+      "loss": 0.5126,
+      "step": 2865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.5244,
+      "step": 2866
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.666193181818182e-05,
+      "loss": 0.4826,
+      "step": 2867
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.665719696969697e-05,
+      "loss": 0.4284,
+      "step": 2868
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.665246212121212e-05,
+      "loss": 0.5261,
+      "step": 2869
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6647727272727274e-05,
+      "loss": 0.5865,
+      "step": 2870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.664299242424243e-05,
+      "loss": 0.5178,
+      "step": 2871
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.663825757575758e-05,
+      "loss": 0.5459,
+      "step": 2872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.663352272727273e-05,
+      "loss": 0.5844,
+      "step": 2873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.662878787878788e-05,
+      "loss": 0.5447,
+      "step": 2874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.662405303030303e-05,
+      "loss": 0.4485,
+      "step": 2875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.661931818181818e-05,
+      "loss": 0.4333,
+      "step": 2876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.661458333333334e-05,
+      "loss": 0.4685,
+      "step": 2877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6609848484848484e-05,
+      "loss": 0.6082,
+      "step": 2878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.660511363636364e-05,
+      "loss": 0.587,
+      "step": 2879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.660037878787879e-05,
+      "loss": 0.5871,
+      "step": 2880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6595643939393945e-05,
+      "loss": 0.5465,
+      "step": 2881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.659090909090909e-05,
+      "loss": 0.5006,
+      "step": 2882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6586174242424246e-05,
+      "loss": 0.6463,
+      "step": 2883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.65814393939394e-05,
+      "loss": 0.5261,
+      "step": 2884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6576704545454547e-05,
+      "loss": 0.5139,
+      "step": 2885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.65719696969697e-05,
+      "loss": 0.5389,
+      "step": 2886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.656723484848485e-05,
+      "loss": 0.4591,
+      "step": 2887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.65625e-05,
+      "loss": 0.5548,
+      "step": 2888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6557765151515155e-05,
+      "loss": 0.541,
+      "step": 2889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.655303030303031e-05,
+      "loss": 0.5718,
+      "step": 2890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6548295454545456e-05,
+      "loss": 0.4777,
+      "step": 2891
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.654356060606061e-05,
+      "loss": 0.5421,
+      "step": 2892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.653882575757576e-05,
+      "loss": 0.4525,
+      "step": 2893
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.653409090909091e-05,
+      "loss": 0.6432,
+      "step": 2894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6529356060606064e-05,
+      "loss": 0.4657,
+      "step": 2895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.652462121212121e-05,
+      "loss": 0.4235,
+      "step": 2896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6519886363636365e-05,
+      "loss": 0.4363,
+      "step": 2897
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.651515151515152e-05,
+      "loss": 0.4753,
+      "step": 2898
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.651041666666667e-05,
+      "loss": 0.5675,
+      "step": 2899
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.650568181818182e-05,
+      "loss": 0.5063,
+      "step": 2900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.650094696969697e-05,
+      "loss": 0.5357,
+      "step": 2901
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6496212121212127e-05,
+      "loss": 0.4206,
+      "step": 2902
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6491477272727274e-05,
+      "loss": 0.4407,
+      "step": 2903
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.648674242424243e-05,
+      "loss": 0.4872,
+      "step": 2904
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6482007575757574e-05,
+      "loss": 0.6134,
+      "step": 2905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.647727272727273e-05,
+      "loss": 0.4153,
+      "step": 2906
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.647253787878788e-05,
+      "loss": 0.4615,
+      "step": 2907
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6467803030303036e-05,
+      "loss": 0.5087,
+      "step": 2908
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.646306818181818e-05,
+      "loss": 0.6878,
+      "step": 2909
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6458333333333336e-05,
+      "loss": 0.4917,
+      "step": 2910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.645359848484849e-05,
+      "loss": 0.5596,
+      "step": 2911
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.644886363636364e-05,
+      "loss": 0.6049,
+      "step": 2912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.644412878787879e-05,
+      "loss": 0.6596,
+      "step": 2913
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.643939393939394e-05,
+      "loss": 0.5716,
+      "step": 2914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.643465909090909e-05,
+      "loss": 0.4706,
+      "step": 2915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6429924242424245e-05,
+      "loss": 0.4773,
+      "step": 2916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.64251893939394e-05,
+      "loss": 0.5438,
+      "step": 2917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6420454545454546e-05,
+      "loss": 0.4387,
+      "step": 2918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.64157196969697e-05,
+      "loss": 0.5874,
+      "step": 2919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6410984848484854e-05,
+      "loss": 0.5523,
+      "step": 2920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.640625e-05,
+      "loss": 0.628,
+      "step": 2921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6401515151515154e-05,
+      "loss": 0.5617,
+      "step": 2922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.63967803030303e-05,
+      "loss": 0.5774,
+      "step": 2923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6392045454545455e-05,
+      "loss": 0.4689,
+      "step": 2924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.638731060606061e-05,
+      "loss": 0.519,
+      "step": 2925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.638257575757576e-05,
+      "loss": 0.4832,
+      "step": 2926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.637784090909091e-05,
+      "loss": 0.5601,
+      "step": 2927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6373106060606063e-05,
+      "loss": 0.5622,
+      "step": 2928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.636837121212122e-05,
+      "loss": 0.5142,
+      "step": 2929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.5465,
+      "step": 2930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.635890151515152e-05,
+      "loss": 0.4928,
+      "step": 2931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6354166666666665e-05,
+      "loss": 0.5007,
+      "step": 2932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.634943181818182e-05,
+      "loss": 0.5263,
+      "step": 2933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.634469696969697e-05,
+      "loss": 0.5036,
+      "step": 2934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6339962121212126e-05,
+      "loss": 0.5395,
+      "step": 2935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.633522727272727e-05,
+      "loss": 0.5271,
+      "step": 2936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.633049242424243e-05,
+      "loss": 0.5097,
+      "step": 2937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.632575757575758e-05,
+      "loss": 0.4404,
+      "step": 2938
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.632102272727273e-05,
+      "loss": 0.4729,
+      "step": 2939
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.631628787878788e-05,
+      "loss": 0.4614,
+      "step": 2940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.631155303030303e-05,
+      "loss": 0.5492,
+      "step": 2941
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.630681818181819e-05,
+      "loss": 0.527,
+      "step": 2942
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6302083333333336e-05,
+      "loss": 0.559,
+      "step": 2943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.629734848484849e-05,
+      "loss": 0.5195,
+      "step": 2944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.629261363636364e-05,
+      "loss": 0.5545,
+      "step": 2945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.628787878787879e-05,
+      "loss": 0.5893,
+      "step": 2946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6283143939393944e-05,
+      "loss": 0.4235,
+      "step": 2947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.627840909090909e-05,
+      "loss": 0.6324,
+      "step": 2948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6273674242424245e-05,
+      "loss": 0.5598,
+      "step": 2949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.626893939393939e-05,
+      "loss": 0.4957,
+      "step": 2950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.626420454545455e-05,
+      "loss": 0.5771,
+      "step": 2951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.62594696969697e-05,
+      "loss": 0.5551,
+      "step": 2952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.625473484848485e-05,
+      "loss": 0.5029,
+      "step": 2953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.625e-05,
+      "loss": 0.5715,
+      "step": 2954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6245265151515154e-05,
+      "loss": 0.5366,
+      "step": 2955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.624053030303031e-05,
+      "loss": 0.5014,
+      "step": 2956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6235795454545455e-05,
+      "loss": 0.6,
+      "step": 2957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.623106060606061e-05,
+      "loss": 0.5417,
+      "step": 2958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6226325757575755e-05,
+      "loss": 0.4308,
+      "step": 2959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6221590909090916e-05,
+      "loss": 0.4655,
+      "step": 2960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.621685606060606e-05,
+      "loss": 0.6563,
+      "step": 2961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.621212121212122e-05,
+      "loss": 0.504,
+      "step": 2962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6207386363636364e-05,
+      "loss": 0.5059,
+      "step": 2963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.620265151515152e-05,
+      "loss": 0.4714,
+      "step": 2964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.619791666666667e-05,
+      "loss": 0.5085,
+      "step": 2965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.619318181818182e-05,
+      "loss": 0.6635,
+      "step": 2966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.618844696969697e-05,
+      "loss": 0.5265,
+      "step": 2967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.618371212121212e-05,
+      "loss": 0.5304,
+      "step": 2968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.617897727272728e-05,
+      "loss": 0.5498,
+      "step": 2969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6174242424242427e-05,
+      "loss": 0.573,
+      "step": 2970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.616950757575758e-05,
+      "loss": 0.5033,
+      "step": 2971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.616477272727273e-05,
+      "loss": 0.5792,
+      "step": 2972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.616003787878788e-05,
+      "loss": 0.5713,
+      "step": 2973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6155303030303035e-05,
+      "loss": 0.6102,
+      "step": 2974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.615056818181818e-05,
+      "loss": 0.5332,
+      "step": 2975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6145833333333336e-05,
+      "loss": 0.4726,
+      "step": 2976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.614109848484848e-05,
+      "loss": 0.3835,
+      "step": 2977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.613636363636364e-05,
+      "loss": 0.6377,
+      "step": 2978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.613162878787879e-05,
+      "loss": 0.549,
+      "step": 2979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6126893939393944e-05,
+      "loss": 0.5941,
+      "step": 2980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.612215909090909e-05,
+      "loss": 0.4604,
+      "step": 2981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6117424242424245e-05,
+      "loss": 0.4707,
+      "step": 2982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.61126893939394e-05,
+      "loss": 0.5318,
+      "step": 2983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6107954545454545e-05,
+      "loss": 0.4543,
+      "step": 2984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.61032196969697e-05,
+      "loss": 0.4758,
+      "step": 2985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6098484848484846e-05,
+      "loss": 0.6117,
+      "step": 2986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6093750000000007e-05,
+      "loss": 0.6701,
+      "step": 2987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6089015151515154e-05,
+      "loss": 0.4995,
+      "step": 2988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.608428030303031e-05,
+      "loss": 0.6077,
+      "step": 2989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6079545454545454e-05,
+      "loss": 0.4402,
+      "step": 2990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.607481060606061e-05,
+      "loss": 0.6705,
+      "step": 2991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.607007575757576e-05,
+      "loss": 0.5896,
+      "step": 2992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.606534090909091e-05,
+      "loss": 0.7837,
+      "step": 2993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.606060606060606e-05,
+      "loss": 0.5233,
+      "step": 2994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.605587121212121e-05,
+      "loss": 0.4956,
+      "step": 2995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.605113636363637e-05,
+      "loss": 0.4545,
+      "step": 2996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.604640151515152e-05,
+      "loss": 0.6273,
+      "step": 2997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.604166666666667e-05,
+      "loss": 0.5359,
+      "step": 2998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.603693181818182e-05,
+      "loss": 0.5501,
+      "step": 2999
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.603219696969697e-05,
+      "loss": 0.5541,
+      "step": 3000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6027462121212125e-05,
+      "loss": 0.4658,
+      "step": 3001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.602272727272727e-05,
+      "loss": 0.5291,
+      "step": 3002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6017992424242426e-05,
+      "loss": 0.4613,
+      "step": 3003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.601325757575757e-05,
+      "loss": 0.5504,
+      "step": 3004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6008522727272734e-05,
+      "loss": 0.5387,
+      "step": 3005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.600378787878788e-05,
+      "loss": 0.5497,
+      "step": 3006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5999053030303034e-05,
+      "loss": 0.5124,
+      "step": 3007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.599431818181818e-05,
+      "loss": 0.4972,
+      "step": 3008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5989583333333335e-05,
+      "loss": 0.5826,
+      "step": 3009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.598484848484849e-05,
+      "loss": 0.5197,
+      "step": 3010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5980113636363636e-05,
+      "loss": 0.4196,
+      "step": 3011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.597537878787879e-05,
+      "loss": 0.5014,
+      "step": 3012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.597064393939394e-05,
+      "loss": 0.5272,
+      "step": 3013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59659090909091e-05,
+      "loss": 0.5386,
+      "step": 3014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5961174242424244e-05,
+      "loss": 0.5144,
+      "step": 3015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59564393939394e-05,
+      "loss": 0.5085,
+      "step": 3016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5951704545454545e-05,
+      "loss": 0.5164,
+      "step": 3017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59469696969697e-05,
+      "loss": 0.5732,
+      "step": 3018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.594223484848485e-05,
+      "loss": 0.5201,
+      "step": 3019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59375e-05,
+      "loss": 0.408,
+      "step": 3020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.593276515151515e-05,
+      "loss": 0.592,
+      "step": 3021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59280303030303e-05,
+      "loss": 0.4475,
+      "step": 3022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.592329545454546e-05,
+      "loss": 0.6202,
+      "step": 3023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.591856060606061e-05,
+      "loss": 0.5634,
+      "step": 3024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.591382575757576e-05,
+      "loss": 0.5678,
+      "step": 3025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.590909090909091e-05,
+      "loss": 0.5059,
+      "step": 3026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.590435606060606e-05,
+      "loss": 0.4985,
+      "step": 3027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5899621212121216e-05,
+      "loss": 0.5329,
+      "step": 3028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.589488636363636e-05,
+      "loss": 0.5411,
+      "step": 3029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.589015151515152e-05,
+      "loss": 0.608,
+      "step": 3030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5885416666666664e-05,
+      "loss": 0.4199,
+      "step": 3031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5880681818181824e-05,
+      "loss": 0.4798,
+      "step": 3032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.587594696969697e-05,
+      "loss": 0.444,
+      "step": 3033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5871212121212125e-05,
+      "loss": 0.4914,
+      "step": 3034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.586647727272727e-05,
+      "loss": 0.5141,
+      "step": 3035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5861742424242426e-05,
+      "loss": 0.4603,
+      "step": 3036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.585700757575758e-05,
+      "loss": 0.4507,
+      "step": 3037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5852272727272726e-05,
+      "loss": 0.5903,
+      "step": 3038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.584753787878788e-05,
+      "loss": 0.5583,
+      "step": 3039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.584280303030303e-05,
+      "loss": 0.5511,
+      "step": 3040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.583806818181819e-05,
+      "loss": 0.445,
+      "step": 3041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5833333333333335e-05,
+      "loss": 0.4764,
+      "step": 3042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.582859848484849e-05,
+      "loss": 0.4962,
+      "step": 3043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5823863636363635e-05,
+      "loss": 0.5115,
+      "step": 3044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.581912878787879e-05,
+      "loss": 0.504,
+      "step": 3045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.581439393939394e-05,
+      "loss": 0.5631,
+      "step": 3046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.580965909090909e-05,
+      "loss": 0.3914,
+      "step": 3047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5804924242424244e-05,
+      "loss": 0.4954,
+      "step": 3048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.58001893939394e-05,
+      "loss": 0.3779,
+      "step": 3049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.579545454545455e-05,
+      "loss": 0.5149,
+      "step": 3050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.57907196969697e-05,
+      "loss": 0.5481,
+      "step": 3051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.578598484848485e-05,
+      "loss": 0.545,
+      "step": 3052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.578125e-05,
+      "loss": 0.7082,
+      "step": 3053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.577651515151515e-05,
+      "loss": 0.4886,
+      "step": 3054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5771780303030307e-05,
+      "loss": 0.5047,
+      "step": 3055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5767045454545453e-05,
+      "loss": 0.4391,
+      "step": 3056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.576231060606061e-05,
+      "loss": 0.5354,
+      "step": 3057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.575757575757576e-05,
+      "loss": 0.5207,
+      "step": 3058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5752840909090915e-05,
+      "loss": 0.5624,
+      "step": 3059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.574810606060606e-05,
+      "loss": 0.4264,
+      "step": 3060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5743371212121216e-05,
+      "loss": 0.6698,
+      "step": 3061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.573863636363636e-05,
+      "loss": 0.6111,
+      "step": 3062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5733901515151516e-05,
+      "loss": 0.589,
+      "step": 3063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.572916666666667e-05,
+      "loss": 0.4896,
+      "step": 3064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.572443181818182e-05,
+      "loss": 0.4643,
+      "step": 3065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.571969696969697e-05,
+      "loss": 0.4829,
+      "step": 3066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5714962121212125e-05,
+      "loss": 0.4911,
+      "step": 3067
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.571022727272728e-05,
+      "loss": 0.5524,
+      "step": 3068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5705492424242425e-05,
+      "loss": 0.4699,
+      "step": 3069
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.570075757575758e-05,
+      "loss": 0.5736,
+      "step": 3070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5696022727272726e-05,
+      "loss": 0.5758,
+      "step": 3071
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.569128787878788e-05,
+      "loss": 0.6085,
+      "step": 3072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5686553030303034e-05,
+      "loss": 0.4728,
+      "step": 3073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.568181818181818e-05,
+      "loss": 0.4843,
+      "step": 3074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5677083333333334e-05,
+      "loss": 0.506,
+      "step": 3075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.567234848484849e-05,
+      "loss": 0.3471,
+      "step": 3076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.566761363636364e-05,
+      "loss": 0.5435,
+      "step": 3077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.566287878787879e-05,
+      "loss": 0.5546,
+      "step": 3078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.565814393939394e-05,
+      "loss": 0.4447,
+      "step": 3079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.565340909090909e-05,
+      "loss": 0.4919,
+      "step": 3080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.564867424242424e-05,
+      "loss": 0.4916,
+      "step": 3081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.56439393939394e-05,
+      "loss": 0.5524,
+      "step": 3082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5639204545454544e-05,
+      "loss": 0.6273,
+      "step": 3083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.56344696969697e-05,
+      "loss": 0.4616,
+      "step": 3084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.562973484848485e-05,
+      "loss": 0.5351,
+      "step": 3085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5625000000000005e-05,
+      "loss": 0.5854,
+      "step": 3086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.562026515151515e-05,
+      "loss": 0.4851,
+      "step": 3087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5615530303030306e-05,
+      "loss": 0.5609,
+      "step": 3088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.561079545454545e-05,
+      "loss": 0.5751,
+      "step": 3089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.560606060606061e-05,
+      "loss": 0.5001,
+      "step": 3090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.560132575757576e-05,
+      "loss": 0.5183,
+      "step": 3091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.559659090909091e-05,
+      "loss": 0.4556,
+      "step": 3092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.559185606060606e-05,
+      "loss": 0.5158,
+      "step": 3093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5587121212121215e-05,
+      "loss": 0.5657,
+      "step": 3094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.558238636363637e-05,
+      "loss": 0.4798,
+      "step": 3095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5577651515151516e-05,
+      "loss": 0.5105,
+      "step": 3096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.557291666666667e-05,
+      "loss": 0.4366,
+      "step": 3097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5568181818181817e-05,
+      "loss": 0.53,
+      "step": 3098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.556344696969697e-05,
+      "loss": 0.5122,
+      "step": 3099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5558712121212124e-05,
+      "loss": 0.4879,
+      "step": 3100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.555397727272727e-05,
+      "loss": 0.6156,
+      "step": 3101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5549242424242425e-05,
+      "loss": 0.5484,
+      "step": 3102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.554450757575758e-05,
+      "loss": 0.6063,
+      "step": 3103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.553977272727273e-05,
+      "loss": 0.5898,
+      "step": 3104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.553503787878788e-05,
+      "loss": 0.4443,
+      "step": 3105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.553030303030303e-05,
+      "loss": 0.5137,
+      "step": 3106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.552556818181819e-05,
+      "loss": 0.491,
+      "step": 3107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5520833333333334e-05,
+      "loss": 0.6093,
+      "step": 3108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.551609848484849e-05,
+      "loss": 0.4431,
+      "step": 3109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5511363636363635e-05,
+      "loss": 0.4484,
+      "step": 3110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.550662878787879e-05,
+      "loss": 0.5502,
+      "step": 3111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.550189393939394e-05,
+      "loss": 0.484,
+      "step": 3112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5497159090909096e-05,
+      "loss": 0.4248,
+      "step": 3113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.549242424242424e-05,
+      "loss": 0.6288,
+      "step": 3114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.54876893939394e-05,
+      "loss": 0.4191,
+      "step": 3115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.548295454545455e-05,
+      "loss": 0.4688,
+      "step": 3116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.54782196969697e-05,
+      "loss": 0.5339,
+      "step": 3117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.547348484848485e-05,
+      "loss": 0.5044,
+      "step": 3118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.546875e-05,
+      "loss": 0.3867,
+      "step": 3119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.546401515151515e-05,
+      "loss": 0.4516,
+      "step": 3120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5459280303030306e-05,
+      "loss": 0.421,
+      "step": 3121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.545454545454546e-05,
+      "loss": 0.5089,
+      "step": 3122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5449810606060606e-05,
+      "loss": 0.568,
+      "step": 3123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.544507575757576e-05,
+      "loss": 0.6369,
+      "step": 3124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5440340909090914e-05,
+      "loss": 0.5424,
+      "step": 3125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.543560606060606e-05,
+      "loss": 0.5316,
+      "step": 3126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5430871212121215e-05,
+      "loss": 0.4915,
+      "step": 3127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.542613636363636e-05,
+      "loss": 0.4405,
+      "step": 3128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5421401515151515e-05,
+      "loss": 0.5542,
+      "step": 3129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.541666666666667e-05,
+      "loss": 0.3933,
+      "step": 3130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.541193181818182e-05,
+      "loss": 0.4593,
+      "step": 3131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.540719696969697e-05,
+      "loss": 0.5755,
+      "step": 3132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5402462121212124e-05,
+      "loss": 0.5688,
+      "step": 3133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.539772727272728e-05,
+      "loss": 0.5924,
+      "step": 3134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5392992424242424e-05,
+      "loss": 0.4485,
+      "step": 3135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.538825757575758e-05,
+      "loss": 0.5235,
+      "step": 3136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5383522727272725e-05,
+      "loss": 0.5059,
+      "step": 3137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.537878787878788e-05,
+      "loss": 0.5961,
+      "step": 3138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.537405303030303e-05,
+      "loss": 0.4888,
+      "step": 3139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5369318181818186e-05,
+      "loss": 0.6072,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5364583333333333e-05,
+      "loss": 0.4981,
+      "step": 3141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.535984848484849e-05,
+      "loss": 0.5874,
+      "step": 3142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.535511363636364e-05,
+      "loss": 0.6323,
+      "step": 3143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.535037878787879e-05,
+      "loss": 0.4801,
+      "step": 3144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.534564393939394e-05,
+      "loss": 0.4878,
+      "step": 3145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.534090909090909e-05,
+      "loss": 0.4768,
+      "step": 3146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.533617424242424e-05,
+      "loss": 0.5645,
+      "step": 3147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5331439393939396e-05,
+      "loss": 0.6692,
+      "step": 3148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.532670454545455e-05,
+      "loss": 0.432,
+      "step": 3149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.53219696969697e-05,
+      "loss": 0.5813,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.531723484848485e-05,
+      "loss": 0.4378,
+      "step": 3151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5312500000000005e-05,
+      "loss": 0.4936,
+      "step": 3152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.530776515151515e-05,
+      "loss": 0.4214,
+      "step": 3153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5303030303030305e-05,
+      "loss": 0.5348,
+      "step": 3154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.529829545454545e-05,
+      "loss": 0.5531,
+      "step": 3155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5293560606060606e-05,
+      "loss": 0.5391,
+      "step": 3156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.528882575757576e-05,
+      "loss": 0.4537,
+      "step": 3157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5284090909090914e-05,
+      "loss": 0.4365,
+      "step": 3158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.527935606060606e-05,
+      "loss": 0.53,
+      "step": 3159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5274621212121214e-05,
+      "loss": 0.6103,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.526988636363637e-05,
+      "loss": 0.5047,
+      "step": 3161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5265151515151515e-05,
+      "loss": 0.531,
+      "step": 3162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.526041666666667e-05,
+      "loss": 0.5069,
+      "step": 3163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5255681818181816e-05,
+      "loss": 0.5097,
+      "step": 3164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5250946969696976e-05,
+      "loss": 0.5554,
+      "step": 3165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.524621212121212e-05,
+      "loss": 0.4481,
+      "step": 3166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.524147727272728e-05,
+      "loss": 0.4832,
+      "step": 3167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5236742424242424e-05,
+      "loss": 0.5849,
+      "step": 3168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.523200757575758e-05,
+      "loss": 0.5812,
+      "step": 3169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.522727272727273e-05,
+      "loss": 0.4892,
+      "step": 3170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.522253787878788e-05,
+      "loss": 0.602,
+      "step": 3171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.521780303030303e-05,
+      "loss": 0.5457,
+      "step": 3172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.521306818181818e-05,
+      "loss": 0.4437,
+      "step": 3173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.520833333333334e-05,
+      "loss": 0.5029,
+      "step": 3174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.520359848484849e-05,
+      "loss": 0.5037,
+      "step": 3175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.519886363636364e-05,
+      "loss": 0.4334,
+      "step": 3176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.519412878787879e-05,
+      "loss": 0.5931,
+      "step": 3177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.518939393939394e-05,
+      "loss": 0.4348,
+      "step": 3178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5184659090909095e-05,
+      "loss": 0.5081,
+      "step": 3179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.517992424242424e-05,
+      "loss": 0.4806,
+      "step": 3180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5175189393939396e-05,
+      "loss": 0.54,
+      "step": 3181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.517045454545454e-05,
+      "loss": 0.5019,
+      "step": 3182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.51657196969697e-05,
+      "loss": 0.4606,
+      "step": 3183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.516098484848485e-05,
+      "loss": 0.6563,
+      "step": 3184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5156250000000004e-05,
+      "loss": 0.4221,
+      "step": 3185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.515151515151515e-05,
+      "loss": 0.5076,
+      "step": 3186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5146780303030305e-05,
+      "loss": 0.3439,
+      "step": 3187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.514204545454546e-05,
+      "loss": 0.5909,
+      "step": 3188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5137310606060606e-05,
+      "loss": 0.4435,
+      "step": 3189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.513257575757576e-05,
+      "loss": 0.4478,
+      "step": 3190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5127840909090906e-05,
+      "loss": 0.4699,
+      "step": 3191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.512310606060607e-05,
+      "loss": 0.467,
+      "step": 3192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5118371212121214e-05,
+      "loss": 0.4917,
+      "step": 3193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.511363636363637e-05,
+      "loss": 0.4049,
+      "step": 3194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5108901515151515e-05,
+      "loss": 0.5209,
+      "step": 3195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.510416666666667e-05,
+      "loss": 0.4518,
+      "step": 3196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.509943181818182e-05,
+      "loss": 0.5022,
+      "step": 3197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.509469696969697e-05,
+      "loss": 0.6867,
+      "step": 3198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.508996212121212e-05,
+      "loss": 0.6444,
+      "step": 3199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.508522727272727e-05,
+      "loss": 0.7055,
+      "step": 3200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.508049242424243e-05,
+      "loss": 0.4621,
+      "step": 3201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.507575757575758e-05,
+      "loss": 0.5921,
+      "step": 3202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.507102272727273e-05,
+      "loss": 0.5077,
+      "step": 3203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.506628787878788e-05,
+      "loss": 0.5771,
+      "step": 3204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.506155303030303e-05,
+      "loss": 0.5569,
+      "step": 3205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5056818181818186e-05,
+      "loss": 0.3983,
+      "step": 3206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.505208333333333e-05,
+      "loss": 0.4788,
+      "step": 3207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5047348484848486e-05,
+      "loss": 0.4998,
+      "step": 3208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5042613636363633e-05,
+      "loss": 0.6135,
+      "step": 3209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5037878787878794e-05,
+      "loss": 0.4972,
+      "step": 3210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.503314393939394e-05,
+      "loss": 0.5152,
+      "step": 3211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5028409090909095e-05,
+      "loss": 0.4801,
+      "step": 3212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.502367424242424e-05,
+      "loss": 0.4763,
+      "step": 3213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5018939393939395e-05,
+      "loss": 0.4567,
+      "step": 3214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.501420454545455e-05,
+      "loss": 0.5145,
+      "step": 3215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5009469696969696e-05,
+      "loss": 0.5358,
+      "step": 3216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.500473484848485e-05,
+      "loss": 0.632,
+      "step": 3217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5e-05,
+      "loss": 0.4024,
+      "step": 3218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.499526515151516e-05,
+      "loss": 0.462,
+      "step": 3219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4990530303030304e-05,
+      "loss": 0.5277,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.498579545454546e-05,
+      "loss": 0.5168,
+      "step": 3221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4981060606060605e-05,
+      "loss": 0.4634,
+      "step": 3222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.497632575757576e-05,
+      "loss": 0.4904,
+      "step": 3223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.497159090909091e-05,
+      "loss": 0.522,
+      "step": 3224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.496685606060606e-05,
+      "loss": 0.5161,
+      "step": 3225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4962121212121213e-05,
+      "loss": 0.4795,
+      "step": 3226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.495738636363636e-05,
+      "loss": 0.5225,
+      "step": 3227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.495265151515152e-05,
+      "loss": 0.4352,
+      "step": 3228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.494791666666667e-05,
+      "loss": 0.5535,
+      "step": 3229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.494318181818182e-05,
+      "loss": 0.6545,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.493844696969697e-05,
+      "loss": 0.601,
+      "step": 3231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.493371212121212e-05,
+      "loss": 0.5581,
+      "step": 3232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4928977272727276e-05,
+      "loss": 0.6027,
+      "step": 3233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.492424242424242e-05,
+      "loss": 0.6047,
+      "step": 3234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.491950757575758e-05,
+      "loss": 0.5682,
+      "step": 3235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4914772727272724e-05,
+      "loss": 0.5807,
+      "step": 3236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4910037878787885e-05,
+      "loss": 0.5483,
+      "step": 3237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.490530303030303e-05,
+      "loss": 0.5373,
+      "step": 3238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4900568181818185e-05,
+      "loss": 0.5424,
+      "step": 3239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.489583333333333e-05,
+      "loss": 0.3743,
+      "step": 3240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4891098484848486e-05,
+      "loss": 0.4805,
+      "step": 3241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.488636363636364e-05,
+      "loss": 0.4769,
+      "step": 3242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.488162878787879e-05,
+      "loss": 0.5337,
+      "step": 3243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.487689393939394e-05,
+      "loss": 0.4676,
+      "step": 3244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.487215909090909e-05,
+      "loss": 0.5555,
+      "step": 3245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.486742424242425e-05,
+      "loss": 0.519,
+      "step": 3246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4862689393939395e-05,
+      "loss": 0.5432,
+      "step": 3247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.485795454545455e-05,
+      "loss": 0.4096,
+      "step": 3248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4853219696969696e-05,
+      "loss": 0.4935,
+      "step": 3249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.484848484848485e-05,
+      "loss": 0.5148,
+      "step": 3250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.484375e-05,
+      "loss": 0.5036,
+      "step": 3251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.483901515151515e-05,
+      "loss": 0.4657,
+      "step": 3252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4834280303030304e-05,
+      "loss": 0.5654,
+      "step": 3253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.482954545454545e-05,
+      "loss": 0.473,
+      "step": 3254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.482481060606061e-05,
+      "loss": 0.5899,
+      "step": 3255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.482007575757576e-05,
+      "loss": 0.6249,
+      "step": 3256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.481534090909091e-05,
+      "loss": 0.5165,
+      "step": 3257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.481060606060606e-05,
+      "loss": 0.4691,
+      "step": 3258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.480587121212121e-05,
+      "loss": 0.6363,
+      "step": 3259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.480113636363637e-05,
+      "loss": 0.4322,
+      "step": 3260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4796401515151514e-05,
+      "loss": 0.426,
+      "step": 3261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.479166666666667e-05,
+      "loss": 0.5075,
+      "step": 3262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4786931818181815e-05,
+      "loss": 0.602,
+      "step": 3263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4782196969696975e-05,
+      "loss": 0.5046,
+      "step": 3264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.477746212121212e-05,
+      "loss": 0.491,
+      "step": 3265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4772727272727276e-05,
+      "loss": 0.4769,
+      "step": 3266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.476799242424242e-05,
+      "loss": 0.5289,
+      "step": 3267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4763257575757577e-05,
+      "loss": 0.3979,
+      "step": 3268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.475852272727273e-05,
+      "loss": 0.4871,
+      "step": 3269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.475378787878788e-05,
+      "loss": 0.4836,
+      "step": 3270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.474905303030303e-05,
+      "loss": 0.4726,
+      "step": 3271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4744318181818185e-05,
+      "loss": 0.4404,
+      "step": 3272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.473958333333334e-05,
+      "loss": 0.5863,
+      "step": 3273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4734848484848486e-05,
+      "loss": 0.4933,
+      "step": 3274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.473011363636364e-05,
+      "loss": 0.5602,
+      "step": 3275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4725378787878786e-05,
+      "loss": 0.42,
+      "step": 3276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.472064393939394e-05,
+      "loss": 0.5645,
+      "step": 3277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4715909090909094e-05,
+      "loss": 0.7036,
+      "step": 3278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.471117424242424e-05,
+      "loss": 0.5325,
+      "step": 3279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4706439393939395e-05,
+      "loss": 0.5914,
+      "step": 3280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.470170454545455e-05,
+      "loss": 0.4492,
+      "step": 3281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46969696969697e-05,
+      "loss": 0.4601,
+      "step": 3282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.469223484848485e-05,
+      "loss": 0.3823,
+      "step": 3283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46875e-05,
+      "loss": 0.5119,
+      "step": 3284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.468276515151515e-05,
+      "loss": 0.4682,
+      "step": 3285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4678030303030304e-05,
+      "loss": 0.608,
+      "step": 3286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.467329545454546e-05,
+      "loss": 0.3982,
+      "step": 3287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4668560606060604e-05,
+      "loss": 0.5369,
+      "step": 3288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.466382575757576e-05,
+      "loss": 0.6,
+      "step": 3289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.465909090909091e-05,
+      "loss": 0.5524,
+      "step": 3290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4654356060606066e-05,
+      "loss": 0.5683,
+      "step": 3291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.464962121212121e-05,
+      "loss": 0.3888,
+      "step": 3292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4644886363636366e-05,
+      "loss": 0.578,
+      "step": 3293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4640151515151513e-05,
+      "loss": 0.5643,
+      "step": 3294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.463541666666667e-05,
+      "loss": 0.5422,
+      "step": 3295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.463068181818182e-05,
+      "loss": 0.5246,
+      "step": 3296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.462594696969697e-05,
+      "loss": 0.4852,
+      "step": 3297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.462121212121212e-05,
+      "loss": 0.5407,
+      "step": 3298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4616477272727275e-05,
+      "loss": 0.5767,
+      "step": 3299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.461174242424243e-05,
+      "loss": 0.5639,
+      "step": 3300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4607007575757576e-05,
+      "loss": 0.5088,
+      "step": 3301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.460227272727273e-05,
+      "loss": 0.5565,
+      "step": 3302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.459753787878788e-05,
+      "loss": 0.4196,
+      "step": 3303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.459280303030303e-05,
+      "loss": 0.4576,
+      "step": 3304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4588068181818184e-05,
+      "loss": 0.4972,
+      "step": 3305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.458333333333333e-05,
+      "loss": 0.5409,
+      "step": 3306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4578598484848485e-05,
+      "loss": 0.4862,
+      "step": 3307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.457386363636364e-05,
+      "loss": 0.4827,
+      "step": 3308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.456912878787879e-05,
+      "loss": 0.5287,
+      "step": 3309
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.456439393939394e-05,
+      "loss": 0.5085,
+      "step": 3310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4559659090909093e-05,
+      "loss": 0.4998,
+      "step": 3311
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.455492424242424e-05,
+      "loss": 0.5994,
+      "step": 3312
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4550189393939394e-05,
+      "loss": 0.4835,
+      "step": 3313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.454545454545455e-05,
+      "loss": 0.387,
+      "step": 3314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4540719696969695e-05,
+      "loss": 0.5082,
+      "step": 3315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.453598484848485e-05,
+      "loss": 0.6062,
+      "step": 3316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.453125e-05,
+      "loss": 0.4298,
+      "step": 3317
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4526515151515156e-05,
+      "loss": 0.4747,
+      "step": 3318
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.45217803030303e-05,
+      "loss": 0.6113,
+      "step": 3319
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.451704545454546e-05,
+      "loss": 0.4289,
+      "step": 3320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4512310606060604e-05,
+      "loss": 0.4863,
+      "step": 3321
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.450757575757576e-05,
+      "loss": 0.44,
+      "step": 3322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.450284090909091e-05,
+      "loss": 0.4962,
+      "step": 3323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.449810606060606e-05,
+      "loss": 0.7254,
+      "step": 3324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.449337121212121e-05,
+      "loss": 0.47,
+      "step": 3325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4488636363636366e-05,
+      "loss": 0.4894,
+      "step": 3326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.448390151515152e-05,
+      "loss": 0.4097,
+      "step": 3327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.447916666666667e-05,
+      "loss": 0.5404,
+      "step": 3328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.447443181818182e-05,
+      "loss": 0.4968,
+      "step": 3329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4469696969696974e-05,
+      "loss": 0.476,
+      "step": 3330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.446496212121212e-05,
+      "loss": 0.6661,
+      "step": 3331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4460227272727275e-05,
+      "loss": 0.5084,
+      "step": 3332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.445549242424242e-05,
+      "loss": 0.5438,
+      "step": 3333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4450757575757576e-05,
+      "loss": 0.5471,
+      "step": 3334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.444602272727273e-05,
+      "loss": 0.4978,
+      "step": 3335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.444128787878788e-05,
+      "loss": 0.6084,
+      "step": 3336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.443655303030303e-05,
+      "loss": 0.5187,
+      "step": 3337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4431818181818184e-05,
+      "loss": 0.5448,
+      "step": 3338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.442708333333334e-05,
+      "loss": 0.4413,
+      "step": 3339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4422348484848485e-05,
+      "loss": 0.4529,
+      "step": 3340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.441761363636364e-05,
+      "loss": 0.5055,
+      "step": 3341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4412878787878786e-05,
+      "loss": 0.3975,
+      "step": 3342
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.440814393939394e-05,
+      "loss": 0.593,
+      "step": 3343
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.440340909090909e-05,
+      "loss": 0.4761,
+      "step": 3344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.439867424242425e-05,
+      "loss": 0.6513,
+      "step": 3345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4393939393939394e-05,
+      "loss": 0.4927,
+      "step": 3346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.438920454545455e-05,
+      "loss": 0.5171,
+      "step": 3347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.43844696969697e-05,
+      "loss": 0.4547,
+      "step": 3348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.437973484848485e-05,
+      "loss": 0.4135,
+      "step": 3349
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4375e-05,
+      "loss": 0.4918,
+      "step": 3350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.437026515151515e-05,
+      "loss": 0.594,
+      "step": 3351
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.43655303030303e-05,
+      "loss": 0.4236,
+      "step": 3352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4360795454545457e-05,
+      "loss": 0.581,
+      "step": 3353
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.435606060606061e-05,
+      "loss": 0.4368,
+      "step": 3354
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.435132575757576e-05,
+      "loss": 0.5783,
+      "step": 3355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.434659090909091e-05,
+      "loss": 0.569,
+      "step": 3356
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4341856060606065e-05,
+      "loss": 0.5526,
+      "step": 3357
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.433712121212121e-05,
+      "loss": 0.478,
+      "step": 3358
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4332386363636366e-05,
+      "loss": 0.6448,
+      "step": 3359
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.432765151515151e-05,
+      "loss": 0.4754,
+      "step": 3360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4322916666666666e-05,
+      "loss": 0.5382,
+      "step": 3361
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.431818181818182e-05,
+      "loss": 0.5127,
+      "step": 3362
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4313446969696974e-05,
+      "loss": 0.61,
+      "step": 3363
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.430871212121212e-05,
+      "loss": 0.5093,
+      "step": 3364
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4303977272727275e-05,
+      "loss": 0.7228,
+      "step": 3365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.429924242424243e-05,
+      "loss": 0.4596,
+      "step": 3366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4294507575757575e-05,
+      "loss": 0.4851,
+      "step": 3367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.428977272727273e-05,
+      "loss": 0.5594,
+      "step": 3368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4285037878787876e-05,
+      "loss": 0.4711,
+      "step": 3369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.428030303030303e-05,
+      "loss": 0.537,
+      "step": 3370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4275568181818184e-05,
+      "loss": 0.5883,
+      "step": 3371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.427083333333334e-05,
+      "loss": 0.5302,
+      "step": 3372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4266098484848484e-05,
+      "loss": 0.4801,
+      "step": 3373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.426136363636364e-05,
+      "loss": 0.5837,
+      "step": 3374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.425662878787879e-05,
+      "loss": 0.5726,
+      "step": 3375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.425189393939394e-05,
+      "loss": 0.5204,
+      "step": 3376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.424715909090909e-05,
+      "loss": 0.4931,
+      "step": 3377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.424242424242424e-05,
+      "loss": 0.4051,
+      "step": 3378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.42376893939394e-05,
+      "loss": 0.5717,
+      "step": 3379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.423295454545455e-05,
+      "loss": 0.5245,
+      "step": 3380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.42282196969697e-05,
+      "loss": 0.5061,
+      "step": 3381
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.422348484848485e-05,
+      "loss": 0.4886,
+      "step": 3382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.421875e-05,
+      "loss": 0.5053,
+      "step": 3383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4214015151515155e-05,
+      "loss": 0.5004,
+      "step": 3384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.42092803030303e-05,
+      "loss": 0.463,
+      "step": 3385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4204545454545456e-05,
+      "loss": 0.3912,
+      "step": 3386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.41998106060606e-05,
+      "loss": 0.4961,
+      "step": 3387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4195075757575764e-05,
+      "loss": 0.4291,
+      "step": 3388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.419034090909091e-05,
+      "loss": 0.5476,
+      "step": 3389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4185606060606064e-05,
+      "loss": 0.4762,
+      "step": 3390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.418087121212121e-05,
+      "loss": 0.5975,
+      "step": 3391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4176136363636365e-05,
+      "loss": 0.4812,
+      "step": 3392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.417140151515152e-05,
+      "loss": 0.4824,
+      "step": 3393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4166666666666666e-05,
+      "loss": 0.5601,
+      "step": 3394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.416193181818182e-05,
+      "loss": 0.5232,
+      "step": 3395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.415719696969697e-05,
+      "loss": 0.5113,
+      "step": 3396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.415246212121213e-05,
+      "loss": 0.3284,
+      "step": 3397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4147727272727274e-05,
+      "loss": 0.3623,
+      "step": 3398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.414299242424243e-05,
+      "loss": 0.478,
+      "step": 3399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4138257575757575e-05,
+      "loss": 0.5263,
+      "step": 3400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.413352272727273e-05,
+      "loss": 0.4252,
+      "step": 3401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.412878787878788e-05,
+      "loss": 0.5293,
+      "step": 3402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.412405303030303e-05,
+      "loss": 0.4371,
+      "step": 3403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.411931818181818e-05,
+      "loss": 0.4299,
+      "step": 3404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.411458333333333e-05,
+      "loss": 0.5546,
+      "step": 3405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.410984848484849e-05,
+      "loss": 0.5199,
+      "step": 3406
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.410511363636364e-05,
+      "loss": 0.4967,
+      "step": 3407
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.410037878787879e-05,
+      "loss": 0.5587,
+      "step": 3408
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.409564393939394e-05,
+      "loss": 0.5453,
+      "step": 3409
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.409090909090909e-05,
+      "loss": 0.5154,
+      "step": 3410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4086174242424246e-05,
+      "loss": 0.4984,
+      "step": 3411
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.408143939393939e-05,
+      "loss": 0.4884,
+      "step": 3412
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.407670454545455e-05,
+      "loss": 0.5057,
+      "step": 3413
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4071969696969694e-05,
+      "loss": 0.3995,
+      "step": 3414
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4067234848484854e-05,
+      "loss": 0.5163,
+      "step": 3415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.40625e-05,
+      "loss": 0.4884,
+      "step": 3416
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4057765151515155e-05,
+      "loss": 0.4538,
+      "step": 3417
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.40530303030303e-05,
+      "loss": 0.5184,
+      "step": 3418
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4048295454545456e-05,
+      "loss": 0.5396,
+      "step": 3419
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.404356060606061e-05,
+      "loss": 0.4338,
+      "step": 3420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4038825757575757e-05,
+      "loss": 0.5045,
+      "step": 3421
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.403409090909091e-05,
+      "loss": 0.373,
+      "step": 3422
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.402935606060606e-05,
+      "loss": 0.4789,
+      "step": 3423
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.402462121212122e-05,
+      "loss": 0.4838,
+      "step": 3424
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4019886363636365e-05,
+      "loss": 0.5432,
+      "step": 3425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.401515151515152e-05,
+      "loss": 0.5043,
+      "step": 3426
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4010416666666666e-05,
+      "loss": 0.4756,
+      "step": 3427
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.400568181818182e-05,
+      "loss": 0.4383,
+      "step": 3428
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.400094696969697e-05,
+      "loss": 0.4704,
+      "step": 3429
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.399621212121212e-05,
+      "loss": 0.4263,
+      "step": 3430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3991477272727274e-05,
+      "loss": 0.5236,
+      "step": 3431
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.398674242424242e-05,
+      "loss": 0.4886,
+      "step": 3432
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.398200757575758e-05,
+      "loss": 0.5447,
+      "step": 3433
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.397727272727273e-05,
+      "loss": 0.5818,
+      "step": 3434
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.397253787878788e-05,
+      "loss": 0.5431,
+      "step": 3435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.396780303030303e-05,
+      "loss": 0.5328,
+      "step": 3436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.396306818181818e-05,
+      "loss": 0.4116,
+      "step": 3437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3958333333333337e-05,
+      "loss": 0.6203,
+      "step": 3438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3953598484848484e-05,
+      "loss": 0.5222,
+      "step": 3439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.394886363636364e-05,
+      "loss": 0.4206,
+      "step": 3440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3944128787878784e-05,
+      "loss": 0.5314,
+      "step": 3441
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3939393939393945e-05,
+      "loss": 0.4511,
+      "step": 3442
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.393465909090909e-05,
+      "loss": 0.5717,
+      "step": 3443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3929924242424246e-05,
+      "loss": 0.5259,
+      "step": 3444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.392518939393939e-05,
+      "loss": 0.4462,
+      "step": 3445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3920454545454546e-05,
+      "loss": 0.4497,
+      "step": 3446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.39157196969697e-05,
+      "loss": 0.4772,
+      "step": 3447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.391098484848485e-05,
+      "loss": 0.4298,
+      "step": 3448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.390625e-05,
+      "loss": 0.4438,
+      "step": 3449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.390151515151515e-05,
+      "loss": 0.5122,
+      "step": 3450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.389678030303031e-05,
+      "loss": 0.5133,
+      "step": 3451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3892045454545455e-05,
+      "loss": 0.5482,
+      "step": 3452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.388731060606061e-05,
+      "loss": 0.4298,
+      "step": 3453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3882575757575756e-05,
+      "loss": 0.5157,
+      "step": 3454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.387784090909091e-05,
+      "loss": 0.5898,
+      "step": 3455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3873106060606064e-05,
+      "loss": 0.5722,
+      "step": 3456
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.386837121212121e-05,
+      "loss": 0.4423,
+      "step": 3457
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3863636363636364e-05,
+      "loss": 0.4289,
+      "step": 3458
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.385890151515151e-05,
+      "loss": 0.5154,
+      "step": 3459
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.385416666666667e-05,
+      "loss": 0.513,
+      "step": 3460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.384943181818182e-05,
+      "loss": 0.5749,
+      "step": 3461
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.384469696969697e-05,
+      "loss": 0.5335,
+      "step": 3462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.383996212121212e-05,
+      "loss": 0.4583,
+      "step": 3463
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.383522727272728e-05,
+      "loss": 0.4399,
+      "step": 3464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.383049242424243e-05,
+      "loss": 0.4941,
+      "step": 3465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3825757575757574e-05,
+      "loss": 0.5046,
+      "step": 3466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.382102272727273e-05,
+      "loss": 0.5423,
+      "step": 3467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3816287878787875e-05,
+      "loss": 0.5452,
+      "step": 3468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3811553030303035e-05,
+      "loss": 0.3782,
+      "step": 3469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.380681818181818e-05,
+      "loss": 0.5425,
+      "step": 3470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3802083333333336e-05,
+      "loss": 0.563,
+      "step": 3471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.379734848484848e-05,
+      "loss": 0.4511,
+      "step": 3472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3792613636363644e-05,
+      "loss": 0.4297,
+      "step": 3473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.378787878787879e-05,
+      "loss": 0.5819,
+      "step": 3474
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.378314393939394e-05,
+      "loss": 0.5052,
+      "step": 3475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.377840909090909e-05,
+      "loss": 0.5695,
+      "step": 3476
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.377367424242424e-05,
+      "loss": 0.4713,
+      "step": 3477
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.37689393939394e-05,
+      "loss": 0.5009,
+      "step": 3478
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3764204545454546e-05,
+      "loss": 0.5262,
+      "step": 3479
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.37594696969697e-05,
+      "loss": 0.4767,
+      "step": 3480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.375473484848485e-05,
+      "loss": 0.4526,
+      "step": 3481
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.375000000000001e-05,
+      "loss": 0.5001,
+      "step": 3482
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3745265151515154e-05,
+      "loss": 0.4434,
+      "step": 3483
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.37405303030303e-05,
+      "loss": 0.5702,
+      "step": 3484
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3735795454545455e-05,
+      "loss": 0.5337,
+      "step": 3485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.37310606060606e-05,
+      "loss": 0.5365,
+      "step": 3486
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.372632575757576e-05,
+      "loss": 0.4575,
+      "step": 3487
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.372159090909091e-05,
+      "loss": 0.5165,
+      "step": 3488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.371685606060606e-05,
+      "loss": 0.428,
+      "step": 3489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.371212121212121e-05,
+      "loss": 0.5206,
+      "step": 3490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.370738636363637e-05,
+      "loss": 0.6206,
+      "step": 3491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.370265151515152e-05,
+      "loss": 0.5537,
+      "step": 3492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3697916666666665e-05,
+      "loss": 0.4451,
+      "step": 3493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.369318181818182e-05,
+      "loss": 0.4511,
+      "step": 3494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.368844696969697e-05,
+      "loss": 0.5335,
+      "step": 3495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3683712121212126e-05,
+      "loss": 0.4586,
+      "step": 3496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.367897727272727e-05,
+      "loss": 0.6572,
+      "step": 3497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.367424242424243e-05,
+      "loss": 0.4886,
+      "step": 3498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3669507575757574e-05,
+      "loss": 0.53,
+      "step": 3499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3664772727272734e-05,
+      "loss": 0.6307,
+      "step": 3500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.366003787878788e-05,
+      "loss": 0.4949,
+      "step": 3501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.365530303030303e-05,
+      "loss": 0.5465,
+      "step": 3502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.365056818181818e-05,
+      "loss": 0.4824,
+      "step": 3503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3645833333333336e-05,
+      "loss": 0.4585,
+      "step": 3504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.364109848484849e-05,
+      "loss": 0.529,
+      "step": 3505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3636363636363636e-05,
+      "loss": 0.42,
+      "step": 3506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.363162878787879e-05,
+      "loss": 0.4296,
+      "step": 3507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.362689393939394e-05,
+      "loss": 0.5189,
+      "step": 3508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.36221590909091e-05,
+      "loss": 0.5069,
+      "step": 3509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3617424242424245e-05,
+      "loss": 0.4573,
+      "step": 3510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.361268939393939e-05,
+      "loss": 0.4862,
+      "step": 3511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3607954545454546e-05,
+      "loss": 0.4657,
+      "step": 3512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.36032196969697e-05,
+      "loss": 0.6407,
+      "step": 3513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.359848484848485e-05,
+      "loss": 0.5097,
+      "step": 3514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.359375e-05,
+      "loss": 0.5982,
+      "step": 3515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3589015151515154e-05,
+      "loss": 0.6457,
+      "step": 3516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.35842803030303e-05,
+      "loss": 0.4192,
+      "step": 3517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.357954545454546e-05,
+      "loss": 0.503,
+      "step": 3518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.357481060606061e-05,
+      "loss": 0.5475,
+      "step": 3519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3570075757575755e-05,
+      "loss": 0.4627,
+      "step": 3520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.356534090909091e-05,
+      "loss": 0.4515,
+      "step": 3521
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.356060606060606e-05,
+      "loss": 0.6479,
+      "step": 3522
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3555871212121217e-05,
+      "loss": 0.4588,
+      "step": 3523
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3551136363636364e-05,
+      "loss": 0.4393,
+      "step": 3524
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.354640151515152e-05,
+      "loss": 0.422,
+      "step": 3525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3541666666666664e-05,
+      "loss": 0.5927,
+      "step": 3526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3536931818181825e-05,
+      "loss": 0.3325,
+      "step": 3527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.353219696969697e-05,
+      "loss": 0.5416,
+      "step": 3528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.352746212121212e-05,
+      "loss": 0.4087,
+      "step": 3529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.352272727272727e-05,
+      "loss": 0.4697,
+      "step": 3530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3517992424242426e-05,
+      "loss": 0.5026,
+      "step": 3531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.351325757575758e-05,
+      "loss": 0.5968,
+      "step": 3532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.350852272727273e-05,
+      "loss": 0.4663,
+      "step": 3533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.350378787878788e-05,
+      "loss": 0.4167,
+      "step": 3534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.349905303030303e-05,
+      "loss": 0.5211,
+      "step": 3535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.349431818181819e-05,
+      "loss": 0.5453,
+      "step": 3536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3489583333333335e-05,
+      "loss": 0.48,
+      "step": 3537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.348484848484848e-05,
+      "loss": 0.4906,
+      "step": 3538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3480113636363636e-05,
+      "loss": 0.4792,
+      "step": 3539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.347537878787879e-05,
+      "loss": 0.5657,
+      "step": 3540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3470643939393944e-05,
+      "loss": 0.5833,
+      "step": 3541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.346590909090909e-05,
+      "loss": 0.4863,
+      "step": 3542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3461174242424244e-05,
+      "loss": 0.5899,
+      "step": 3543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.34564393939394e-05,
+      "loss": 0.4619,
+      "step": 3544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.345170454545455e-05,
+      "loss": 0.4451,
+      "step": 3545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.34469696969697e-05,
+      "loss": 0.456,
+      "step": 3546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3442234848484846e-05,
+      "loss": 0.4757,
+      "step": 3547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.34375e-05,
+      "loss": 0.4598,
+      "step": 3548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.343276515151515e-05,
+      "loss": 0.5791,
+      "step": 3549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.342803030303031e-05,
+      "loss": 0.5328,
+      "step": 3550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3423295454545454e-05,
+      "loss": 0.5823,
+      "step": 3551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.341856060606061e-05,
+      "loss": 0.4468,
+      "step": 3552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.341382575757576e-05,
+      "loss": 0.5866,
+      "step": 3553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3409090909090915e-05,
+      "loss": 0.5662,
+      "step": 3554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.340435606060606e-05,
+      "loss": 0.3798,
+      "step": 3555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.339962121212121e-05,
+      "loss": 0.4835,
+      "step": 3556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.339488636363636e-05,
+      "loss": 0.4917,
+      "step": 3557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.339015151515152e-05,
+      "loss": 0.5573,
+      "step": 3558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.338541666666667e-05,
+      "loss": 0.5785,
+      "step": 3559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.338068181818182e-05,
+      "loss": 0.5156,
+      "step": 3560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.337594696969697e-05,
+      "loss": 0.5541,
+      "step": 3561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3371212121212125e-05,
+      "loss": 0.4968,
+      "step": 3562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.336647727272728e-05,
+      "loss": 0.5456,
+      "step": 3563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3361742424242426e-05,
+      "loss": 0.6666,
+      "step": 3564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.335700757575757e-05,
+      "loss": 0.5873,
+      "step": 3565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.335227272727273e-05,
+      "loss": 0.48,
+      "step": 3566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.334753787878788e-05,
+      "loss": 0.5103,
+      "step": 3567
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3342803030303034e-05,
+      "loss": 0.5686,
+      "step": 3568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.333806818181818e-05,
+      "loss": 0.4884,
+      "step": 3569
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.4314,
+      "step": 3570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.332859848484849e-05,
+      "loss": 0.5167,
+      "step": 3571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.332386363636364e-05,
+      "loss": 0.4123,
+      "step": 3572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.331912878787879e-05,
+      "loss": 0.5142,
+      "step": 3573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3314393939393936e-05,
+      "loss": 0.5246,
+      "step": 3574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.330965909090909e-05,
+      "loss": 0.6209,
+      "step": 3575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3304924242424244e-05,
+      "loss": 0.5562,
+      "step": 3576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.33001893939394e-05,
+      "loss": 0.4731,
+      "step": 3577
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3295454545454545e-05,
+      "loss": 0.6154,
+      "step": 3578
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.32907196969697e-05,
+      "loss": 0.4992,
+      "step": 3579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.328598484848485e-05,
+      "loss": 0.4791,
+      "step": 3580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3281250000000006e-05,
+      "loss": 0.5962,
+      "step": 3581
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.327651515151515e-05,
+      "loss": 0.6025,
+      "step": 3582
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.32717803030303e-05,
+      "loss": 0.5809,
+      "step": 3583
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3267045454545454e-05,
+      "loss": 0.5269,
+      "step": 3584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.326231060606061e-05,
+      "loss": 0.4832,
+      "step": 3585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.325757575757576e-05,
+      "loss": 0.5487,
+      "step": 3586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.325284090909091e-05,
+      "loss": 0.4726,
+      "step": 3587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.324810606060606e-05,
+      "loss": 0.5207,
+      "step": 3588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3243371212121216e-05,
+      "loss": 0.4583,
+      "step": 3589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.323863636363637e-05,
+      "loss": 0.5796,
+      "step": 3590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3233901515151516e-05,
+      "loss": 0.5837,
+      "step": 3591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3229166666666663e-05,
+      "loss": 0.531,
+      "step": 3592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.322443181818182e-05,
+      "loss": 0.4528,
+      "step": 3593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.321969696969697e-05,
+      "loss": 0.5199,
+      "step": 3594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3214962121212125e-05,
+      "loss": 0.4951,
+      "step": 3595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.321022727272727e-05,
+      "loss": 0.5336,
+      "step": 3596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3205492424242426e-05,
+      "loss": 0.4923,
+      "step": 3597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.320075757575758e-05,
+      "loss": 0.522,
+      "step": 3598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.319602272727273e-05,
+      "loss": 0.4671,
+      "step": 3599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.319128787878788e-05,
+      "loss": 0.5876,
+      "step": 3600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.318655303030303e-05,
+      "loss": 0.5299,
+      "step": 3601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.318181818181819e-05,
+      "loss": 0.5408,
+      "step": 3602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3177083333333335e-05,
+      "loss": 0.4515,
+      "step": 3603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.317234848484849e-05,
+      "loss": 0.5491,
+      "step": 3604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3167613636363635e-05,
+      "loss": 0.4744,
+      "step": 3605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.316287878787879e-05,
+      "loss": 0.5779,
+      "step": 3606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.315814393939394e-05,
+      "loss": 0.5063,
+      "step": 3607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3153409090909097e-05,
+      "loss": 0.52,
+      "step": 3608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3148674242424244e-05,
+      "loss": 0.5601,
+      "step": 3609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.314393939393939e-05,
+      "loss": 0.5487,
+      "step": 3610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.313920454545455e-05,
+      "loss": 0.4418,
+      "step": 3611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.31344696969697e-05,
+      "loss": 0.4627,
+      "step": 3612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.312973484848485e-05,
+      "loss": 0.5996,
+      "step": 3613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3125e-05,
+      "loss": 0.4401,
+      "step": 3614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.312026515151515e-05,
+      "loss": 0.4358,
+      "step": 3615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3115530303030306e-05,
+      "loss": 0.6703,
+      "step": 3616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.311079545454546e-05,
+      "loss": 0.4397,
+      "step": 3617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.310606060606061e-05,
+      "loss": 0.4638,
+      "step": 3618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3101325757575754e-05,
+      "loss": 0.5986,
+      "step": 3619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3096590909090915e-05,
+      "loss": 0.6124,
+      "step": 3620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.309185606060606e-05,
+      "loss": 0.4281,
+      "step": 3621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3087121212121215e-05,
+      "loss": 0.4619,
+      "step": 3622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.308238636363636e-05,
+      "loss": 0.5096,
+      "step": 3623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3077651515151516e-05,
+      "loss": 0.5275,
+      "step": 3624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.307291666666667e-05,
+      "loss": 0.4543,
+      "step": 3625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3068181818181824e-05,
+      "loss": 0.4903,
+      "step": 3626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.306344696969697e-05,
+      "loss": 0.5599,
+      "step": 3627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.305871212121212e-05,
+      "loss": 0.5148,
+      "step": 3628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.305397727272728e-05,
+      "loss": 0.4459,
+      "step": 3629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3049242424242425e-05,
+      "loss": 0.3882,
+      "step": 3630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.304450757575758e-05,
+      "loss": 0.4117,
+      "step": 3631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3039772727272726e-05,
+      "loss": 0.6101,
+      "step": 3632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.303503787878788e-05,
+      "loss": 0.3954,
+      "step": 3633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.303030303030303e-05,
+      "loss": 0.4629,
+      "step": 3634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.302556818181819e-05,
+      "loss": 0.5446,
+      "step": 3635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3020833333333334e-05,
+      "loss": 0.559,
+      "step": 3636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.301609848484848e-05,
+      "loss": 0.4007,
+      "step": 3637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.301136363636364e-05,
+      "loss": 0.4448,
+      "step": 3638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.300662878787879e-05,
+      "loss": 0.5856,
+      "step": 3639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.300189393939394e-05,
+      "loss": 0.5141,
+      "step": 3640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.299715909090909e-05,
+      "loss": 0.4278,
+      "step": 3641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.299242424242424e-05,
+      "loss": 0.5197,
+      "step": 3642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.29876893939394e-05,
+      "loss": 0.4757,
+      "step": 3643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.298295454545455e-05,
+      "loss": 0.3859,
+      "step": 3644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.29782196969697e-05,
+      "loss": 0.547,
+      "step": 3645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2973484848484845e-05,
+      "loss": 0.577,
+      "step": 3646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2968750000000005e-05,
+      "loss": 0.4906,
+      "step": 3647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.296401515151515e-05,
+      "loss": 0.6473,
+      "step": 3648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2959280303030306e-05,
+      "loss": 0.5377,
+      "step": 3649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.295454545454545e-05,
+      "loss": 0.465,
+      "step": 3650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.294981060606061e-05,
+      "loss": 0.4828,
+      "step": 3651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.294507575757576e-05,
+      "loss": 0.4162,
+      "step": 3652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2940340909090914e-05,
+      "loss": 0.5736,
+      "step": 3653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.293560606060606e-05,
+      "loss": 0.4227,
+      "step": 3654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.293087121212121e-05,
+      "loss": 0.5793,
+      "step": 3655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.292613636363637e-05,
+      "loss": 0.5242,
+      "step": 3656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2921401515151516e-05,
+      "loss": 0.4144,
+      "step": 3657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.291666666666667e-05,
+      "loss": 0.4859,
+      "step": 3658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2911931818181816e-05,
+      "loss": 0.4904,
+      "step": 3659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.290719696969698e-05,
+      "loss": 0.4691,
+      "step": 3660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2902462121212124e-05,
+      "loss": 0.451,
+      "step": 3661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.289772727272728e-05,
+      "loss": 0.5051,
+      "step": 3662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2892992424242425e-05,
+      "loss": 0.5509,
+      "step": 3663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.288825757575757e-05,
+      "loss": 0.404,
+      "step": 3664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.288352272727273e-05,
+      "loss": 0.5585,
+      "step": 3665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.287878787878788e-05,
+      "loss": 0.4912,
+      "step": 3666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.287405303030303e-05,
+      "loss": 0.5169,
+      "step": 3667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.286931818181818e-05,
+      "loss": 0.3482,
+      "step": 3668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.286458333333334e-05,
+      "loss": 0.3988,
+      "step": 3669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.285984848484849e-05,
+      "loss": 0.4113,
+      "step": 3670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.285511363636364e-05,
+      "loss": 0.5653,
+      "step": 3671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.285037878787879e-05,
+      "loss": 0.5221,
+      "step": 3672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2845643939393935e-05,
+      "loss": 0.3704,
+      "step": 3673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2840909090909096e-05,
+      "loss": 0.638,
+      "step": 3674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.283617424242424e-05,
+      "loss": 0.4723,
+      "step": 3675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2831439393939396e-05,
+      "loss": 0.6256,
+      "step": 3676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2826704545454543e-05,
+      "loss": 0.6134,
+      "step": 3677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2821969696969704e-05,
+      "loss": 0.5255,
+      "step": 3678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.281723484848485e-05,
+      "loss": 0.5148,
+      "step": 3679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2812500000000005e-05,
+      "loss": 0.4981,
+      "step": 3680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.280776515151515e-05,
+      "loss": 0.5987,
+      "step": 3681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2803030303030305e-05,
+      "loss": 0.5866,
+      "step": 3682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.279829545454546e-05,
+      "loss": 0.4485,
+      "step": 3683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2793560606060606e-05,
+      "loss": 0.4346,
+      "step": 3684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.278882575757576e-05,
+      "loss": 0.504,
+      "step": 3685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.278409090909091e-05,
+      "loss": 0.4496,
+      "step": 3686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.277935606060607e-05,
+      "loss": 0.5113,
+      "step": 3687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2774621212121215e-05,
+      "loss": 0.5442,
+      "step": 3688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.276988636363637e-05,
+      "loss": 0.5101,
+      "step": 3689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2765151515151515e-05,
+      "loss": 0.5611,
+      "step": 3690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.276041666666667e-05,
+      "loss": 0.3863,
+      "step": 3691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.275568181818182e-05,
+      "loss": 0.587,
+      "step": 3692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.275094696969697e-05,
+      "loss": 0.5515,
+      "step": 3693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2746212121212124e-05,
+      "loss": 0.497,
+      "step": 3694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.274147727272727e-05,
+      "loss": 0.5746,
+      "step": 3695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.273674242424243e-05,
+      "loss": 0.5359,
+      "step": 3696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.273200757575758e-05,
+      "loss": 0.5933,
+      "step": 3697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.272727272727273e-05,
+      "loss": 0.6051,
+      "step": 3698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.272253787878788e-05,
+      "loss": 0.5153,
+      "step": 3699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.271780303030303e-05,
+      "loss": 0.4692,
+      "step": 3700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2713068181818186e-05,
+      "loss": 0.4981,
+      "step": 3701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.270833333333333e-05,
+      "loss": 0.5732,
+      "step": 3702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.270359848484849e-05,
+      "loss": 0.5525,
+      "step": 3703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2698863636363634e-05,
+      "loss": 0.5359,
+      "step": 3704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2694128787878795e-05,
+      "loss": 0.4975,
+      "step": 3705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.268939393939394e-05,
+      "loss": 0.5241,
+      "step": 3706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2684659090909095e-05,
+      "loss": 0.5569,
+      "step": 3707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.267992424242424e-05,
+      "loss": 0.5428,
+      "step": 3708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2675189393939396e-05,
+      "loss": 0.5415,
+      "step": 3709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.267045454545455e-05,
+      "loss": 0.558,
+      "step": 3710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.26657196969697e-05,
+      "loss": 0.5008,
+      "step": 3711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.266098484848485e-05,
+      "loss": 0.4341,
+      "step": 3712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.265625e-05,
+      "loss": 0.6105,
+      "step": 3713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.265151515151516e-05,
+      "loss": 0.5419,
+      "step": 3714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2646780303030305e-05,
+      "loss": 0.4812,
+      "step": 3715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.264204545454546e-05,
+      "loss": 0.4918,
+      "step": 3716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2637310606060606e-05,
+      "loss": 0.5221,
+      "step": 3717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.263257575757576e-05,
+      "loss": 0.5035,
+      "step": 3718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.262784090909091e-05,
+      "loss": 0.4566,
+      "step": 3719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.262310606060606e-05,
+      "loss": 0.4742,
+      "step": 3720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2618371212121214e-05,
+      "loss": 0.5018,
+      "step": 3721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.261363636363636e-05,
+      "loss": 0.5907,
+      "step": 3722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.260890151515152e-05,
+      "loss": 0.5144,
+      "step": 3723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.260416666666667e-05,
+      "loss": 0.4344,
+      "step": 3724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.259943181818182e-05,
+      "loss": 0.5542,
+      "step": 3725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.259469696969697e-05,
+      "loss": 0.4881,
+      "step": 3726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.258996212121212e-05,
+      "loss": 0.5664,
+      "step": 3727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.258522727272728e-05,
+      "loss": 0.4225,
+      "step": 3728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2580492424242424e-05,
+      "loss": 0.559,
+      "step": 3729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.257575757575758e-05,
+      "loss": 0.4942,
+      "step": 3730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2571022727272725e-05,
+      "loss": 0.5092,
+      "step": 3731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2566287878787885e-05,
+      "loss": 0.4583,
+      "step": 3732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.256155303030303e-05,
+      "loss": 0.562,
+      "step": 3733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2556818181818186e-05,
+      "loss": 0.6446,
+      "step": 3734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.255208333333333e-05,
+      "loss": 0.4734,
+      "step": 3735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.254734848484849e-05,
+      "loss": 0.4536,
+      "step": 3736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.254261363636364e-05,
+      "loss": 0.5649,
+      "step": 3737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.253787878787879e-05,
+      "loss": 0.5074,
+      "step": 3738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.253314393939394e-05,
+      "loss": 0.4588,
+      "step": 3739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.252840909090909e-05,
+      "loss": 0.4664,
+      "step": 3740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.252367424242425e-05,
+      "loss": 0.5228,
+      "step": 3741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2518939393939396e-05,
+      "loss": 0.3758,
+      "step": 3742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.251420454545455e-05,
+      "loss": 0.5467,
+      "step": 3743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2509469696969696e-05,
+      "loss": 0.4985,
+      "step": 3744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.250473484848485e-05,
+      "loss": 0.459,
+      "step": 3745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.4068,
+      "step": 3746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.249526515151515e-05,
+      "loss": 0.503,
+      "step": 3747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2490530303030305e-05,
+      "loss": 0.5975,
+      "step": 3748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.248579545454545e-05,
+      "loss": 0.5495,
+      "step": 3749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.248106060606061e-05,
+      "loss": 0.5838,
+      "step": 3750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.247632575757576e-05,
+      "loss": 0.481,
+      "step": 3751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.247159090909091e-05,
+      "loss": 0.5158,
+      "step": 3752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.246685606060606e-05,
+      "loss": 0.583,
+      "step": 3753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2462121212121214e-05,
+      "loss": 0.5199,
+      "step": 3754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.245738636363637e-05,
+      "loss": 0.4348,
+      "step": 3755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2452651515151514e-05,
+      "loss": 0.5266,
+      "step": 3756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.244791666666667e-05,
+      "loss": 0.4494,
+      "step": 3757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2443181818181815e-05,
+      "loss": 0.4862,
+      "step": 3758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2438446969696976e-05,
+      "loss": 0.512,
+      "step": 3759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.243371212121212e-05,
+      "loss": 0.4571,
+      "step": 3760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2428977272727276e-05,
+      "loss": 0.5056,
+      "step": 3761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2424242424242423e-05,
+      "loss": 0.5387,
+      "step": 3762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.241950757575758e-05,
+      "loss": 0.4869,
+      "step": 3763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.241477272727273e-05,
+      "loss": 0.6719,
+      "step": 3764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.241003787878788e-05,
+      "loss": 0.6557,
+      "step": 3765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.240530303030303e-05,
+      "loss": 0.3874,
+      "step": 3766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2400568181818185e-05,
+      "loss": 0.5152,
+      "step": 3767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.239583333333334e-05,
+      "loss": 0.4796,
+      "step": 3768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2391098484848486e-05,
+      "loss": 0.4971,
+      "step": 3769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.238636363636364e-05,
+      "loss": 0.4758,
+      "step": 3770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.238162878787879e-05,
+      "loss": 0.4912,
+      "step": 3771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.237689393939394e-05,
+      "loss": 0.6162,
+      "step": 3772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2372159090909095e-05,
+      "loss": 0.4084,
+      "step": 3773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.236742424242424e-05,
+      "loss": 0.4923,
+      "step": 3774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2362689393939395e-05,
+      "loss": 0.5658,
+      "step": 3775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.235795454545455e-05,
+      "loss": 0.5739,
+      "step": 3776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.23532196969697e-05,
+      "loss": 0.4965,
+      "step": 3777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.234848484848485e-05,
+      "loss": 0.4276,
+      "step": 3778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2343750000000004e-05,
+      "loss": 0.3668,
+      "step": 3779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.233901515151515e-05,
+      "loss": 0.4639,
+      "step": 3780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2334280303030304e-05,
+      "loss": 0.5111,
+      "step": 3781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.232954545454546e-05,
+      "loss": 0.4524,
+      "step": 3782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2324810606060605e-05,
+      "loss": 0.5144,
+      "step": 3783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.232007575757576e-05,
+      "loss": 0.5498,
+      "step": 3784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.231534090909091e-05,
+      "loss": 0.3443,
+      "step": 3785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2310606060606066e-05,
+      "loss": 0.4823,
+      "step": 3786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.230587121212121e-05,
+      "loss": 0.4483,
+      "step": 3787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.230113636363637e-05,
+      "loss": 0.5422,
+      "step": 3788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2296401515151514e-05,
+      "loss": 0.4018,
+      "step": 3789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.229166666666667e-05,
+      "loss": 0.5596,
+      "step": 3790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.228693181818182e-05,
+      "loss": 0.4586,
+      "step": 3791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.228219696969697e-05,
+      "loss": 0.4562,
+      "step": 3792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.227746212121212e-05,
+      "loss": 0.5301,
+      "step": 3793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2272727272727276e-05,
+      "loss": 0.4742,
+      "step": 3794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.226799242424243e-05,
+      "loss": 0.5135,
+      "step": 3795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.226325757575758e-05,
+      "loss": 0.4909,
+      "step": 3796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.225852272727273e-05,
+      "loss": 0.4306,
+      "step": 3797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.225378787878788e-05,
+      "loss": 0.4398,
+      "step": 3798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.224905303030303e-05,
+      "loss": 0.5955,
+      "step": 3799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2244318181818185e-05,
+      "loss": 0.3898,
+      "step": 3800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.223958333333333e-05,
+      "loss": 0.4629,
+      "step": 3801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2234848484848486e-05,
+      "loss": 0.4477,
+      "step": 3802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.223011363636364e-05,
+      "loss": 0.3861,
+      "step": 3803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.222537878787879e-05,
+      "loss": 0.3928,
+      "step": 3804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.222064393939394e-05,
+      "loss": 0.511,
+      "step": 3805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2215909090909094e-05,
+      "loss": 0.4496,
+      "step": 3806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.221117424242424e-05,
+      "loss": 0.5847,
+      "step": 3807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2206439393939395e-05,
+      "loss": 0.455,
+      "step": 3808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.220170454545455e-05,
+      "loss": 0.4476,
+      "step": 3809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2196969696969696e-05,
+      "loss": 0.4641,
+      "step": 3810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.219223484848485e-05,
+      "loss": 0.515,
+      "step": 3811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.21875e-05,
+      "loss": 0.4968,
+      "step": 3812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.218276515151516e-05,
+      "loss": 0.3774,
+      "step": 3813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2178030303030304e-05,
+      "loss": 0.5186,
+      "step": 3814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.217329545454546e-05,
+      "loss": 0.5327,
+      "step": 3815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2168560606060605e-05,
+      "loss": 0.4846,
+      "step": 3816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.216382575757576e-05,
+      "loss": 0.4642,
+      "step": 3817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.215909090909091e-05,
+      "loss": 0.4596,
+      "step": 3818
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.215435606060606e-05,
+      "loss": 0.5475,
+      "step": 3819
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.214962121212121e-05,
+      "loss": 0.584,
+      "step": 3820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.214488636363637e-05,
+      "loss": 0.443,
+      "step": 3821
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.214015151515152e-05,
+      "loss": 0.4831,
+      "step": 3822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.213541666666667e-05,
+      "loss": 0.489,
+      "step": 3823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.213068181818182e-05,
+      "loss": 0.5035,
+      "step": 3824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2125946969696975e-05,
+      "loss": 0.5026,
+      "step": 3825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.212121212121212e-05,
+      "loss": 0.5005,
+      "step": 3826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2116477272727276e-05,
+      "loss": 0.6353,
+      "step": 3827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.211174242424242e-05,
+      "loss": 0.4703,
+      "step": 3828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2107007575757576e-05,
+      "loss": 0.5658,
+      "step": 3829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.210227272727273e-05,
+      "loss": 0.5828,
+      "step": 3830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2097537878787884e-05,
+      "loss": 0.5131,
+      "step": 3831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.209280303030303e-05,
+      "loss": 0.4164,
+      "step": 3832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2088068181818185e-05,
+      "loss": 0.4587,
+      "step": 3833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.208333333333334e-05,
+      "loss": 0.4298,
+      "step": 3834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2078598484848485e-05,
+      "loss": 0.5479,
+      "step": 3835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.207386363636364e-05,
+      "loss": 0.5472,
+      "step": 3836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2069128787878786e-05,
+      "loss": 0.3889,
+      "step": 3837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.206439393939394e-05,
+      "loss": 0.5129,
+      "step": 3838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2059659090909094e-05,
+      "loss": 0.5152,
+      "step": 3839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.205492424242425e-05,
+      "loss": 0.5518,
+      "step": 3840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2050189393939394e-05,
+      "loss": 0.6159,
+      "step": 3841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.204545454545455e-05,
+      "loss": 0.4394,
+      "step": 3842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.20407196969697e-05,
+      "loss": 0.5032,
+      "step": 3843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.203598484848485e-05,
+      "loss": 0.6421,
+      "step": 3844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.203125e-05,
+      "loss": 0.4556,
+      "step": 3845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.202651515151515e-05,
+      "loss": 0.4684,
+      "step": 3846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2021780303030303e-05,
+      "loss": 0.5731,
+      "step": 3847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.201704545454546e-05,
+      "loss": 0.4772,
+      "step": 3848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.201231060606061e-05,
+      "loss": 0.5397,
+      "step": 3849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.200757575757576e-05,
+      "loss": 0.4938,
+      "step": 3850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.200284090909091e-05,
+      "loss": 0.5191,
+      "step": 3851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1998106060606065e-05,
+      "loss": 0.4221,
+      "step": 3852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.199337121212121e-05,
+      "loss": 0.4346,
+      "step": 3853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1988636363636366e-05,
+      "loss": 0.4299,
+      "step": 3854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.198390151515151e-05,
+      "loss": 0.56,
+      "step": 3855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.197916666666667e-05,
+      "loss": 0.4933,
+      "step": 3856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.197443181818182e-05,
+      "loss": 0.4523,
+      "step": 3857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1969696969696974e-05,
+      "loss": 0.521,
+      "step": 3858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.196496212121212e-05,
+      "loss": 0.5724,
+      "step": 3859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1960227272727275e-05,
+      "loss": 0.5467,
+      "step": 3860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.195549242424243e-05,
+      "loss": 0.5629,
+      "step": 3861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1950757575757576e-05,
+      "loss": 0.4807,
+      "step": 3862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.194602272727273e-05,
+      "loss": 0.5133,
+      "step": 3863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.194128787878788e-05,
+      "loss": 0.5963,
+      "step": 3864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.193655303030303e-05,
+      "loss": 0.536,
+      "step": 3865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1931818181818184e-05,
+      "loss": 0.4255,
+      "step": 3866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.192708333333334e-05,
+      "loss": 0.4989,
+      "step": 3867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1922348484848485e-05,
+      "loss": 0.4897,
+      "step": 3868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.191761363636364e-05,
+      "loss": 0.515,
+      "step": 3869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.191287878787879e-05,
+      "loss": 0.6016,
+      "step": 3870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.190814393939394e-05,
+      "loss": 0.4859,
+      "step": 3871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.190340909090909e-05,
+      "loss": 0.4715,
+      "step": 3872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.189867424242424e-05,
+      "loss": 0.458,
+      "step": 3873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.18939393939394e-05,
+      "loss": 0.401,
+      "step": 3874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.188920454545455e-05,
+      "loss": 0.5207,
+      "step": 3875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.18844696969697e-05,
+      "loss": 0.4265,
+      "step": 3876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.187973484848485e-05,
+      "loss": 0.3533,
+      "step": 3877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1875e-05,
+      "loss": 0.4019,
+      "step": 3878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1870265151515156e-05,
+      "loss": 0.442,
+      "step": 3879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.18655303030303e-05,
+      "loss": 0.6082,
+      "step": 3880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.186079545454546e-05,
+      "loss": 0.4936,
+      "step": 3881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1856060606060604e-05,
+      "loss": 0.5869,
+      "step": 3882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1851325757575764e-05,
+      "loss": 0.399,
+      "step": 3883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.184659090909091e-05,
+      "loss": 0.4635,
+      "step": 3884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1841856060606065e-05,
+      "loss": 0.6263,
+      "step": 3885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.183712121212121e-05,
+      "loss": 0.4924,
+      "step": 3886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1832386363636366e-05,
+      "loss": 0.3468,
+      "step": 3887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.182765151515152e-05,
+      "loss": 0.683,
+      "step": 3888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1822916666666667e-05,
+      "loss": 0.4639,
+      "step": 3889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 0.5398,
+      "step": 3890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.181344696969697e-05,
+      "loss": 0.4617,
+      "step": 3891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.180871212121213e-05,
+      "loss": 0.496,
+      "step": 3892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1803977272727275e-05,
+      "loss": 0.68,
+      "step": 3893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.179924242424243e-05,
+      "loss": 0.4303,
+      "step": 3894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1794507575757576e-05,
+      "loss": 0.5517,
+      "step": 3895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.178977272727273e-05,
+      "loss": 0.4265,
+      "step": 3896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.178503787878788e-05,
+      "loss": 0.529,
+      "step": 3897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.178030303030303e-05,
+      "loss": 0.4949,
+      "step": 3898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1775568181818184e-05,
+      "loss": 0.4655,
+      "step": 3899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.177083333333333e-05,
+      "loss": 0.495,
+      "step": 3900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.176609848484849e-05,
+      "loss": 0.4992,
+      "step": 3901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.176136363636364e-05,
+      "loss": 0.6239,
+      "step": 3902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.175662878787879e-05,
+      "loss": 0.4732,
+      "step": 3903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.175189393939394e-05,
+      "loss": 0.4947,
+      "step": 3904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.174715909090909e-05,
+      "loss": 0.4032,
+      "step": 3905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.174242424242425e-05,
+      "loss": 0.44,
+      "step": 3906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1737689393939394e-05,
+      "loss": 0.4427,
+      "step": 3907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.173295454545455e-05,
+      "loss": 0.4658,
+      "step": 3908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1728219696969694e-05,
+      "loss": 0.6081,
+      "step": 3909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1723484848484855e-05,
+      "loss": 0.4231,
+      "step": 3910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.171875e-05,
+      "loss": 0.4611,
+      "step": 3911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1714015151515156e-05,
+      "loss": 0.7634,
+      "step": 3912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.17092803030303e-05,
+      "loss": 0.5726,
+      "step": 3913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1704545454545456e-05,
+      "loss": 0.4251,
+      "step": 3914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.169981060606061e-05,
+      "loss": 0.4984,
+      "step": 3915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.169507575757576e-05,
+      "loss": 0.5584,
+      "step": 3916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.169034090909091e-05,
+      "loss": 0.4672,
+      "step": 3917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.168560606060606e-05,
+      "loss": 0.6336,
+      "step": 3918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.168087121212122e-05,
+      "loss": 0.5386,
+      "step": 3919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1676136363636365e-05,
+      "loss": 0.382,
+      "step": 3920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.167140151515152e-05,
+      "loss": 0.4841,
+      "step": 3921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.5321,
+      "step": 3922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.166193181818182e-05,
+      "loss": 0.4426,
+      "step": 3923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1657196969696974e-05,
+      "loss": 0.5648,
+      "step": 3924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.165246212121212e-05,
+      "loss": 0.4846,
+      "step": 3925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1647727272727274e-05,
+      "loss": 0.3845,
+      "step": 3926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.164299242424242e-05,
+      "loss": 0.5407,
+      "step": 3927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.163825757575758e-05,
+      "loss": 0.4649,
+      "step": 3928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.163352272727273e-05,
+      "loss": 0.5624,
+      "step": 3929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.162878787878788e-05,
+      "loss": 0.5519,
+      "step": 3930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.162405303030303e-05,
+      "loss": 0.6077,
+      "step": 3931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1619318181818183e-05,
+      "loss": 0.5712,
+      "step": 3932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.161458333333334e-05,
+      "loss": 0.3789,
+      "step": 3933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1609848484848484e-05,
+      "loss": 0.468,
+      "step": 3934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.160511363636364e-05,
+      "loss": 0.4942,
+      "step": 3935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1600378787878785e-05,
+      "loss": 0.5054,
+      "step": 3936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1595643939393945e-05,
+      "loss": 0.5155,
+      "step": 3937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.159090909090909e-05,
+      "loss": 0.4786,
+      "step": 3938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1586174242424246e-05,
+      "loss": 0.4986,
+      "step": 3939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.158143939393939e-05,
+      "loss": 0.4099,
+      "step": 3940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.157670454545455e-05,
+      "loss": 0.5383,
+      "step": 3941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.15719696969697e-05,
+      "loss": 0.4454,
+      "step": 3942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.156723484848485e-05,
+      "loss": 0.4901,
+      "step": 3943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.15625e-05,
+      "loss": 0.341,
+      "step": 3944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.155776515151515e-05,
+      "loss": 0.4901,
+      "step": 3945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.155303030303031e-05,
+      "loss": 0.5865,
+      "step": 3946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1548295454545456e-05,
+      "loss": 0.5078,
+      "step": 3947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.154356060606061e-05,
+      "loss": 0.6298,
+      "step": 3948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.153882575757576e-05,
+      "loss": 0.4935,
+      "step": 3949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.153409090909091e-05,
+      "loss": 0.4499,
+      "step": 3950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1529356060606064e-05,
+      "loss": 0.4684,
+      "step": 3951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.152462121212121e-05,
+      "loss": 0.4954,
+      "step": 3952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1519886363636365e-05,
+      "loss": 0.5489,
+      "step": 3953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.151515151515151e-05,
+      "loss": 0.4215,
+      "step": 3954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.151041666666667e-05,
+      "loss": 0.3854,
+      "step": 3955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.150568181818182e-05,
+      "loss": 0.4937,
+      "step": 3956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.150094696969697e-05,
+      "loss": 0.5615,
+      "step": 3957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.149621212121212e-05,
+      "loss": 0.371,
+      "step": 3958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1491477272727274e-05,
+      "loss": 0.4749,
+      "step": 3959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.148674242424243e-05,
+      "loss": 0.6348,
+      "step": 3960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1482007575757575e-05,
+      "loss": 0.4534,
+      "step": 3961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.147727272727273e-05,
+      "loss": 0.4268,
+      "step": 3962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1472537878787876e-05,
+      "loss": 0.4827,
+      "step": 3963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1467803030303036e-05,
+      "loss": 0.6882,
+      "step": 3964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.146306818181818e-05,
+      "loss": 0.4928,
+      "step": 3965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.145833333333334e-05,
+      "loss": 0.5208,
+      "step": 3966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1453598484848484e-05,
+      "loss": 0.4756,
+      "step": 3967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.144886363636364e-05,
+      "loss": 0.6085,
+      "step": 3968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.144412878787879e-05,
+      "loss": 0.4053,
+      "step": 3969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.143939393939394e-05,
+      "loss": 0.5676,
+      "step": 3970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.143465909090909e-05,
+      "loss": 0.4319,
+      "step": 3971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.142992424242424e-05,
+      "loss": 0.5038,
+      "step": 3972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.14251893939394e-05,
+      "loss": 0.4589,
+      "step": 3973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1420454545454547e-05,
+      "loss": 0.4401,
+      "step": 3974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.14157196969697e-05,
+      "loss": 0.4271,
+      "step": 3975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.141098484848485e-05,
+      "loss": 0.4278,
+      "step": 3976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.140625e-05,
+      "loss": 0.5426,
+      "step": 3977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1401515151515155e-05,
+      "loss": 0.4748,
+      "step": 3978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.13967803030303e-05,
+      "loss": 0.4178,
+      "step": 3979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1392045454545456e-05,
+      "loss": 0.4161,
+      "step": 3980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.13873106060606e-05,
+      "loss": 0.3737,
+      "step": 3981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.138257575757576e-05,
+      "loss": 0.5517,
+      "step": 3982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.137784090909091e-05,
+      "loss": 0.5152,
+      "step": 3983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1373106060606064e-05,
+      "loss": 0.4273,
+      "step": 3984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.136837121212121e-05,
+      "loss": 0.6216,
+      "step": 3985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1363636363636365e-05,
+      "loss": 0.4314,
+      "step": 3986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.135890151515152e-05,
+      "loss": 0.6914,
+      "step": 3987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1354166666666665e-05,
+      "loss": 0.447,
+      "step": 3988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.134943181818182e-05,
+      "loss": 0.5318,
+      "step": 3989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.134469696969697e-05,
+      "loss": 0.5983,
+      "step": 3990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.133996212121213e-05,
+      "loss": 0.5484,
+      "step": 3991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1335227272727274e-05,
+      "loss": 0.4881,
+      "step": 3992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.133049242424243e-05,
+      "loss": 0.5195,
+      "step": 3993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1325757575757574e-05,
+      "loss": 0.5247,
+      "step": 3994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.132102272727273e-05,
+      "loss": 0.4757,
+      "step": 3995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.131628787878788e-05,
+      "loss": 0.5207,
+      "step": 3996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.131155303030303e-05,
+      "loss": 0.5585,
+      "step": 3997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.130681818181818e-05,
+      "loss": 0.5211,
+      "step": 3998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1302083333333336e-05,
+      "loss": 0.4036,
+      "step": 3999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.129734848484849e-05,
+      "loss": 0.4707,
+      "step": 4000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.129261363636364e-05,
+      "loss": 0.4594,
+      "step": 4001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.128787878787879e-05,
+      "loss": 0.4911,
+      "step": 4002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.128314393939394e-05,
+      "loss": 0.444,
+      "step": 4003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.127840909090909e-05,
+      "loss": 0.5651,
+      "step": 4004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1273674242424245e-05,
+      "loss": 0.5356,
+      "step": 4005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.126893939393939e-05,
+      "loss": 0.4804,
+      "step": 4006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1264204545454546e-05,
+      "loss": 0.5442,
+      "step": 4007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.12594696969697e-05,
+      "loss": 0.4415,
+      "step": 4008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1254734848484854e-05,
+      "loss": 0.5732,
+      "step": 4009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.125e-05,
+      "loss": 0.466,
+      "step": 4010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1245265151515154e-05,
+      "loss": 0.396,
+      "step": 4011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.12405303030303e-05,
+      "loss": 0.3776,
+      "step": 4012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1235795454545455e-05,
+      "loss": 0.4202,
+      "step": 4013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.123106060606061e-05,
+      "loss": 0.5229,
+      "step": 4014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1226325757575756e-05,
+      "loss": 0.4955,
+      "step": 4015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.122159090909091e-05,
+      "loss": 0.4196,
+      "step": 4016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1216856060606063e-05,
+      "loss": 0.4773,
+      "step": 4017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.121212121212122e-05,
+      "loss": 0.6151,
+      "step": 4018
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1207386363636364e-05,
+      "loss": 0.5371,
+      "step": 4019
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.120265151515152e-05,
+      "loss": 0.431,
+      "step": 4020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1197916666666665e-05,
+      "loss": 0.5478,
+      "step": 4021
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.119318181818182e-05,
+      "loss": 0.3356,
+      "step": 4022
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.118844696969697e-05,
+      "loss": 0.5245,
+      "step": 4023
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.118371212121212e-05,
+      "loss": 0.4865,
+      "step": 4024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.117897727272727e-05,
+      "loss": 0.462,
+      "step": 4025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.117424242424243e-05,
+      "loss": 0.5702,
+      "step": 4026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.116950757575758e-05,
+      "loss": 0.4427,
+      "step": 4027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.116477272727273e-05,
+      "loss": 0.464,
+      "step": 4028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.116003787878788e-05,
+      "loss": 0.5101,
+      "step": 4029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.115530303030303e-05,
+      "loss": 0.5512,
+      "step": 4030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.115056818181818e-05,
+      "loss": 0.4183,
+      "step": 4031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1145833333333336e-05,
+      "loss": 0.4791,
+      "step": 4032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.114109848484848e-05,
+      "loss": 0.4845,
+      "step": 4033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.113636363636364e-05,
+      "loss": 0.5549,
+      "step": 4034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.113162878787879e-05,
+      "loss": 0.3949,
+      "step": 4035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1126893939393944e-05,
+      "loss": 0.5009,
+      "step": 4036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.112215909090909e-05,
+      "loss": 0.4456,
+      "step": 4037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1117424242424245e-05,
+      "loss": 0.4536,
+      "step": 4038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.11126893939394e-05,
+      "loss": 0.5172,
+      "step": 4039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1107954545454546e-05,
+      "loss": 0.5519,
+      "step": 4040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.11032196969697e-05,
+      "loss": 0.4819,
+      "step": 4041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1098484848484846e-05,
+      "loss": 0.4143,
+      "step": 4042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.109375e-05,
+      "loss": 0.467,
+      "step": 4043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1089015151515154e-05,
+      "loss": 0.5614,
+      "step": 4044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.108428030303031e-05,
+      "loss": 0.5111,
+      "step": 4045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1079545454545455e-05,
+      "loss": 0.3843,
+      "step": 4046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.107481060606061e-05,
+      "loss": 0.413,
+      "step": 4047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.107007575757576e-05,
+      "loss": 0.5379,
+      "step": 4048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.106534090909091e-05,
+      "loss": 0.5075,
+      "step": 4049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.106060606060606e-05,
+      "loss": 0.4836,
+      "step": 4050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.105587121212121e-05,
+      "loss": 0.5634,
+      "step": 4051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1051136363636364e-05,
+      "loss": 0.4667,
+      "step": 4052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.104640151515152e-05,
+      "loss": 0.6048,
+      "step": 4053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.104166666666667e-05,
+      "loss": 0.4809,
+      "step": 4054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.103693181818182e-05,
+      "loss": 0.5563,
+      "step": 4055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.103219696969697e-05,
+      "loss": 0.4792,
+      "step": 4056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1027462121212126e-05,
+      "loss": 0.4601,
+      "step": 4057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.102272727272727e-05,
+      "loss": 0.4802,
+      "step": 4058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1017992424242427e-05,
+      "loss": 0.722,
+      "step": 4059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1013257575757574e-05,
+      "loss": 0.4626,
+      "step": 4060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.100852272727273e-05,
+      "loss": 0.4477,
+      "step": 4061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.100378787878788e-05,
+      "loss": 0.4984,
+      "step": 4062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0999053030303035e-05,
+      "loss": 0.5253,
+      "step": 4063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.099431818181818e-05,
+      "loss": 0.4516,
+      "step": 4064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0989583333333336e-05,
+      "loss": 0.6417,
+      "step": 4065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.098484848484849e-05,
+      "loss": 0.496,
+      "step": 4066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0980113636363636e-05,
+      "loss": 0.4671,
+      "step": 4067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.097537878787879e-05,
+      "loss": 0.4723,
+      "step": 4068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.097064393939394e-05,
+      "loss": 0.5446,
+      "step": 4069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.096590909090909e-05,
+      "loss": 0.5601,
+      "step": 4070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0961174242424245e-05,
+      "loss": 0.5739,
+      "step": 4071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09564393939394e-05,
+      "loss": 0.4861,
+      "step": 4072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0951704545454545e-05,
+      "loss": 0.6166,
+      "step": 4073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09469696969697e-05,
+      "loss": 0.4693,
+      "step": 4074
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.094223484848485e-05,
+      "loss": 0.5699,
+      "step": 4075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09375e-05,
+      "loss": 0.4359,
+      "step": 4076
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0932765151515154e-05,
+      "loss": 0.4442,
+      "step": 4077
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09280303030303e-05,
+      "loss": 0.5165,
+      "step": 4078
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0923295454545454e-05,
+      "loss": 0.5159,
+      "step": 4079
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.091856060606061e-05,
+      "loss": 0.5024,
+      "step": 4080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.091382575757576e-05,
+      "loss": 0.455,
+      "step": 4081
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.090909090909091e-05,
+      "loss": 0.4772,
+      "step": 4082
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.090435606060606e-05,
+      "loss": 0.4576,
+      "step": 4083
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0899621212121216e-05,
+      "loss": 0.4671,
+      "step": 4084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.089488636363636e-05,
+      "loss": 0.5451,
+      "step": 4085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.089015151515152e-05,
+      "loss": 0.4509,
+      "step": 4086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0885416666666664e-05,
+      "loss": 0.5537,
+      "step": 4087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.088068181818182e-05,
+      "loss": 0.3596,
+      "step": 4088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.087594696969697e-05,
+      "loss": 0.4872,
+      "step": 4089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0871212121212125e-05,
+      "loss": 0.5845,
+      "step": 4090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.086647727272727e-05,
+      "loss": 0.4269,
+      "step": 4091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0861742424242426e-05,
+      "loss": 0.4853,
+      "step": 4092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.085700757575758e-05,
+      "loss": 0.3929,
+      "step": 4093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.085227272727273e-05,
+      "loss": 0.5036,
+      "step": 4094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.084753787878788e-05,
+      "loss": 0.482,
+      "step": 4095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.084280303030303e-05,
+      "loss": 0.5195,
+      "step": 4096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.083806818181819e-05,
+      "loss": 0.4938,
+      "step": 4097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0833333333333335e-05,
+      "loss": 0.5437,
+      "step": 4098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.082859848484849e-05,
+      "loss": 0.4761,
+      "step": 4099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0823863636363636e-05,
+      "loss": 0.6041,
+      "step": 4100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.081912878787879e-05,
+      "loss": 0.4974,
+      "step": 4101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0814393939393943e-05,
+      "loss": 0.5093,
+      "step": 4102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.080965909090909e-05,
+      "loss": 0.3713,
+      "step": 4103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0804924242424244e-05,
+      "loss": 0.4588,
+      "step": 4104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.080018939393939e-05,
+      "loss": 0.4301,
+      "step": 4105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.079545454545455e-05,
+      "loss": 0.5422,
+      "step": 4106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.07907196969697e-05,
+      "loss": 0.4288,
+      "step": 4107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.078598484848485e-05,
+      "loss": 0.5199,
+      "step": 4108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.078125e-05,
+      "loss": 0.6395,
+      "step": 4109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.077651515151515e-05,
+      "loss": 0.4815,
+      "step": 4110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.077178030303031e-05,
+      "loss": 0.4611,
+      "step": 4111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0767045454545454e-05,
+      "loss": 0.4861,
+      "step": 4112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.076231060606061e-05,
+      "loss": 0.4355,
+      "step": 4113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0757575757575755e-05,
+      "loss": 0.4415,
+      "step": 4114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0752840909090915e-05,
+      "loss": 0.5353,
+      "step": 4115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.074810606060606e-05,
+      "loss": 0.4212,
+      "step": 4116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0743371212121216e-05,
+      "loss": 0.5029,
+      "step": 4117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.073863636363636e-05,
+      "loss": 0.3484,
+      "step": 4118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.073390151515152e-05,
+      "loss": 0.5479,
+      "step": 4119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.072916666666667e-05,
+      "loss": 0.4262,
+      "step": 4120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.072443181818182e-05,
+      "loss": 0.4358,
+      "step": 4121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.071969696969697e-05,
+      "loss": 0.5434,
+      "step": 4122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.071496212121212e-05,
+      "loss": 0.424,
+      "step": 4123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.071022727272728e-05,
+      "loss": 0.535,
+      "step": 4124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0705492424242426e-05,
+      "loss": 0.5488,
+      "step": 4125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.070075757575758e-05,
+      "loss": 0.4048,
+      "step": 4126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0696022727272726e-05,
+      "loss": 0.7135,
+      "step": 4127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.069128787878788e-05,
+      "loss": 0.605,
+      "step": 4128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0686553030303034e-05,
+      "loss": 0.4381,
+      "step": 4129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.068181818181818e-05,
+      "loss": 0.4123,
+      "step": 4130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0677083333333335e-05,
+      "loss": 0.5576,
+      "step": 4131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.067234848484848e-05,
+      "loss": 0.6694,
+      "step": 4132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.066761363636364e-05,
+      "loss": 0.6425,
+      "step": 4133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.066287878787879e-05,
+      "loss": 0.6873,
+      "step": 4134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.065814393939394e-05,
+      "loss": 0.3818,
+      "step": 4135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.065340909090909e-05,
+      "loss": 0.4536,
+      "step": 4136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0648674242424244e-05,
+      "loss": 0.4315,
+      "step": 4137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.06439393939394e-05,
+      "loss": 0.463,
+      "step": 4138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0639204545454545e-05,
+      "loss": 0.5282,
+      "step": 4139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.06344696969697e-05,
+      "loss": 0.4983,
+      "step": 4140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0629734848484845e-05,
+      "loss": 0.3876,
+      "step": 4141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0625000000000006e-05,
+      "loss": 0.6507,
+      "step": 4142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.062026515151515e-05,
+      "loss": 0.6311,
+      "step": 4143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0615530303030307e-05,
+      "loss": 0.3745,
+      "step": 4144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0610795454545454e-05,
+      "loss": 0.4578,
+      "step": 4145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.060606060606061e-05,
+      "loss": 0.5316,
+      "step": 4146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.060132575757576e-05,
+      "loss": 0.476,
+      "step": 4147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.059659090909091e-05,
+      "loss": 0.5414,
+      "step": 4148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.059185606060606e-05,
+      "loss": 0.4611,
+      "step": 4149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.058712121212121e-05,
+      "loss": 0.5231,
+      "step": 4150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.058238636363637e-05,
+      "loss": 0.4721,
+      "step": 4151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0577651515151516e-05,
+      "loss": 0.5015,
+      "step": 4152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.057291666666667e-05,
+      "loss": 0.5164,
+      "step": 4153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.056818181818182e-05,
+      "loss": 0.4703,
+      "step": 4154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.056344696969697e-05,
+      "loss": 0.5881,
+      "step": 4155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0558712121212125e-05,
+      "loss": 0.5477,
+      "step": 4156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.055397727272727e-05,
+      "loss": 0.5755,
+      "step": 4157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0549242424242425e-05,
+      "loss": 0.4308,
+      "step": 4158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.054450757575757e-05,
+      "loss": 0.4406,
+      "step": 4159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.053977272727273e-05,
+      "loss": 0.4569,
+      "step": 4160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.053503787878788e-05,
+      "loss": 0.4969,
+      "step": 4161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0530303030303034e-05,
+      "loss": 0.4786,
+      "step": 4162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.052556818181818e-05,
+      "loss": 0.5185,
+      "step": 4163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0520833333333334e-05,
+      "loss": 0.4081,
+      "step": 4164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0516098484848488e-05,
+      "loss": 0.4211,
+      "step": 4165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.051136363636364e-05,
+      "loss": 0.5978,
+      "step": 4166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.050662878787879e-05,
+      "loss": 0.3861,
+      "step": 4167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.050189393939394e-05,
+      "loss": 0.3884,
+      "step": 4168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0497159090909093e-05,
+      "loss": 0.4299,
+      "step": 4169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0492424242424243e-05,
+      "loss": 0.3977,
+      "step": 4170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0487689393939394e-05,
+      "loss": 0.5826,
+      "step": 4171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0482954545454544e-05,
+      "loss": 0.459,
+      "step": 4172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.04782196969697e-05,
+      "loss": 0.57,
+      "step": 4173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.047348484848485e-05,
+      "loss": 0.4916,
+      "step": 4174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0468750000000002e-05,
+      "loss": 0.456,
+      "step": 4175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0464015151515152e-05,
+      "loss": 0.4539,
+      "step": 4176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0459280303030303e-05,
+      "loss": 0.5906,
+      "step": 4177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0454545454545456e-05,
+      "loss": 0.4647,
+      "step": 4178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0449810606060607e-05,
+      "loss": 0.4305,
+      "step": 4179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0445075757575757e-05,
+      "loss": 0.4644,
+      "step": 4180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0440340909090908e-05,
+      "loss": 0.5841,
+      "step": 4181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0435606060606065e-05,
+      "loss": 0.4361,
+      "step": 4182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0430871212121215e-05,
+      "loss": 0.663,
+      "step": 4183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0426136363636366e-05,
+      "loss": 0.5879,
+      "step": 4184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0421401515151516e-05,
+      "loss": 0.456,
+      "step": 4185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0416666666666666e-05,
+      "loss": 0.4934,
+      "step": 4186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.041193181818182e-05,
+      "loss": 0.4187,
+      "step": 4187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.040719696969697e-05,
+      "loss": 0.3732,
+      "step": 4188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.040246212121212e-05,
+      "loss": 0.5651,
+      "step": 4189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.039772727272727e-05,
+      "loss": 0.426,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0392992424242428e-05,
+      "loss": 0.5111,
+      "step": 4191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.038825757575758e-05,
+      "loss": 0.5218,
+      "step": 4192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.038352272727273e-05,
+      "loss": 0.4932,
+      "step": 4193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.037878787878788e-05,
+      "loss": 0.5507,
+      "step": 4194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.037405303030303e-05,
+      "loss": 0.533,
+      "step": 4195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0369318181818184e-05,
+      "loss": 0.4239,
+      "step": 4196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0364583333333334e-05,
+      "loss": 0.5783,
+      "step": 4197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0359848484848484e-05,
+      "loss": 0.6165,
+      "step": 4198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0355113636363635e-05,
+      "loss": 0.4446,
+      "step": 4199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0350378787878792e-05,
+      "loss": 0.4826,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0345643939393942e-05,
+      "loss": 0.4743,
+      "step": 4201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0340909090909093e-05,
+      "loss": 0.461,
+      "step": 4202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0336174242424243e-05,
+      "loss": 0.4498,
+      "step": 4203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0331439393939397e-05,
+      "loss": 0.4662,
+      "step": 4204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0326704545454547e-05,
+      "loss": 0.4802,
+      "step": 4205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0321969696969697e-05,
+      "loss": 0.4685,
+      "step": 4206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0317234848484848e-05,
+      "loss": 0.4777,
+      "step": 4207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0312499999999998e-05,
+      "loss": 0.4399,
+      "step": 4208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0307765151515155e-05,
+      "loss": 0.5019,
+      "step": 4209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 0.4669,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0298295454545456e-05,
+      "loss": 0.4124,
+      "step": 4211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0293560606060606e-05,
+      "loss": 0.4178,
+      "step": 4212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.028882575757576e-05,
+      "loss": 0.5543,
+      "step": 4213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.028409090909091e-05,
+      "loss": 0.5067,
+      "step": 4214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.027935606060606e-05,
+      "loss": 0.4737,
+      "step": 4215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.027462121212121e-05,
+      "loss": 0.5991,
+      "step": 4216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0269886363636362e-05,
+      "loss": 0.463,
+      "step": 4217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.026515151515152e-05,
+      "loss": 0.4807,
+      "step": 4218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.026041666666667e-05,
+      "loss": 0.542,
+      "step": 4219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.025568181818182e-05,
+      "loss": 0.5117,
+      "step": 4220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.025094696969697e-05,
+      "loss": 0.5689,
+      "step": 4221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0246212121212124e-05,
+      "loss": 0.4582,
+      "step": 4222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0241477272727274e-05,
+      "loss": 0.5475,
+      "step": 4223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0236742424242424e-05,
+      "loss": 0.568,
+      "step": 4224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0232007575757575e-05,
+      "loss": 0.5027,
+      "step": 4225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0227272727272725e-05,
+      "loss": 0.4786,
+      "step": 4226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0222537878787882e-05,
+      "loss": 0.4914,
+      "step": 4227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0217803030303033e-05,
+      "loss": 0.4846,
+      "step": 4228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0213068181818183e-05,
+      "loss": 0.4809,
+      "step": 4229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0208333333333334e-05,
+      "loss": 0.5212,
+      "step": 4230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0203598484848487e-05,
+      "loss": 0.5137,
+      "step": 4231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0198863636363638e-05,
+      "loss": 0.5495,
+      "step": 4232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0194128787878788e-05,
+      "loss": 0.5603,
+      "step": 4233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.018939393939394e-05,
+      "loss": 0.4552,
+      "step": 4234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.018465909090909e-05,
+      "loss": 0.4687,
+      "step": 4235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0179924242424246e-05,
+      "loss": 0.5364,
+      "step": 4236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0175189393939396e-05,
+      "loss": 0.4473,
+      "step": 4237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0170454545454547e-05,
+      "loss": 0.4389,
+      "step": 4238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0165719696969697e-05,
+      "loss": 0.4884,
+      "step": 4239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.016098484848485e-05,
+      "loss": 0.4465,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.015625e-05,
+      "loss": 0.5471,
+      "step": 4241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.015151515151515e-05,
+      "loss": 0.54,
+      "step": 4242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0146780303030302e-05,
+      "loss": 0.4067,
+      "step": 4243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0142045454545452e-05,
+      "loss": 0.526,
+      "step": 4244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.013731060606061e-05,
+      "loss": 0.6455,
+      "step": 4245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.013257575757576e-05,
+      "loss": 0.4944,
+      "step": 4246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.012784090909091e-05,
+      "loss": 0.4354,
+      "step": 4247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.012310606060606e-05,
+      "loss": 0.4549,
+      "step": 4248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0118371212121214e-05,
+      "loss": 0.386,
+      "step": 4249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0113636363636365e-05,
+      "loss": 0.5037,
+      "step": 4250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0108901515151515e-05,
+      "loss": 0.5209,
+      "step": 4251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0104166666666665e-05,
+      "loss": 0.3781,
+      "step": 4252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0099431818181816e-05,
+      "loss": 0.4866,
+      "step": 4253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0094696969696973e-05,
+      "loss": 0.4094,
+      "step": 4254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0089962121212123e-05,
+      "loss": 0.5219,
+      "step": 4255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0085227272727274e-05,
+      "loss": 0.6249,
+      "step": 4256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0080492424242424e-05,
+      "loss": 0.4902,
+      "step": 4257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0075757575757578e-05,
+      "loss": 0.4586,
+      "step": 4258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0071022727272728e-05,
+      "loss": 0.5096,
+      "step": 4259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.006628787878788e-05,
+      "loss": 0.5441,
+      "step": 4260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.006155303030303e-05,
+      "loss": 0.4084,
+      "step": 4261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0056818181818186e-05,
+      "loss": 0.5582,
+      "step": 4262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0052083333333336e-05,
+      "loss": 0.4761,
+      "step": 4263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0047348484848487e-05,
+      "loss": 0.5572,
+      "step": 4264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0042613636363637e-05,
+      "loss": 0.5434,
+      "step": 4265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0037878787878788e-05,
+      "loss": 0.4134,
+      "step": 4266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.003314393939394e-05,
+      "loss": 0.5656,
+      "step": 4267
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0028409090909092e-05,
+      "loss": 0.4775,
+      "step": 4268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0023674242424242e-05,
+      "loss": 0.5067,
+      "step": 4269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0018939393939392e-05,
+      "loss": 0.5205,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.001420454545455e-05,
+      "loss": 0.4999,
+      "step": 4271
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.00094696969697e-05,
+      "loss": 0.4491,
+      "step": 4272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.000473484848485e-05,
+      "loss": 0.5623,
+      "step": 4273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3e-05,
+      "loss": 0.5593,
+      "step": 4274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.999526515151515e-05,
+      "loss": 0.5221,
+      "step": 4275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9990530303030305e-05,
+      "loss": 0.566,
+      "step": 4276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9985795454545455e-05,
+      "loss": 0.4395,
+      "step": 4277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9981060606060606e-05,
+      "loss": 0.5196,
+      "step": 4278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9976325757575756e-05,
+      "loss": 0.4521,
+      "step": 4279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9971590909090913e-05,
+      "loss": 0.4888,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9966856060606064e-05,
+      "loss": 0.5198,
+      "step": 4281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9962121212121214e-05,
+      "loss": 0.4182,
+      "step": 4282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9957386363636364e-05,
+      "loss": 0.4359,
+      "step": 4283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9952651515151515e-05,
+      "loss": 0.4927,
+      "step": 4284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.994791666666667e-05,
+      "loss": 0.5877,
+      "step": 4285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.994318181818182e-05,
+      "loss": 0.4888,
+      "step": 4286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.993844696969697e-05,
+      "loss": 0.4787,
+      "step": 4287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.993371212121212e-05,
+      "loss": 0.4388,
+      "step": 4288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9928977272727277e-05,
+      "loss": 0.4749,
+      "step": 4289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9924242424242427e-05,
+      "loss": 0.3653,
+      "step": 4290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9919507575757577e-05,
+      "loss": 0.616,
+      "step": 4291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9914772727272728e-05,
+      "loss": 0.498,
+      "step": 4292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9910037878787878e-05,
+      "loss": 0.4632,
+      "step": 4293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9905303030303032e-05,
+      "loss": 0.456,
+      "step": 4294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9900568181818182e-05,
+      "loss": 0.4413,
+      "step": 4295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9895833333333333e-05,
+      "loss": 0.5292,
+      "step": 4296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9891098484848483e-05,
+      "loss": 0.3848,
+      "step": 4297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.988636363636364e-05,
+      "loss": 0.5353,
+      "step": 4298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.988162878787879e-05,
+      "loss": 0.3869,
+      "step": 4299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.987689393939394e-05,
+      "loss": 0.5096,
+      "step": 4300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.987215909090909e-05,
+      "loss": 0.3458,
+      "step": 4301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9867424242424242e-05,
+      "loss": 0.4571,
+      "step": 4302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9862689393939395e-05,
+      "loss": 0.5766,
+      "step": 4303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9857954545454546e-05,
+      "loss": 0.5049,
+      "step": 4304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9853219696969696e-05,
+      "loss": 0.4577,
+      "step": 4305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9848484848484847e-05,
+      "loss": 0.4741,
+      "step": 4306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9843750000000004e-05,
+      "loss": 0.4623,
+      "step": 4307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9839015151515154e-05,
+      "loss": 0.5954,
+      "step": 4308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9834280303030304e-05,
+      "loss": 0.3981,
+      "step": 4309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9829545454545455e-05,
+      "loss": 0.503,
+      "step": 4310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9824810606060605e-05,
+      "loss": 0.4295,
+      "step": 4311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.982007575757576e-05,
+      "loss": 0.4728,
+      "step": 4312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.981534090909091e-05,
+      "loss": 0.4818,
+      "step": 4313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.981060606060606e-05,
+      "loss": 0.4997,
+      "step": 4314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.980587121212121e-05,
+      "loss": 0.5661,
+      "step": 4315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9801136363636367e-05,
+      "loss": 0.5319,
+      "step": 4316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9796401515151518e-05,
+      "loss": 0.5366,
+      "step": 4317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9791666666666668e-05,
+      "loss": 0.4807,
+      "step": 4318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.978693181818182e-05,
+      "loss": 0.4967,
+      "step": 4319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9782196969696972e-05,
+      "loss": 0.4913,
+      "step": 4320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9777462121212123e-05,
+      "loss": 0.4629,
+      "step": 4321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9772727272727273e-05,
+      "loss": 0.5014,
+      "step": 4322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9767992424242423e-05,
+      "loss": 0.4367,
+      "step": 4323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9763257575757574e-05,
+      "loss": 0.4326,
+      "step": 4324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.975852272727273e-05,
+      "loss": 0.45,
+      "step": 4325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.975378787878788e-05,
+      "loss": 0.5038,
+      "step": 4326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.974905303030303e-05,
+      "loss": 0.5791,
+      "step": 4327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9744318181818182e-05,
+      "loss": 0.4442,
+      "step": 4328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9739583333333336e-05,
+      "loss": 0.3664,
+      "step": 4329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9734848484848486e-05,
+      "loss": 0.3994,
+      "step": 4330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9730113636363636e-05,
+      "loss": 0.5344,
+      "step": 4331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9725378787878787e-05,
+      "loss": 0.4047,
+      "step": 4332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9720643939393937e-05,
+      "loss": 0.4154,
+      "step": 4333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9715909090909094e-05,
+      "loss": 0.5741,
+      "step": 4334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9711174242424245e-05,
+      "loss": 0.4048,
+      "step": 4335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9706439393939395e-05,
+      "loss": 0.5535,
+      "step": 4336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9701704545454545e-05,
+      "loss": 0.4361,
+      "step": 4337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.96969696969697e-05,
+      "loss": 0.4801,
+      "step": 4338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.969223484848485e-05,
+      "loss": 0.5064,
+      "step": 4339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.96875e-05,
+      "loss": 0.5319,
+      "step": 4340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968276515151515e-05,
+      "loss": 0.3263,
+      "step": 4341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.96780303030303e-05,
+      "loss": 0.5105,
+      "step": 4342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9673295454545458e-05,
+      "loss": 0.3915,
+      "step": 4343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9668560606060608e-05,
+      "loss": 0.3917,
+      "step": 4344
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.966382575757576e-05,
+      "loss": 0.4532,
+      "step": 4345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.965909090909091e-05,
+      "loss": 0.4341,
+      "step": 4346
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9654356060606066e-05,
+      "loss": 0.5123,
+      "step": 4347
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9649621212121213e-05,
+      "loss": 0.4836,
+      "step": 4348
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9644886363636363e-05,
+      "loss": 0.4583,
+      "step": 4349
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9640151515151514e-05,
+      "loss": 0.5244,
+      "step": 4350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9635416666666664e-05,
+      "loss": 0.6549,
+      "step": 4351
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.963068181818182e-05,
+      "loss": 0.6103,
+      "step": 4352
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9625946969696972e-05,
+      "loss": 0.5071,
+      "step": 4353
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9621212121212122e-05,
+      "loss": 0.4758,
+      "step": 4354
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9616477272727272e-05,
+      "loss": 0.5088,
+      "step": 4355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.961174242424243e-05,
+      "loss": 0.5442,
+      "step": 4356
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9607007575757577e-05,
+      "loss": 0.6593,
+      "step": 4357
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9602272727272727e-05,
+      "loss": 0.5056,
+      "step": 4358
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9597537878787877e-05,
+      "loss": 0.4937,
+      "step": 4359
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9592803030303028e-05,
+      "loss": 0.5168,
+      "step": 4360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9588068181818185e-05,
+      "loss": 0.4984,
+      "step": 4361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9583333333333335e-05,
+      "loss": 0.4874,
+      "step": 4362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9578598484848486e-05,
+      "loss": 0.4254,
+      "step": 4363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9573863636363636e-05,
+      "loss": 0.4905,
+      "step": 4364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9569128787878793e-05,
+      "loss": 0.5376,
+      "step": 4365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.956439393939394e-05,
+      "loss": 0.3964,
+      "step": 4366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.955965909090909e-05,
+      "loss": 0.4785,
+      "step": 4367
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.955492424242424e-05,
+      "loss": 0.4893,
+      "step": 4368
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9550189393939398e-05,
+      "loss": 0.522,
+      "step": 4369
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.954545454545455e-05,
+      "loss": 0.4475,
+      "step": 4370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.95407196969697e-05,
+      "loss": 0.4534,
+      "step": 4371
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.953598484848485e-05,
+      "loss": 0.4206,
+      "step": 4372
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.953125e-05,
+      "loss": 0.5435,
+      "step": 4373
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9526515151515157e-05,
+      "loss": 0.4833,
+      "step": 4374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9521780303030304e-05,
+      "loss": 0.4879,
+      "step": 4375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9517045454545454e-05,
+      "loss": 0.496,
+      "step": 4376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9512310606060604e-05,
+      "loss": 0.438,
+      "step": 4377
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.950757575757576e-05,
+      "loss": 0.5735,
+      "step": 4378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9502840909090912e-05,
+      "loss": 0.4257,
+      "step": 4379
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9498106060606062e-05,
+      "loss": 0.4022,
+      "step": 4380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9493371212121213e-05,
+      "loss": 0.366,
+      "step": 4381
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9488636363636363e-05,
+      "loss": 0.4859,
+      "step": 4382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.948390151515152e-05,
+      "loss": 0.4534,
+      "step": 4383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9479166666666667e-05,
+      "loss": 0.4845,
+      "step": 4384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9474431818181818e-05,
+      "loss": 0.5815,
+      "step": 4385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9469696969696968e-05,
+      "loss": 0.5198,
+      "step": 4386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9464962121212125e-05,
+      "loss": 0.4938,
+      "step": 4387
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9460227272727275e-05,
+      "loss": 0.5136,
+      "step": 4388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9455492424242426e-05,
+      "loss": 0.4789,
+      "step": 4389
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9450757575757576e-05,
+      "loss": 0.4957,
+      "step": 4390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9446022727272727e-05,
+      "loss": 0.412,
+      "step": 4391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9441287878787884e-05,
+      "loss": 0.5335,
+      "step": 4392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.943655303030303e-05,
+      "loss": 0.4409,
+      "step": 4393
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.943181818181818e-05,
+      "loss": 0.5287,
+      "step": 4394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.942708333333333e-05,
+      "loss": 0.5807,
+      "step": 4395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.942234848484849e-05,
+      "loss": 0.4165,
+      "step": 4396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.941761363636364e-05,
+      "loss": 0.4475,
+      "step": 4397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.941287878787879e-05,
+      "loss": 0.4422,
+      "step": 4398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.940814393939394e-05,
+      "loss": 0.5549,
+      "step": 4399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.940340909090909e-05,
+      "loss": 0.4482,
+      "step": 4400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9398674242424247e-05,
+      "loss": 0.4639,
+      "step": 4401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9393939393939394e-05,
+      "loss": 0.4901,
+      "step": 4402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9389204545454545e-05,
+      "loss": 0.5806,
+      "step": 4403
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9384469696969695e-05,
+      "loss": 0.6082,
+      "step": 4404
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9379734848484852e-05,
+      "loss": 0.5226,
+      "step": 4405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9375000000000003e-05,
+      "loss": 0.541,
+      "step": 4406
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9370265151515153e-05,
+      "loss": 0.4108,
+      "step": 4407
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9365530303030303e-05,
+      "loss": 0.5186,
+      "step": 4408
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9360795454545454e-05,
+      "loss": 0.4543,
+      "step": 4409
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.935606060606061e-05,
+      "loss": 0.5238,
+      "step": 4410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9351325757575758e-05,
+      "loss": 0.5006,
+      "step": 4411
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9346590909090908e-05,
+      "loss": 0.6731,
+      "step": 4412
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.934185606060606e-05,
+      "loss": 0.5127,
+      "step": 4413
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9337121212121216e-05,
+      "loss": 0.485,
+      "step": 4414
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9332386363636366e-05,
+      "loss": 0.514,
+      "step": 4415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9327651515151516e-05,
+      "loss": 0.4118,
+      "step": 4416
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9322916666666667e-05,
+      "loss": 0.4794,
+      "step": 4417
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9318181818181817e-05,
+      "loss": 0.4509,
+      "step": 4418
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9313446969696974e-05,
+      "loss": 0.4702,
+      "step": 4419
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.930871212121212e-05,
+      "loss": 0.5644,
+      "step": 4420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.930397727272727e-05,
+      "loss": 0.4535,
+      "step": 4421
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9299242424242422e-05,
+      "loss": 0.4817,
+      "step": 4422
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.929450757575758e-05,
+      "loss": 0.5674,
+      "step": 4423
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.928977272727273e-05,
+      "loss": 0.4308,
+      "step": 4424
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.928503787878788e-05,
+      "loss": 0.4101,
+      "step": 4425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.928030303030303e-05,
+      "loss": 0.4874,
+      "step": 4426
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9275568181818187e-05,
+      "loss": 0.4432,
+      "step": 4427
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9270833333333338e-05,
+      "loss": 0.4322,
+      "step": 4428
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9266098484848485e-05,
+      "loss": 0.4777,
+      "step": 4429
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9261363636363635e-05,
+      "loss": 0.6065,
+      "step": 4430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9256628787878786e-05,
+      "loss": 0.4546,
+      "step": 4431
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9251893939393943e-05,
+      "loss": 0.4199,
+      "step": 4432
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9247159090909093e-05,
+      "loss": 0.4944,
+      "step": 4433
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9242424242424243e-05,
+      "loss": 0.4877,
+      "step": 4434
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9237689393939394e-05,
+      "loss": 0.4774,
+      "step": 4435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.923295454545455e-05,
+      "loss": 0.5502,
+      "step": 4436
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.92282196969697e-05,
+      "loss": 0.4537,
+      "step": 4437
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.922348484848485e-05,
+      "loss": 0.512,
+      "step": 4438
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.921875e-05,
+      "loss": 0.4762,
+      "step": 4439
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.921401515151515e-05,
+      "loss": 0.563,
+      "step": 4440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9209280303030306e-05,
+      "loss": 0.5623,
+      "step": 4441
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9204545454545457e-05,
+      "loss": 0.491,
+      "step": 4442
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9199810606060607e-05,
+      "loss": 0.5262,
+      "step": 4443
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9195075757575757e-05,
+      "loss": 0.4852,
+      "step": 4444
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9190340909090915e-05,
+      "loss": 0.4124,
+      "step": 4445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9185606060606065e-05,
+      "loss": 0.5387,
+      "step": 4446
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9180871212121212e-05,
+      "loss": 0.3792,
+      "step": 4447
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9176136363636362e-05,
+      "loss": 0.5058,
+      "step": 4448
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9171401515151513e-05,
+      "loss": 0.454,
+      "step": 4449
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.4976,
+      "step": 4450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.916193181818182e-05,
+      "loss": 0.4433,
+      "step": 4451
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.915719696969697e-05,
+      "loss": 0.4749,
+      "step": 4452
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.915246212121212e-05,
+      "loss": 0.49,
+      "step": 4453
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9147727272727278e-05,
+      "loss": 0.3851,
+      "step": 4454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.914299242424243e-05,
+      "loss": 0.5562,
+      "step": 4455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.913825757575758e-05,
+      "loss": 0.7761,
+      "step": 4456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9133522727272726e-05,
+      "loss": 0.4534,
+      "step": 4457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9128787878787876e-05,
+      "loss": 0.4361,
+      "step": 4458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9124053030303033e-05,
+      "loss": 0.5844,
+      "step": 4459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9119318181818184e-05,
+      "loss": 0.5763,
+      "step": 4460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9114583333333334e-05,
+      "loss": 0.651,
+      "step": 4461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9109848484848484e-05,
+      "loss": 0.5093,
+      "step": 4462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.910511363636364e-05,
+      "loss": 0.4914,
+      "step": 4463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9100378787878792e-05,
+      "loss": 0.4199,
+      "step": 4464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9095643939393942e-05,
+      "loss": 0.4509,
+      "step": 4465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.909090909090909e-05,
+      "loss": 0.4224,
+      "step": 4466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.908617424242424e-05,
+      "loss": 0.4263,
+      "step": 4467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9081439393939397e-05,
+      "loss": 0.5127,
+      "step": 4468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9076704545454547e-05,
+      "loss": 0.4274,
+      "step": 4469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9071969696969698e-05,
+      "loss": 0.4429,
+      "step": 4470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9067234848484848e-05,
+      "loss": 0.5937,
+      "step": 4471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9062500000000005e-05,
+      "loss": 0.5903,
+      "step": 4472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9057765151515155e-05,
+      "loss": 0.4806,
+      "step": 4473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9053030303030306e-05,
+      "loss": 0.4478,
+      "step": 4474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9048295454545453e-05,
+      "loss": 0.5006,
+      "step": 4475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9043560606060603e-05,
+      "loss": 0.4149,
+      "step": 4476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.903882575757576e-05,
+      "loss": 0.5738,
+      "step": 4477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.903409090909091e-05,
+      "loss": 0.4813,
+      "step": 4478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.902935606060606e-05,
+      "loss": 0.4171,
+      "step": 4479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.902462121212121e-05,
+      "loss": 0.391,
+      "step": 4480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.901988636363637e-05,
+      "loss": 0.4733,
+      "step": 4481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.901515151515152e-05,
+      "loss": 0.5054,
+      "step": 4482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.901041666666667e-05,
+      "loss": 0.3753,
+      "step": 4483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9005681818181816e-05,
+      "loss": 0.567,
+      "step": 4484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9000946969696973e-05,
+      "loss": 0.5089,
+      "step": 4485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8996212121212124e-05,
+      "loss": 0.4372,
+      "step": 4486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8991477272727274e-05,
+      "loss": 0.5179,
+      "step": 4487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8986742424242425e-05,
+      "loss": 0.5751,
+      "step": 4488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8982007575757575e-05,
+      "loss": 0.3762,
+      "step": 4489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8977272727272732e-05,
+      "loss": 0.4458,
+      "step": 4490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8972537878787882e-05,
+      "loss": 0.5272,
+      "step": 4491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8967803030303033e-05,
+      "loss": 0.49,
+      "step": 4492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.896306818181818e-05,
+      "loss": 0.5094,
+      "step": 4493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8958333333333337e-05,
+      "loss": 0.4827,
+      "step": 4494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8953598484848487e-05,
+      "loss": 0.6704,
+      "step": 4495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8948863636363638e-05,
+      "loss": 0.3847,
+      "step": 4496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8944128787878788e-05,
+      "loss": 0.4719,
+      "step": 4497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.893939393939394e-05,
+      "loss": 0.5078,
+      "step": 4498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8934659090909096e-05,
+      "loss": 0.4858,
+      "step": 4499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8929924242424246e-05,
+      "loss": 0.5489,
+      "step": 4500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8925189393939396e-05,
+      "loss": 0.5469,
+      "step": 4501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8920454545454543e-05,
+      "loss": 0.4808,
+      "step": 4502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.89157196969697e-05,
+      "loss": 0.5251,
+      "step": 4503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.891098484848485e-05,
+      "loss": 0.4339,
+      "step": 4504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.890625e-05,
+      "loss": 0.4407,
+      "step": 4505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.890151515151515e-05,
+      "loss": 0.4581,
+      "step": 4506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8896780303030302e-05,
+      "loss": 0.4873,
+      "step": 4507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.889204545454546e-05,
+      "loss": 0.5201,
+      "step": 4508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.888731060606061e-05,
+      "loss": 0.5305,
+      "step": 4509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.888257575757576e-05,
+      "loss": 0.4352,
+      "step": 4510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8877840909090907e-05,
+      "loss": 0.384,
+      "step": 4511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8873106060606064e-05,
+      "loss": 0.5136,
+      "step": 4512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8868371212121214e-05,
+      "loss": 0.5587,
+      "step": 4513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8863636363636365e-05,
+      "loss": 0.584,
+      "step": 4514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8858901515151515e-05,
+      "loss": 0.4558,
+      "step": 4515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8854166666666666e-05,
+      "loss": 0.5135,
+      "step": 4516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8849431818181823e-05,
+      "loss": 0.4477,
+      "step": 4517
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8844696969696973e-05,
+      "loss": 0.4176,
+      "step": 4518
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8839962121212123e-05,
+      "loss": 0.4758,
+      "step": 4519
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.883522727272727e-05,
+      "loss": 0.4471,
+      "step": 4520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8830492424242428e-05,
+      "loss": 0.3669,
+      "step": 4521
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8825757575757578e-05,
+      "loss": 0.6101,
+      "step": 4522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.882102272727273e-05,
+      "loss": 0.6412,
+      "step": 4523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.881628787878788e-05,
+      "loss": 0.4786,
+      "step": 4524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.881155303030303e-05,
+      "loss": 0.5052,
+      "step": 4525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8806818181818186e-05,
+      "loss": 0.5213,
+      "step": 4526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8802083333333337e-05,
+      "loss": 0.3589,
+      "step": 4527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8797348484848487e-05,
+      "loss": 0.3671,
+      "step": 4528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8792613636363634e-05,
+      "loss": 0.3646,
+      "step": 4529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.878787878787879e-05,
+      "loss": 0.5398,
+      "step": 4530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.878314393939394e-05,
+      "loss": 0.6119,
+      "step": 4531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8778409090909092e-05,
+      "loss": 0.4343,
+      "step": 4532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8773674242424242e-05,
+      "loss": 0.4425,
+      "step": 4533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.87689393939394e-05,
+      "loss": 0.5377,
+      "step": 4534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.876420454545455e-05,
+      "loss": 0.4383,
+      "step": 4535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.87594696969697e-05,
+      "loss": 0.4176,
+      "step": 4536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.875473484848485e-05,
+      "loss": 0.4063,
+      "step": 4537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8749999999999997e-05,
+      "loss": 0.4068,
+      "step": 4538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8745265151515155e-05,
+      "loss": 0.4438,
+      "step": 4539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8740530303030305e-05,
+      "loss": 0.5693,
+      "step": 4540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8735795454545455e-05,
+      "loss": 0.4196,
+      "step": 4541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8731060606060606e-05,
+      "loss": 0.4465,
+      "step": 4542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8726325757575763e-05,
+      "loss": 0.4982,
+      "step": 4543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8721590909090913e-05,
+      "loss": 0.5105,
+      "step": 4544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8716856060606064e-05,
+      "loss": 0.4706,
+      "step": 4545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8712121212121214e-05,
+      "loss": 0.5362,
+      "step": 4546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.870738636363636e-05,
+      "loss": 0.6187,
+      "step": 4547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8702651515151518e-05,
+      "loss": 0.4878,
+      "step": 4548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.869791666666667e-05,
+      "loss": 0.5077,
+      "step": 4549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.869318181818182e-05,
+      "loss": 0.5155,
+      "step": 4550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.868844696969697e-05,
+      "loss": 0.3968,
+      "step": 4551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8683712121212126e-05,
+      "loss": 0.5236,
+      "step": 4552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8678977272727277e-05,
+      "loss": 0.4693,
+      "step": 4553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8674242424242427e-05,
+      "loss": 0.6198,
+      "step": 4554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8669507575757578e-05,
+      "loss": 0.5832,
+      "step": 4555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8664772727272725e-05,
+      "loss": 0.526,
+      "step": 4556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.866003787878788e-05,
+      "loss": 0.4071,
+      "step": 4557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8655303030303032e-05,
+      "loss": 0.5468,
+      "step": 4558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8650568181818182e-05,
+      "loss": 0.466,
+      "step": 4559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8645833333333333e-05,
+      "loss": 0.5238,
+      "step": 4560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.864109848484849e-05,
+      "loss": 0.5251,
+      "step": 4561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.863636363636364e-05,
+      "loss": 0.4483,
+      "step": 4562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.863162878787879e-05,
+      "loss": 0.4543,
+      "step": 4563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.862689393939394e-05,
+      "loss": 0.4788,
+      "step": 4564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.862215909090909e-05,
+      "loss": 0.5102,
+      "step": 4565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8617424242424245e-05,
+      "loss": 0.5482,
+      "step": 4566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8612689393939396e-05,
+      "loss": 0.4829,
+      "step": 4567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8607954545454546e-05,
+      "loss": 0.5012,
+      "step": 4568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8603219696969696e-05,
+      "loss": 0.5646,
+      "step": 4569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8598484848484853e-05,
+      "loss": 0.5198,
+      "step": 4570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8593750000000004e-05,
+      "loss": 0.5217,
+      "step": 4571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8589015151515154e-05,
+      "loss": 0.4556,
+      "step": 4572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8584280303030305e-05,
+      "loss": 0.5984,
+      "step": 4573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8579545454545455e-05,
+      "loss": 0.4192,
+      "step": 4574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.857481060606061e-05,
+      "loss": 0.4869,
+      "step": 4575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.857007575757576e-05,
+      "loss": 0.531,
+      "step": 4576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.856534090909091e-05,
+      "loss": 0.5403,
+      "step": 4577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.856060606060606e-05,
+      "loss": 0.484,
+      "step": 4578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8555871212121217e-05,
+      "loss": 0.5031,
+      "step": 4579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8551136363636367e-05,
+      "loss": 0.5514,
+      "step": 4580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8546401515151518e-05,
+      "loss": 0.5076,
+      "step": 4581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8541666666666668e-05,
+      "loss": 0.486,
+      "step": 4582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.853693181818182e-05,
+      "loss": 0.5432,
+      "step": 4583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8532196969696972e-05,
+      "loss": 0.4802,
+      "step": 4584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8527462121212123e-05,
+      "loss": 0.616,
+      "step": 4585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8522727272727273e-05,
+      "loss": 0.5267,
+      "step": 4586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8517992424242423e-05,
+      "loss": 0.4374,
+      "step": 4587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.851325757575758e-05,
+      "loss": 0.5001,
+      "step": 4588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.850852272727273e-05,
+      "loss": 0.4681,
+      "step": 4589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.850378787878788e-05,
+      "loss": 0.5175,
+      "step": 4590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.849905303030303e-05,
+      "loss": 0.511,
+      "step": 4591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8494318181818185e-05,
+      "loss": 0.5055,
+      "step": 4592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8489583333333336e-05,
+      "loss": 0.4624,
+      "step": 4593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8484848484848486e-05,
+      "loss": 0.5149,
+      "step": 4594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8480113636363637e-05,
+      "loss": 0.5087,
+      "step": 4595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8475378787878787e-05,
+      "loss": 0.5036,
+      "step": 4596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8470643939393944e-05,
+      "loss": 0.3988,
+      "step": 4597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8465909090909094e-05,
+      "loss": 0.4457,
+      "step": 4598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8461174242424245e-05,
+      "loss": 0.5416,
+      "step": 4599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8456439393939395e-05,
+      "loss": 0.4572,
+      "step": 4600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.845170454545455e-05,
+      "loss": 0.3819,
+      "step": 4601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.84469696969697e-05,
+      "loss": 0.3754,
+      "step": 4602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.844223484848485e-05,
+      "loss": 0.3612,
+      "step": 4603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.84375e-05,
+      "loss": 0.4558,
+      "step": 4604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.843276515151515e-05,
+      "loss": 0.4597,
+      "step": 4605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8428030303030308e-05,
+      "loss": 0.4666,
+      "step": 4606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8423295454545458e-05,
+      "loss": 0.4271,
+      "step": 4607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.841856060606061e-05,
+      "loss": 0.4238,
+      "step": 4608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.841382575757576e-05,
+      "loss": 0.3787,
+      "step": 4609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8409090909090912e-05,
+      "loss": 0.4441,
+      "step": 4610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8404356060606063e-05,
+      "loss": 0.5258,
+      "step": 4611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8399621212121213e-05,
+      "loss": 0.3269,
+      "step": 4612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8394886363636364e-05,
+      "loss": 0.4397,
+      "step": 4613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8390151515151514e-05,
+      "loss": 0.4571,
+      "step": 4614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.838541666666667e-05,
+      "loss": 0.4408,
+      "step": 4615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.838068181818182e-05,
+      "loss": 0.4325,
+      "step": 4616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8375946969696972e-05,
+      "loss": 0.4656,
+      "step": 4617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8371212121212122e-05,
+      "loss": 0.446,
+      "step": 4618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8366477272727276e-05,
+      "loss": 0.5025,
+      "step": 4619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8361742424242426e-05,
+      "loss": 0.5797,
+      "step": 4620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8357007575757577e-05,
+      "loss": 0.6226,
+      "step": 4621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8352272727272727e-05,
+      "loss": 0.3498,
+      "step": 4622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8347537878787877e-05,
+      "loss": 0.5716,
+      "step": 4623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8342803030303035e-05,
+      "loss": 0.3811,
+      "step": 4624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8338068181818185e-05,
+      "loss": 0.5319,
+      "step": 4625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8333333333333335e-05,
+      "loss": 0.4609,
+      "step": 4626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8328598484848486e-05,
+      "loss": 0.3766,
+      "step": 4627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.832386363636364e-05,
+      "loss": 0.5044,
+      "step": 4628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.831912878787879e-05,
+      "loss": 0.4569,
+      "step": 4629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.831439393939394e-05,
+      "loss": 0.4694,
+      "step": 4630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.830965909090909e-05,
+      "loss": 0.4528,
+      "step": 4631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.830492424242424e-05,
+      "loss": 0.4346,
+      "step": 4632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8300189393939398e-05,
+      "loss": 0.5185,
+      "step": 4633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.829545454545455e-05,
+      "loss": 0.4696,
+      "step": 4634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.82907196969697e-05,
+      "loss": 0.4234,
+      "step": 4635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.828598484848485e-05,
+      "loss": 0.4799,
+      "step": 4636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8281250000000003e-05,
+      "loss": 0.5579,
+      "step": 4637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8276515151515153e-05,
+      "loss": 0.4964,
+      "step": 4638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8271780303030304e-05,
+      "loss": 0.3718,
+      "step": 4639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8267045454545454e-05,
+      "loss": 0.4777,
+      "step": 4640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8262310606060605e-05,
+      "loss": 0.5233,
+      "step": 4641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.825757575757576e-05,
+      "loss": 0.5244,
+      "step": 4642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8252840909090912e-05,
+      "loss": 0.5016,
+      "step": 4643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8248106060606062e-05,
+      "loss": 0.4257,
+      "step": 4644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8243371212121213e-05,
+      "loss": 0.4094,
+      "step": 4645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8238636363636367e-05,
+      "loss": 0.4181,
+      "step": 4646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8233901515151517e-05,
+      "loss": 0.4925,
+      "step": 4647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8229166666666667e-05,
+      "loss": 0.4253,
+      "step": 4648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8224431818181818e-05,
+      "loss": 0.4678,
+      "step": 4649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8219696969696975e-05,
+      "loss": 0.4805,
+      "step": 4650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8214962121212125e-05,
+      "loss": 0.5167,
+      "step": 4651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8210227272727276e-05,
+      "loss": 0.4901,
+      "step": 4652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8205492424242426e-05,
+      "loss": 0.7313,
+      "step": 4653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8200757575757576e-05,
+      "loss": 0.4355,
+      "step": 4654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.819602272727273e-05,
+      "loss": 0.4554,
+      "step": 4655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.819128787878788e-05,
+      "loss": 0.4763,
+      "step": 4656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.818655303030303e-05,
+      "loss": 0.5208,
+      "step": 4657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.818181818181818e-05,
+      "loss": 0.6359,
+      "step": 4658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.817708333333334e-05,
+      "loss": 0.5009,
+      "step": 4659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.817234848484849e-05,
+      "loss": 0.4863,
+      "step": 4660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.816761363636364e-05,
+      "loss": 0.5789,
+      "step": 4661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.816287878787879e-05,
+      "loss": 0.4629,
+      "step": 4662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.815814393939394e-05,
+      "loss": 0.6198,
+      "step": 4663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8153409090909094e-05,
+      "loss": 0.5452,
+      "step": 4664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8148674242424244e-05,
+      "loss": 0.4992,
+      "step": 4665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8143939393939394e-05,
+      "loss": 0.4666,
+      "step": 4666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8139204545454545e-05,
+      "loss": 0.4735,
+      "step": 4667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8134469696969702e-05,
+      "loss": 0.4595,
+      "step": 4668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8129734848484852e-05,
+      "loss": 0.4902,
+      "step": 4669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 0.5819,
+      "step": 4670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8120265151515153e-05,
+      "loss": 0.505,
+      "step": 4671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8115530303030303e-05,
+      "loss": 0.5409,
+      "step": 4672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8110795454545457e-05,
+      "loss": 0.5452,
+      "step": 4673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8106060606060607e-05,
+      "loss": 0.4524,
+      "step": 4674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8101325757575758e-05,
+      "loss": 0.5112,
+      "step": 4675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8096590909090908e-05,
+      "loss": 0.4438,
+      "step": 4676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8091856060606065e-05,
+      "loss": 0.5017,
+      "step": 4677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8087121212121216e-05,
+      "loss": 0.4405,
+      "step": 4678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8082386363636366e-05,
+      "loss": 0.4023,
+      "step": 4679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8077651515151517e-05,
+      "loss": 0.4032,
+      "step": 4680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8072916666666667e-05,
+      "loss": 0.4988,
+      "step": 4681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.806818181818182e-05,
+      "loss": 0.4791,
+      "step": 4682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.806344696969697e-05,
+      "loss": 0.5361,
+      "step": 4683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.805871212121212e-05,
+      "loss": 0.4167,
+      "step": 4684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8053977272727272e-05,
+      "loss": 0.4668,
+      "step": 4685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.804924242424243e-05,
+      "loss": 0.4495,
+      "step": 4686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.804450757575758e-05,
+      "loss": 0.4672,
+      "step": 4687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.803977272727273e-05,
+      "loss": 0.4924,
+      "step": 4688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.803503787878788e-05,
+      "loss": 0.4275,
+      "step": 4689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.803030303030303e-05,
+      "loss": 0.372,
+      "step": 4690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8025568181818184e-05,
+      "loss": 0.4462,
+      "step": 4691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8020833333333335e-05,
+      "loss": 0.3712,
+      "step": 4692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8016098484848485e-05,
+      "loss": 0.5754,
+      "step": 4693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8011363636363635e-05,
+      "loss": 0.4952,
+      "step": 4694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8006628787878792e-05,
+      "loss": 0.309,
+      "step": 4695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8001893939393943e-05,
+      "loss": 0.5234,
+      "step": 4696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7997159090909093e-05,
+      "loss": 0.4323,
+      "step": 4697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7992424242424244e-05,
+      "loss": 0.5376,
+      "step": 4698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7987689393939397e-05,
+      "loss": 0.5095,
+      "step": 4699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7982954545454548e-05,
+      "loss": 0.3902,
+      "step": 4700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7978219696969698e-05,
+      "loss": 0.4199,
+      "step": 4701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.797348484848485e-05,
+      "loss": 0.5568,
+      "step": 4702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.796875e-05,
+      "loss": 0.4453,
+      "step": 4703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7964015151515156e-05,
+      "loss": 0.3947,
+      "step": 4704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7959280303030306e-05,
+      "loss": 0.4917,
+      "step": 4705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7954545454545457e-05,
+      "loss": 0.6374,
+      "step": 4706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7949810606060607e-05,
+      "loss": 0.3969,
+      "step": 4707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.794507575757576e-05,
+      "loss": 0.5455,
+      "step": 4708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.794034090909091e-05,
+      "loss": 0.5374,
+      "step": 4709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.793560606060606e-05,
+      "loss": 0.455,
+      "step": 4710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7930871212121212e-05,
+      "loss": 0.3891,
+      "step": 4711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7926136363636362e-05,
+      "loss": 0.4512,
+      "step": 4712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.792140151515152e-05,
+      "loss": 0.4747,
+      "step": 4713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.791666666666667e-05,
+      "loss": 0.4535,
+      "step": 4714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.791193181818182e-05,
+      "loss": 0.5015,
+      "step": 4715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.790719696969697e-05,
+      "loss": 0.5638,
+      "step": 4716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7902462121212124e-05,
+      "loss": 0.5249,
+      "step": 4717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7897727272727275e-05,
+      "loss": 0.4691,
+      "step": 4718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7892992424242425e-05,
+      "loss": 0.5037,
+      "step": 4719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7888257575757575e-05,
+      "loss": 0.4588,
+      "step": 4720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7883522727272726e-05,
+      "loss": 0.6004,
+      "step": 4721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7878787878787883e-05,
+      "loss": 0.5146,
+      "step": 4722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7874053030303033e-05,
+      "loss": 0.5483,
+      "step": 4723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7869318181818184e-05,
+      "loss": 0.5202,
+      "step": 4724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7864583333333334e-05,
+      "loss": 0.6536,
+      "step": 4725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7859848484848488e-05,
+      "loss": 0.4701,
+      "step": 4726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7855113636363638e-05,
+      "loss": 0.5883,
+      "step": 4727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.785037878787879e-05,
+      "loss": 0.4032,
+      "step": 4728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.784564393939394e-05,
+      "loss": 0.458,
+      "step": 4729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.784090909090909e-05,
+      "loss": 0.4252,
+      "step": 4730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7836174242424247e-05,
+      "loss": 0.4526,
+      "step": 4731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7831439393939397e-05,
+      "loss": 0.4594,
+      "step": 4732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7826704545454547e-05,
+      "loss": 0.4376,
+      "step": 4733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7821969696969698e-05,
+      "loss": 0.3883,
+      "step": 4734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.781723484848485e-05,
+      "loss": 0.5301,
+      "step": 4735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7812500000000002e-05,
+      "loss": 0.5138,
+      "step": 4736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7807765151515152e-05,
+      "loss": 0.5249,
+      "step": 4737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7803030303030303e-05,
+      "loss": 0.4835,
+      "step": 4738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7798295454545453e-05,
+      "loss": 0.5185,
+      "step": 4739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.779356060606061e-05,
+      "loss": 0.4574,
+      "step": 4740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.778882575757576e-05,
+      "loss": 0.4585,
+      "step": 4741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.778409090909091e-05,
+      "loss": 0.5608,
+      "step": 4742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.777935606060606e-05,
+      "loss": 0.4709,
+      "step": 4743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7774621212121215e-05,
+      "loss": 0.6284,
+      "step": 4744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7769886363636365e-05,
+      "loss": 0.5327,
+      "step": 4745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7765151515151516e-05,
+      "loss": 0.4688,
+      "step": 4746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7760416666666666e-05,
+      "loss": 0.4843,
+      "step": 4747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7755681818181816e-05,
+      "loss": 0.4656,
+      "step": 4748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7750946969696974e-05,
+      "loss": 0.5054,
+      "step": 4749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7746212121212124e-05,
+      "loss": 0.5624,
+      "step": 4750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7741477272727274e-05,
+      "loss": 0.4771,
+      "step": 4751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7736742424242425e-05,
+      "loss": 0.4,
+      "step": 4752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.773200757575758e-05,
+      "loss": 0.5717,
+      "step": 4753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.772727272727273e-05,
+      "loss": 0.47,
+      "step": 4754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.772253787878788e-05,
+      "loss": 0.4193,
+      "step": 4755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.771780303030303e-05,
+      "loss": 0.4655,
+      "step": 4756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7713068181818187e-05,
+      "loss": 0.4712,
+      "step": 4757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7708333333333337e-05,
+      "loss": 0.4366,
+      "step": 4758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7703598484848487e-05,
+      "loss": 0.5575,
+      "step": 4759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7698863636363638e-05,
+      "loss": 0.5144,
+      "step": 4760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7694128787878788e-05,
+      "loss": 0.409,
+      "step": 4761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7689393939393942e-05,
+      "loss": 0.4756,
+      "step": 4762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7684659090909092e-05,
+      "loss": 0.4333,
+      "step": 4763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7679924242424243e-05,
+      "loss": 0.4829,
+      "step": 4764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7675189393939393e-05,
+      "loss": 0.6419,
+      "step": 4765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.767045454545455e-05,
+      "loss": 0.6825,
+      "step": 4766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.76657196969697e-05,
+      "loss": 0.472,
+      "step": 4767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.766098484848485e-05,
+      "loss": 0.4964,
+      "step": 4768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.765625e-05,
+      "loss": 0.4471,
+      "step": 4769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7651515151515152e-05,
+      "loss": 0.4416,
+      "step": 4770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7646780303030306e-05,
+      "loss": 0.5241,
+      "step": 4771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7642045454545456e-05,
+      "loss": 0.3162,
+      "step": 4772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7637310606060606e-05,
+      "loss": 0.4651,
+      "step": 4773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7632575757575757e-05,
+      "loss": 0.3734,
+      "step": 4774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7627840909090914e-05,
+      "loss": 0.4525,
+      "step": 4775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7623106060606064e-05,
+      "loss": 0.4495,
+      "step": 4776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7618371212121215e-05,
+      "loss": 0.4285,
+      "step": 4777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7613636363636365e-05,
+      "loss": 0.4529,
+      "step": 4778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7608901515151515e-05,
+      "loss": 0.5535,
+      "step": 4779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.760416666666667e-05,
+      "loss": 0.4878,
+      "step": 4780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.759943181818182e-05,
+      "loss": 0.4986,
+      "step": 4781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.759469696969697e-05,
+      "loss": 0.5154,
+      "step": 4782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.758996212121212e-05,
+      "loss": 0.467,
+      "step": 4783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7585227272727277e-05,
+      "loss": 0.4173,
+      "step": 4784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7580492424242428e-05,
+      "loss": 0.3028,
+      "step": 4785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7575757575757578e-05,
+      "loss": 0.4929,
+      "step": 4786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.757102272727273e-05,
+      "loss": 0.4583,
+      "step": 4787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.756628787878788e-05,
+      "loss": 0.4206,
+      "step": 4788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7561553030303033e-05,
+      "loss": 0.5448,
+      "step": 4789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7556818181818183e-05,
+      "loss": 0.5404,
+      "step": 4790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7552083333333333e-05,
+      "loss": 0.5833,
+      "step": 4791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7547348484848484e-05,
+      "loss": 0.3345,
+      "step": 4792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.754261363636364e-05,
+      "loss": 0.4459,
+      "step": 4793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.753787878787879e-05,
+      "loss": 0.4176,
+      "step": 4794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.753314393939394e-05,
+      "loss": 0.5831,
+      "step": 4795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7528409090909092e-05,
+      "loss": 0.4178,
+      "step": 4796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7523674242424242e-05,
+      "loss": 0.4256,
+      "step": 4797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7518939393939396e-05,
+      "loss": 0.4198,
+      "step": 4798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7514204545454546e-05,
+      "loss": 0.408,
+      "step": 4799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7509469696969697e-05,
+      "loss": 0.5038,
+      "step": 4800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7504734848484847e-05,
+      "loss": 0.6038,
+      "step": 4801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.5709,
+      "step": 4802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7495265151515155e-05,
+      "loss": 0.5172,
+      "step": 4803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7490530303030305e-05,
+      "loss": 0.5023,
+      "step": 4804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7485795454545455e-05,
+      "loss": 0.4469,
+      "step": 4805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7481060606060606e-05,
+      "loss": 0.4076,
+      "step": 4806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.747632575757576e-05,
+      "loss": 0.3826,
+      "step": 4807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.747159090909091e-05,
+      "loss": 0.401,
+      "step": 4808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.746685606060606e-05,
+      "loss": 0.3389,
+      "step": 4809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.746212121212121e-05,
+      "loss": 0.4973,
+      "step": 4810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7457386363636368e-05,
+      "loss": 0.4779,
+      "step": 4811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7452651515151518e-05,
+      "loss": 0.5717,
+      "step": 4812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.744791666666667e-05,
+      "loss": 0.4339,
+      "step": 4813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.744318181818182e-05,
+      "loss": 0.6301,
+      "step": 4814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7438446969696973e-05,
+      "loss": 0.4057,
+      "step": 4815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7433712121212123e-05,
+      "loss": 0.4135,
+      "step": 4816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7428977272727274e-05,
+      "loss": 0.4636,
+      "step": 4817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7424242424242424e-05,
+      "loss": 0.5565,
+      "step": 4818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7419507575757574e-05,
+      "loss": 0.4318,
+      "step": 4819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.741477272727273e-05,
+      "loss": 0.5608,
+      "step": 4820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7410037878787882e-05,
+      "loss": 0.5679,
+      "step": 4821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7405303030303032e-05,
+      "loss": 0.3579,
+      "step": 4822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7400568181818183e-05,
+      "loss": 0.42,
+      "step": 4823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7395833333333336e-05,
+      "loss": 0.3002,
+      "step": 4824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7391098484848487e-05,
+      "loss": 0.383,
+      "step": 4825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7386363636363637e-05,
+      "loss": 0.3747,
+      "step": 4826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7381628787878787e-05,
+      "loss": 0.5097,
+      "step": 4827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7376893939393938e-05,
+      "loss": 0.4226,
+      "step": 4828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7372159090909095e-05,
+      "loss": 0.4707,
+      "step": 4829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7367424242424245e-05,
+      "loss": 0.4195,
+      "step": 4830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7362689393939396e-05,
+      "loss": 0.4152,
+      "step": 4831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7357954545454546e-05,
+      "loss": 0.3754,
+      "step": 4832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.73532196969697e-05,
+      "loss": 0.4536,
+      "step": 4833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.734848484848485e-05,
+      "loss": 0.2816,
+      "step": 4834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.734375e-05,
+      "loss": 0.4203,
+      "step": 4835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.733901515151515e-05,
+      "loss": 0.5338,
+      "step": 4836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.73342803030303e-05,
+      "loss": 0.3277,
+      "step": 4837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.732954545454546e-05,
+      "loss": 0.4485,
+      "step": 4838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.732481060606061e-05,
+      "loss": 0.572,
+      "step": 4839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.732007575757576e-05,
+      "loss": 0.4222,
+      "step": 4840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.731534090909091e-05,
+      "loss": 0.5013,
+      "step": 4841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7310606060606063e-05,
+      "loss": 0.4775,
+      "step": 4842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7305871212121214e-05,
+      "loss": 0.5902,
+      "step": 4843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7301136363636364e-05,
+      "loss": 0.448,
+      "step": 4844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7296401515151514e-05,
+      "loss": 0.4409,
+      "step": 4845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7291666666666665e-05,
+      "loss": 0.3657,
+      "step": 4846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7286931818181822e-05,
+      "loss": 0.5142,
+      "step": 4847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7282196969696972e-05,
+      "loss": 0.3859,
+      "step": 4848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7277462121212123e-05,
+      "loss": 0.4078,
+      "step": 4849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 0.3598,
+      "step": 4850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7267992424242427e-05,
+      "loss": 0.4686,
+      "step": 4851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7263257575757577e-05,
+      "loss": 0.492,
+      "step": 4852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7258522727272728e-05,
+      "loss": 0.5087,
+      "step": 4853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7253787878787878e-05,
+      "loss": 0.3555,
+      "step": 4854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.724905303030303e-05,
+      "loss": 0.4358,
+      "step": 4855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7244318181818186e-05,
+      "loss": 0.606,
+      "step": 4856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7239583333333336e-05,
+      "loss": 0.7466,
+      "step": 4857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7234848484848486e-05,
+      "loss": 0.4817,
+      "step": 4858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7230113636363637e-05,
+      "loss": 0.4484,
+      "step": 4859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.722537878787879e-05,
+      "loss": 0.4092,
+      "step": 4860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.722064393939394e-05,
+      "loss": 0.6327,
+      "step": 4861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.721590909090909e-05,
+      "loss": 0.3781,
+      "step": 4862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.721117424242424e-05,
+      "loss": 0.5254,
+      "step": 4863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.72064393939394e-05,
+      "loss": 0.3984,
+      "step": 4864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.720170454545455e-05,
+      "loss": 0.5288,
+      "step": 4865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.71969696969697e-05,
+      "loss": 0.3818,
+      "step": 4866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.719223484848485e-05,
+      "loss": 0.3748,
+      "step": 4867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.71875e-05,
+      "loss": 0.3795,
+      "step": 4868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7182765151515154e-05,
+      "loss": 0.4401,
+      "step": 4869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7178030303030304e-05,
+      "loss": 0.4215,
+      "step": 4870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7173295454545455e-05,
+      "loss": 0.2914,
+      "step": 4871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7168560606060605e-05,
+      "loss": 0.4026,
+      "step": 4872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7163825757575762e-05,
+      "loss": 0.6022,
+      "step": 4873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7159090909090913e-05,
+      "loss": 0.429,
+      "step": 4874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7154356060606063e-05,
+      "loss": 0.4563,
+      "step": 4875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7149621212121213e-05,
+      "loss": 0.4114,
+      "step": 4876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7144886363636364e-05,
+      "loss": 0.5012,
+      "step": 4877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7140151515151517e-05,
+      "loss": 0.3969,
+      "step": 4878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7135416666666668e-05,
+      "loss": 0.4493,
+      "step": 4879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7130681818181818e-05,
+      "loss": 0.362,
+      "step": 4880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.712594696969697e-05,
+      "loss": 0.4399,
+      "step": 4881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7121212121212126e-05,
+      "loss": 0.4098,
+      "step": 4882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7116477272727276e-05,
+      "loss": 0.6726,
+      "step": 4883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7111742424242426e-05,
+      "loss": 0.471,
+      "step": 4884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7107007575757577e-05,
+      "loss": 0.4298,
+      "step": 4885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7102272727272727e-05,
+      "loss": 0.5572,
+      "step": 4886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.709753787878788e-05,
+      "loss": 0.5764,
+      "step": 4887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.709280303030303e-05,
+      "loss": 0.3932,
+      "step": 4888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7088068181818182e-05,
+      "loss": 0.487,
+      "step": 4889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7083333333333332e-05,
+      "loss": 0.4208,
+      "step": 4890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.707859848484849e-05,
+      "loss": 0.6619,
+      "step": 4891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.707386363636364e-05,
+      "loss": 0.4239,
+      "step": 4892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.706912878787879e-05,
+      "loss": 0.4409,
+      "step": 4893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.706439393939394e-05,
+      "loss": 0.4459,
+      "step": 4894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.705965909090909e-05,
+      "loss": 0.3764,
+      "step": 4895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7054924242424244e-05,
+      "loss": 0.4722,
+      "step": 4896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7050189393939395e-05,
+      "loss": 0.3192,
+      "step": 4897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7045454545454545e-05,
+      "loss": 0.5091,
+      "step": 4898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7040719696969696e-05,
+      "loss": 0.6137,
+      "step": 4899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7035984848484853e-05,
+      "loss": 0.4308,
+      "step": 4900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7031250000000003e-05,
+      "loss": 0.373,
+      "step": 4901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7026515151515154e-05,
+      "loss": 0.4739,
+      "step": 4902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7021780303030304e-05,
+      "loss": 0.4322,
+      "step": 4903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7017045454545454e-05,
+      "loss": 0.4349,
+      "step": 4904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7012310606060608e-05,
+      "loss": 0.4294,
+      "step": 4905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.700757575757576e-05,
+      "loss": 0.5302,
+      "step": 4906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.700284090909091e-05,
+      "loss": 0.4833,
+      "step": 4907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.699810606060606e-05,
+      "loss": 0.4563,
+      "step": 4908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6993371212121216e-05,
+      "loss": 0.4703,
+      "step": 4909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6988636363636367e-05,
+      "loss": 0.4479,
+      "step": 4910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6983901515151517e-05,
+      "loss": 0.458,
+      "step": 4911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6979166666666667e-05,
+      "loss": 0.4738,
+      "step": 4912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6974431818181818e-05,
+      "loss": 0.422,
+      "step": 4913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.696969696969697e-05,
+      "loss": 0.6431,
+      "step": 4914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6964962121212122e-05,
+      "loss": 0.5167,
+      "step": 4915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6960227272727272e-05,
+      "loss": 0.5382,
+      "step": 4916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6955492424242423e-05,
+      "loss": 0.5912,
+      "step": 4917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.695075757575758e-05,
+      "loss": 0.5334,
+      "step": 4918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.694602272727273e-05,
+      "loss": 0.5334,
+      "step": 4919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.694128787878788e-05,
+      "loss": 0.5782,
+      "step": 4920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.693655303030303e-05,
+      "loss": 0.4687,
+      "step": 4921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6931818181818185e-05,
+      "loss": 0.3868,
+      "step": 4922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6927083333333335e-05,
+      "loss": 0.5012,
+      "step": 4923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6922348484848485e-05,
+      "loss": 0.5088,
+      "step": 4924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6917613636363636e-05,
+      "loss": 0.5279,
+      "step": 4925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6912878787878786e-05,
+      "loss": 0.5254,
+      "step": 4926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6908143939393943e-05,
+      "loss": 0.4955,
+      "step": 4927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6903409090909094e-05,
+      "loss": 0.5709,
+      "step": 4928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6898674242424244e-05,
+      "loss": 0.4854,
+      "step": 4929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6893939393939394e-05,
+      "loss": 0.4094,
+      "step": 4930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6889204545454548e-05,
+      "loss": 0.4089,
+      "step": 4931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.68844696969697e-05,
+      "loss": 0.5103,
+      "step": 4932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.687973484848485e-05,
+      "loss": 0.5201,
+      "step": 4933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6875e-05,
+      "loss": 0.4593,
+      "step": 4934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.687026515151515e-05,
+      "loss": 0.4719,
+      "step": 4935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6865530303030307e-05,
+      "loss": 0.4739,
+      "step": 4936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6860795454545457e-05,
+      "loss": 0.4182,
+      "step": 4937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6856060606060608e-05,
+      "loss": 0.4752,
+      "step": 4938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6851325757575758e-05,
+      "loss": 0.4514,
+      "step": 4939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6846590909090912e-05,
+      "loss": 0.439,
+      "step": 4940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6841856060606062e-05,
+      "loss": 0.5792,
+      "step": 4941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6837121212121212e-05,
+      "loss": 0.4694,
+      "step": 4942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6832386363636363e-05,
+      "loss": 0.4517,
+      "step": 4943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6827651515151513e-05,
+      "loss": 0.3523,
+      "step": 4944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.682291666666667e-05,
+      "loss": 0.4208,
+      "step": 4945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.681818181818182e-05,
+      "loss": 0.5835,
+      "step": 4946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.681344696969697e-05,
+      "loss": 0.468,
+      "step": 4947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.680871212121212e-05,
+      "loss": 0.4255,
+      "step": 4948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6803977272727275e-05,
+      "loss": 0.503,
+      "step": 4949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6799242424242426e-05,
+      "loss": 0.4281,
+      "step": 4950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6794507575757576e-05,
+      "loss": 0.5703,
+      "step": 4951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6789772727272726e-05,
+      "loss": 0.4612,
+      "step": 4952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6785037878787877e-05,
+      "loss": 0.5778,
+      "step": 4953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6780303030303034e-05,
+      "loss": 0.5044,
+      "step": 4954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6775568181818184e-05,
+      "loss": 0.5119,
+      "step": 4955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6770833333333335e-05,
+      "loss": 0.3968,
+      "step": 4956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6766098484848485e-05,
+      "loss": 0.48,
+      "step": 4957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.676136363636364e-05,
+      "loss": 0.4606,
+      "step": 4958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.675662878787879e-05,
+      "loss": 0.4427,
+      "step": 4959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.675189393939394e-05,
+      "loss": 0.4846,
+      "step": 4960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.674715909090909e-05,
+      "loss": 0.3693,
+      "step": 4961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.674242424242424e-05,
+      "loss": 0.4791,
+      "step": 4962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6737689393939397e-05,
+      "loss": 0.3193,
+      "step": 4963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6732954545454548e-05,
+      "loss": 0.3522,
+      "step": 4964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6728219696969698e-05,
+      "loss": 0.4461,
+      "step": 4965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.672348484848485e-05,
+      "loss": 0.4505,
+      "step": 4966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6718750000000002e-05,
+      "loss": 0.5478,
+      "step": 4967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6714015151515153e-05,
+      "loss": 0.4334,
+      "step": 4968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6709280303030303e-05,
+      "loss": 0.3757,
+      "step": 4969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6704545454545453e-05,
+      "loss": 0.4032,
+      "step": 4970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6699810606060604e-05,
+      "loss": 0.3185,
+      "step": 4971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.669507575757576e-05,
+      "loss": 0.5652,
+      "step": 4972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.669034090909091e-05,
+      "loss": 0.499,
+      "step": 4973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6685606060606062e-05,
+      "loss": 0.3735,
+      "step": 4974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6680871212121212e-05,
+      "loss": 0.4581,
+      "step": 4975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6676136363636366e-05,
+      "loss": 0.5495,
+      "step": 4976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6671401515151516e-05,
+      "loss": 0.4089,
+      "step": 4977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.4965,
+      "step": 4978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6661931818181817e-05,
+      "loss": 0.5561,
+      "step": 4979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6657196969696974e-05,
+      "loss": 0.5295,
+      "step": 4980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6652462121212124e-05,
+      "loss": 0.4715,
+      "step": 4981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6647727272727275e-05,
+      "loss": 0.5217,
+      "step": 4982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6642992424242425e-05,
+      "loss": 0.4878,
+      "step": 4983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6638257575757576e-05,
+      "loss": 0.4825,
+      "step": 4984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.663352272727273e-05,
+      "loss": 0.5845,
+      "step": 4985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.662878787878788e-05,
+      "loss": 0.5431,
+      "step": 4986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.662405303030303e-05,
+      "loss": 0.4776,
+      "step": 4987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.661931818181818e-05,
+      "loss": 0.4778,
+      "step": 4988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6614583333333338e-05,
+      "loss": 0.3795,
+      "step": 4989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6609848484848488e-05,
+      "loss": 0.5901,
+      "step": 4990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.660511363636364e-05,
+      "loss": 0.483,
+      "step": 4991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.660037878787879e-05,
+      "loss": 0.4597,
+      "step": 4992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.659564393939394e-05,
+      "loss": 0.4591,
+      "step": 4993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6590909090909093e-05,
+      "loss": 0.5491,
+      "step": 4994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6586174242424243e-05,
+      "loss": 0.4061,
+      "step": 4995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6581439393939394e-05,
+      "loss": 0.466,
+      "step": 4996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6576704545454544e-05,
+      "loss": 0.4489,
+      "step": 4997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.65719696969697e-05,
+      "loss": 0.404,
+      "step": 4998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.656723484848485e-05,
+      "loss": 0.4852,
+      "step": 4999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6562500000000002e-05,
+      "loss": 0.4765,
+      "step": 5000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6557765151515152e-05,
+      "loss": 0.6408,
+      "step": 5001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6553030303030303e-05,
+      "loss": 0.4618,
+      "step": 5002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6548295454545456e-05,
+      "loss": 0.5525,
+      "step": 5003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6543560606060607e-05,
+      "loss": 0.4701,
+      "step": 5004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6538825757575757e-05,
+      "loss": 0.4713,
+      "step": 5005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6534090909090908e-05,
+      "loss": 0.5141,
+      "step": 5006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6529356060606065e-05,
+      "loss": 0.4483,
+      "step": 5007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6524621212121215e-05,
+      "loss": 0.4517,
+      "step": 5008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6519886363636365e-05,
+      "loss": 0.4216,
+      "step": 5009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6515151515151516e-05,
+      "loss": 0.4388,
+      "step": 5010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6510416666666666e-05,
+      "loss": 0.5006,
+      "step": 5011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.650568181818182e-05,
+      "loss": 0.4341,
+      "step": 5012
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.650094696969697e-05,
+      "loss": 0.2986,
+      "step": 5013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.649621212121212e-05,
+      "loss": 0.3818,
+      "step": 5014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.649147727272727e-05,
+      "loss": 0.4254,
+      "step": 5015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6486742424242428e-05,
+      "loss": 0.3663,
+      "step": 5016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.648200757575758e-05,
+      "loss": 0.4065,
+      "step": 5017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.647727272727273e-05,
+      "loss": 0.4949,
+      "step": 5018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.647253787878788e-05,
+      "loss": 0.4851,
+      "step": 5019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.646780303030303e-05,
+      "loss": 0.5149,
+      "step": 5020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6463068181818183e-05,
+      "loss": 0.3379,
+      "step": 5021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6458333333333334e-05,
+      "loss": 0.4349,
+      "step": 5022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6453598484848484e-05,
+      "loss": 0.3914,
+      "step": 5023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6448863636363635e-05,
+      "loss": 0.4567,
+      "step": 5024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6444128787878792e-05,
+      "loss": 0.4849,
+      "step": 5025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6439393939393942e-05,
+      "loss": 0.3418,
+      "step": 5026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6434659090909092e-05,
+      "loss": 0.46,
+      "step": 5027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6429924242424243e-05,
+      "loss": 0.4897,
+      "step": 5028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6425189393939397e-05,
+      "loss": 0.4335,
+      "step": 5029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6420454545454547e-05,
+      "loss": 0.3974,
+      "step": 5030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6415719696969697e-05,
+      "loss": 0.4227,
+      "step": 5031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6410984848484848e-05,
+      "loss": 0.4493,
+      "step": 5032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6406249999999998e-05,
+      "loss": 0.3476,
+      "step": 5033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6401515151515155e-05,
+      "loss": 0.601,
+      "step": 5034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6396780303030306e-05,
+      "loss": 0.512,
+      "step": 5035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6392045454545456e-05,
+      "loss": 0.4018,
+      "step": 5036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6387310606060606e-05,
+      "loss": 0.5083,
+      "step": 5037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.638257575757576e-05,
+      "loss": 0.6166,
+      "step": 5038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.637784090909091e-05,
+      "loss": 0.5046,
+      "step": 5039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.637310606060606e-05,
+      "loss": 0.5072,
+      "step": 5040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.636837121212121e-05,
+      "loss": 0.4558,
+      "step": 5041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.636363636363636e-05,
+      "loss": 0.4579,
+      "step": 5042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.635890151515152e-05,
+      "loss": 0.4722,
+      "step": 5043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.635416666666667e-05,
+      "loss": 0.4348,
+      "step": 5044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.634943181818182e-05,
+      "loss": 0.577,
+      "step": 5045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.634469696969697e-05,
+      "loss": 0.517,
+      "step": 5046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6339962121212124e-05,
+      "loss": 0.4411,
+      "step": 5047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6335227272727274e-05,
+      "loss": 0.4024,
+      "step": 5048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6330492424242424e-05,
+      "loss": 0.3846,
+      "step": 5049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6325757575757575e-05,
+      "loss": 0.3641,
+      "step": 5050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6321022727272725e-05,
+      "loss": 0.469,
+      "step": 5051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6316287878787882e-05,
+      "loss": 0.4043,
+      "step": 5052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6311553030303033e-05,
+      "loss": 0.3712,
+      "step": 5053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6306818181818183e-05,
+      "loss": 0.3281,
+      "step": 5054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6302083333333333e-05,
+      "loss": 0.4434,
+      "step": 5055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6297348484848487e-05,
+      "loss": 0.4957,
+      "step": 5056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6292613636363638e-05,
+      "loss": 0.5337,
+      "step": 5057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6287878787878788e-05,
+      "loss": 0.4706,
+      "step": 5058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.628314393939394e-05,
+      "loss": 0.5717,
+      "step": 5059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.627840909090909e-05,
+      "loss": 0.4309,
+      "step": 5060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6273674242424246e-05,
+      "loss": 0.5203,
+      "step": 5061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6268939393939396e-05,
+      "loss": 0.6175,
+      "step": 5062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6264204545454547e-05,
+      "loss": 0.6165,
+      "step": 5063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6259469696969697e-05,
+      "loss": 0.3919,
+      "step": 5064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.625473484848485e-05,
+      "loss": 0.433,
+      "step": 5065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.625e-05,
+      "loss": 0.4432,
+      "step": 5066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.624526515151515e-05,
+      "loss": 0.5431,
+      "step": 5067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6240530303030302e-05,
+      "loss": 0.4262,
+      "step": 5068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6235795454545452e-05,
+      "loss": 0.4652,
+      "step": 5069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.623106060606061e-05,
+      "loss": 0.388,
+      "step": 5070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.622632575757576e-05,
+      "loss": 0.4488,
+      "step": 5071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.622159090909091e-05,
+      "loss": 0.5156,
+      "step": 5072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.621685606060606e-05,
+      "loss": 0.4314,
+      "step": 5073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6212121212121214e-05,
+      "loss": 0.5511,
+      "step": 5074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6207386363636365e-05,
+      "loss": 0.441,
+      "step": 5075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6202651515151515e-05,
+      "loss": 0.5168,
+      "step": 5076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6197916666666665e-05,
+      "loss": 0.3953,
+      "step": 5077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6193181818181816e-05,
+      "loss": 0.4516,
+      "step": 5078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6188446969696973e-05,
+      "loss": 0.4409,
+      "step": 5079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6183712121212123e-05,
+      "loss": 0.4874,
+      "step": 5080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6178977272727274e-05,
+      "loss": 0.4769,
+      "step": 5081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6174242424242424e-05,
+      "loss": 0.4964,
+      "step": 5082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6169507575757578e-05,
+      "loss": 0.5493,
+      "step": 5083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6164772727272728e-05,
+      "loss": 0.3977,
+      "step": 5084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.616003787878788e-05,
+      "loss": 0.5244,
+      "step": 5085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.615530303030303e-05,
+      "loss": 0.4476,
+      "step": 5086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6150568181818186e-05,
+      "loss": 0.5009,
+      "step": 5087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6145833333333336e-05,
+      "loss": 0.4497,
+      "step": 5088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6141098484848487e-05,
+      "loss": 0.523,
+      "step": 5089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6136363636363637e-05,
+      "loss": 0.4357,
+      "step": 5090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6131628787878788e-05,
+      "loss": 0.4803,
+      "step": 5091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.612689393939394e-05,
+      "loss": 0.4687,
+      "step": 5092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.612215909090909e-05,
+      "loss": 0.417,
+      "step": 5093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6117424242424242e-05,
+      "loss": 0.4598,
+      "step": 5094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6112689393939392e-05,
+      "loss": 0.5474,
+      "step": 5095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.610795454545455e-05,
+      "loss": 0.5206,
+      "step": 5096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.61032196969697e-05,
+      "loss": 0.5002,
+      "step": 5097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.609848484848485e-05,
+      "loss": 0.5289,
+      "step": 5098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.609375e-05,
+      "loss": 0.4559,
+      "step": 5099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.608901515151515e-05,
+      "loss": 0.4494,
+      "step": 5100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6084280303030305e-05,
+      "loss": 0.4425,
+      "step": 5101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6079545454545455e-05,
+      "loss": 0.548,
+      "step": 5102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6074810606060606e-05,
+      "loss": 0.4791,
+      "step": 5103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6070075757575756e-05,
+      "loss": 0.4961,
+      "step": 5104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6065340909090913e-05,
+      "loss": 0.4615,
+      "step": 5105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6060606060606063e-05,
+      "loss": 0.4736,
+      "step": 5106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6055871212121214e-05,
+      "loss": 0.463,
+      "step": 5107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6051136363636364e-05,
+      "loss": 0.5497,
+      "step": 5108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6046401515151515e-05,
+      "loss": 0.4358,
+      "step": 5109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.604166666666667e-05,
+      "loss": 0.4843,
+      "step": 5110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.603693181818182e-05,
+      "loss": 0.5642,
+      "step": 5111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.603219696969697e-05,
+      "loss": 0.3848,
+      "step": 5112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.602746212121212e-05,
+      "loss": 0.576,
+      "step": 5113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6022727272727277e-05,
+      "loss": 0.5779,
+      "step": 5114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6017992424242427e-05,
+      "loss": 0.3674,
+      "step": 5115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6013257575757577e-05,
+      "loss": 0.4018,
+      "step": 5116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6008522727272728e-05,
+      "loss": 0.3861,
+      "step": 5117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6003787878787878e-05,
+      "loss": 0.4668,
+      "step": 5118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5999053030303032e-05,
+      "loss": 0.3923,
+      "step": 5119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5994318181818182e-05,
+      "loss": 0.3585,
+      "step": 5120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5989583333333333e-05,
+      "loss": 0.5356,
+      "step": 5121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5984848484848483e-05,
+      "loss": 0.5201,
+      "step": 5122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.598011363636364e-05,
+      "loss": 0.3509,
+      "step": 5123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.597537878787879e-05,
+      "loss": 0.4137,
+      "step": 5124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.597064393939394e-05,
+      "loss": 0.3808,
+      "step": 5125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.596590909090909e-05,
+      "loss": 0.3594,
+      "step": 5126
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.596117424242424e-05,
+      "loss": 0.4536,
+      "step": 5127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5956439393939395e-05,
+      "loss": 0.4996,
+      "step": 5128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5951704545454546e-05,
+      "loss": 0.4146,
+      "step": 5129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5946969696969696e-05,
+      "loss": 0.4778,
+      "step": 5130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5942234848484847e-05,
+      "loss": 0.5261,
+      "step": 5131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5937500000000004e-05,
+      "loss": 0.5192,
+      "step": 5132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5932765151515154e-05,
+      "loss": 0.4379,
+      "step": 5133
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5928030303030304e-05,
+      "loss": 0.3761,
+      "step": 5134
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5923295454545455e-05,
+      "loss": 0.5298,
+      "step": 5135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5918560606060605e-05,
+      "loss": 0.4856,
+      "step": 5136
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.591382575757576e-05,
+      "loss": 0.5047,
+      "step": 5137
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.590909090909091e-05,
+      "loss": 0.4093,
+      "step": 5138
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.590435606060606e-05,
+      "loss": 0.391,
+      "step": 5139
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.589962121212121e-05,
+      "loss": 0.4813,
+      "step": 5140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5894886363636367e-05,
+      "loss": 0.4829,
+      "step": 5141
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5890151515151518e-05,
+      "loss": 0.5641,
+      "step": 5142
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5885416666666668e-05,
+      "loss": 0.5416,
+      "step": 5143
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5880681818181818e-05,
+      "loss": 0.4975,
+      "step": 5144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5875946969696972e-05,
+      "loss": 0.4687,
+      "step": 5145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5871212121212122e-05,
+      "loss": 0.4337,
+      "step": 5146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5866477272727273e-05,
+      "loss": 0.6112,
+      "step": 5147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5861742424242423e-05,
+      "loss": 0.5084,
+      "step": 5148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5857007575757574e-05,
+      "loss": 0.3851,
+      "step": 5149
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.585227272727273e-05,
+      "loss": 0.5471,
+      "step": 5150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.584753787878788e-05,
+      "loss": 0.478,
+      "step": 5151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.584280303030303e-05,
+      "loss": 0.4087,
+      "step": 5152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5838068181818182e-05,
+      "loss": 0.4987,
+      "step": 5153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5833333333333336e-05,
+      "loss": 0.5577,
+      "step": 5154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5828598484848486e-05,
+      "loss": 0.4935,
+      "step": 5155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5823863636363636e-05,
+      "loss": 0.5237,
+      "step": 5156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5819128787878787e-05,
+      "loss": 0.3775,
+      "step": 5157
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5814393939393937e-05,
+      "loss": 0.5312,
+      "step": 5158
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5809659090909094e-05,
+      "loss": 0.4578,
+      "step": 5159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5804924242424245e-05,
+      "loss": 0.5007,
+      "step": 5160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5800189393939395e-05,
+      "loss": 0.4482,
+      "step": 5161
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5795454545454545e-05,
+      "loss": 0.5276,
+      "step": 5162
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.57907196969697e-05,
+      "loss": 0.5483,
+      "step": 5163
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.578598484848485e-05,
+      "loss": 0.5571,
+      "step": 5164
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.578125e-05,
+      "loss": 0.411,
+      "step": 5165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.577651515151515e-05,
+      "loss": 0.4752,
+      "step": 5166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.57717803030303e-05,
+      "loss": 0.3691,
+      "step": 5167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5767045454545458e-05,
+      "loss": 0.5538,
+      "step": 5168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5762310606060608e-05,
+      "loss": 0.3956,
+      "step": 5169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.575757575757576e-05,
+      "loss": 0.4276,
+      "step": 5170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.575284090909091e-05,
+      "loss": 0.5224,
+      "step": 5171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5748106060606063e-05,
+      "loss": 0.4431,
+      "step": 5172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5743371212121213e-05,
+      "loss": 0.4581,
+      "step": 5173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5738636363636363e-05,
+      "loss": 0.5102,
+      "step": 5174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5733901515151514e-05,
+      "loss": 0.3836,
+      "step": 5175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5729166666666664e-05,
+      "loss": 0.571,
+      "step": 5176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.572443181818182e-05,
+      "loss": 0.4184,
+      "step": 5177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.571969696969697e-05,
+      "loss": 0.4499,
+      "step": 5178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5714962121212122e-05,
+      "loss": 0.3915,
+      "step": 5179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5710227272727272e-05,
+      "loss": 0.4582,
+      "step": 5180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5705492424242426e-05,
+      "loss": 0.4331,
+      "step": 5181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5700757575757577e-05,
+      "loss": 0.3248,
+      "step": 5182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5696022727272727e-05,
+      "loss": 0.3222,
+      "step": 5183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5691287878787877e-05,
+      "loss": 0.4494,
+      "step": 5184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5686553030303028e-05,
+      "loss": 0.3251,
+      "step": 5185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5681818181818185e-05,
+      "loss": 0.5356,
+      "step": 5186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5677083333333335e-05,
+      "loss": 0.464,
+      "step": 5187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5672348484848486e-05,
+      "loss": 0.5465,
+      "step": 5188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5667613636363636e-05,
+      "loss": 0.4604,
+      "step": 5189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.566287878787879e-05,
+      "loss": 0.5233,
+      "step": 5190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.565814393939394e-05,
+      "loss": 0.5264,
+      "step": 5191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.565340909090909e-05,
+      "loss": 0.5131,
+      "step": 5192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.564867424242424e-05,
+      "loss": 0.4536,
+      "step": 5193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5643939393939398e-05,
+      "loss": 0.5386,
+      "step": 5194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.563920454545455e-05,
+      "loss": 0.4182,
+      "step": 5195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.56344696969697e-05,
+      "loss": 0.3356,
+      "step": 5196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.562973484848485e-05,
+      "loss": 0.4736,
+      "step": 5197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5625e-05,
+      "loss": 0.5654,
+      "step": 5198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5620265151515153e-05,
+      "loss": 0.4833,
+      "step": 5199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5615530303030304e-05,
+      "loss": 0.4683,
+      "step": 5200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5610795454545454e-05,
+      "loss": 0.5375,
+      "step": 5201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5606060606060604e-05,
+      "loss": 0.3064,
+      "step": 5202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.560132575757576e-05,
+      "loss": 0.3086,
+      "step": 5203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5596590909090912e-05,
+      "loss": 0.497,
+      "step": 5204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5591856060606062e-05,
+      "loss": 0.4562,
+      "step": 5205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5587121212121213e-05,
+      "loss": 0.5336,
+      "step": 5206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5582386363636363e-05,
+      "loss": 0.4379,
+      "step": 5207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5577651515151517e-05,
+      "loss": 0.4938,
+      "step": 5208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5572916666666667e-05,
+      "loss": 0.5029,
+      "step": 5209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5568181818181817e-05,
+      "loss": 0.5396,
+      "step": 5210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5563446969696968e-05,
+      "loss": 0.4515,
+      "step": 5211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5558712121212125e-05,
+      "loss": 0.3911,
+      "step": 5212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5553977272727275e-05,
+      "loss": 0.3832,
+      "step": 5213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5549242424242426e-05,
+      "loss": 0.5377,
+      "step": 5214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5544507575757576e-05,
+      "loss": 0.3395,
+      "step": 5215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5539772727272726e-05,
+      "loss": 0.4942,
+      "step": 5216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.553503787878788e-05,
+      "loss": 0.4723,
+      "step": 5217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.553030303030303e-05,
+      "loss": 0.5281,
+      "step": 5218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.552556818181818e-05,
+      "loss": 0.3893,
+      "step": 5219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.552083333333333e-05,
+      "loss": 0.4726,
+      "step": 5220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.551609848484849e-05,
+      "loss": 0.3504,
+      "step": 5221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.551136363636364e-05,
+      "loss": 0.5665,
+      "step": 5222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.550662878787879e-05,
+      "loss": 0.5469,
+      "step": 5223
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.550189393939394e-05,
+      "loss": 0.3614,
+      "step": 5224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.549715909090909e-05,
+      "loss": 0.4414,
+      "step": 5225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5492424242424244e-05,
+      "loss": 0.3674,
+      "step": 5226
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5487689393939394e-05,
+      "loss": 0.4469,
+      "step": 5227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5482954545454545e-05,
+      "loss": 0.3963,
+      "step": 5228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5478219696969695e-05,
+      "loss": 0.3074,
+      "step": 5229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5473484848484852e-05,
+      "loss": 0.4145,
+      "step": 5230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5468750000000002e-05,
+      "loss": 0.4512,
+      "step": 5231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5464015151515153e-05,
+      "loss": 0.4392,
+      "step": 5232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5459280303030303e-05,
+      "loss": 0.5573,
+      "step": 5233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5454545454545454e-05,
+      "loss": 0.4677,
+      "step": 5234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5449810606060607e-05,
+      "loss": 0.4403,
+      "step": 5235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5445075757575758e-05,
+      "loss": 0.4586,
+      "step": 5236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5440340909090908e-05,
+      "loss": 0.5955,
+      "step": 5237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.543560606060606e-05,
+      "loss": 0.6194,
+      "step": 5238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5430871212121216e-05,
+      "loss": 0.4938,
+      "step": 5239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5426136363636366e-05,
+      "loss": 0.3706,
+      "step": 5240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5421401515151516e-05,
+      "loss": 0.3838,
+      "step": 5241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5416666666666667e-05,
+      "loss": 0.4068,
+      "step": 5242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5411931818181817e-05,
+      "loss": 0.4375,
+      "step": 5243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.540719696969697e-05,
+      "loss": 0.4394,
+      "step": 5244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.540246212121212e-05,
+      "loss": 0.4249,
+      "step": 5245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.539772727272727e-05,
+      "loss": 0.4777,
+      "step": 5246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5392992424242422e-05,
+      "loss": 0.5462,
+      "step": 5247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.538825757575758e-05,
+      "loss": 0.4358,
+      "step": 5248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.538352272727273e-05,
+      "loss": 0.4859,
+      "step": 5249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.537878787878788e-05,
+      "loss": 0.3645,
+      "step": 5250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.537405303030303e-05,
+      "loss": 0.4572,
+      "step": 5251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5369318181818187e-05,
+      "loss": 0.487,
+      "step": 5252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5364583333333334e-05,
+      "loss": 0.4091,
+      "step": 5253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5359848484848485e-05,
+      "loss": 0.4918,
+      "step": 5254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5355113636363635e-05,
+      "loss": 0.4669,
+      "step": 5255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5350378787878785e-05,
+      "loss": 0.4595,
+      "step": 5256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5345643939393943e-05,
+      "loss": 0.6341,
+      "step": 5257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5340909090909093e-05,
+      "loss": 0.4138,
+      "step": 5258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5336174242424243e-05,
+      "loss": 0.5325,
+      "step": 5259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5331439393939394e-05,
+      "loss": 0.5007,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.532670454545455e-05,
+      "loss": 0.4866,
+      "step": 5261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5321969696969698e-05,
+      "loss": 0.4521,
+      "step": 5262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5317234848484848e-05,
+      "loss": 0.431,
+      "step": 5263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.53125e-05,
+      "loss": 0.4773,
+      "step": 5264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.530776515151515e-05,
+      "loss": 0.4992,
+      "step": 5265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5303030303030306e-05,
+      "loss": 0.4243,
+      "step": 5266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5298295454545457e-05,
+      "loss": 0.6077,
+      "step": 5267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5293560606060607e-05,
+      "loss": 0.3563,
+      "step": 5268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5288825757575757e-05,
+      "loss": 0.5197,
+      "step": 5269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5284090909090914e-05,
+      "loss": 0.3959,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.527935606060606e-05,
+      "loss": 0.3602,
+      "step": 5271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5274621212121212e-05,
+      "loss": 0.4858,
+      "step": 5272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5269886363636362e-05,
+      "loss": 0.5562,
+      "step": 5273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5265151515151513e-05,
+      "loss": 0.3439,
+      "step": 5274
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.526041666666667e-05,
+      "loss": 0.4376,
+      "step": 5275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.525568181818182e-05,
+      "loss": 0.4785,
+      "step": 5276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.525094696969697e-05,
+      "loss": 0.3136,
+      "step": 5277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.524621212121212e-05,
+      "loss": 0.5091,
+      "step": 5278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5241477272727278e-05,
+      "loss": 0.4581,
+      "step": 5279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.523674242424243e-05,
+      "loss": 0.3739,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5232007575757575e-05,
+      "loss": 0.3913,
+      "step": 5281
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5227272727272726e-05,
+      "loss": 0.4525,
+      "step": 5282
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5222537878787876e-05,
+      "loss": 0.4892,
+      "step": 5283
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5217803030303033e-05,
+      "loss": 0.3572,
+      "step": 5284
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5213068181818184e-05,
+      "loss": 0.5074,
+      "step": 5285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5208333333333334e-05,
+      "loss": 0.5112,
+      "step": 5286
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5203598484848484e-05,
+      "loss": 0.4003,
+      "step": 5287
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.519886363636364e-05,
+      "loss": 0.4832,
+      "step": 5288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5194128787878792e-05,
+      "loss": 0.4581,
+      "step": 5289
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.518939393939394e-05,
+      "loss": 0.5734,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.518465909090909e-05,
+      "loss": 0.4187,
+      "step": 5291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.517992424242424e-05,
+      "loss": 0.4482,
+      "step": 5292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5175189393939397e-05,
+      "loss": 0.3798,
+      "step": 5293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5170454545454547e-05,
+      "loss": 0.494,
+      "step": 5294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5165719696969697e-05,
+      "loss": 0.4289,
+      "step": 5295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5160984848484848e-05,
+      "loss": 0.5019,
+      "step": 5296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5156250000000005e-05,
+      "loss": 0.5637,
+      "step": 5297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5151515151515155e-05,
+      "loss": 0.6389,
+      "step": 5298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5146780303030302e-05,
+      "loss": 0.5244,
+      "step": 5299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5142045454545453e-05,
+      "loss": 0.5001,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5137310606060603e-05,
+      "loss": 0.3436,
+      "step": 5301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.513257575757576e-05,
+      "loss": 0.4776,
+      "step": 5302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.512784090909091e-05,
+      "loss": 0.4012,
+      "step": 5303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.512310606060606e-05,
+      "loss": 0.4169,
+      "step": 5304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.511837121212121e-05,
+      "loss": 0.4259,
+      "step": 5305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.511363636363637e-05,
+      "loss": 0.5207,
+      "step": 5306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.510890151515152e-05,
+      "loss": 0.4761,
+      "step": 5307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5104166666666666e-05,
+      "loss": 0.5121,
+      "step": 5308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5099431818181816e-05,
+      "loss": 0.4895,
+      "step": 5309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5094696969696973e-05,
+      "loss": 0.3741,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5089962121212124e-05,
+      "loss": 0.4714,
+      "step": 5311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5085227272727274e-05,
+      "loss": 0.4133,
+      "step": 5312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5080492424242425e-05,
+      "loss": 0.4991,
+      "step": 5313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5075757575757575e-05,
+      "loss": 0.3869,
+      "step": 5314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5071022727272732e-05,
+      "loss": 0.5122,
+      "step": 5315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5066287878787882e-05,
+      "loss": 0.436,
+      "step": 5316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.506155303030303e-05,
+      "loss": 0.3828,
+      "step": 5317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.505681818181818e-05,
+      "loss": 0.4807,
+      "step": 5318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5052083333333337e-05,
+      "loss": 0.3869,
+      "step": 5319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5047348484848487e-05,
+      "loss": 0.467,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5042613636363638e-05,
+      "loss": 0.45,
+      "step": 5321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5037878787878788e-05,
+      "loss": 0.4399,
+      "step": 5322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.503314393939394e-05,
+      "loss": 0.4689,
+      "step": 5323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5028409090909096e-05,
+      "loss": 0.4776,
+      "step": 5324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5023674242424246e-05,
+      "loss": 0.4536,
+      "step": 5325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5018939393939393e-05,
+      "loss": 0.4896,
+      "step": 5326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5014204545454543e-05,
+      "loss": 0.4227,
+      "step": 5327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.50094696969697e-05,
+      "loss": 0.5555,
+      "step": 5328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.500473484848485e-05,
+      "loss": 0.465,
+      "step": 5329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5879,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.499526515151515e-05,
+      "loss": 0.5266,
+      "step": 5331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4990530303030305e-05,
+      "loss": 0.4426,
+      "step": 5332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4985795454545456e-05,
+      "loss": 0.4364,
+      "step": 5333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.498106060606061e-05,
+      "loss": 0.4545,
+      "step": 5334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4976325757575756e-05,
+      "loss": 0.3925,
+      "step": 5335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.497159090909091e-05,
+      "loss": 0.4438,
+      "step": 5336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.496685606060606e-05,
+      "loss": 0.4273,
+      "step": 5337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4962121212121214e-05,
+      "loss": 0.4769,
+      "step": 5338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4957386363636365e-05,
+      "loss": 0.4422,
+      "step": 5339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4952651515151515e-05,
+      "loss": 0.4416,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.494791666666667e-05,
+      "loss": 0.4374,
+      "step": 5341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.494318181818182e-05,
+      "loss": 0.3986,
+      "step": 5342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4938446969696973e-05,
+      "loss": 0.6257,
+      "step": 5343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.493371212121212e-05,
+      "loss": 0.4865,
+      "step": 5344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4928977272727274e-05,
+      "loss": 0.4753,
+      "step": 5345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4924242424242424e-05,
+      "loss": 0.4574,
+      "step": 5346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4919507575757578e-05,
+      "loss": 0.5016,
+      "step": 5347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4914772727272728e-05,
+      "loss": 0.5201,
+      "step": 5348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4910037878787882e-05,
+      "loss": 0.5084,
+      "step": 5349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4905303030303032e-05,
+      "loss": 0.6034,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4900568181818183e-05,
+      "loss": 0.4342,
+      "step": 5351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4895833333333337e-05,
+      "loss": 0.4485,
+      "step": 5352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4891098484848484e-05,
+      "loss": 0.4878,
+      "step": 5353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4886363636363637e-05,
+      "loss": 0.512,
+      "step": 5354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4881628787878788e-05,
+      "loss": 0.4233,
+      "step": 5355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.487689393939394e-05,
+      "loss": 0.4462,
+      "step": 5356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4872159090909092e-05,
+      "loss": 0.4692,
+      "step": 5357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4867424242424246e-05,
+      "loss": 0.5111,
+      "step": 5358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4862689393939396e-05,
+      "loss": 0.4951,
+      "step": 5359
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4857954545454546e-05,
+      "loss": 0.4039,
+      "step": 5360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.48532196969697e-05,
+      "loss": 0.6001,
+      "step": 5361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4848484848484847e-05,
+      "loss": 0.3798,
+      "step": 5362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.484375e-05,
+      "loss": 0.4527,
+      "step": 5363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.483901515151515e-05,
+      "loss": 0.4563,
+      "step": 5364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4834280303030305e-05,
+      "loss": 0.3193,
+      "step": 5365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4829545454545455e-05,
+      "loss": 0.3348,
+      "step": 5366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.482481060606061e-05,
+      "loss": 0.4454,
+      "step": 5367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.482007575757576e-05,
+      "loss": 0.4511,
+      "step": 5368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.481534090909091e-05,
+      "loss": 0.5406,
+      "step": 5369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4810606060606064e-05,
+      "loss": 0.5773,
+      "step": 5370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.480587121212121e-05,
+      "loss": 0.3876,
+      "step": 5371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4801136363636364e-05,
+      "loss": 0.3535,
+      "step": 5372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4796401515151515e-05,
+      "loss": 0.4659,
+      "step": 5373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.479166666666667e-05,
+      "loss": 0.4364,
+      "step": 5374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.478693181818182e-05,
+      "loss": 0.5814,
+      "step": 5375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4782196969696973e-05,
+      "loss": 0.4878,
+      "step": 5376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4777462121212123e-05,
+      "loss": 0.4837,
+      "step": 5377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4772727272727277e-05,
+      "loss": 0.4356,
+      "step": 5378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4767992424242427e-05,
+      "loss": 0.5092,
+      "step": 5379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4763257575757574e-05,
+      "loss": 0.3873,
+      "step": 5380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4758522727272728e-05,
+      "loss": 0.4508,
+      "step": 5381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4753787878787878e-05,
+      "loss": 0.6459,
+      "step": 5382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4749053030303032e-05,
+      "loss": 0.584,
+      "step": 5383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4744318181818182e-05,
+      "loss": 0.578,
+      "step": 5384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4739583333333336e-05,
+      "loss": 0.4381,
+      "step": 5385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4734848484848486e-05,
+      "loss": 0.4823,
+      "step": 5386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.473011363636364e-05,
+      "loss": 0.4384,
+      "step": 5387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.472537878787879e-05,
+      "loss": 0.4358,
+      "step": 5388
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.472064393939394e-05,
+      "loss": 0.4443,
+      "step": 5389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.471590909090909e-05,
+      "loss": 0.2956,
+      "step": 5390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4711174242424242e-05,
+      "loss": 0.532,
+      "step": 5391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4706439393939395e-05,
+      "loss": 0.4104,
+      "step": 5392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4701704545454546e-05,
+      "loss": 0.4173,
+      "step": 5393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.46969696969697e-05,
+      "loss": 0.4156,
+      "step": 5394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.469223484848485e-05,
+      "loss": 0.3737,
+      "step": 5395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4687500000000004e-05,
+      "loss": 0.4249,
+      "step": 5396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4682765151515154e-05,
+      "loss": 0.4348,
+      "step": 5397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4678030303030305e-05,
+      "loss": 0.4093,
+      "step": 5398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4673295454545455e-05,
+      "loss": 0.4522,
+      "step": 5399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4668560606060605e-05,
+      "loss": 0.5591,
+      "step": 5400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.466382575757576e-05,
+      "loss": 0.4186,
+      "step": 5401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.465909090909091e-05,
+      "loss": 0.4149,
+      "step": 5402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4654356060606063e-05,
+      "loss": 0.4887,
+      "step": 5403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4649621212121214e-05,
+      "loss": 0.3853,
+      "step": 5404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4644886363636367e-05,
+      "loss": 0.369,
+      "step": 5405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4640151515151518e-05,
+      "loss": 0.5504,
+      "step": 5406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4635416666666668e-05,
+      "loss": 0.3919,
+      "step": 5407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.463068181818182e-05,
+      "loss": 0.3133,
+      "step": 5408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.462594696969697e-05,
+      "loss": 0.3933,
+      "step": 5409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4621212121212123e-05,
+      "loss": 0.5091,
+      "step": 5410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4616477272727273e-05,
+      "loss": 0.4032,
+      "step": 5411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4611742424242427e-05,
+      "loss": 0.4729,
+      "step": 5412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4607007575757577e-05,
+      "loss": 0.423,
+      "step": 5413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.460227272727273e-05,
+      "loss": 0.394,
+      "step": 5414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.459753787878788e-05,
+      "loss": 0.4937,
+      "step": 5415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.459280303030303e-05,
+      "loss": 0.4258,
+      "step": 5416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4588068181818182e-05,
+      "loss": 0.4784,
+      "step": 5417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4583333333333332e-05,
+      "loss": 0.5226,
+      "step": 5418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4578598484848486e-05,
+      "loss": 0.5106,
+      "step": 5419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4573863636363636e-05,
+      "loss": 0.5231,
+      "step": 5420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.456912878787879e-05,
+      "loss": 0.4933,
+      "step": 5421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.456439393939394e-05,
+      "loss": 0.4439,
+      "step": 5422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4559659090909094e-05,
+      "loss": 0.5093,
+      "step": 5423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4554924242424245e-05,
+      "loss": 0.528,
+      "step": 5424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4550189393939395e-05,
+      "loss": 0.4959,
+      "step": 5425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4545454545454545e-05,
+      "loss": 0.5173,
+      "step": 5426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.45407196969697e-05,
+      "loss": 0.4597,
+      "step": 5427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.453598484848485e-05,
+      "loss": 0.4459,
+      "step": 5428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.453125e-05,
+      "loss": 0.4987,
+      "step": 5429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4526515151515154e-05,
+      "loss": 0.4642,
+      "step": 5430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4521780303030304e-05,
+      "loss": 0.4166,
+      "step": 5431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4517045454545458e-05,
+      "loss": 0.3877,
+      "step": 5432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4512310606060608e-05,
+      "loss": 0.3397,
+      "step": 5433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.450757575757576e-05,
+      "loss": 0.4722,
+      "step": 5434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.450284090909091e-05,
+      "loss": 0.5211,
+      "step": 5435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4498106060606063e-05,
+      "loss": 0.5159,
+      "step": 5436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4493371212121213e-05,
+      "loss": 0.6235,
+      "step": 5437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4488636363636363e-05,
+      "loss": 0.3827,
+      "step": 5438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4483901515151517e-05,
+      "loss": 0.4394,
+      "step": 5439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4479166666666668e-05,
+      "loss": 0.5316,
+      "step": 5440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.447443181818182e-05,
+      "loss": 0.4249,
+      "step": 5441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4469696969696972e-05,
+      "loss": 0.4062,
+      "step": 5442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4464962121212122e-05,
+      "loss": 0.5216,
+      "step": 5443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4460227272727273e-05,
+      "loss": 0.3357,
+      "step": 5444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4455492424242426e-05,
+      "loss": 0.5105,
+      "step": 5445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4450757575757577e-05,
+      "loss": 0.4422,
+      "step": 5446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4446022727272727e-05,
+      "loss": 0.5076,
+      "step": 5447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.444128787878788e-05,
+      "loss": 0.5901,
+      "step": 5448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.443655303030303e-05,
+      "loss": 0.5358,
+      "step": 5449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4431818181818185e-05,
+      "loss": 0.4462,
+      "step": 5450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4427083333333335e-05,
+      "loss": 0.4187,
+      "step": 5451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4422348484848486e-05,
+      "loss": 0.4672,
+      "step": 5452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4417613636363636e-05,
+      "loss": 0.4296,
+      "step": 5453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.441287878787879e-05,
+      "loss": 0.6333,
+      "step": 5454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.440814393939394e-05,
+      "loss": 0.4536,
+      "step": 5455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4403409090909094e-05,
+      "loss": 0.4846,
+      "step": 5456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4398674242424244e-05,
+      "loss": 0.4531,
+      "step": 5457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4393939393939395e-05,
+      "loss": 0.4862,
+      "step": 5458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.438920454545455e-05,
+      "loss": 0.4324,
+      "step": 5459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.43844696969697e-05,
+      "loss": 0.5161,
+      "step": 5460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.437973484848485e-05,
+      "loss": 0.4195,
+      "step": 5461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4375e-05,
+      "loss": 0.491,
+      "step": 5462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4370265151515153e-05,
+      "loss": 0.5294,
+      "step": 5463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4365530303030304e-05,
+      "loss": 0.4501,
+      "step": 5464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4360795454545457e-05,
+      "loss": 0.4358,
+      "step": 5465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4356060606060608e-05,
+      "loss": 0.375,
+      "step": 5466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4351325757575758e-05,
+      "loss": 0.4564,
+      "step": 5467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4346590909090912e-05,
+      "loss": 0.3948,
+      "step": 5468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4341856060606062e-05,
+      "loss": 0.4788,
+      "step": 5469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4337121212121213e-05,
+      "loss": 0.3718,
+      "step": 5470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4332386363636363e-05,
+      "loss": 0.4169,
+      "step": 5471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4327651515151517e-05,
+      "loss": 0.4431,
+      "step": 5472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4322916666666667e-05,
+      "loss": 0.4462,
+      "step": 5473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.431818181818182e-05,
+      "loss": 0.5827,
+      "step": 5474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.431344696969697e-05,
+      "loss": 0.3978,
+      "step": 5475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4308712121212122e-05,
+      "loss": 0.5106,
+      "step": 5476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4303977272727275e-05,
+      "loss": 0.4726,
+      "step": 5477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4299242424242426e-05,
+      "loss": 0.5429,
+      "step": 5478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4294507575757576e-05,
+      "loss": 0.4786,
+      "step": 5479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4289772727272727e-05,
+      "loss": 0.4025,
+      "step": 5480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.428503787878788e-05,
+      "loss": 0.4755,
+      "step": 5481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.428030303030303e-05,
+      "loss": 0.3324,
+      "step": 5482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4275568181818185e-05,
+      "loss": 0.5145,
+      "step": 5483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4270833333333335e-05,
+      "loss": 0.541,
+      "step": 5484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.426609848484849e-05,
+      "loss": 0.3353,
+      "step": 5485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.426136363636364e-05,
+      "loss": 0.411,
+      "step": 5486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.425662878787879e-05,
+      "loss": 0.3193,
+      "step": 5487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.425189393939394e-05,
+      "loss": 0.4514,
+      "step": 5488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.424715909090909e-05,
+      "loss": 0.3567,
+      "step": 5489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4242424242424244e-05,
+      "loss": 0.3584,
+      "step": 5490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4237689393939394e-05,
+      "loss": 0.3805,
+      "step": 5491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4232954545454548e-05,
+      "loss": 0.4811,
+      "step": 5492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.42282196969697e-05,
+      "loss": 0.5589,
+      "step": 5493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4223484848484852e-05,
+      "loss": 0.5586,
+      "step": 5494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4218750000000003e-05,
+      "loss": 0.4737,
+      "step": 5495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4214015151515153e-05,
+      "loss": 0.5736,
+      "step": 5496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4209280303030303e-05,
+      "loss": 0.534,
+      "step": 5497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4204545454545454e-05,
+      "loss": 0.5165,
+      "step": 5498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4199810606060607e-05,
+      "loss": 0.4824,
+      "step": 5499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4195075757575758e-05,
+      "loss": 0.5705,
+      "step": 5500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.419034090909091e-05,
+      "loss": 0.4245,
+      "step": 5501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4185606060606062e-05,
+      "loss": 0.5544,
+      "step": 5502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4180871212121216e-05,
+      "loss": 0.3939,
+      "step": 5503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4176136363636366e-05,
+      "loss": 0.3759,
+      "step": 5504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4171401515151516e-05,
+      "loss": 0.3597,
+      "step": 5505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4166666666666667e-05,
+      "loss": 0.4201,
+      "step": 5506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4161931818181817e-05,
+      "loss": 0.3891,
+      "step": 5507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.415719696969697e-05,
+      "loss": 0.4619,
+      "step": 5508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.415246212121212e-05,
+      "loss": 0.4816,
+      "step": 5509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4147727272727275e-05,
+      "loss": 0.3772,
+      "step": 5510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4142992424242425e-05,
+      "loss": 0.4199,
+      "step": 5511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.413825757575758e-05,
+      "loss": 0.38,
+      "step": 5512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.413352272727273e-05,
+      "loss": 0.4398,
+      "step": 5513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.412878787878788e-05,
+      "loss": 0.5436,
+      "step": 5514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.412405303030303e-05,
+      "loss": 0.462,
+      "step": 5515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.411931818181818e-05,
+      "loss": 0.5426,
+      "step": 5516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4114583333333334e-05,
+      "loss": 0.5542,
+      "step": 5517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4109848484848485e-05,
+      "loss": 0.5323,
+      "step": 5518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.410511363636364e-05,
+      "loss": 0.4431,
+      "step": 5519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.410037878787879e-05,
+      "loss": 0.5199,
+      "step": 5520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4095643939393943e-05,
+      "loss": 0.4657,
+      "step": 5521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4090909090909093e-05,
+      "loss": 0.4537,
+      "step": 5522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4086174242424243e-05,
+      "loss": 0.467,
+      "step": 5523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4081439393939394e-05,
+      "loss": 0.4979,
+      "step": 5524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4076704545454544e-05,
+      "loss": 0.5114,
+      "step": 5525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4071969696969698e-05,
+      "loss": 0.4615,
+      "step": 5526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.406723484848485e-05,
+      "loss": 0.5235,
+      "step": 5527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4062500000000002e-05,
+      "loss": 0.498,
+      "step": 5528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4057765151515153e-05,
+      "loss": 0.4901,
+      "step": 5529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4053030303030306e-05,
+      "loss": 0.523,
+      "step": 5530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4048295454545457e-05,
+      "loss": 0.4336,
+      "step": 5531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4043560606060607e-05,
+      "loss": 0.475,
+      "step": 5532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4038825757575757e-05,
+      "loss": 0.41,
+      "step": 5533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4034090909090908e-05,
+      "loss": 0.4872,
+      "step": 5534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.402935606060606e-05,
+      "loss": 0.3418,
+      "step": 5535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4024621212121212e-05,
+      "loss": 0.4013,
+      "step": 5536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4019886363636366e-05,
+      "loss": 0.5095,
+      "step": 5537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4015151515151516e-05,
+      "loss": 0.4829,
+      "step": 5538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.401041666666667e-05,
+      "loss": 0.3645,
+      "step": 5539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.400568181818182e-05,
+      "loss": 0.3843,
+      "step": 5540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.400094696969697e-05,
+      "loss": 0.6705,
+      "step": 5541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.399621212121212e-05,
+      "loss": 0.5944,
+      "step": 5542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3991477272727275e-05,
+      "loss": 0.4932,
+      "step": 5543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3986742424242425e-05,
+      "loss": 0.4425,
+      "step": 5544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3982007575757575e-05,
+      "loss": 0.5007,
+      "step": 5545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.397727272727273e-05,
+      "loss": 0.5007,
+      "step": 5546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.397253787878788e-05,
+      "loss": 0.4524,
+      "step": 5547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3967803030303033e-05,
+      "loss": 0.3948,
+      "step": 5548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3963068181818184e-05,
+      "loss": 0.4998,
+      "step": 5549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3958333333333334e-05,
+      "loss": 0.5146,
+      "step": 5550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3953598484848484e-05,
+      "loss": 0.4083,
+      "step": 5551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3948863636363638e-05,
+      "loss": 0.2924,
+      "step": 5552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.394412878787879e-05,
+      "loss": 0.4793,
+      "step": 5553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.393939393939394e-05,
+      "loss": 0.3909,
+      "step": 5554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3934659090909093e-05,
+      "loss": 0.3982,
+      "step": 5555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3929924242424243e-05,
+      "loss": 0.487,
+      "step": 5556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3925189393939397e-05,
+      "loss": 0.3608,
+      "step": 5557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3920454545454547e-05,
+      "loss": 0.4461,
+      "step": 5558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3915719696969698e-05,
+      "loss": 0.383,
+      "step": 5559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3910984848484848e-05,
+      "loss": 0.4261,
+      "step": 5560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3906250000000002e-05,
+      "loss": 0.3477,
+      "step": 5561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3901515151515152e-05,
+      "loss": 0.4843,
+      "step": 5562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3896780303030302e-05,
+      "loss": 0.4984,
+      "step": 5563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3892045454545456e-05,
+      "loss": 0.6455,
+      "step": 5564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3887310606060607e-05,
+      "loss": 0.4229,
+      "step": 5565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.388257575757576e-05,
+      "loss": 0.3647,
+      "step": 5566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.387784090909091e-05,
+      "loss": 0.4925,
+      "step": 5567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.387310606060606e-05,
+      "loss": 0.4093,
+      "step": 5568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.386837121212121e-05,
+      "loss": 0.4646,
+      "step": 5569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3863636363636365e-05,
+      "loss": 0.5033,
+      "step": 5570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3858901515151516e-05,
+      "loss": 0.4683,
+      "step": 5571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.385416666666667e-05,
+      "loss": 0.5891,
+      "step": 5572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.384943181818182e-05,
+      "loss": 0.5017,
+      "step": 5573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.384469696969697e-05,
+      "loss": 0.4023,
+      "step": 5574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3839962121212124e-05,
+      "loss": 0.4462,
+      "step": 5575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3835227272727274e-05,
+      "loss": 0.5965,
+      "step": 5576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3830492424242425e-05,
+      "loss": 0.4631,
+      "step": 5577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3825757575757575e-05,
+      "loss": 0.4017,
+      "step": 5578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.382102272727273e-05,
+      "loss": 0.3667,
+      "step": 5579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.381628787878788e-05,
+      "loss": 0.4986,
+      "step": 5580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3811553030303033e-05,
+      "loss": 0.4756,
+      "step": 5581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3806818181818183e-05,
+      "loss": 0.4216,
+      "step": 5582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3802083333333334e-05,
+      "loss": 0.4824,
+      "step": 5583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3797348484848487e-05,
+      "loss": 0.4549,
+      "step": 5584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3792613636363638e-05,
+      "loss": 0.3528,
+      "step": 5585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3787878787878788e-05,
+      "loss": 0.4909,
+      "step": 5586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.378314393939394e-05,
+      "loss": 0.6284,
+      "step": 5587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3778409090909092e-05,
+      "loss": 0.4499,
+      "step": 5588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3773674242424243e-05,
+      "loss": 0.5231,
+      "step": 5589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3768939393939396e-05,
+      "loss": 0.373,
+      "step": 5590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3764204545454547e-05,
+      "loss": 0.4581,
+      "step": 5591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.37594696969697e-05,
+      "loss": 0.4516,
+      "step": 5592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.375473484848485e-05,
+      "loss": 0.4606,
+      "step": 5593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.375e-05,
+      "loss": 0.5861,
+      "step": 5594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.374526515151515e-05,
+      "loss": 0.4798,
+      "step": 5595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3740530303030302e-05,
+      "loss": 0.3985,
+      "step": 5596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3735795454545456e-05,
+      "loss": 0.5294,
+      "step": 5597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3731060606060606e-05,
+      "loss": 0.5231,
+      "step": 5598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.372632575757576e-05,
+      "loss": 0.39,
+      "step": 5599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.372159090909091e-05,
+      "loss": 0.4397,
+      "step": 5600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3716856060606064e-05,
+      "loss": 0.4513,
+      "step": 5601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3712121212121214e-05,
+      "loss": 0.4863,
+      "step": 5602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3707386363636365e-05,
+      "loss": 0.5018,
+      "step": 5603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3702651515151515e-05,
+      "loss": 0.3314,
+      "step": 5604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3697916666666666e-05,
+      "loss": 0.4396,
+      "step": 5605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.369318181818182e-05,
+      "loss": 0.467,
+      "step": 5606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.368844696969697e-05,
+      "loss": 0.4117,
+      "step": 5607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3683712121212123e-05,
+      "loss": 0.5068,
+      "step": 5608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3678977272727274e-05,
+      "loss": 0.4376,
+      "step": 5609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3674242424242428e-05,
+      "loss": 0.4816,
+      "step": 5610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3669507575757578e-05,
+      "loss": 0.4061,
+      "step": 5611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.366477272727273e-05,
+      "loss": 0.5431,
+      "step": 5612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.366003787878788e-05,
+      "loss": 0.3689,
+      "step": 5613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.365530303030303e-05,
+      "loss": 0.3883,
+      "step": 5614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3650568181818183e-05,
+      "loss": 0.4114,
+      "step": 5615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3645833333333333e-05,
+      "loss": 0.5038,
+      "step": 5616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3641098484848487e-05,
+      "loss": 0.3931,
+      "step": 5617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3636363636363637e-05,
+      "loss": 0.373,
+      "step": 5618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.363162878787879e-05,
+      "loss": 0.6009,
+      "step": 5619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.362689393939394e-05,
+      "loss": 0.3198,
+      "step": 5620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3622159090909092e-05,
+      "loss": 0.401,
+      "step": 5621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3617424242424242e-05,
+      "loss": 0.4712,
+      "step": 5622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3612689393939393e-05,
+      "loss": 0.5577,
+      "step": 5623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3607954545454546e-05,
+      "loss": 0.46,
+      "step": 5624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3603219696969697e-05,
+      "loss": 0.3638,
+      "step": 5625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.359848484848485e-05,
+      "loss": 0.5059,
+      "step": 5626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.359375e-05,
+      "loss": 0.3992,
+      "step": 5627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3589015151515155e-05,
+      "loss": 0.4481,
+      "step": 5628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3584280303030305e-05,
+      "loss": 0.5006,
+      "step": 5629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3579545454545455e-05,
+      "loss": 0.4268,
+      "step": 5630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3574810606060606e-05,
+      "loss": 0.4126,
+      "step": 5631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3570075757575756e-05,
+      "loss": 0.4976,
+      "step": 5632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.356534090909091e-05,
+      "loss": 0.5179,
+      "step": 5633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.356060606060606e-05,
+      "loss": 0.389,
+      "step": 5634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3555871212121214e-05,
+      "loss": 0.4495,
+      "step": 5635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3551136363636364e-05,
+      "loss": 0.3998,
+      "step": 5636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3546401515151518e-05,
+      "loss": 0.4693,
+      "step": 5637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.354166666666667e-05,
+      "loss": 0.4001,
+      "step": 5638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.353693181818182e-05,
+      "loss": 0.5349,
+      "step": 5639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.353219696969697e-05,
+      "loss": 0.4449,
+      "step": 5640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.352746212121212e-05,
+      "loss": 0.4691,
+      "step": 5641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3522727272727273e-05,
+      "loss": 0.4875,
+      "step": 5642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3517992424242424e-05,
+      "loss": 0.5568,
+      "step": 5643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3513257575757578e-05,
+      "loss": 0.4535,
+      "step": 5644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3508522727272728e-05,
+      "loss": 0.4891,
+      "step": 5645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3503787878787882e-05,
+      "loss": 0.5642,
+      "step": 5646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3499053030303032e-05,
+      "loss": 0.4672,
+      "step": 5647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3494318181818182e-05,
+      "loss": 0.5068,
+      "step": 5648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3489583333333333e-05,
+      "loss": 0.416,
+      "step": 5649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3484848484848487e-05,
+      "loss": 0.4053,
+      "step": 5650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3480113636363637e-05,
+      "loss": 0.4879,
+      "step": 5651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3475378787878787e-05,
+      "loss": 0.4311,
+      "step": 5652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.347064393939394e-05,
+      "loss": 0.385,
+      "step": 5653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.346590909090909e-05,
+      "loss": 0.4326,
+      "step": 5654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3461174242424245e-05,
+      "loss": 0.5029,
+      "step": 5655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3456439393939396e-05,
+      "loss": 0.4237,
+      "step": 5656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3451704545454546e-05,
+      "loss": 0.4861,
+      "step": 5657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3446969696969696e-05,
+      "loss": 0.3877,
+      "step": 5658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.344223484848485e-05,
+      "loss": 0.4519,
+      "step": 5659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.34375e-05,
+      "loss": 0.4332,
+      "step": 5660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.343276515151515e-05,
+      "loss": 0.4038,
+      "step": 5661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3428030303030305e-05,
+      "loss": 0.4357,
+      "step": 5662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3423295454545455e-05,
+      "loss": 0.4975,
+      "step": 5663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.341856060606061e-05,
+      "loss": 0.4085,
+      "step": 5664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.341382575757576e-05,
+      "loss": 0.4836,
+      "step": 5665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.340909090909091e-05,
+      "loss": 0.43,
+      "step": 5666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.340435606060606e-05,
+      "loss": 0.453,
+      "step": 5667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3399621212121214e-05,
+      "loss": 0.457,
+      "step": 5668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3394886363636364e-05,
+      "loss": 0.4094,
+      "step": 5669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3390151515151514e-05,
+      "loss": 0.4843,
+      "step": 5670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3385416666666668e-05,
+      "loss": 0.4181,
+      "step": 5671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.338068181818182e-05,
+      "loss": 0.4525,
+      "step": 5672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3375946969696972e-05,
+      "loss": 0.461,
+      "step": 5673
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3371212121212123e-05,
+      "loss": 0.4036,
+      "step": 5674
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3366477272727273e-05,
+      "loss": 0.4195,
+      "step": 5675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3361742424242423e-05,
+      "loss": 0.6489,
+      "step": 5676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3357007575757577e-05,
+      "loss": 0.3447,
+      "step": 5677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3352272727272728e-05,
+      "loss": 0.3643,
+      "step": 5678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.334753787878788e-05,
+      "loss": 0.4612,
+      "step": 5679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.334280303030303e-05,
+      "loss": 0.4782,
+      "step": 5680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3338068181818182e-05,
+      "loss": 0.5679,
+      "step": 5681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.516,
+      "step": 5682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3328598484848486e-05,
+      "loss": 0.4369,
+      "step": 5683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3323863636363637e-05,
+      "loss": 0.5482,
+      "step": 5684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3319128787878787e-05,
+      "loss": 0.3645,
+      "step": 5685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.331439393939394e-05,
+      "loss": 0.393,
+      "step": 5686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.330965909090909e-05,
+      "loss": 0.409,
+      "step": 5687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3304924242424245e-05,
+      "loss": 0.3958,
+      "step": 5688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3300189393939395e-05,
+      "loss": 0.5436,
+      "step": 5689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3295454545454546e-05,
+      "loss": 0.4686,
+      "step": 5690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.32907196969697e-05,
+      "loss": 0.4737,
+      "step": 5691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.328598484848485e-05,
+      "loss": 0.4231,
+      "step": 5692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.328125e-05,
+      "loss": 0.3967,
+      "step": 5693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.327651515151515e-05,
+      "loss": 0.4857,
+      "step": 5694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3271780303030304e-05,
+      "loss": 0.4502,
+      "step": 5695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3267045454545455e-05,
+      "loss": 0.3891,
+      "step": 5696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.326231060606061e-05,
+      "loss": 0.3847,
+      "step": 5697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.325757575757576e-05,
+      "loss": 0.416,
+      "step": 5698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.325284090909091e-05,
+      "loss": 0.453,
+      "step": 5699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3248106060606063e-05,
+      "loss": 0.3507,
+      "step": 5700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3243371212121213e-05,
+      "loss": 0.3387,
+      "step": 5701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3238636363636364e-05,
+      "loss": 0.4126,
+      "step": 5702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3233901515151514e-05,
+      "loss": 0.3883,
+      "step": 5703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3229166666666668e-05,
+      "loss": 0.4247,
+      "step": 5704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3224431818181818e-05,
+      "loss": 0.4255,
+      "step": 5705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3219696969696972e-05,
+      "loss": 0.5091,
+      "step": 5706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3214962121212122e-05,
+      "loss": 0.5045,
+      "step": 5707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3210227272727276e-05,
+      "loss": 0.4349,
+      "step": 5708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3205492424242426e-05,
+      "loss": 0.4657,
+      "step": 5709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3200757575757577e-05,
+      "loss": 0.4178,
+      "step": 5710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3196022727272727e-05,
+      "loss": 0.3576,
+      "step": 5711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3191287878787878e-05,
+      "loss": 0.533,
+      "step": 5712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.318655303030303e-05,
+      "loss": 0.5329,
+      "step": 5713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.318181818181818e-05,
+      "loss": 0.5911,
+      "step": 5714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3177083333333335e-05,
+      "loss": 0.5204,
+      "step": 5715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3172348484848486e-05,
+      "loss": 0.4377,
+      "step": 5716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.316761363636364e-05,
+      "loss": 0.3179,
+      "step": 5717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.316287878787879e-05,
+      "loss": 0.4556,
+      "step": 5718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.315814393939394e-05,
+      "loss": 0.4552,
+      "step": 5719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.315340909090909e-05,
+      "loss": 0.3398,
+      "step": 5720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.314867424242424e-05,
+      "loss": 0.4,
+      "step": 5721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3143939393939395e-05,
+      "loss": 0.4796,
+      "step": 5722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3139204545454545e-05,
+      "loss": 0.5377,
+      "step": 5723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.31344696969697e-05,
+      "loss": 0.4254,
+      "step": 5724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.312973484848485e-05,
+      "loss": 0.5664,
+      "step": 5725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3125000000000003e-05,
+      "loss": 0.4587,
+      "step": 5726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3120265151515153e-05,
+      "loss": 0.545,
+      "step": 5727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3115530303030304e-05,
+      "loss": 0.4672,
+      "step": 5728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3110795454545454e-05,
+      "loss": 0.5761,
+      "step": 5729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3106060606060605e-05,
+      "loss": 0.4557,
+      "step": 5730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.310132575757576e-05,
+      "loss": 0.4334,
+      "step": 5731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.309659090909091e-05,
+      "loss": 0.4123,
+      "step": 5732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3091856060606062e-05,
+      "loss": 0.5079,
+      "step": 5733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3087121212121213e-05,
+      "loss": 0.4189,
+      "step": 5734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3082386363636367e-05,
+      "loss": 0.4613,
+      "step": 5735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3077651515151517e-05,
+      "loss": 0.4245,
+      "step": 5736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.307291666666667e-05,
+      "loss": 0.4608,
+      "step": 5737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3068181818181818e-05,
+      "loss": 0.5505,
+      "step": 5738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3063446969696968e-05,
+      "loss": 0.4856,
+      "step": 5739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3058712121212122e-05,
+      "loss": 0.4541,
+      "step": 5740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3053977272727272e-05,
+      "loss": 0.4449,
+      "step": 5741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3049242424242426e-05,
+      "loss": 0.5206,
+      "step": 5742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3044507575757576e-05,
+      "loss": 0.4745,
+      "step": 5743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.303977272727273e-05,
+      "loss": 0.4547,
+      "step": 5744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.303503787878788e-05,
+      "loss": 0.4285,
+      "step": 5745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3030303030303034e-05,
+      "loss": 0.4709,
+      "step": 5746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.302556818181818e-05,
+      "loss": 0.5817,
+      "step": 5747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.302083333333333e-05,
+      "loss": 0.4211,
+      "step": 5748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3016098484848485e-05,
+      "loss": 0.4349,
+      "step": 5749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3011363636363636e-05,
+      "loss": 0.3678,
+      "step": 5750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.300662878787879e-05,
+      "loss": 0.6067,
+      "step": 5751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.300189393939394e-05,
+      "loss": 0.6955,
+      "step": 5752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2997159090909094e-05,
+      "loss": 0.4559,
+      "step": 5753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2992424242424244e-05,
+      "loss": 0.3627,
+      "step": 5754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2987689393939398e-05,
+      "loss": 0.4269,
+      "step": 5755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2982954545454545e-05,
+      "loss": 0.4769,
+      "step": 5756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.29782196969697e-05,
+      "loss": 0.4719,
+      "step": 5757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.297348484848485e-05,
+      "loss": 0.4162,
+      "step": 5758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.296875e-05,
+      "loss": 0.4665,
+      "step": 5759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2964015151515153e-05,
+      "loss": 0.347,
+      "step": 5760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2959280303030303e-05,
+      "loss": 0.333,
+      "step": 5761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2954545454545457e-05,
+      "loss": 0.3348,
+      "step": 5762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2949810606060608e-05,
+      "loss": 0.4733,
+      "step": 5763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.294507575757576e-05,
+      "loss": 0.5658,
+      "step": 5764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2940340909090908e-05,
+      "loss": 0.4028,
+      "step": 5765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2935606060606062e-05,
+      "loss": 0.4072,
+      "step": 5766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2930871212121212e-05,
+      "loss": 0.4187,
+      "step": 5767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2926136363636363e-05,
+      "loss": 0.4996,
+      "step": 5768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2921401515151517e-05,
+      "loss": 0.6232,
+      "step": 5769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2916666666666667e-05,
+      "loss": 0.5219,
+      "step": 5770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.291193181818182e-05,
+      "loss": 0.4563,
+      "step": 5771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.290719696969697e-05,
+      "loss": 0.4697,
+      "step": 5772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2902462121212125e-05,
+      "loss": 0.4274,
+      "step": 5773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2897727272727272e-05,
+      "loss": 0.4171,
+      "step": 5774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2892992424242426e-05,
+      "loss": 0.4526,
+      "step": 5775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2888257575757576e-05,
+      "loss": 0.4679,
+      "step": 5776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2883522727272726e-05,
+      "loss": 0.3918,
+      "step": 5777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.287878787878788e-05,
+      "loss": 0.3353,
+      "step": 5778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.287405303030303e-05,
+      "loss": 0.4517,
+      "step": 5779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2869318181818184e-05,
+      "loss": 0.4801,
+      "step": 5780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2864583333333335e-05,
+      "loss": 0.3613,
+      "step": 5781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.285984848484849e-05,
+      "loss": 0.4927,
+      "step": 5782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2855113636363635e-05,
+      "loss": 0.6563,
+      "step": 5783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.285037878787879e-05,
+      "loss": 0.3063,
+      "step": 5784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.284564393939394e-05,
+      "loss": 0.4563,
+      "step": 5785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2840909090909093e-05,
+      "loss": 0.4855,
+      "step": 5786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2836174242424244e-05,
+      "loss": 0.496,
+      "step": 5787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2831439393939394e-05,
+      "loss": 0.48,
+      "step": 5788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2826704545454548e-05,
+      "loss": 0.3912,
+      "step": 5789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2821969696969698e-05,
+      "loss": 0.3648,
+      "step": 5790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2817234848484852e-05,
+      "loss": 0.6171,
+      "step": 5791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.28125e-05,
+      "loss": 0.596,
+      "step": 5792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2807765151515153e-05,
+      "loss": 0.4114,
+      "step": 5793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2803030303030303e-05,
+      "loss": 0.504,
+      "step": 5794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2798295454545457e-05,
+      "loss": 0.3825,
+      "step": 5795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2793560606060607e-05,
+      "loss": 0.4789,
+      "step": 5796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2788825757575757e-05,
+      "loss": 0.4822,
+      "step": 5797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.278409090909091e-05,
+      "loss": 0.4803,
+      "step": 5798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.277935606060606e-05,
+      "loss": 0.6778,
+      "step": 5799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2774621212121215e-05,
+      "loss": 0.492,
+      "step": 5800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2769886363636366e-05,
+      "loss": 0.5655,
+      "step": 5801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2765151515151516e-05,
+      "loss": 0.4865,
+      "step": 5802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2760416666666667e-05,
+      "loss": 0.5557,
+      "step": 5803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.275568181818182e-05,
+      "loss": 0.4102,
+      "step": 5804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.275094696969697e-05,
+      "loss": 0.5074,
+      "step": 5805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.274621212121212e-05,
+      "loss": 0.4967,
+      "step": 5806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2741477272727275e-05,
+      "loss": 0.3787,
+      "step": 5807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2736742424242425e-05,
+      "loss": 0.5107,
+      "step": 5808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.273200757575758e-05,
+      "loss": 0.5029,
+      "step": 5809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 0.4551,
+      "step": 5810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.272253787878788e-05,
+      "loss": 0.4595,
+      "step": 5811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.271780303030303e-05,
+      "loss": 0.3924,
+      "step": 5812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2713068181818184e-05,
+      "loss": 0.4691,
+      "step": 5813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2708333333333334e-05,
+      "loss": 0.4292,
+      "step": 5814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2703598484848488e-05,
+      "loss": 0.4025,
+      "step": 5815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2698863636363638e-05,
+      "loss": 0.4132,
+      "step": 5816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.269412878787879e-05,
+      "loss": 0.4263,
+      "step": 5817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2689393939393942e-05,
+      "loss": 0.4486,
+      "step": 5818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2684659090909093e-05,
+      "loss": 0.4336,
+      "step": 5819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2679924242424243e-05,
+      "loss": 0.5569,
+      "step": 5820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2675189393939394e-05,
+      "loss": 0.3778,
+      "step": 5821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2670454545454547e-05,
+      "loss": 0.4629,
+      "step": 5822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2665719696969698e-05,
+      "loss": 0.5098,
+      "step": 5823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.266098484848485e-05,
+      "loss": 0.46,
+      "step": 5824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2656250000000002e-05,
+      "loss": 0.4749,
+      "step": 5825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2651515151515152e-05,
+      "loss": 0.4439,
+      "step": 5826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2646780303030306e-05,
+      "loss": 0.4589,
+      "step": 5827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2642045454545456e-05,
+      "loss": 0.3896,
+      "step": 5828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2637310606060607e-05,
+      "loss": 0.402,
+      "step": 5829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2632575757575757e-05,
+      "loss": 0.4087,
+      "step": 5830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.262784090909091e-05,
+      "loss": 0.5904,
+      "step": 5831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.262310606060606e-05,
+      "loss": 0.3604,
+      "step": 5832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2618371212121215e-05,
+      "loss": 0.4268,
+      "step": 5833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2613636363636365e-05,
+      "loss": 0.4554,
+      "step": 5834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2608901515151516e-05,
+      "loss": 0.5077,
+      "step": 5835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.260416666666667e-05,
+      "loss": 0.434,
+      "step": 5836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.259943181818182e-05,
+      "loss": 0.4652,
+      "step": 5837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.259469696969697e-05,
+      "loss": 0.4711,
+      "step": 5838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.258996212121212e-05,
+      "loss": 0.4235,
+      "step": 5839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2585227272727274e-05,
+      "loss": 0.3764,
+      "step": 5840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2580492424242425e-05,
+      "loss": 0.4336,
+      "step": 5841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.257575757575758e-05,
+      "loss": 0.4638,
+      "step": 5842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.257102272727273e-05,
+      "loss": 0.4295,
+      "step": 5843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2566287878787883e-05,
+      "loss": 0.4749,
+      "step": 5844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2561553030303033e-05,
+      "loss": 0.4059,
+      "step": 5845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2556818181818183e-05,
+      "loss": 0.4215,
+      "step": 5846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2552083333333334e-05,
+      "loss": 0.478,
+      "step": 5847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2547348484848484e-05,
+      "loss": 0.3918,
+      "step": 5848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2542613636363638e-05,
+      "loss": 0.4202,
+      "step": 5849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2537878787878788e-05,
+      "loss": 0.4791,
+      "step": 5850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2533143939393942e-05,
+      "loss": 0.3587,
+      "step": 5851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2528409090909092e-05,
+      "loss": 0.5716,
+      "step": 5852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2523674242424246e-05,
+      "loss": 0.4008,
+      "step": 5853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2518939393939397e-05,
+      "loss": 0.4578,
+      "step": 5854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2514204545454547e-05,
+      "loss": 0.3086,
+      "step": 5855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2509469696969697e-05,
+      "loss": 0.3683,
+      "step": 5856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2504734848484848e-05,
+      "loss": 0.4872,
+      "step": 5857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.25e-05,
+      "loss": 0.4855,
+      "step": 5858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2495265151515152e-05,
+      "loss": 0.4218,
+      "step": 5859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2490530303030306e-05,
+      "loss": 0.5161,
+      "step": 5860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2485795454545456e-05,
+      "loss": 0.4009,
+      "step": 5861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.248106060606061e-05,
+      "loss": 0.3522,
+      "step": 5862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.247632575757576e-05,
+      "loss": 0.3441,
+      "step": 5863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.247159090909091e-05,
+      "loss": 0.4639,
+      "step": 5864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.246685606060606e-05,
+      "loss": 0.3873,
+      "step": 5865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.246212121212121e-05,
+      "loss": 0.4754,
+      "step": 5866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2457386363636365e-05,
+      "loss": 0.4577,
+      "step": 5867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2452651515151515e-05,
+      "loss": 0.5141,
+      "step": 5868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.244791666666667e-05,
+      "loss": 0.4647,
+      "step": 5869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.244318181818182e-05,
+      "loss": 0.4997,
+      "step": 5870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2438446969696973e-05,
+      "loss": 0.5028,
+      "step": 5871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2433712121212124e-05,
+      "loss": 0.5112,
+      "step": 5872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2428977272727274e-05,
+      "loss": 0.339,
+      "step": 5873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2424242424242424e-05,
+      "loss": 0.3397,
+      "step": 5874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2419507575757575e-05,
+      "loss": 0.4597,
+      "step": 5875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.241477272727273e-05,
+      "loss": 0.4686,
+      "step": 5876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.241003787878788e-05,
+      "loss": 0.4572,
+      "step": 5877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2405303030303033e-05,
+      "loss": 0.3435,
+      "step": 5878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2400568181818183e-05,
+      "loss": 0.4349,
+      "step": 5879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2395833333333337e-05,
+      "loss": 0.4703,
+      "step": 5880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2391098484848487e-05,
+      "loss": 0.3576,
+      "step": 5881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2386363636363637e-05,
+      "loss": 0.3596,
+      "step": 5882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2381628787878788e-05,
+      "loss": 0.566,
+      "step": 5883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2376893939393938e-05,
+      "loss": 0.4115,
+      "step": 5884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2372159090909092e-05,
+      "loss": 0.4376,
+      "step": 5885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2367424242424242e-05,
+      "loss": 0.4949,
+      "step": 5886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2362689393939396e-05,
+      "loss": 0.3908,
+      "step": 5887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2357954545454546e-05,
+      "loss": 0.4707,
+      "step": 5888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.23532196969697e-05,
+      "loss": 0.443,
+      "step": 5889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.234848484848485e-05,
+      "loss": 0.4234,
+      "step": 5890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.234375e-05,
+      "loss": 0.5028,
+      "step": 5891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.233901515151515e-05,
+      "loss": 0.4458,
+      "step": 5892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2334280303030302e-05,
+      "loss": 0.4402,
+      "step": 5893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2329545454545456e-05,
+      "loss": 0.4503,
+      "step": 5894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2324810606060606e-05,
+      "loss": 0.3804,
+      "step": 5895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.232007575757576e-05,
+      "loss": 0.4172,
+      "step": 5896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.231534090909091e-05,
+      "loss": 0.3391,
+      "step": 5897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2310606060606064e-05,
+      "loss": 0.4924,
+      "step": 5898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2305871212121214e-05,
+      "loss": 0.3101,
+      "step": 5899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2301136363636365e-05,
+      "loss": 0.7032,
+      "step": 5900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2296401515151515e-05,
+      "loss": 0.3395,
+      "step": 5901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.229166666666667e-05,
+      "loss": 0.4527,
+      "step": 5902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.228693181818182e-05,
+      "loss": 0.5159,
+      "step": 5903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.228219696969697e-05,
+      "loss": 0.3032,
+      "step": 5904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2277462121212123e-05,
+      "loss": 0.4268,
+      "step": 5905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2272727272727274e-05,
+      "loss": 0.5303,
+      "step": 5906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2267992424242427e-05,
+      "loss": 0.5852,
+      "step": 5907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2263257575757578e-05,
+      "loss": 0.3096,
+      "step": 5908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2258522727272728e-05,
+      "loss": 0.4529,
+      "step": 5909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.225378787878788e-05,
+      "loss": 0.3789,
+      "step": 5910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2249053030303032e-05,
+      "loss": 0.351,
+      "step": 5911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2244318181818183e-05,
+      "loss": 0.4763,
+      "step": 5912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2239583333333333e-05,
+      "loss": 0.3568,
+      "step": 5913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2234848484848487e-05,
+      "loss": 0.3766,
+      "step": 5914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2230113636363637e-05,
+      "loss": 0.4851,
+      "step": 5915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.222537878787879e-05,
+      "loss": 0.3366,
+      "step": 5916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.222064393939394e-05,
+      "loss": 0.4459,
+      "step": 5917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.221590909090909e-05,
+      "loss": 0.5764,
+      "step": 5918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2211174242424242e-05,
+      "loss": 0.4644,
+      "step": 5919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2206439393939396e-05,
+      "loss": 0.4563,
+      "step": 5920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2201704545454546e-05,
+      "loss": 0.2957,
+      "step": 5921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.21969696969697e-05,
+      "loss": 0.5738,
+      "step": 5922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.219223484848485e-05,
+      "loss": 0.3345,
+      "step": 5923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.21875e-05,
+      "loss": 0.3747,
+      "step": 5924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2182765151515154e-05,
+      "loss": 0.4339,
+      "step": 5925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2178030303030305e-05,
+      "loss": 0.4349,
+      "step": 5926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2173295454545455e-05,
+      "loss": 0.3958,
+      "step": 5927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2168560606060605e-05,
+      "loss": 0.5232,
+      "step": 5928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.216382575757576e-05,
+      "loss": 0.4299,
+      "step": 5929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.215909090909091e-05,
+      "loss": 0.3248,
+      "step": 5930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2154356060606063e-05,
+      "loss": 0.4414,
+      "step": 5931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2149621212121214e-05,
+      "loss": 0.4959,
+      "step": 5932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2144886363636364e-05,
+      "loss": 0.5757,
+      "step": 5933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2140151515151518e-05,
+      "loss": 0.5502,
+      "step": 5934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2135416666666668e-05,
+      "loss": 0.4382,
+      "step": 5935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.213068181818182e-05,
+      "loss": 0.457,
+      "step": 5936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.212594696969697e-05,
+      "loss": 0.4695,
+      "step": 5937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2121212121212123e-05,
+      "loss": 0.4611,
+      "step": 5938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2116477272727273e-05,
+      "loss": 0.3951,
+      "step": 5939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2111742424242427e-05,
+      "loss": 0.4687,
+      "step": 5940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2107007575757577e-05,
+      "loss": 0.6048,
+      "step": 5941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2102272727272728e-05,
+      "loss": 0.4265,
+      "step": 5942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.209753787878788e-05,
+      "loss": 0.3588,
+      "step": 5943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2092803030303032e-05,
+      "loss": 0.3916,
+      "step": 5944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2088068181818182e-05,
+      "loss": 0.3941,
+      "step": 5945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2083333333333333e-05,
+      "loss": 0.4909,
+      "step": 5946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2078598484848486e-05,
+      "loss": 0.3009,
+      "step": 5947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2073863636363637e-05,
+      "loss": 0.3323,
+      "step": 5948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.206912878787879e-05,
+      "loss": 0.3724,
+      "step": 5949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.206439393939394e-05,
+      "loss": 0.4262,
+      "step": 5950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2059659090909095e-05,
+      "loss": 0.555,
+      "step": 5951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2054924242424245e-05,
+      "loss": 0.4606,
+      "step": 5952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2050189393939395e-05,
+      "loss": 0.5319,
+      "step": 5953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2045454545454546e-05,
+      "loss": 0.4734,
+      "step": 5954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2040719696969696e-05,
+      "loss": 0.4766,
+      "step": 5955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.203598484848485e-05,
+      "loss": 0.5076,
+      "step": 5956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.203125e-05,
+      "loss": 0.3385,
+      "step": 5957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2026515151515154e-05,
+      "loss": 0.3579,
+      "step": 5958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2021780303030304e-05,
+      "loss": 0.4367,
+      "step": 5959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2017045454545458e-05,
+      "loss": 0.4624,
+      "step": 5960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.201231060606061e-05,
+      "loss": 0.4069,
+      "step": 5961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.200757575757576e-05,
+      "loss": 0.4815,
+      "step": 5962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.200284090909091e-05,
+      "loss": 0.4233,
+      "step": 5963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.199810606060606e-05,
+      "loss": 0.4287,
+      "step": 5964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1993371212121213e-05,
+      "loss": 0.3716,
+      "step": 5965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1988636363636364e-05,
+      "loss": 0.4719,
+      "step": 5966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1983901515151517e-05,
+      "loss": 0.4335,
+      "step": 5967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1979166666666668e-05,
+      "loss": 0.4085,
+      "step": 5968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.197443181818182e-05,
+      "loss": 0.53,
+      "step": 5969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1969696969696972e-05,
+      "loss": 0.3153,
+      "step": 5970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1964962121212122e-05,
+      "loss": 0.4781,
+      "step": 5971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1960227272727273e-05,
+      "loss": 0.4244,
+      "step": 5972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1955492424242423e-05,
+      "loss": 0.3537,
+      "step": 5973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1950757575757577e-05,
+      "loss": 0.4655,
+      "step": 5974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1946022727272727e-05,
+      "loss": 0.4563,
+      "step": 5975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.194128787878788e-05,
+      "loss": 0.5539,
+      "step": 5976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.193655303030303e-05,
+      "loss": 0.5151,
+      "step": 5977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1931818181818185e-05,
+      "loss": 0.4541,
+      "step": 5978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1927083333333336e-05,
+      "loss": 0.4539,
+      "step": 5979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1922348484848486e-05,
+      "loss": 0.5291,
+      "step": 5980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1917613636363636e-05,
+      "loss": 0.4761,
+      "step": 5981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1912878787878787e-05,
+      "loss": 0.5339,
+      "step": 5982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.190814393939394e-05,
+      "loss": 0.4447,
+      "step": 5983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.190340909090909e-05,
+      "loss": 0.5483,
+      "step": 5984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1898674242424245e-05,
+      "loss": 0.3845,
+      "step": 5985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1893939393939395e-05,
+      "loss": 0.4091,
+      "step": 5986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.188920454545455e-05,
+      "loss": 0.3236,
+      "step": 5987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.18844696969697e-05,
+      "loss": 0.431,
+      "step": 5988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.187973484848485e-05,
+      "loss": 0.5259,
+      "step": 5989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1875e-05,
+      "loss": 0.3907,
+      "step": 5990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.187026515151515e-05,
+      "loss": 0.415,
+      "step": 5991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1865530303030304e-05,
+      "loss": 0.4741,
+      "step": 5992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1860795454545454e-05,
+      "loss": 0.4495,
+      "step": 5993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1856060606060608e-05,
+      "loss": 0.3203,
+      "step": 5994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.185132575757576e-05,
+      "loss": 0.3866,
+      "step": 5995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1846590909090912e-05,
+      "loss": 0.479,
+      "step": 5996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1841856060606063e-05,
+      "loss": 0.4603,
+      "step": 5997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1837121212121213e-05,
+      "loss": 0.4039,
+      "step": 5998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1832386363636363e-05,
+      "loss": 0.4007,
+      "step": 5999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1827651515151514e-05,
+      "loss": 0.3432,
+      "step": 6000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1822916666666667e-05,
+      "loss": 0.4918,
+      "step": 6001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1818181818181818e-05,
+      "loss": 0.4183,
+      "step": 6002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.181344696969697e-05,
+      "loss": 0.3898,
+      "step": 6003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1808712121212122e-05,
+      "loss": 0.5069,
+      "step": 6004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1803977272727276e-05,
+      "loss": 0.5909,
+      "step": 6005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1799242424242426e-05,
+      "loss": 0.4732,
+      "step": 6006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1794507575757576e-05,
+      "loss": 0.524,
+      "step": 6007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1789772727272727e-05,
+      "loss": 0.462,
+      "step": 6008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.178503787878788e-05,
+      "loss": 0.4246,
+      "step": 6009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.178030303030303e-05,
+      "loss": 0.539,
+      "step": 6010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.177556818181818e-05,
+      "loss": 0.3879,
+      "step": 6011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1770833333333335e-05,
+      "loss": 0.4269,
+      "step": 6012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1766098484848485e-05,
+      "loss": 0.4738,
+      "step": 6013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.176136363636364e-05,
+      "loss": 0.4772,
+      "step": 6014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.175662878787879e-05,
+      "loss": 0.3195,
+      "step": 6015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.175189393939394e-05,
+      "loss": 0.5058,
+      "step": 6016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.174715909090909e-05,
+      "loss": 0.3953,
+      "step": 6017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1742424242424244e-05,
+      "loss": 0.376,
+      "step": 6018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1737689393939394e-05,
+      "loss": 0.484,
+      "step": 6019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1732954545454545e-05,
+      "loss": 0.4293,
+      "step": 6020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.17282196969697e-05,
+      "loss": 0.4001,
+      "step": 6021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.172348484848485e-05,
+      "loss": 0.5091,
+      "step": 6022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1718750000000003e-05,
+      "loss": 0.3986,
+      "step": 6023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1714015151515153e-05,
+      "loss": 0.3975,
+      "step": 6024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1709280303030304e-05,
+      "loss": 0.5603,
+      "step": 6025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1704545454545454e-05,
+      "loss": 0.3356,
+      "step": 6026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1699810606060608e-05,
+      "loss": 0.509,
+      "step": 6027
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1695075757575758e-05,
+      "loss": 0.4093,
+      "step": 6028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.169034090909091e-05,
+      "loss": 0.4728,
+      "step": 6029
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1685606060606062e-05,
+      "loss": 0.5134,
+      "step": 6030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1680871212121213e-05,
+      "loss": 0.4247,
+      "step": 6031
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1676136363636366e-05,
+      "loss": 0.4823,
+      "step": 6032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1671401515151517e-05,
+      "loss": 0.4474,
+      "step": 6033
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 0.5434,
+      "step": 6034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1661931818181817e-05,
+      "loss": 0.4292,
+      "step": 6035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.165719696969697e-05,
+      "loss": 0.3651,
+      "step": 6036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.165246212121212e-05,
+      "loss": 0.4333,
+      "step": 6037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1647727272727275e-05,
+      "loss": 0.3608,
+      "step": 6038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1642992424242426e-05,
+      "loss": 0.3846,
+      "step": 6039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1638257575757576e-05,
+      "loss": 0.4259,
+      "step": 6040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.163352272727273e-05,
+      "loss": 0.5015,
+      "step": 6041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.162878787878788e-05,
+      "loss": 0.482,
+      "step": 6042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.162405303030303e-05,
+      "loss": 0.3772,
+      "step": 6043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.161931818181818e-05,
+      "loss": 0.4233,
+      "step": 6044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1614583333333335e-05,
+      "loss": 0.4206,
+      "step": 6045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1609848484848485e-05,
+      "loss": 0.411,
+      "step": 6046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.160511363636364e-05,
+      "loss": 0.3346,
+      "step": 6047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.160037878787879e-05,
+      "loss": 0.4574,
+      "step": 6048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.159564393939394e-05,
+      "loss": 0.4783,
+      "step": 6049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1590909090909093e-05,
+      "loss": 0.5519,
+      "step": 6050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1586174242424244e-05,
+      "loss": 0.4612,
+      "step": 6051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1581439393939394e-05,
+      "loss": 0.5871,
+      "step": 6052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1576704545454544e-05,
+      "loss": 0.394,
+      "step": 6053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1571969696969698e-05,
+      "loss": 0.5255,
+      "step": 6054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.156723484848485e-05,
+      "loss": 0.4224,
+      "step": 6055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1562500000000002e-05,
+      "loss": 0.5056,
+      "step": 6056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1557765151515153e-05,
+      "loss": 0.4547,
+      "step": 6057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1553030303030303e-05,
+      "loss": 0.5515,
+      "step": 6058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1548295454545457e-05,
+      "loss": 0.5037,
+      "step": 6059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1543560606060607e-05,
+      "loss": 0.4894,
+      "step": 6060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1538825757575758e-05,
+      "loss": 0.4418,
+      "step": 6061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1534090909090908e-05,
+      "loss": 0.5169,
+      "step": 6062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1529356060606062e-05,
+      "loss": 0.5725,
+      "step": 6063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1524621212121212e-05,
+      "loss": 0.5622,
+      "step": 6064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1519886363636366e-05,
+      "loss": 0.5622,
+      "step": 6065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1515151515151516e-05,
+      "loss": 0.4937,
+      "step": 6066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.151041666666667e-05,
+      "loss": 0.3864,
+      "step": 6067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.150568181818182e-05,
+      "loss": 0.5438,
+      "step": 6068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.150094696969697e-05,
+      "loss": 0.4944,
+      "step": 6069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.149621212121212e-05,
+      "loss": 0.4901,
+      "step": 6070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.149147727272727e-05,
+      "loss": 0.4462,
+      "step": 6071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1486742424242425e-05,
+      "loss": 0.4001,
+      "step": 6072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1482007575757576e-05,
+      "loss": 0.4049,
+      "step": 6073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.147727272727273e-05,
+      "loss": 0.4698,
+      "step": 6074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.147253787878788e-05,
+      "loss": 0.4924,
+      "step": 6075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1467803030303034e-05,
+      "loss": 0.5709,
+      "step": 6076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1463068181818184e-05,
+      "loss": 0.5408,
+      "step": 6077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1458333333333334e-05,
+      "loss": 0.5334,
+      "step": 6078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1453598484848485e-05,
+      "loss": 0.4622,
+      "step": 6079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1448863636363635e-05,
+      "loss": 0.5155,
+      "step": 6080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.144412878787879e-05,
+      "loss": 0.3808,
+      "step": 6081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.143939393939394e-05,
+      "loss": 0.4454,
+      "step": 6082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1434659090909093e-05,
+      "loss": 0.5077,
+      "step": 6083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1429924242424243e-05,
+      "loss": 0.3945,
+      "step": 6084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1425189393939397e-05,
+      "loss": 0.4321,
+      "step": 6085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1420454545454547e-05,
+      "loss": 0.4462,
+      "step": 6086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1415719696969698e-05,
+      "loss": 0.553,
+      "step": 6087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1410984848484848e-05,
+      "loss": 0.3544,
+      "step": 6088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.140625e-05,
+      "loss": 0.4176,
+      "step": 6089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1401515151515152e-05,
+      "loss": 0.4634,
+      "step": 6090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1396780303030303e-05,
+      "loss": 0.4237,
+      "step": 6091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1392045454545456e-05,
+      "loss": 0.3704,
+      "step": 6092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1387310606060607e-05,
+      "loss": 0.4572,
+      "step": 6093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.138257575757576e-05,
+      "loss": 0.4822,
+      "step": 6094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.137784090909091e-05,
+      "loss": 0.4788,
+      "step": 6095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.137310606060606e-05,
+      "loss": 0.5357,
+      "step": 6096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1368371212121212e-05,
+      "loss": 0.4785,
+      "step": 6097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1363636363636362e-05,
+      "loss": 0.3614,
+      "step": 6098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1358901515151516e-05,
+      "loss": 0.5216,
+      "step": 6099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1354166666666666e-05,
+      "loss": 0.4628,
+      "step": 6100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.134943181818182e-05,
+      "loss": 0.5103,
+      "step": 6101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.134469696969697e-05,
+      "loss": 0.401,
+      "step": 6102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1339962121212124e-05,
+      "loss": 0.561,
+      "step": 6103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1335227272727274e-05,
+      "loss": 0.369,
+      "step": 6104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1330492424242425e-05,
+      "loss": 0.4565,
+      "step": 6105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1325757575757575e-05,
+      "loss": 0.4575,
+      "step": 6106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1321022727272726e-05,
+      "loss": 0.3149,
+      "step": 6107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.131628787878788e-05,
+      "loss": 0.3811,
+      "step": 6108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.131155303030303e-05,
+      "loss": 0.3631,
+      "step": 6109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1306818181818183e-05,
+      "loss": 0.5372,
+      "step": 6110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1302083333333334e-05,
+      "loss": 0.3536,
+      "step": 6111
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1297348484848488e-05,
+      "loss": 0.3998,
+      "step": 6112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1292613636363638e-05,
+      "loss": 0.4562,
+      "step": 6113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.128787878787879e-05,
+      "loss": 0.3317,
+      "step": 6114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.128314393939394e-05,
+      "loss": 0.5028,
+      "step": 6115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1278409090909093e-05,
+      "loss": 0.3831,
+      "step": 6116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1273674242424243e-05,
+      "loss": 0.3016,
+      "step": 6117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1268939393939393e-05,
+      "loss": 0.4282,
+      "step": 6118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1264204545454547e-05,
+      "loss": 0.3159,
+      "step": 6119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1259469696969697e-05,
+      "loss": 0.4556,
+      "step": 6120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.125473484848485e-05,
+      "loss": 0.5214,
+      "step": 6121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.125e-05,
+      "loss": 0.5921,
+      "step": 6122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1245265151515152e-05,
+      "loss": 0.3673,
+      "step": 6123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1240530303030302e-05,
+      "loss": 0.3425,
+      "step": 6124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1235795454545456e-05,
+      "loss": 0.3984,
+      "step": 6125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1231060606060606e-05,
+      "loss": 0.463,
+      "step": 6126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1226325757575757e-05,
+      "loss": 0.4376,
+      "step": 6127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.122159090909091e-05,
+      "loss": 0.5966,
+      "step": 6128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.121685606060606e-05,
+      "loss": 0.4402,
+      "step": 6129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1212121212121215e-05,
+      "loss": 0.5215,
+      "step": 6130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1207386363636365e-05,
+      "loss": 0.4811,
+      "step": 6131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1202651515151515e-05,
+      "loss": 0.3186,
+      "step": 6132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1197916666666666e-05,
+      "loss": 0.2714,
+      "step": 6133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.119318181818182e-05,
+      "loss": 0.3005,
+      "step": 6134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.118844696969697e-05,
+      "loss": 0.5196,
+      "step": 6135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.118371212121212e-05,
+      "loss": 0.3082,
+      "step": 6136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1178977272727274e-05,
+      "loss": 0.4953,
+      "step": 6137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1174242424242424e-05,
+      "loss": 0.3144,
+      "step": 6138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1169507575757578e-05,
+      "loss": 0.5309,
+      "step": 6139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.116477272727273e-05,
+      "loss": 0.4603,
+      "step": 6140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.116003787878788e-05,
+      "loss": 0.413,
+      "step": 6141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.115530303030303e-05,
+      "loss": 0.3509,
+      "step": 6142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1150568181818183e-05,
+      "loss": 0.5377,
+      "step": 6143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1145833333333333e-05,
+      "loss": 0.3546,
+      "step": 6144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1141098484848487e-05,
+      "loss": 0.484,
+      "step": 6145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1136363636363638e-05,
+      "loss": 0.6052,
+      "step": 6146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1131628787878788e-05,
+      "loss": 0.5215,
+      "step": 6147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1126893939393942e-05,
+      "loss": 0.4326,
+      "step": 6148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1122159090909092e-05,
+      "loss": 0.4535,
+      "step": 6149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1117424242424242e-05,
+      "loss": 0.2811,
+      "step": 6150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1112689393939393e-05,
+      "loss": 0.4727,
+      "step": 6151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1107954545454547e-05,
+      "loss": 0.34,
+      "step": 6152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1103219696969697e-05,
+      "loss": 0.4074,
+      "step": 6153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.109848484848485e-05,
+      "loss": 0.5392,
+      "step": 6154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.109375e-05,
+      "loss": 0.3924,
+      "step": 6155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.108901515151515e-05,
+      "loss": 0.3543,
+      "step": 6156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1084280303030305e-05,
+      "loss": 0.3855,
+      "step": 6157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1079545454545456e-05,
+      "loss": 0.4782,
+      "step": 6158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1074810606060606e-05,
+      "loss": 0.505,
+      "step": 6159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1070075757575756e-05,
+      "loss": 0.444,
+      "step": 6160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.106534090909091e-05,
+      "loss": 0.5535,
+      "step": 6161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.106060606060606e-05,
+      "loss": 0.3033,
+      "step": 6162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1055871212121214e-05,
+      "loss": 0.4192,
+      "step": 6163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1051136363636365e-05,
+      "loss": 0.4297,
+      "step": 6164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1046401515151515e-05,
+      "loss": 0.4857,
+      "step": 6165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.104166666666667e-05,
+      "loss": 0.5062,
+      "step": 6166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.103693181818182e-05,
+      "loss": 0.443,
+      "step": 6167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.103219696969697e-05,
+      "loss": 0.4349,
+      "step": 6168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.102746212121212e-05,
+      "loss": 0.4678,
+      "step": 6169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1022727272727274e-05,
+      "loss": 0.4833,
+      "step": 6170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1017992424242424e-05,
+      "loss": 0.4937,
+      "step": 6171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1013257575757578e-05,
+      "loss": 0.4913,
+      "step": 6172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1008522727272728e-05,
+      "loss": 0.4994,
+      "step": 6173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1003787878787882e-05,
+      "loss": 0.3548,
+      "step": 6174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0999053030303032e-05,
+      "loss": 0.3588,
+      "step": 6175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0994318181818183e-05,
+      "loss": 0.434,
+      "step": 6176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0989583333333333e-05,
+      "loss": 0.4585,
+      "step": 6177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0984848484848483e-05,
+      "loss": 0.4482,
+      "step": 6178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0980113636363637e-05,
+      "loss": 0.4442,
+      "step": 6179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0975378787878788e-05,
+      "loss": 0.4192,
+      "step": 6180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.097064393939394e-05,
+      "loss": 0.4659,
+      "step": 6181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.096590909090909e-05,
+      "loss": 0.5073,
+      "step": 6182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0961174242424245e-05,
+      "loss": 0.4132,
+      "step": 6183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0956439393939396e-05,
+      "loss": 0.5139,
+      "step": 6184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0951704545454546e-05,
+      "loss": 0.4601,
+      "step": 6185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0946969696969697e-05,
+      "loss": 0.4517,
+      "step": 6186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0942234848484847e-05,
+      "loss": 0.3392,
+      "step": 6187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.09375e-05,
+      "loss": 0.3767,
+      "step": 6188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.093276515151515e-05,
+      "loss": 0.4199,
+      "step": 6189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0928030303030305e-05,
+      "loss": 0.3393,
+      "step": 6190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0923295454545455e-05,
+      "loss": 0.3725,
+      "step": 6191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.091856060606061e-05,
+      "loss": 0.4169,
+      "step": 6192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.091382575757576e-05,
+      "loss": 0.4634,
+      "step": 6193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.090909090909091e-05,
+      "loss": 0.4122,
+      "step": 6194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.090435606060606e-05,
+      "loss": 0.4154,
+      "step": 6195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.089962121212121e-05,
+      "loss": 0.4734,
+      "step": 6196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0894886363636364e-05,
+      "loss": 0.3855,
+      "step": 6197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0890151515151515e-05,
+      "loss": 0.4881,
+      "step": 6198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.088541666666667e-05,
+      "loss": 0.3997,
+      "step": 6199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.088068181818182e-05,
+      "loss": 0.3117,
+      "step": 6200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0875946969696973e-05,
+      "loss": 0.378,
+      "step": 6201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0871212121212123e-05,
+      "loss": 0.5117,
+      "step": 6202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0866477272727277e-05,
+      "loss": 0.388,
+      "step": 6203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0861742424242424e-05,
+      "loss": 0.4541,
+      "step": 6204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0857007575757574e-05,
+      "loss": 0.4985,
+      "step": 6205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0852272727272728e-05,
+      "loss": 0.3922,
+      "step": 6206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0847537878787878e-05,
+      "loss": 0.4262,
+      "step": 6207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0842803030303032e-05,
+      "loss": 0.3903,
+      "step": 6208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0838068181818182e-05,
+      "loss": 0.5457,
+      "step": 6209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.36,
+      "step": 6210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0828598484848486e-05,
+      "loss": 0.3653,
+      "step": 6211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.082386363636364e-05,
+      "loss": 0.4063,
+      "step": 6212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.081912878787879e-05,
+      "loss": 0.4879,
+      "step": 6213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0814393939393938e-05,
+      "loss": 0.455,
+      "step": 6214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.080965909090909e-05,
+      "loss": 0.5002,
+      "step": 6215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.080492424242424e-05,
+      "loss": 0.2965,
+      "step": 6216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0800189393939395e-05,
+      "loss": 0.4862,
+      "step": 6217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0795454545454546e-05,
+      "loss": 0.5911,
+      "step": 6218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.07907196969697e-05,
+      "loss": 0.4326,
+      "step": 6219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.078598484848485e-05,
+      "loss": 0.3909,
+      "step": 6220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0781250000000004e-05,
+      "loss": 0.4758,
+      "step": 6221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0776515151515154e-05,
+      "loss": 0.5125,
+      "step": 6222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.07717803030303e-05,
+      "loss": 0.3168,
+      "step": 6223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0767045454545455e-05,
+      "loss": 0.5936,
+      "step": 6224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0762310606060605e-05,
+      "loss": 0.5135,
+      "step": 6225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.075757575757576e-05,
+      "loss": 0.397,
+      "step": 6226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.075284090909091e-05,
+      "loss": 0.5328,
+      "step": 6227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0748106060606063e-05,
+      "loss": 0.4227,
+      "step": 6228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0743371212121213e-05,
+      "loss": 0.3898,
+      "step": 6229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0738636363636367e-05,
+      "loss": 0.3659,
+      "step": 6230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0733901515151518e-05,
+      "loss": 0.3989,
+      "step": 6231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0729166666666668e-05,
+      "loss": 0.4119,
+      "step": 6232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.072443181818182e-05,
+      "loss": 0.4821,
+      "step": 6233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.071969696969697e-05,
+      "loss": 0.4132,
+      "step": 6234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0714962121212122e-05,
+      "loss": 0.4008,
+      "step": 6235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0710227272727273e-05,
+      "loss": 0.4651,
+      "step": 6236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0705492424242427e-05,
+      "loss": 0.5015,
+      "step": 6237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0700757575757577e-05,
+      "loss": 0.4559,
+      "step": 6238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.069602272727273e-05,
+      "loss": 0.4189,
+      "step": 6239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.069128787878788e-05,
+      "loss": 0.3559,
+      "step": 6240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.068655303030303e-05,
+      "loss": 0.5143,
+      "step": 6241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0681818181818182e-05,
+      "loss": 0.5406,
+      "step": 6242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0677083333333332e-05,
+      "loss": 0.5949,
+      "step": 6243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0672348484848486e-05,
+      "loss": 0.5304,
+      "step": 6244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0667613636363636e-05,
+      "loss": 0.447,
+      "step": 6245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.066287878787879e-05,
+      "loss": 0.3041,
+      "step": 6246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.065814393939394e-05,
+      "loss": 0.3579,
+      "step": 6247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0653409090909094e-05,
+      "loss": 0.4428,
+      "step": 6248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0648674242424245e-05,
+      "loss": 0.4493,
+      "step": 6249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0643939393939395e-05,
+      "loss": 0.3527,
+      "step": 6250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0639204545454545e-05,
+      "loss": 0.4645,
+      "step": 6251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.06344696969697e-05,
+      "loss": 0.4438,
+      "step": 6252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.062973484848485e-05,
+      "loss": 0.5249,
+      "step": 6253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0625e-05,
+      "loss": 0.3607,
+      "step": 6254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0620265151515154e-05,
+      "loss": 0.4645,
+      "step": 6255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0615530303030304e-05,
+      "loss": 0.36,
+      "step": 6256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0610795454545458e-05,
+      "loss": 0.5723,
+      "step": 6257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0606060606060608e-05,
+      "loss": 0.3743,
+      "step": 6258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.060132575757576e-05,
+      "loss": 0.4969,
+      "step": 6259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.059659090909091e-05,
+      "loss": 0.4374,
+      "step": 6260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0591856060606063e-05,
+      "loss": 0.4217,
+      "step": 6261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0587121212121213e-05,
+      "loss": 0.3992,
+      "step": 6262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0582386363636363e-05,
+      "loss": 0.509,
+      "step": 6263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0577651515151517e-05,
+      "loss": 0.4397,
+      "step": 6264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0572916666666668e-05,
+      "loss": 0.384,
+      "step": 6265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.056818181818182e-05,
+      "loss": 0.6138,
+      "step": 6266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.056344696969697e-05,
+      "loss": 0.4523,
+      "step": 6267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0558712121212122e-05,
+      "loss": 0.4853,
+      "step": 6268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0553977272727272e-05,
+      "loss": 0.2708,
+      "step": 6269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0549242424242426e-05,
+      "loss": 0.3997,
+      "step": 6270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0544507575757577e-05,
+      "loss": 0.496,
+      "step": 6271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0539772727272727e-05,
+      "loss": 0.4802,
+      "step": 6272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.053503787878788e-05,
+      "loss": 0.3405,
+      "step": 6273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.053030303030303e-05,
+      "loss": 0.4845,
+      "step": 6274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0525568181818185e-05,
+      "loss": 0.3883,
+      "step": 6275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0520833333333335e-05,
+      "loss": 0.3508,
+      "step": 6276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0516098484848486e-05,
+      "loss": 0.6221,
+      "step": 6277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0511363636363636e-05,
+      "loss": 0.4168,
+      "step": 6278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.050662878787879e-05,
+      "loss": 0.3755,
+      "step": 6279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.050189393939394e-05,
+      "loss": 0.5146,
+      "step": 6280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0497159090909094e-05,
+      "loss": 0.3707,
+      "step": 6281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0492424242424244e-05,
+      "loss": 0.5212,
+      "step": 6282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0487689393939395e-05,
+      "loss": 0.4971,
+      "step": 6283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.048295454545455e-05,
+      "loss": 0.3775,
+      "step": 6284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.04782196969697e-05,
+      "loss": 0.3343,
+      "step": 6285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.047348484848485e-05,
+      "loss": 0.3958,
+      "step": 6286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.046875e-05,
+      "loss": 0.4936,
+      "step": 6287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0464015151515153e-05,
+      "loss": 0.4694,
+      "step": 6288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0459280303030304e-05,
+      "loss": 0.4934,
+      "step": 6289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0454545454545457e-05,
+      "loss": 0.3307,
+      "step": 6290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0449810606060608e-05,
+      "loss": 0.4253,
+      "step": 6291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0445075757575758e-05,
+      "loss": 0.3889,
+      "step": 6292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0440340909090912e-05,
+      "loss": 0.3026,
+      "step": 6293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0435606060606062e-05,
+      "loss": 0.444,
+      "step": 6294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0430871212121213e-05,
+      "loss": 0.438,
+      "step": 6295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0426136363636363e-05,
+      "loss": 0.4696,
+      "step": 6296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0421401515151517e-05,
+      "loss": 0.4287,
+      "step": 6297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0416666666666667e-05,
+      "loss": 0.4612,
+      "step": 6298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.041193181818182e-05,
+      "loss": 0.433,
+      "step": 6299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.040719696969697e-05,
+      "loss": 0.3274,
+      "step": 6300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.040246212121212e-05,
+      "loss": 0.3603,
+      "step": 6301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0397727272727275e-05,
+      "loss": 0.3421,
+      "step": 6302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0392992424242426e-05,
+      "loss": 0.3547,
+      "step": 6303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0388257575757576e-05,
+      "loss": 0.4174,
+      "step": 6304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0383522727272727e-05,
+      "loss": 0.5691,
+      "step": 6305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.037878787878788e-05,
+      "loss": 0.5017,
+      "step": 6306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.037405303030303e-05,
+      "loss": 0.4225,
+      "step": 6307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0369318181818184e-05,
+      "loss": 0.4776,
+      "step": 6308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0364583333333335e-05,
+      "loss": 0.4898,
+      "step": 6309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.035984848484849e-05,
+      "loss": 0.5686,
+      "step": 6310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.035511363636364e-05,
+      "loss": 0.3657,
+      "step": 6311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.035037878787879e-05,
+      "loss": 0.5465,
+      "step": 6312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.034564393939394e-05,
+      "loss": 0.4291,
+      "step": 6313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.034090909090909e-05,
+      "loss": 0.5603,
+      "step": 6314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0336174242424244e-05,
+      "loss": 0.3656,
+      "step": 6315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0331439393939394e-05,
+      "loss": 0.2884,
+      "step": 6316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0326704545454548e-05,
+      "loss": 0.3357,
+      "step": 6317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.03219696969697e-05,
+      "loss": 0.4333,
+      "step": 6318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0317234848484852e-05,
+      "loss": 0.4864,
+      "step": 6319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0312500000000002e-05,
+      "loss": 0.3488,
+      "step": 6320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0307765151515153e-05,
+      "loss": 0.5082,
+      "step": 6321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0303030303030303e-05,
+      "loss": 0.5089,
+      "step": 6322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0298295454545454e-05,
+      "loss": 0.5075,
+      "step": 6323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0293560606060607e-05,
+      "loss": 0.3975,
+      "step": 6324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0288825757575758e-05,
+      "loss": 0.4102,
+      "step": 6325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.028409090909091e-05,
+      "loss": 0.5146,
+      "step": 6326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0279356060606062e-05,
+      "loss": 0.5548,
+      "step": 6327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0274621212121216e-05,
+      "loss": 0.4309,
+      "step": 6328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0269886363636366e-05,
+      "loss": 0.4173,
+      "step": 6329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0265151515151516e-05,
+      "loss": 0.466,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0260416666666667e-05,
+      "loss": 0.5049,
+      "step": 6331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0255681818181817e-05,
+      "loss": 0.4116,
+      "step": 6332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.025094696969697e-05,
+      "loss": 0.4582,
+      "step": 6333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.024621212121212e-05,
+      "loss": 0.3588,
+      "step": 6334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0241477272727275e-05,
+      "loss": 0.3911,
+      "step": 6335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0236742424242425e-05,
+      "loss": 0.3516,
+      "step": 6336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.023200757575758e-05,
+      "loss": 0.3369,
+      "step": 6337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.022727272727273e-05,
+      "loss": 0.4259,
+      "step": 6338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.022253787878788e-05,
+      "loss": 0.4455,
+      "step": 6339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.021780303030303e-05,
+      "loss": 0.3521,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.021306818181818e-05,
+      "loss": 0.5094,
+      "step": 6341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0208333333333334e-05,
+      "loss": 0.3609,
+      "step": 6342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0203598484848485e-05,
+      "loss": 0.4481,
+      "step": 6343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.019886363636364e-05,
+      "loss": 0.4866,
+      "step": 6344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.019412878787879e-05,
+      "loss": 0.414,
+      "step": 6345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0189393939393943e-05,
+      "loss": 0.4675,
+      "step": 6346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0184659090909093e-05,
+      "loss": 0.502,
+      "step": 6347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0179924242424243e-05,
+      "loss": 0.4037,
+      "step": 6348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0175189393939394e-05,
+      "loss": 0.4392,
+      "step": 6349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0170454545454544e-05,
+      "loss": 0.4481,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0165719696969698e-05,
+      "loss": 0.3528,
+      "step": 6351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0160984848484848e-05,
+      "loss": 0.4422,
+      "step": 6352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0156250000000002e-05,
+      "loss": 0.5116,
+      "step": 6353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0151515151515152e-05,
+      "loss": 0.4842,
+      "step": 6354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0146780303030306e-05,
+      "loss": 0.4276,
+      "step": 6355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0142045454545457e-05,
+      "loss": 0.3497,
+      "step": 6356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0137310606060607e-05,
+      "loss": 0.5725,
+      "step": 6357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0132575757575757e-05,
+      "loss": 0.4469,
+      "step": 6358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0127840909090908e-05,
+      "loss": 0.353,
+      "step": 6359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.012310606060606e-05,
+      "loss": 0.4299,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0118371212121212e-05,
+      "loss": 0.4373,
+      "step": 6361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0113636363636366e-05,
+      "loss": 0.3856,
+      "step": 6362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0108901515151516e-05,
+      "loss": 0.5446,
+      "step": 6363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.010416666666667e-05,
+      "loss": 0.4887,
+      "step": 6364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.009943181818182e-05,
+      "loss": 0.415,
+      "step": 6365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.009469696969697e-05,
+      "loss": 0.3056,
+      "step": 6366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.008996212121212e-05,
+      "loss": 0.3353,
+      "step": 6367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0085227272727275e-05,
+      "loss": 0.4176,
+      "step": 6368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0080492424242425e-05,
+      "loss": 0.4334,
+      "step": 6369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0075757575757575e-05,
+      "loss": 0.4627,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.007102272727273e-05,
+      "loss": 0.5779,
+      "step": 6371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.006628787878788e-05,
+      "loss": 0.4507,
+      "step": 6372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0061553030303033e-05,
+      "loss": 0.312,
+      "step": 6373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0056818181818184e-05,
+      "loss": 0.5392,
+      "step": 6374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0052083333333334e-05,
+      "loss": 0.3868,
+      "step": 6375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0047348484848484e-05,
+      "loss": 0.4786,
+      "step": 6376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0042613636363638e-05,
+      "loss": 0.3985,
+      "step": 6377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.003787878787879e-05,
+      "loss": 0.2863,
+      "step": 6378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.003314393939394e-05,
+      "loss": 0.449,
+      "step": 6379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0028409090909093e-05,
+      "loss": 0.357,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0023674242424243e-05,
+      "loss": 0.4792,
+      "step": 6381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0018939393939397e-05,
+      "loss": 0.3966,
+      "step": 6382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0014204545454547e-05,
+      "loss": 0.4042,
+      "step": 6383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0009469696969698e-05,
+      "loss": 0.5659,
+      "step": 6384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0004734848484848e-05,
+      "loss": 0.4082,
+      "step": 6385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.4733,
+      "step": 6386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9995265151515152e-05,
+      "loss": 0.5364,
+      "step": 6387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9990530303030302e-05,
+      "loss": 0.3887,
+      "step": 6388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9985795454545456e-05,
+      "loss": 0.5273,
+      "step": 6389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9981060606060607e-05,
+      "loss": 0.2907,
+      "step": 6390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997632575757576e-05,
+      "loss": 0.4907,
+      "step": 6391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.997159090909091e-05,
+      "loss": 0.4694,
+      "step": 6392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.996685606060606e-05,
+      "loss": 0.6049,
+      "step": 6393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.996212121212121e-05,
+      "loss": 0.3793,
+      "step": 6394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9957386363636365e-05,
+      "loss": 0.4486,
+      "step": 6395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9952651515151516e-05,
+      "loss": 0.5061,
+      "step": 6396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.994791666666667e-05,
+      "loss": 0.5134,
+      "step": 6397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.994318181818182e-05,
+      "loss": 0.45,
+      "step": 6398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.993844696969697e-05,
+      "loss": 0.3767,
+      "step": 6399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9933712121212124e-05,
+      "loss": 0.4393,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9928977272727274e-05,
+      "loss": 0.355,
+      "step": 6401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9924242424242425e-05,
+      "loss": 0.5215,
+      "step": 6402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9919507575757575e-05,
+      "loss": 0.4654,
+      "step": 6403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.991477272727273e-05,
+      "loss": 0.4414,
+      "step": 6404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.991003787878788e-05,
+      "loss": 0.461,
+      "step": 6405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9905303030303033e-05,
+      "loss": 0.3782,
+      "step": 6406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9900568181818183e-05,
+      "loss": 0.3771,
+      "step": 6407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9895833333333334e-05,
+      "loss": 0.4142,
+      "step": 6408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9891098484848487e-05,
+      "loss": 0.5334,
+      "step": 6409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9886363636363638e-05,
+      "loss": 0.4284,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9881628787878788e-05,
+      "loss": 0.5137,
+      "step": 6411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.987689393939394e-05,
+      "loss": 0.5205,
+      "step": 6412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9872159090909092e-05,
+      "loss": 0.4606,
+      "step": 6413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9867424242424243e-05,
+      "loss": 0.4321,
+      "step": 6414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9862689393939396e-05,
+      "loss": 0.3917,
+      "step": 6415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9857954545454547e-05,
+      "loss": 0.4411,
+      "step": 6416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.98532196969697e-05,
+      "loss": 0.3666,
+      "step": 6417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.984848484848485e-05,
+      "loss": 0.4975,
+      "step": 6418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.984375e-05,
+      "loss": 0.3642,
+      "step": 6419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.983901515151515e-05,
+      "loss": 0.4262,
+      "step": 6420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9834280303030302e-05,
+      "loss": 0.3824,
+      "step": 6421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9829545454545456e-05,
+      "loss": 0.463,
+      "step": 6422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9824810606060606e-05,
+      "loss": 0.4084,
+      "step": 6423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.982007575757576e-05,
+      "loss": 0.4393,
+      "step": 6424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.981534090909091e-05,
+      "loss": 0.362,
+      "step": 6425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9810606060606064e-05,
+      "loss": 0.4377,
+      "step": 6426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9805871212121214e-05,
+      "loss": 0.4113,
+      "step": 6427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9801136363636365e-05,
+      "loss": 0.3354,
+      "step": 6428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9796401515151515e-05,
+      "loss": 0.3455,
+      "step": 6429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9791666666666665e-05,
+      "loss": 0.3074,
+      "step": 6430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.978693181818182e-05,
+      "loss": 0.453,
+      "step": 6431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.978219696969697e-05,
+      "loss": 0.4674,
+      "step": 6432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9777462121212123e-05,
+      "loss": 0.4489,
+      "step": 6433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9772727272727274e-05,
+      "loss": 0.2577,
+      "step": 6434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9767992424242428e-05,
+      "loss": 0.3546,
+      "step": 6435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9763257575757578e-05,
+      "loss": 0.4425,
+      "step": 6436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9758522727272728e-05,
+      "loss": 0.4596,
+      "step": 6437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.975378787878788e-05,
+      "loss": 0.3641,
+      "step": 6438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.974905303030303e-05,
+      "loss": 0.4754,
+      "step": 6439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9744318181818183e-05,
+      "loss": 0.4095,
+      "step": 6440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9739583333333333e-05,
+      "loss": 0.4204,
+      "step": 6441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9734848484848487e-05,
+      "loss": 0.4928,
+      "step": 6442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9730113636363637e-05,
+      "loss": 0.4818,
+      "step": 6443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.972537878787879e-05,
+      "loss": 0.3269,
+      "step": 6444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.972064393939394e-05,
+      "loss": 0.5517,
+      "step": 6445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9715909090909092e-05,
+      "loss": 0.462,
+      "step": 6446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9711174242424242e-05,
+      "loss": 0.4125,
+      "step": 6447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9706439393939393e-05,
+      "loss": 0.3336,
+      "step": 6448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9701704545454546e-05,
+      "loss": 0.2947,
+      "step": 6449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9696969696969697e-05,
+      "loss": 0.3242,
+      "step": 6450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.969223484848485e-05,
+      "loss": 0.5129,
+      "step": 6451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.96875e-05,
+      "loss": 0.4822,
+      "step": 6452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9682765151515155e-05,
+      "loss": 0.4799,
+      "step": 6453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9678030303030305e-05,
+      "loss": 0.497,
+      "step": 6454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9673295454545455e-05,
+      "loss": 0.3643,
+      "step": 6455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9668560606060606e-05,
+      "loss": 0.5052,
+      "step": 6456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9663825757575756e-05,
+      "loss": 0.4237,
+      "step": 6457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.965909090909091e-05,
+      "loss": 0.5606,
+      "step": 6458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.965435606060606e-05,
+      "loss": 0.4764,
+      "step": 6459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9649621212121214e-05,
+      "loss": 0.34,
+      "step": 6460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9644886363636364e-05,
+      "loss": 0.4244,
+      "step": 6461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9640151515151518e-05,
+      "loss": 0.3526,
+      "step": 6462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.963541666666667e-05,
+      "loss": 0.4392,
+      "step": 6463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.963068181818182e-05,
+      "loss": 0.3441,
+      "step": 6464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.962594696969697e-05,
+      "loss": 0.3232,
+      "step": 6465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.962121212121212e-05,
+      "loss": 0.4627,
+      "step": 6466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9616477272727273e-05,
+      "loss": 0.4433,
+      "step": 6467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9611742424242424e-05,
+      "loss": 0.5302,
+      "step": 6468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9607007575757577e-05,
+      "loss": 0.3386,
+      "step": 6469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9602272727272728e-05,
+      "loss": 0.339,
+      "step": 6470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.959753787878788e-05,
+      "loss": 0.417,
+      "step": 6471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9592803030303032e-05,
+      "loss": 0.4379,
+      "step": 6472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9588068181818182e-05,
+      "loss": 0.4609,
+      "step": 6473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9583333333333333e-05,
+      "loss": 0.5126,
+      "step": 6474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9578598484848487e-05,
+      "loss": 0.3537,
+      "step": 6475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9573863636363637e-05,
+      "loss": 0.4252,
+      "step": 6476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9569128787878787e-05,
+      "loss": 0.3941,
+      "step": 6477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956439393939394e-05,
+      "loss": 0.3114,
+      "step": 6478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955965909090909e-05,
+      "loss": 0.565,
+      "step": 6479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554924242424245e-05,
+      "loss": 0.6338,
+      "step": 6480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550189393939396e-05,
+      "loss": 0.4828,
+      "step": 6481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9545454545454546e-05,
+      "loss": 0.4469,
+      "step": 6482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9540719696969696e-05,
+      "loss": 0.4667,
+      "step": 6483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.953598484848485e-05,
+      "loss": 0.3561,
+      "step": 6484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.953125e-05,
+      "loss": 0.3837,
+      "step": 6485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.952651515151515e-05,
+      "loss": 0.4534,
+      "step": 6486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9521780303030305e-05,
+      "loss": 0.3514,
+      "step": 6487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9517045454545455e-05,
+      "loss": 0.4811,
+      "step": 6488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.951231060606061e-05,
+      "loss": 0.3943,
+      "step": 6489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.950757575757576e-05,
+      "loss": 0.4699,
+      "step": 6490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.950284090909091e-05,
+      "loss": 0.4879,
+      "step": 6491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.949810606060606e-05,
+      "loss": 0.4107,
+      "step": 6492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9493371212121214e-05,
+      "loss": 0.4921,
+      "step": 6493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9488636363636364e-05,
+      "loss": 0.4371,
+      "step": 6494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9483901515151514e-05,
+      "loss": 0.4516,
+      "step": 6495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9479166666666668e-05,
+      "loss": 0.4077,
+      "step": 6496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.947443181818182e-05,
+      "loss": 0.4104,
+      "step": 6497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9469696969696972e-05,
+      "loss": 0.4994,
+      "step": 6498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9464962121212123e-05,
+      "loss": 0.5735,
+      "step": 6499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9460227272727273e-05,
+      "loss": 0.4618,
+      "step": 6500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9455492424242423e-05,
+      "loss": 0.4655,
+      "step": 6501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9450757575757577e-05,
+      "loss": 0.4377,
+      "step": 6502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9446022727272727e-05,
+      "loss": 0.5388,
+      "step": 6503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.944128787878788e-05,
+      "loss": 0.3372,
+      "step": 6504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.943655303030303e-05,
+      "loss": 0.3116,
+      "step": 6505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9431818181818182e-05,
+      "loss": 0.4221,
+      "step": 6506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9427083333333336e-05,
+      "loss": 0.3726,
+      "step": 6507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9422348484848486e-05,
+      "loss": 0.3977,
+      "step": 6508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9417613636363636e-05,
+      "loss": 0.5353,
+      "step": 6509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9412878787878787e-05,
+      "loss": 0.4727,
+      "step": 6510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.940814393939394e-05,
+      "loss": 0.6558,
+      "step": 6511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.940340909090909e-05,
+      "loss": 0.5938,
+      "step": 6512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9398674242424245e-05,
+      "loss": 0.399,
+      "step": 6513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 0.239,
+      "step": 6514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9389204545454545e-05,
+      "loss": 0.4849,
+      "step": 6515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.93844696969697e-05,
+      "loss": 0.5043,
+      "step": 6516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.937973484848485e-05,
+      "loss": 0.3279,
+      "step": 6517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9375e-05,
+      "loss": 0.4161,
+      "step": 6518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.937026515151515e-05,
+      "loss": 0.6142,
+      "step": 6519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9365530303030304e-05,
+      "loss": 0.4757,
+      "step": 6520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9360795454545455e-05,
+      "loss": 0.5145,
+      "step": 6521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9356060606060608e-05,
+      "loss": 0.4676,
+      "step": 6522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.935132575757576e-05,
+      "loss": 0.4253,
+      "step": 6523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.934659090909091e-05,
+      "loss": 0.4286,
+      "step": 6524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9341856060606063e-05,
+      "loss": 0.405,
+      "step": 6525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9337121212121213e-05,
+      "loss": 0.4452,
+      "step": 6526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9332386363636364e-05,
+      "loss": 0.4752,
+      "step": 6527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9327651515151514e-05,
+      "loss": 0.4019,
+      "step": 6528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9322916666666668e-05,
+      "loss": 0.3878,
+      "step": 6529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9318181818181818e-05,
+      "loss": 0.2714,
+      "step": 6530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9313446969696972e-05,
+      "loss": 0.4094,
+      "step": 6531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9308712121212122e-05,
+      "loss": 0.4096,
+      "step": 6532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9303977272727276e-05,
+      "loss": 0.377,
+      "step": 6533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9299242424242426e-05,
+      "loss": 0.5228,
+      "step": 6534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9294507575757577e-05,
+      "loss": 0.3786,
+      "step": 6535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9289772727272727e-05,
+      "loss": 0.2948,
+      "step": 6536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9285037878787877e-05,
+      "loss": 0.4096,
+      "step": 6537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.928030303030303e-05,
+      "loss": 0.4574,
+      "step": 6538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.927556818181818e-05,
+      "loss": 0.412,
+      "step": 6539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9270833333333335e-05,
+      "loss": 0.3094,
+      "step": 6540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9266098484848486e-05,
+      "loss": 0.4482,
+      "step": 6541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.926136363636364e-05,
+      "loss": 0.4463,
+      "step": 6542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.925662878787879e-05,
+      "loss": 0.4649,
+      "step": 6543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.925189393939394e-05,
+      "loss": 0.4301,
+      "step": 6544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.924715909090909e-05,
+      "loss": 0.3786,
+      "step": 6545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.924242424242424e-05,
+      "loss": 0.5196,
+      "step": 6546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9237689393939395e-05,
+      "loss": 0.3887,
+      "step": 6547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9232954545454545e-05,
+      "loss": 0.3248,
+      "step": 6548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.92282196969697e-05,
+      "loss": 0.3441,
+      "step": 6549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.922348484848485e-05,
+      "loss": 0.3533,
+      "step": 6550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9218750000000003e-05,
+      "loss": 0.4358,
+      "step": 6551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9214015151515153e-05,
+      "loss": 0.5872,
+      "step": 6552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9209280303030304e-05,
+      "loss": 0.3238,
+      "step": 6553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9204545454545454e-05,
+      "loss": 0.4439,
+      "step": 6554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9199810606060604e-05,
+      "loss": 0.4932,
+      "step": 6555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9195075757575758e-05,
+      "loss": 0.4372,
+      "step": 6556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.919034090909091e-05,
+      "loss": 0.4848,
+      "step": 6557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9185606060606062e-05,
+      "loss": 0.4615,
+      "step": 6558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9180871212121213e-05,
+      "loss": 0.3636,
+      "step": 6559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9176136363636366e-05,
+      "loss": 0.3472,
+      "step": 6560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9171401515151517e-05,
+      "loss": 0.5945,
+      "step": 6561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9166666666666667e-05,
+      "loss": 0.4948,
+      "step": 6562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9161931818181818e-05,
+      "loss": 0.3893,
+      "step": 6563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9157196969696968e-05,
+      "loss": 0.3879,
+      "step": 6564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9152462121212122e-05,
+      "loss": 0.5881,
+      "step": 6565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9147727272727272e-05,
+      "loss": 0.428,
+      "step": 6566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9142992424242426e-05,
+      "loss": 0.3862,
+      "step": 6567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9138257575757576e-05,
+      "loss": 0.3578,
+      "step": 6568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.913352272727273e-05,
+      "loss": 0.4275,
+      "step": 6569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.912878787878788e-05,
+      "loss": 0.3452,
+      "step": 6570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.912405303030303e-05,
+      "loss": 0.4367,
+      "step": 6571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.911931818181818e-05,
+      "loss": 0.4286,
+      "step": 6572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.911458333333333e-05,
+      "loss": 0.3076,
+      "step": 6573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9109848484848485e-05,
+      "loss": 0.5266,
+      "step": 6574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9105113636363636e-05,
+      "loss": 0.4619,
+      "step": 6575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.910037878787879e-05,
+      "loss": 0.3701,
+      "step": 6576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.909564393939394e-05,
+      "loss": 0.3721,
+      "step": 6577
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 0.4739,
+      "step": 6578
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9086174242424244e-05,
+      "loss": 0.5092,
+      "step": 6579
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9081439393939394e-05,
+      "loss": 0.3816,
+      "step": 6580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9076704545454545e-05,
+      "loss": 0.2807,
+      "step": 6581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.90719696969697e-05,
+      "loss": 0.5223,
+      "step": 6582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.906723484848485e-05,
+      "loss": 0.3587,
+      "step": 6583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.90625e-05,
+      "loss": 0.2754,
+      "step": 6584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9057765151515153e-05,
+      "loss": 0.5158,
+      "step": 6585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9053030303030303e-05,
+      "loss": 0.5002,
+      "step": 6586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9048295454545457e-05,
+      "loss": 0.4875,
+      "step": 6587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9043560606060607e-05,
+      "loss": 0.2704,
+      "step": 6588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9038825757575758e-05,
+      "loss": 0.5268,
+      "step": 6589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9034090909090908e-05,
+      "loss": 0.5404,
+      "step": 6590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9029356060606062e-05,
+      "loss": 0.3106,
+      "step": 6591
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9024621212121212e-05,
+      "loss": 0.5426,
+      "step": 6592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9019886363636363e-05,
+      "loss": 0.4918,
+      "step": 6593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9015151515151516e-05,
+      "loss": 0.3594,
+      "step": 6594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9010416666666667e-05,
+      "loss": 0.5496,
+      "step": 6595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.900568181818182e-05,
+      "loss": 0.4478,
+      "step": 6596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.900094696969697e-05,
+      "loss": 0.4863,
+      "step": 6597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.899621212121212e-05,
+      "loss": 0.4155,
+      "step": 6598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8991477272727272e-05,
+      "loss": 0.3777,
+      "step": 6599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8986742424242425e-05,
+      "loss": 0.4473,
+      "step": 6600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8982007575757576e-05,
+      "loss": 0.3349,
+      "step": 6601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8977272727272726e-05,
+      "loss": 0.3897,
+      "step": 6602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.897253787878788e-05,
+      "loss": 0.4829,
+      "step": 6603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.896780303030303e-05,
+      "loss": 0.3208,
+      "step": 6604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8963068181818184e-05,
+      "loss": 0.3897,
+      "step": 6605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8958333333333334e-05,
+      "loss": 0.5162,
+      "step": 6606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8953598484848485e-05,
+      "loss": 0.5564,
+      "step": 6607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8948863636363635e-05,
+      "loss": 0.4228,
+      "step": 6608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.894412878787879e-05,
+      "loss": 0.3526,
+      "step": 6609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.893939393939394e-05,
+      "loss": 0.5618,
+      "step": 6610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8934659090909093e-05,
+      "loss": 0.518,
+      "step": 6611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8929924242424244e-05,
+      "loss": 0.4853,
+      "step": 6612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8925189393939394e-05,
+      "loss": 0.5033,
+      "step": 6613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8920454545454548e-05,
+      "loss": 0.5327,
+      "step": 6614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8915719696969698e-05,
+      "loss": 0.4882,
+      "step": 6615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.891098484848485e-05,
+      "loss": 0.3013,
+      "step": 6616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.890625e-05,
+      "loss": 0.4485,
+      "step": 6617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8901515151515153e-05,
+      "loss": 0.5029,
+      "step": 6618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8896780303030303e-05,
+      "loss": 0.3622,
+      "step": 6619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8892045454545457e-05,
+      "loss": 0.4114,
+      "step": 6620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8887310606060607e-05,
+      "loss": 0.3978,
+      "step": 6621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8882575757575757e-05,
+      "loss": 0.4052,
+      "step": 6622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.887784090909091e-05,
+      "loss": 0.4387,
+      "step": 6623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.887310606060606e-05,
+      "loss": 0.5463,
+      "step": 6624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8868371212121215e-05,
+      "loss": 0.5774,
+      "step": 6625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8863636363636362e-05,
+      "loss": 0.3966,
+      "step": 6626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8858901515151516e-05,
+      "loss": 0.432,
+      "step": 6627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8854166666666666e-05,
+      "loss": 0.3203,
+      "step": 6628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.884943181818182e-05,
+      "loss": 0.4814,
+      "step": 6629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.884469696969697e-05,
+      "loss": 0.4456,
+      "step": 6630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.883996212121212e-05,
+      "loss": 0.4395,
+      "step": 6631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8835227272727275e-05,
+      "loss": 0.3802,
+      "step": 6632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8830492424242425e-05,
+      "loss": 0.4066,
+      "step": 6633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882575757575758e-05,
+      "loss": 0.3757,
+      "step": 6634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8821022727272726e-05,
+      "loss": 0.5469,
+      "step": 6635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.881628787878788e-05,
+      "loss": 0.3757,
+      "step": 6636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.881155303030303e-05,
+      "loss": 0.5182,
+      "step": 6637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8806818181818184e-05,
+      "loss": 0.4076,
+      "step": 6638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8802083333333334e-05,
+      "loss": 0.3478,
+      "step": 6639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8797348484848488e-05,
+      "loss": 0.4316,
+      "step": 6640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8792613636363638e-05,
+      "loss": 0.4801,
+      "step": 6641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.878787878787879e-05,
+      "loss": 0.4337,
+      "step": 6642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8783143939393942e-05,
+      "loss": 0.4993,
+      "step": 6643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.877840909090909e-05,
+      "loss": 0.3613,
+      "step": 6644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8773674242424243e-05,
+      "loss": 0.375,
+      "step": 6645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8768939393939393e-05,
+      "loss": 0.3965,
+      "step": 6646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8764204545454547e-05,
+      "loss": 0.444,
+      "step": 6647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8759469696969698e-05,
+      "loss": 0.432,
+      "step": 6648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.875473484848485e-05,
+      "loss": 0.4277,
+      "step": 6649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.4728,
+      "step": 6650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8745265151515152e-05,
+      "loss": 0.3496,
+      "step": 6651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8740530303030306e-05,
+      "loss": 0.4722,
+      "step": 6652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8735795454545453e-05,
+      "loss": 0.4294,
+      "step": 6653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8731060606060607e-05,
+      "loss": 0.5073,
+      "step": 6654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8726325757575757e-05,
+      "loss": 0.4351,
+      "step": 6655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.872159090909091e-05,
+      "loss": 0.5083,
+      "step": 6656
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.871685606060606e-05,
+      "loss": 0.5358,
+      "step": 6657
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8712121212121215e-05,
+      "loss": 0.4242,
+      "step": 6658
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8707386363636365e-05,
+      "loss": 0.4016,
+      "step": 6659
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8702651515151516e-05,
+      "loss": 0.5139,
+      "step": 6660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.869791666666667e-05,
+      "loss": 0.6557,
+      "step": 6661
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8693181818181816e-05,
+      "loss": 0.419,
+      "step": 6662
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.868844696969697e-05,
+      "loss": 0.3518,
+      "step": 6663
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.868371212121212e-05,
+      "loss": 0.4168,
+      "step": 6664
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8678977272727274e-05,
+      "loss": 0.5679,
+      "step": 6665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8674242424242425e-05,
+      "loss": 0.3669,
+      "step": 6666
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.866950757575758e-05,
+      "loss": 0.4637,
+      "step": 6667
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.866477272727273e-05,
+      "loss": 0.4312,
+      "step": 6668
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8660037878787883e-05,
+      "loss": 0.4548,
+      "step": 6669
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8655303030303033e-05,
+      "loss": 0.4051,
+      "step": 6670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.865056818181818e-05,
+      "loss": 0.3277,
+      "step": 6671
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8645833333333334e-05,
+      "loss": 0.4033,
+      "step": 6672
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8641098484848484e-05,
+      "loss": 0.4588,
+      "step": 6673
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8636363636363638e-05,
+      "loss": 0.3454,
+      "step": 6674
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8631628787878788e-05,
+      "loss": 0.496,
+      "step": 6675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8626893939393942e-05,
+      "loss": 0.3509,
+      "step": 6676
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8622159090909092e-05,
+      "loss": 0.4044,
+      "step": 6677
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8617424242424246e-05,
+      "loss": 0.4449,
+      "step": 6678
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8612689393939396e-05,
+      "loss": 0.4138,
+      "step": 6679
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8607954545454543e-05,
+      "loss": 0.3322,
+      "step": 6680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8603219696969697e-05,
+      "loss": 0.3885,
+      "step": 6681
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8598484848484848e-05,
+      "loss": 0.3098,
+      "step": 6682
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.859375e-05,
+      "loss": 0.3607,
+      "step": 6683
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8589015151515152e-05,
+      "loss": 0.3773,
+      "step": 6684
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8584280303030305e-05,
+      "loss": 0.5255,
+      "step": 6685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8579545454545456e-05,
+      "loss": 0.4347,
+      "step": 6686
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.857481060606061e-05,
+      "loss": 0.3573,
+      "step": 6687
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.857007575757576e-05,
+      "loss": 0.3757,
+      "step": 6688
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8565340909090907e-05,
+      "loss": 0.2905,
+      "step": 6689
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.856060606060606e-05,
+      "loss": 0.5524,
+      "step": 6690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.855587121212121e-05,
+      "loss": 0.4693,
+      "step": 6691
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8551136363636365e-05,
+      "loss": 0.3851,
+      "step": 6692
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8546401515151515e-05,
+      "loss": 0.4201,
+      "step": 6693
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.854166666666667e-05,
+      "loss": 0.4994,
+      "step": 6694
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.853693181818182e-05,
+      "loss": 0.496,
+      "step": 6695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8532196969696973e-05,
+      "loss": 0.2625,
+      "step": 6696
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8527462121212124e-05,
+      "loss": 0.4396,
+      "step": 6697
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8522727272727274e-05,
+      "loss": 0.3934,
+      "step": 6698
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8517992424242424e-05,
+      "loss": 0.394,
+      "step": 6699
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8513257575757575e-05,
+      "loss": 0.4481,
+      "step": 6700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.850852272727273e-05,
+      "loss": 0.3901,
+      "step": 6701
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.850378787878788e-05,
+      "loss": 0.3964,
+      "step": 6702
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8499053030303033e-05,
+      "loss": 0.3602,
+      "step": 6703
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8494318181818183e-05,
+      "loss": 0.467,
+      "step": 6704
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8489583333333337e-05,
+      "loss": 0.4591,
+      "step": 6705
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8484848484848487e-05,
+      "loss": 0.5346,
+      "step": 6706
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8480113636363637e-05,
+      "loss": 0.3828,
+      "step": 6707
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8475378787878788e-05,
+      "loss": 0.5973,
+      "step": 6708
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8470643939393938e-05,
+      "loss": 0.4963,
+      "step": 6709
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8465909090909092e-05,
+      "loss": 0.4118,
+      "step": 6710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8461174242424242e-05,
+      "loss": 0.4415,
+      "step": 6711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8456439393939396e-05,
+      "loss": 0.397,
+      "step": 6712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8451704545454546e-05,
+      "loss": 0.2618,
+      "step": 6713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.84469696969697e-05,
+      "loss": 0.3275,
+      "step": 6714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.844223484848485e-05,
+      "loss": 0.4323,
+      "step": 6715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.84375e-05,
+      "loss": 0.49,
+      "step": 6716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.843276515151515e-05,
+      "loss": 0.4482,
+      "step": 6717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.84280303030303e-05,
+      "loss": 0.276,
+      "step": 6718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8423295454545455e-05,
+      "loss": 0.4345,
+      "step": 6719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8418560606060606e-05,
+      "loss": 0.3705,
+      "step": 6720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.841382575757576e-05,
+      "loss": 0.4688,
+      "step": 6721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.840909090909091e-05,
+      "loss": 0.4052,
+      "step": 6722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8404356060606064e-05,
+      "loss": 0.5046,
+      "step": 6723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8399621212121214e-05,
+      "loss": 0.5073,
+      "step": 6724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8394886363636364e-05,
+      "loss": 0.3221,
+      "step": 6725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8390151515151515e-05,
+      "loss": 0.3464,
+      "step": 6726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.838541666666667e-05,
+      "loss": 0.4896,
+      "step": 6727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.838068181818182e-05,
+      "loss": 0.4112,
+      "step": 6728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837594696969697e-05,
+      "loss": 0.5692,
+      "step": 6729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8371212121212123e-05,
+      "loss": 0.3721,
+      "step": 6730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366477272727273e-05,
+      "loss": 0.4483,
+      "step": 6731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361742424242427e-05,
+      "loss": 0.4533,
+      "step": 6732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8357007575757578e-05,
+      "loss": 0.3091,
+      "step": 6733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8352272727272728e-05,
+      "loss": 0.4442,
+      "step": 6734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834753787878788e-05,
+      "loss": 0.3568,
+      "step": 6735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8342803030303032e-05,
+      "loss": 0.3282,
+      "step": 6736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8338068181818182e-05,
+      "loss": 0.3297,
+      "step": 6737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.3576,
+      "step": 6738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8328598484848487e-05,
+      "loss": 0.4282,
+      "step": 6739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8323863636363637e-05,
+      "loss": 0.4546,
+      "step": 6740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831912878787879e-05,
+      "loss": 0.4507,
+      "step": 6741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831439393939394e-05,
+      "loss": 0.3358,
+      "step": 6742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830965909090909e-05,
+      "loss": 0.3765,
+      "step": 6743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304924242424242e-05,
+      "loss": 0.3789,
+      "step": 6744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300189393939396e-05,
+      "loss": 0.4514,
+      "step": 6745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8295454545454546e-05,
+      "loss": 0.3065,
+      "step": 6746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82907196969697e-05,
+      "loss": 0.2937,
+      "step": 6747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828598484848485e-05,
+      "loss": 0.3734,
+      "step": 6748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828125e-05,
+      "loss": 0.3943,
+      "step": 6749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8276515151515154e-05,
+      "loss": 0.3667,
+      "step": 6750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271780303030305e-05,
+      "loss": 0.4907,
+      "step": 6751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267045454545455e-05,
+      "loss": 0.446,
+      "step": 6752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262310606060605e-05,
+      "loss": 0.3489,
+      "step": 6753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.825757575757576e-05,
+      "loss": 0.3262,
+      "step": 6754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.825284090909091e-05,
+      "loss": 0.3333,
+      "step": 6755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8248106060606063e-05,
+      "loss": 0.4285,
+      "step": 6756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8243371212121214e-05,
+      "loss": 0.4752,
+      "step": 6757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8238636363636364e-05,
+      "loss": 0.4079,
+      "step": 6758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8233901515151518e-05,
+      "loss": 0.4283,
+      "step": 6759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8229166666666668e-05,
+      "loss": 0.4256,
+      "step": 6760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.822443181818182e-05,
+      "loss": 0.3661,
+      "step": 6761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.821969696969697e-05,
+      "loss": 0.524,
+      "step": 6762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8214962121212123e-05,
+      "loss": 0.3873,
+      "step": 6763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8210227272727273e-05,
+      "loss": 0.5465,
+      "step": 6764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8205492424242427e-05,
+      "loss": 0.4473,
+      "step": 6765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8200757575757577e-05,
+      "loss": 0.3577,
+      "step": 6766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8196022727272728e-05,
+      "loss": 0.3297,
+      "step": 6767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.819128787878788e-05,
+      "loss": 0.3993,
+      "step": 6768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8186553030303032e-05,
+      "loss": 0.3628,
+      "step": 6769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.4283,
+      "step": 6770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8177083333333332e-05,
+      "loss": 0.411,
+      "step": 6771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8172348484848486e-05,
+      "loss": 0.5317,
+      "step": 6772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8167613636363637e-05,
+      "loss": 0.4912,
+      "step": 6773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.816287878787879e-05,
+      "loss": 0.345,
+      "step": 6774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.815814393939394e-05,
+      "loss": 0.3602,
+      "step": 6775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8153409090909094e-05,
+      "loss": 0.4601,
+      "step": 6776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8148674242424245e-05,
+      "loss": 0.5106,
+      "step": 6777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8143939393939395e-05,
+      "loss": 0.3197,
+      "step": 6778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8139204545454546e-05,
+      "loss": 0.4446,
+      "step": 6779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8134469696969696e-05,
+      "loss": 0.4629,
+      "step": 6780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.812973484848485e-05,
+      "loss": 0.4198,
+      "step": 6781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8125e-05,
+      "loss": 0.3518,
+      "step": 6782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8120265151515154e-05,
+      "loss": 0.3789,
+      "step": 6783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8115530303030304e-05,
+      "loss": 0.4386,
+      "step": 6784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8110795454545458e-05,
+      "loss": 0.4061,
+      "step": 6785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.810606060606061e-05,
+      "loss": 0.4254,
+      "step": 6786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.810132575757576e-05,
+      "loss": 0.4992,
+      "step": 6787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.809659090909091e-05,
+      "loss": 0.3847,
+      "step": 6788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.809185606060606e-05,
+      "loss": 0.4075,
+      "step": 6789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8087121212121213e-05,
+      "loss": 0.443,
+      "step": 6790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8082386363636364e-05,
+      "loss": 0.4942,
+      "step": 6791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8077651515151517e-05,
+      "loss": 0.5746,
+      "step": 6792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8072916666666668e-05,
+      "loss": 0.4271,
+      "step": 6793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.806818181818182e-05,
+      "loss": 0.3958,
+      "step": 6794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8063446969696972e-05,
+      "loss": 0.4905,
+      "step": 6795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8058712121212122e-05,
+      "loss": 0.4936,
+      "step": 6796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8053977272727273e-05,
+      "loss": 0.3742,
+      "step": 6797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8049242424242423e-05,
+      "loss": 0.3913,
+      "step": 6798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8044507575757577e-05,
+      "loss": 0.5264,
+      "step": 6799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8039772727272727e-05,
+      "loss": 0.3866,
+      "step": 6800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.803503787878788e-05,
+      "loss": 0.4035,
+      "step": 6801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.803030303030303e-05,
+      "loss": 0.4413,
+      "step": 6802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8025568181818185e-05,
+      "loss": 0.3593,
+      "step": 6803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8020833333333335e-05,
+      "loss": 0.4869,
+      "step": 6804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8016098484848486e-05,
+      "loss": 0.4756,
+      "step": 6805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8011363636363636e-05,
+      "loss": 0.459,
+      "step": 6806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8006628787878787e-05,
+      "loss": 0.4126,
+      "step": 6807
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.800189393939394e-05,
+      "loss": 0.3832,
+      "step": 6808
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.799715909090909e-05,
+      "loss": 0.4295,
+      "step": 6809
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7992424242424244e-05,
+      "loss": 0.3904,
+      "step": 6810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7987689393939395e-05,
+      "loss": 0.3963,
+      "step": 6811
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.798295454545455e-05,
+      "loss": 0.4352,
+      "step": 6812
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.79782196969697e-05,
+      "loss": 0.4954,
+      "step": 6813
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.797348484848485e-05,
+      "loss": 0.5843,
+      "step": 6814
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.796875e-05,
+      "loss": 0.6076,
+      "step": 6815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.796401515151515e-05,
+      "loss": 0.2656,
+      "step": 6816
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7959280303030304e-05,
+      "loss": 0.6356,
+      "step": 6817
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7954545454545454e-05,
+      "loss": 0.4601,
+      "step": 6818
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7949810606060608e-05,
+      "loss": 0.4434,
+      "step": 6819
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.794507575757576e-05,
+      "loss": 0.543,
+      "step": 6820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7940340909090912e-05,
+      "loss": 0.3901,
+      "step": 6821
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7935606060606062e-05,
+      "loss": 0.4264,
+      "step": 6822
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7930871212121213e-05,
+      "loss": 0.5126,
+      "step": 6823
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7926136363636363e-05,
+      "loss": 0.47,
+      "step": 6824
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7921401515151514e-05,
+      "loss": 0.4825,
+      "step": 6825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7916666666666667e-05,
+      "loss": 0.333,
+      "step": 6826
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7911931818181818e-05,
+      "loss": 0.3789,
+      "step": 6827
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.790719696969697e-05,
+      "loss": 0.3437,
+      "step": 6828
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7902462121212122e-05,
+      "loss": 0.4001,
+      "step": 6829
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7897727272727276e-05,
+      "loss": 0.2777,
+      "step": 6830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7892992424242426e-05,
+      "loss": 0.4328,
+      "step": 6831
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7888257575757576e-05,
+      "loss": 0.4909,
+      "step": 6832
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7883522727272727e-05,
+      "loss": 0.3206,
+      "step": 6833
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.787878787878788e-05,
+      "loss": 0.3927,
+      "step": 6834
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.787405303030303e-05,
+      "loss": 0.4036,
+      "step": 6835
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.786931818181818e-05,
+      "loss": 0.421,
+      "step": 6836
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7864583333333335e-05,
+      "loss": 0.4525,
+      "step": 6837
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7859848484848485e-05,
+      "loss": 0.4248,
+      "step": 6838
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.785511363636364e-05,
+      "loss": 0.5352,
+      "step": 6839
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.785037878787879e-05,
+      "loss": 0.3723,
+      "step": 6840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.784564393939394e-05,
+      "loss": 0.4264,
+      "step": 6841
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.784090909090909e-05,
+      "loss": 0.4752,
+      "step": 6842
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7836174242424244e-05,
+      "loss": 0.5194,
+      "step": 6843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7831439393939394e-05,
+      "loss": 0.4446,
+      "step": 6844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7826704545454545e-05,
+      "loss": 0.337,
+      "step": 6845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.78219696969697e-05,
+      "loss": 0.3991,
+      "step": 6846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.781723484848485e-05,
+      "loss": 0.4663,
+      "step": 6847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7812500000000003e-05,
+      "loss": 0.3683,
+      "step": 6848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7807765151515153e-05,
+      "loss": 0.4858,
+      "step": 6849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7803030303030303e-05,
+      "loss": 0.4395,
+      "step": 6850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7798295454545454e-05,
+      "loss": 0.3999,
+      "step": 6851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7793560606060608e-05,
+      "loss": 0.3988,
+      "step": 6852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7788825757575758e-05,
+      "loss": 0.5398,
+      "step": 6853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7784090909090908e-05,
+      "loss": 0.3641,
+      "step": 6854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7779356060606062e-05,
+      "loss": 0.471,
+      "step": 6855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7774621212121212e-05,
+      "loss": 0.3576,
+      "step": 6856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7769886363636366e-05,
+      "loss": 0.4595,
+      "step": 6857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7765151515151517e-05,
+      "loss": 0.3813,
+      "step": 6858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7760416666666667e-05,
+      "loss": 0.4124,
+      "step": 6859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7755681818181817e-05,
+      "loss": 0.323,
+      "step": 6860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.775094696969697e-05,
+      "loss": 0.4276,
+      "step": 6861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.774621212121212e-05,
+      "loss": 0.4665,
+      "step": 6862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7741477272727275e-05,
+      "loss": 0.3722,
+      "step": 6863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7736742424242426e-05,
+      "loss": 0.4418,
+      "step": 6864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7732007575757576e-05,
+      "loss": 0.4319,
+      "step": 6865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.772727272727273e-05,
+      "loss": 0.4586,
+      "step": 6866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.772253787878788e-05,
+      "loss": 0.4139,
+      "step": 6867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.771780303030303e-05,
+      "loss": 0.4564,
+      "step": 6868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.771306818181818e-05,
+      "loss": 0.4181,
+      "step": 6869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7708333333333335e-05,
+      "loss": 0.4784,
+      "step": 6870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7703598484848485e-05,
+      "loss": 0.3942,
+      "step": 6871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.769886363636364e-05,
+      "loss": 0.4429,
+      "step": 6872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.769412878787879e-05,
+      "loss": 0.4414,
+      "step": 6873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.768939393939394e-05,
+      "loss": 0.5513,
+      "step": 6874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7684659090909093e-05,
+      "loss": 0.4093,
+      "step": 6875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7679924242424244e-05,
+      "loss": 0.2998,
+      "step": 6876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7675189393939394e-05,
+      "loss": 0.4758,
+      "step": 6877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7670454545454544e-05,
+      "loss": 0.4449,
+      "step": 6878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7665719696969698e-05,
+      "loss": 0.4986,
+      "step": 6879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.766098484848485e-05,
+      "loss": 0.4105,
+      "step": 6880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7656250000000002e-05,
+      "loss": 0.4622,
+      "step": 6881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7651515151515153e-05,
+      "loss": 0.491,
+      "step": 6882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7646780303030303e-05,
+      "loss": 0.4237,
+      "step": 6883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7642045454545457e-05,
+      "loss": 0.4242,
+      "step": 6884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7637310606060607e-05,
+      "loss": 0.5139,
+      "step": 6885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7632575757575758e-05,
+      "loss": 0.3118,
+      "step": 6886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7627840909090908e-05,
+      "loss": 0.3564,
+      "step": 6887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.762310606060606e-05,
+      "loss": 0.4644,
+      "step": 6888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7618371212121212e-05,
+      "loss": 0.531,
+      "step": 6889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7613636363636366e-05,
+      "loss": 0.4144,
+      "step": 6890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7608901515151516e-05,
+      "loss": 0.3415,
+      "step": 6891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.760416666666667e-05,
+      "loss": 0.3681,
+      "step": 6892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.759943181818182e-05,
+      "loss": 0.3924,
+      "step": 6893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.759469696969697e-05,
+      "loss": 0.3392,
+      "step": 6894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.758996212121212e-05,
+      "loss": 0.3371,
+      "step": 6895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.758522727272727e-05,
+      "loss": 0.3975,
+      "step": 6896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7580492424242425e-05,
+      "loss": 0.3579,
+      "step": 6897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7575757575757576e-05,
+      "loss": 0.352,
+      "step": 6898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.757102272727273e-05,
+      "loss": 0.4447,
+      "step": 6899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.756628787878788e-05,
+      "loss": 0.5279,
+      "step": 6900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7561553030303033e-05,
+      "loss": 0.4869,
+      "step": 6901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7556818181818184e-05,
+      "loss": 0.3497,
+      "step": 6902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7552083333333334e-05,
+      "loss": 0.4305,
+      "step": 6903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7547348484848485e-05,
+      "loss": 0.4014,
+      "step": 6904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7542613636363635e-05,
+      "loss": 0.4927,
+      "step": 6905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.753787878787879e-05,
+      "loss": 0.485,
+      "step": 6906
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.753314393939394e-05,
+      "loss": 0.3544,
+      "step": 6907
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7528409090909093e-05,
+      "loss": 0.2806,
+      "step": 6908
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7523674242424243e-05,
+      "loss": 0.4783,
+      "step": 6909
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7518939393939397e-05,
+      "loss": 0.4532,
+      "step": 6910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7514204545454547e-05,
+      "loss": 0.4263,
+      "step": 6911
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7509469696969698e-05,
+      "loss": 0.3461,
+      "step": 6912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7504734848484848e-05,
+      "loss": 0.4115,
+      "step": 6913
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.75e-05,
+      "loss": 0.4611,
+      "step": 6914
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7495265151515152e-05,
+      "loss": 0.4329,
+      "step": 6915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7490530303030303e-05,
+      "loss": 0.4023,
+      "step": 6916
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7485795454545456e-05,
+      "loss": 0.4175,
+      "step": 6917
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7481060606060607e-05,
+      "loss": 0.4871,
+      "step": 6918
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.747632575757576e-05,
+      "loss": 0.4329,
+      "step": 6919
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.747159090909091e-05,
+      "loss": 0.5798,
+      "step": 6920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.746685606060606e-05,
+      "loss": 0.408,
+      "step": 6921
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.746212121212121e-05,
+      "loss": 0.4273,
+      "step": 6922
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7457386363636362e-05,
+      "loss": 0.5544,
+      "step": 6923
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7452651515151516e-05,
+      "loss": 0.4127,
+      "step": 6924
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7447916666666666e-05,
+      "loss": 0.4044,
+      "step": 6925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.744318181818182e-05,
+      "loss": 0.3708,
+      "step": 6926
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.743844696969697e-05,
+      "loss": 0.4451,
+      "step": 6927
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7433712121212124e-05,
+      "loss": 0.4511,
+      "step": 6928
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7428977272727274e-05,
+      "loss": 0.3224,
+      "step": 6929
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7424242424242425e-05,
+      "loss": 0.4458,
+      "step": 6930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7419507575757575e-05,
+      "loss": 0.4008,
+      "step": 6931
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7414772727272726e-05,
+      "loss": 0.3869,
+      "step": 6932
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.741003787878788e-05,
+      "loss": 0.4889,
+      "step": 6933
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.740530303030303e-05,
+      "loss": 0.5431,
+      "step": 6934
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7400568181818183e-05,
+      "loss": 0.4706,
+      "step": 6935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7395833333333334e-05,
+      "loss": 0.4013,
+      "step": 6936
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7391098484848488e-05,
+      "loss": 0.3778,
+      "step": 6937
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7386363636363638e-05,
+      "loss": 0.5273,
+      "step": 6938
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7381628787878788e-05,
+      "loss": 0.6661,
+      "step": 6939
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.737689393939394e-05,
+      "loss": 0.4076,
+      "step": 6940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7372159090909092e-05,
+      "loss": 0.4515,
+      "step": 6941
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7367424242424243e-05,
+      "loss": 0.3954,
+      "step": 6942
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7362689393939393e-05,
+      "loss": 0.3926,
+      "step": 6943
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7357954545454547e-05,
+      "loss": 0.34,
+      "step": 6944
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7353219696969697e-05,
+      "loss": 0.492,
+      "step": 6945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.734848484848485e-05,
+      "loss": 0.4018,
+      "step": 6946
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.734375e-05,
+      "loss": 0.4468,
+      "step": 6947
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7339015151515152e-05,
+      "loss": 0.438,
+      "step": 6948
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7334280303030302e-05,
+      "loss": 0.3763,
+      "step": 6949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7329545454545456e-05,
+      "loss": 0.4529,
+      "step": 6950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7324810606060606e-05,
+      "loss": 0.4542,
+      "step": 6951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7320075757575757e-05,
+      "loss": 0.386,
+      "step": 6952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.731534090909091e-05,
+      "loss": 0.4004,
+      "step": 6953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.731060606060606e-05,
+      "loss": 0.4125,
+      "step": 6954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7305871212121215e-05,
+      "loss": 0.3429,
+      "step": 6955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7301136363636365e-05,
+      "loss": 0.4563,
+      "step": 6956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7296401515151515e-05,
+      "loss": 0.3914,
+      "step": 6957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7291666666666666e-05,
+      "loss": 0.3427,
+      "step": 6958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.728693181818182e-05,
+      "loss": 0.3547,
+      "step": 6959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.728219696969697e-05,
+      "loss": 0.3864,
+      "step": 6960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.727746212121212e-05,
+      "loss": 0.3926,
+      "step": 6961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 0.4701,
+      "step": 6962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7267992424242424e-05,
+      "loss": 0.4246,
+      "step": 6963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7263257575757578e-05,
+      "loss": 0.3819,
+      "step": 6964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.725852272727273e-05,
+      "loss": 0.4788,
+      "step": 6965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.725378787878788e-05,
+      "loss": 0.3239,
+      "step": 6966
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.724905303030303e-05,
+      "loss": 0.4347,
+      "step": 6967
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7244318181818183e-05,
+      "loss": 0.3234,
+      "step": 6968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7239583333333333e-05,
+      "loss": 0.3125,
+      "step": 6969
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7234848484848487e-05,
+      "loss": 0.5802,
+      "step": 6970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7230113636363638e-05,
+      "loss": 0.4384,
+      "step": 6971
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7225378787878788e-05,
+      "loss": 0.3112,
+      "step": 6972
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.722064393939394e-05,
+      "loss": 0.5045,
+      "step": 6973
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7215909090909092e-05,
+      "loss": 0.4496,
+      "step": 6974
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7211174242424242e-05,
+      "loss": 0.4656,
+      "step": 6975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7206439393939393e-05,
+      "loss": 0.4737,
+      "step": 6976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7201704545454547e-05,
+      "loss": 0.4386,
+      "step": 6977
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7196969696969697e-05,
+      "loss": 0.4635,
+      "step": 6978
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.719223484848485e-05,
+      "loss": 0.3592,
+      "step": 6979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.71875e-05,
+      "loss": 0.3719,
+      "step": 6980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.718276515151515e-05,
+      "loss": 0.3652,
+      "step": 6981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7178030303030305e-05,
+      "loss": 0.3775,
+      "step": 6982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7173295454545456e-05,
+      "loss": 0.373,
+      "step": 6983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7168560606060606e-05,
+      "loss": 0.3677,
+      "step": 6984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7163825757575756e-05,
+      "loss": 0.4608,
+      "step": 6985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.715909090909091e-05,
+      "loss": 0.3683,
+      "step": 6986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.715435606060606e-05,
+      "loss": 0.4323,
+      "step": 6987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7149621212121214e-05,
+      "loss": 0.4742,
+      "step": 6988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7144886363636365e-05,
+      "loss": 0.3658,
+      "step": 6989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7140151515151515e-05,
+      "loss": 0.4272,
+      "step": 6990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.713541666666667e-05,
+      "loss": 0.3197,
+      "step": 6991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.713068181818182e-05,
+      "loss": 0.4319,
+      "step": 6992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.712594696969697e-05,
+      "loss": 0.4764,
+      "step": 6993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.712121212121212e-05,
+      "loss": 0.4655,
+      "step": 6994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7116477272727274e-05,
+      "loss": 0.4423,
+      "step": 6995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7111742424242424e-05,
+      "loss": 0.3688,
+      "step": 6996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7107007575757578e-05,
+      "loss": 0.4334,
+      "step": 6997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7102272727272728e-05,
+      "loss": 0.3603,
+      "step": 6998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7097537878787882e-05,
+      "loss": 0.4243,
+      "step": 6999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7092803030303032e-05,
+      "loss": 0.4687,
+      "step": 7000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7088068181818183e-05,
+      "loss": 0.3641,
+      "step": 7001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7083333333333333e-05,
+      "loss": 0.4925,
+      "step": 7002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7078598484848483e-05,
+      "loss": 0.5205,
+      "step": 7003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7073863636363637e-05,
+      "loss": 0.4716,
+      "step": 7004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7069128787878787e-05,
+      "loss": 0.4598,
+      "step": 7005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.706439393939394e-05,
+      "loss": 0.4162,
+      "step": 7006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.705965909090909e-05,
+      "loss": 0.3534,
+      "step": 7007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7054924242424245e-05,
+      "loss": 0.4088,
+      "step": 7008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7050189393939396e-05,
+      "loss": 0.391,
+      "step": 7009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7045454545454546e-05,
+      "loss": 0.4093,
+      "step": 7010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7040719696969696e-05,
+      "loss": 0.5061,
+      "step": 7011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7035984848484847e-05,
+      "loss": 0.2784,
+      "step": 7012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.703125e-05,
+      "loss": 0.4181,
+      "step": 7013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.702651515151515e-05,
+      "loss": 0.497,
+      "step": 7014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7021780303030305e-05,
+      "loss": 0.5576,
+      "step": 7015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7017045454545455e-05,
+      "loss": 0.4869,
+      "step": 7016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.701231060606061e-05,
+      "loss": 0.4456,
+      "step": 7017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.700757575757576e-05,
+      "loss": 0.2308,
+      "step": 7018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.700284090909091e-05,
+      "loss": 0.3524,
+      "step": 7019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.699810606060606e-05,
+      "loss": 0.3499,
+      "step": 7020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.699337121212121e-05,
+      "loss": 0.3956,
+      "step": 7021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6988636363636364e-05,
+      "loss": 0.3888,
+      "step": 7022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6983901515151515e-05,
+      "loss": 0.4285,
+      "step": 7023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6979166666666668e-05,
+      "loss": 0.3841,
+      "step": 7024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.697443181818182e-05,
+      "loss": 0.3633,
+      "step": 7025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6969696969696972e-05,
+      "loss": 0.4645,
+      "step": 7026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6964962121212123e-05,
+      "loss": 0.3376,
+      "step": 7027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6960227272727273e-05,
+      "loss": 0.3029,
+      "step": 7028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6955492424242424e-05,
+      "loss": 0.4229,
+      "step": 7029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6950757575757574e-05,
+      "loss": 0.3226,
+      "step": 7030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6946022727272728e-05,
+      "loss": 0.3762,
+      "step": 7031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6941287878787878e-05,
+      "loss": 0.521,
+      "step": 7032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6936553030303032e-05,
+      "loss": 0.3838,
+      "step": 7033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6931818181818182e-05,
+      "loss": 0.5467,
+      "step": 7034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6927083333333336e-05,
+      "loss": 0.4833,
+      "step": 7035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6922348484848486e-05,
+      "loss": 0.3757,
+      "step": 7036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.691761363636364e-05,
+      "loss": 0.6216,
+      "step": 7037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6912878787878787e-05,
+      "loss": 0.3966,
+      "step": 7038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6908143939393937e-05,
+      "loss": 0.4243,
+      "step": 7039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.690340909090909e-05,
+      "loss": 0.3634,
+      "step": 7040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.689867424242424e-05,
+      "loss": 0.4364,
+      "step": 7041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6893939393939395e-05,
+      "loss": 0.3859,
+      "step": 7042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6889204545454546e-05,
+      "loss": 0.3482,
+      "step": 7043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.68844696969697e-05,
+      "loss": 0.4744,
+      "step": 7044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.687973484848485e-05,
+      "loss": 0.4068,
+      "step": 7045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6875000000000004e-05,
+      "loss": 0.3941,
+      "step": 7046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.687026515151515e-05,
+      "loss": 0.3275,
+      "step": 7047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.68655303030303e-05,
+      "loss": 0.3892,
+      "step": 7048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6860795454545455e-05,
+      "loss": 0.2928,
+      "step": 7049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6856060606060605e-05,
+      "loss": 0.3934,
+      "step": 7050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.685132575757576e-05,
+      "loss": 0.5273,
+      "step": 7051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.684659090909091e-05,
+      "loss": 0.4961,
+      "step": 7052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6841856060606063e-05,
+      "loss": 0.5076,
+      "step": 7053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6837121212121213e-05,
+      "loss": 0.3906,
+      "step": 7054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6832386363636367e-05,
+      "loss": 0.3946,
+      "step": 7055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6827651515151514e-05,
+      "loss": 0.4104,
+      "step": 7056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6822916666666668e-05,
+      "loss": 0.4179,
+      "step": 7057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6818181818181818e-05,
+      "loss": 0.3573,
+      "step": 7058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.681344696969697e-05,
+      "loss": 0.4368,
+      "step": 7059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6808712121212122e-05,
+      "loss": 0.5051,
+      "step": 7060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6803977272727273e-05,
+      "loss": 0.4534,
+      "step": 7061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6799242424242427e-05,
+      "loss": 0.4078,
+      "step": 7062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6794507575757577e-05,
+      "loss": 0.3605,
+      "step": 7063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.678977272727273e-05,
+      "loss": 0.4943,
+      "step": 7064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6785037878787878e-05,
+      "loss": 0.3333,
+      "step": 7065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.678030303030303e-05,
+      "loss": 0.3469,
+      "step": 7066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6775568181818182e-05,
+      "loss": 0.4048,
+      "step": 7067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6770833333333332e-05,
+      "loss": 0.3644,
+      "step": 7068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6766098484848486e-05,
+      "loss": 0.4217,
+      "step": 7069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6761363636363636e-05,
+      "loss": 0.3384,
+      "step": 7070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.675662878787879e-05,
+      "loss": 0.3911,
+      "step": 7071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.675189393939394e-05,
+      "loss": 0.5112,
+      "step": 7072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6747159090909094e-05,
+      "loss": 0.6303,
+      "step": 7073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.674242424242424e-05,
+      "loss": 0.3358,
+      "step": 7074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6737689393939395e-05,
+      "loss": 0.422,
+      "step": 7075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6732954545454545e-05,
+      "loss": 0.445,
+      "step": 7076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.67282196969697e-05,
+      "loss": 0.4125,
+      "step": 7077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.672348484848485e-05,
+      "loss": 0.2739,
+      "step": 7078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.671875e-05,
+      "loss": 0.3794,
+      "step": 7079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6714015151515154e-05,
+      "loss": 0.4228,
+      "step": 7080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6709280303030304e-05,
+      "loss": 0.492,
+      "step": 7081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6704545454545458e-05,
+      "loss": 0.4138,
+      "step": 7082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6699810606060605e-05,
+      "loss": 0.5721,
+      "step": 7083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.669507575757576e-05,
+      "loss": 0.4345,
+      "step": 7084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.669034090909091e-05,
+      "loss": 0.3796,
+      "step": 7085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6685606060606063e-05,
+      "loss": 0.3736,
+      "step": 7086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6680871212121213e-05,
+      "loss": 0.4261,
+      "step": 7087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6676136363636363e-05,
+      "loss": 0.4781,
+      "step": 7088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6671401515151517e-05,
+      "loss": 0.3841,
+      "step": 7089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.4348,
+      "step": 7090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.666193181818182e-05,
+      "loss": 0.486,
+      "step": 7091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6657196969696968e-05,
+      "loss": 0.2975,
+      "step": 7092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6652462121212122e-05,
+      "loss": 0.3634,
+      "step": 7093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6647727272727272e-05,
+      "loss": 0.334,
+      "step": 7094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6642992424242426e-05,
+      "loss": 0.4555,
+      "step": 7095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6638257575757576e-05,
+      "loss": 0.448,
+      "step": 7096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6633522727272727e-05,
+      "loss": 0.2879,
+      "step": 7097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.662878787878788e-05,
+      "loss": 0.4585,
+      "step": 7098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.662405303030303e-05,
+      "loss": 0.4244,
+      "step": 7099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6619318181818185e-05,
+      "loss": 0.3813,
+      "step": 7100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6614583333333332e-05,
+      "loss": 0.3178,
+      "step": 7101
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6609848484848485e-05,
+      "loss": 0.4237,
+      "step": 7102
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6605113636363636e-05,
+      "loss": 0.3839,
+      "step": 7103
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.660037878787879e-05,
+      "loss": 0.3131,
+      "step": 7104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.659564393939394e-05,
+      "loss": 0.3706,
+      "step": 7105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6590909090909094e-05,
+      "loss": 0.3833,
+      "step": 7106
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6586174242424244e-05,
+      "loss": 0.3668,
+      "step": 7107
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6581439393939395e-05,
+      "loss": 0.3388,
+      "step": 7108
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6576704545454548e-05,
+      "loss": 0.3692,
+      "step": 7109
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6571969696969695e-05,
+      "loss": 0.3903,
+      "step": 7110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.656723484848485e-05,
+      "loss": 0.4425,
+      "step": 7111
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.65625e-05,
+      "loss": 0.4683,
+      "step": 7112
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6557765151515153e-05,
+      "loss": 0.4632,
+      "step": 7113
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6553030303030304e-05,
+      "loss": 0.5409,
+      "step": 7114
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6548295454545457e-05,
+      "loss": 0.3524,
+      "step": 7115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6543560606060608e-05,
+      "loss": 0.3716,
+      "step": 7116
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6538825757575758e-05,
+      "loss": 0.3167,
+      "step": 7117
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6534090909090912e-05,
+      "loss": 0.4002,
+      "step": 7118
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.652935606060606e-05,
+      "loss": 0.4745,
+      "step": 7119
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6524621212121213e-05,
+      "loss": 0.4262,
+      "step": 7120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6519886363636363e-05,
+      "loss": 0.5866,
+      "step": 7121
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6515151515151517e-05,
+      "loss": 0.3796,
+      "step": 7122
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6510416666666667e-05,
+      "loss": 0.4243,
+      "step": 7123
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.650568181818182e-05,
+      "loss": 0.4109,
+      "step": 7124
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.650094696969697e-05,
+      "loss": 0.3142,
+      "step": 7125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.649621212121212e-05,
+      "loss": 0.3062,
+      "step": 7126
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6491477272727275e-05,
+      "loss": 0.4484,
+      "step": 7127
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6486742424242422e-05,
+      "loss": 0.534,
+      "step": 7128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6482007575757576e-05,
+      "loss": 0.3886,
+      "step": 7129
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6477272727272726e-05,
+      "loss": 0.4205,
+      "step": 7130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.647253787878788e-05,
+      "loss": 0.4229,
+      "step": 7131
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.646780303030303e-05,
+      "loss": 0.3226,
+      "step": 7132
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6463068181818184e-05,
+      "loss": 0.3442,
+      "step": 7133
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6458333333333335e-05,
+      "loss": 0.3386,
+      "step": 7134
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.645359848484849e-05,
+      "loss": 0.3478,
+      "step": 7135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.644886363636364e-05,
+      "loss": 0.4185,
+      "step": 7136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6444128787878786e-05,
+      "loss": 0.3945,
+      "step": 7137
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.643939393939394e-05,
+      "loss": 0.5571,
+      "step": 7138
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.643465909090909e-05,
+      "loss": 0.4451,
+      "step": 7139
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6429924242424244e-05,
+      "loss": 0.4962,
+      "step": 7140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6425189393939394e-05,
+      "loss": 0.4174,
+      "step": 7141
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6420454545454548e-05,
+      "loss": 0.3996,
+      "step": 7142
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6415719696969698e-05,
+      "loss": 0.4214,
+      "step": 7143
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6410984848484852e-05,
+      "loss": 0.4116,
+      "step": 7144
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6406250000000002e-05,
+      "loss": 0.3281,
+      "step": 7145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6401515151515153e-05,
+      "loss": 0.3491,
+      "step": 7146
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6396780303030303e-05,
+      "loss": 0.5238,
+      "step": 7147
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6392045454545453e-05,
+      "loss": 0.4536,
+      "step": 7148
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6387310606060607e-05,
+      "loss": 0.4375,
+      "step": 7149
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6382575757575758e-05,
+      "loss": 0.4075,
+      "step": 7150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.637784090909091e-05,
+      "loss": 0.3527,
+      "step": 7151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6373106060606062e-05,
+      "loss": 0.5142,
+      "step": 7152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6368371212121216e-05,
+      "loss": 0.2709,
+      "step": 7153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 0.3935,
+      "step": 7154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6358901515151516e-05,
+      "loss": 0.3161,
+      "step": 7155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6354166666666667e-05,
+      "loss": 0.3434,
+      "step": 7156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6349431818181817e-05,
+      "loss": 0.3966,
+      "step": 7157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.634469696969697e-05,
+      "loss": 0.399,
+      "step": 7158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.633996212121212e-05,
+      "loss": 0.4772,
+      "step": 7159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6335227272727275e-05,
+      "loss": 0.4632,
+      "step": 7160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6330492424242425e-05,
+      "loss": 0.3635,
+      "step": 7161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.632575757575758e-05,
+      "loss": 0.4803,
+      "step": 7162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.632102272727273e-05,
+      "loss": 0.3884,
+      "step": 7163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.631628787878788e-05,
+      "loss": 0.375,
+      "step": 7164
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.631155303030303e-05,
+      "loss": 0.3078,
+      "step": 7165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.630681818181818e-05,
+      "loss": 0.3307,
+      "step": 7166
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6302083333333334e-05,
+      "loss": 0.4012,
+      "step": 7167
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6297348484848485e-05,
+      "loss": 0.4952,
+      "step": 7168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.629261363636364e-05,
+      "loss": 0.4766,
+      "step": 7169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.628787878787879e-05,
+      "loss": 0.3717,
+      "step": 7170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6283143939393943e-05,
+      "loss": 0.3337,
+      "step": 7171
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6278409090909093e-05,
+      "loss": 0.4188,
+      "step": 7172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6273674242424243e-05,
+      "loss": 0.398,
+      "step": 7173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6268939393939394e-05,
+      "loss": 0.4157,
+      "step": 7174
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6264204545454544e-05,
+      "loss": 0.3971,
+      "step": 7175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6259469696969698e-05,
+      "loss": 0.3935,
+      "step": 7176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6254734848484848e-05,
+      "loss": 0.5781,
+      "step": 7177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 0.4736,
+      "step": 7178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6245265151515152e-05,
+      "loss": 0.36,
+      "step": 7179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6240530303030306e-05,
+      "loss": 0.2615,
+      "step": 7180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6235795454545456e-05,
+      "loss": 0.4503,
+      "step": 7181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6231060606060607e-05,
+      "loss": 0.3913,
+      "step": 7182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6226325757575757e-05,
+      "loss": 0.5009,
+      "step": 7183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6221590909090908e-05,
+      "loss": 0.466,
+      "step": 7184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.621685606060606e-05,
+      "loss": 0.3532,
+      "step": 7185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6212121212121212e-05,
+      "loss": 0.4039,
+      "step": 7186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6207386363636365e-05,
+      "loss": 0.3414,
+      "step": 7187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6202651515151516e-05,
+      "loss": 0.3855,
+      "step": 7188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.619791666666667e-05,
+      "loss": 0.4124,
+      "step": 7189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.619318181818182e-05,
+      "loss": 0.557,
+      "step": 7190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.618844696969697e-05,
+      "loss": 0.3278,
+      "step": 7191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.618371212121212e-05,
+      "loss": 0.5164,
+      "step": 7192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6178977272727275e-05,
+      "loss": 0.3107,
+      "step": 7193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6174242424242425e-05,
+      "loss": 0.334,
+      "step": 7194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6169507575757575e-05,
+      "loss": 0.3547,
+      "step": 7195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.616477272727273e-05,
+      "loss": 0.3866,
+      "step": 7196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.616003787878788e-05,
+      "loss": 0.5374,
+      "step": 7197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6155303030303033e-05,
+      "loss": 0.4105,
+      "step": 7198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6150568181818184e-05,
+      "loss": 0.4619,
+      "step": 7199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6145833333333334e-05,
+      "loss": 0.4009,
+      "step": 7200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6141098484848484e-05,
+      "loss": 0.4583,
+      "step": 7201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6136363636363638e-05,
+      "loss": 0.3775,
+      "step": 7202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.613162878787879e-05,
+      "loss": 0.4414,
+      "step": 7203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.612689393939394e-05,
+      "loss": 0.3839,
+      "step": 7204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6122159090909093e-05,
+      "loss": 0.407,
+      "step": 7205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6117424242424243e-05,
+      "loss": 0.4813,
+      "step": 7206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6112689393939397e-05,
+      "loss": 0.4463,
+      "step": 7207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6107954545454547e-05,
+      "loss": 0.3016,
+      "step": 7208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6103219696969697e-05,
+      "loss": 0.5617,
+      "step": 7209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6098484848484848e-05,
+      "loss": 0.4141,
+      "step": 7210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.609375e-05,
+      "loss": 0.5333,
+      "step": 7211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6089015151515152e-05,
+      "loss": 0.4139,
+      "step": 7212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6084280303030302e-05,
+      "loss": 0.4188,
+      "step": 7213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6079545454545456e-05,
+      "loss": 0.4736,
+      "step": 7214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6074810606060606e-05,
+      "loss": 0.4636,
+      "step": 7215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.607007575757576e-05,
+      "loss": 0.4105,
+      "step": 7216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.606534090909091e-05,
+      "loss": 0.426,
+      "step": 7217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.606060606060606e-05,
+      "loss": 0.3458,
+      "step": 7218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.605587121212121e-05,
+      "loss": 0.4409,
+      "step": 7219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6051136363636365e-05,
+      "loss": 0.334,
+      "step": 7220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6046401515151515e-05,
+      "loss": 0.3535,
+      "step": 7221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.604166666666667e-05,
+      "loss": 0.422,
+      "step": 7222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.603693181818182e-05,
+      "loss": 0.4135,
+      "step": 7223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.603219696969697e-05,
+      "loss": 0.3766,
+      "step": 7224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6027462121212124e-05,
+      "loss": 0.4144,
+      "step": 7225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6022727272727274e-05,
+      "loss": 0.2797,
+      "step": 7226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6017992424242424e-05,
+      "loss": 0.4058,
+      "step": 7227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6013257575757575e-05,
+      "loss": 0.4085,
+      "step": 7228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.600852272727273e-05,
+      "loss": 0.4254,
+      "step": 7229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.600378787878788e-05,
+      "loss": 0.4202,
+      "step": 7230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5999053030303033e-05,
+      "loss": 0.4114,
+      "step": 7231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5994318181818183e-05,
+      "loss": 0.3781,
+      "step": 7232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5989583333333333e-05,
+      "loss": 0.4305,
+      "step": 7233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5984848484848487e-05,
+      "loss": 0.4848,
+      "step": 7234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5980113636363638e-05,
+      "loss": 0.3954,
+      "step": 7235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5975378787878788e-05,
+      "loss": 0.3804,
+      "step": 7236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.597064393939394e-05,
+      "loss": 0.5151,
+      "step": 7237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5965909090909092e-05,
+      "loss": 0.5004,
+      "step": 7238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5961174242424243e-05,
+      "loss": 0.4763,
+      "step": 7239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5956439393939396e-05,
+      "loss": 0.4615,
+      "step": 7240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5951704545454547e-05,
+      "loss": 0.3209,
+      "step": 7241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.59469696969697e-05,
+      "loss": 0.3656,
+      "step": 7242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.594223484848485e-05,
+      "loss": 0.3324,
+      "step": 7243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.59375e-05,
+      "loss": 0.569,
+      "step": 7244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.593276515151515e-05,
+      "loss": 0.2867,
+      "step": 7245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5928030303030302e-05,
+      "loss": 0.4524,
+      "step": 7246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5923295454545456e-05,
+      "loss": 0.3554,
+      "step": 7247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5918560606060606e-05,
+      "loss": 0.2858,
+      "step": 7248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.591382575757576e-05,
+      "loss": 0.5212,
+      "step": 7249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 0.3272,
+      "step": 7250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5904356060606064e-05,
+      "loss": 0.4768,
+      "step": 7251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5899621212121214e-05,
+      "loss": 0.4469,
+      "step": 7252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5894886363636365e-05,
+      "loss": 0.414,
+      "step": 7253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5890151515151515e-05,
+      "loss": 0.3964,
+      "step": 7254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5885416666666665e-05,
+      "loss": 0.5213,
+      "step": 7255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.588068181818182e-05,
+      "loss": 0.3919,
+      "step": 7256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.587594696969697e-05,
+      "loss": 0.6115,
+      "step": 7257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5871212121212123e-05,
+      "loss": 0.5049,
+      "step": 7258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5866477272727274e-05,
+      "loss": 0.4509,
+      "step": 7259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5861742424242427e-05,
+      "loss": 0.4274,
+      "step": 7260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5857007575757578e-05,
+      "loss": 0.4425,
+      "step": 7261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5852272727272728e-05,
+      "loss": 0.4776,
+      "step": 7262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.584753787878788e-05,
+      "loss": 0.2678,
+      "step": 7263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.584280303030303e-05,
+      "loss": 0.3452,
+      "step": 7264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5838068181818183e-05,
+      "loss": 0.5455,
+      "step": 7265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.3858,
+      "step": 7266
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5828598484848487e-05,
+      "loss": 0.455,
+      "step": 7267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5823863636363637e-05,
+      "loss": 0.3912,
+      "step": 7268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.581912878787879e-05,
+      "loss": 0.412,
+      "step": 7269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.581439393939394e-05,
+      "loss": 0.4435,
+      "step": 7270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5809659090909092e-05,
+      "loss": 0.3715,
+      "step": 7271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5804924242424242e-05,
+      "loss": 0.3845,
+      "step": 7272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5800189393939392e-05,
+      "loss": 0.5123,
+      "step": 7273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5795454545454546e-05,
+      "loss": 0.4793,
+      "step": 7274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5790719696969697e-05,
+      "loss": 0.4028,
+      "step": 7275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.578598484848485e-05,
+      "loss": 0.4259,
+      "step": 7276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.578125e-05,
+      "loss": 0.3729,
+      "step": 7277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5776515151515154e-05,
+      "loss": 0.316,
+      "step": 7278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5771780303030305e-05,
+      "loss": 0.4073,
+      "step": 7279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5767045454545455e-05,
+      "loss": 0.539,
+      "step": 7280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5762310606060606e-05,
+      "loss": 0.441,
+      "step": 7281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5757575757575756e-05,
+      "loss": 0.571,
+      "step": 7282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.575284090909091e-05,
+      "loss": 0.413,
+      "step": 7283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.574810606060606e-05,
+      "loss": 0.4877,
+      "step": 7284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5743371212121214e-05,
+      "loss": 0.4291,
+      "step": 7285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5738636363636364e-05,
+      "loss": 0.2602,
+      "step": 7286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5733901515151518e-05,
+      "loss": 0.4702,
+      "step": 7287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.572916666666667e-05,
+      "loss": 0.3388,
+      "step": 7288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.572443181818182e-05,
+      "loss": 0.366,
+      "step": 7289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.571969696969697e-05,
+      "loss": 0.4976,
+      "step": 7290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.571496212121212e-05,
+      "loss": 0.5712,
+      "step": 7291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5710227272727273e-05,
+      "loss": 0.4978,
+      "step": 7292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5705492424242424e-05,
+      "loss": 0.4458,
+      "step": 7293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5700757575757577e-05,
+      "loss": 0.4202,
+      "step": 7294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5696022727272728e-05,
+      "loss": 0.4157,
+      "step": 7295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.569128787878788e-05,
+      "loss": 0.4683,
+      "step": 7296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5686553030303032e-05,
+      "loss": 0.3746,
+      "step": 7297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5681818181818182e-05,
+      "loss": 0.3535,
+      "step": 7298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5677083333333333e-05,
+      "loss": 0.5265,
+      "step": 7299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5672348484848486e-05,
+      "loss": 0.5194,
+      "step": 7300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5667613636363637e-05,
+      "loss": 0.3815,
+      "step": 7301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5662878787878787e-05,
+      "loss": 0.4791,
+      "step": 7302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.565814393939394e-05,
+      "loss": 0.3974,
+      "step": 7303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.565340909090909e-05,
+      "loss": 0.365,
+      "step": 7304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5648674242424245e-05,
+      "loss": 0.3671,
+      "step": 7305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5643939393939395e-05,
+      "loss": 0.3976,
+      "step": 7306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5639204545454546e-05,
+      "loss": 0.3842,
+      "step": 7307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5634469696969696e-05,
+      "loss": 0.4794,
+      "step": 7308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.562973484848485e-05,
+      "loss": 0.3724,
+      "step": 7309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.36,
+      "step": 7310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.562026515151515e-05,
+      "loss": 0.3803,
+      "step": 7311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5615530303030304e-05,
+      "loss": 0.5049,
+      "step": 7312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5610795454545455e-05,
+      "loss": 0.3823,
+      "step": 7313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.560606060606061e-05,
+      "loss": 0.371,
+      "step": 7314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.560132575757576e-05,
+      "loss": 0.4655,
+      "step": 7315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.559659090909091e-05,
+      "loss": 0.3747,
+      "step": 7316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.559185606060606e-05,
+      "loss": 0.4726,
+      "step": 7317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5587121212121213e-05,
+      "loss": 0.4927,
+      "step": 7318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5582386363636364e-05,
+      "loss": 0.4497,
+      "step": 7319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5577651515151514e-05,
+      "loss": 0.3135,
+      "step": 7320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5572916666666668e-05,
+      "loss": 0.6661,
+      "step": 7321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.556818181818182e-05,
+      "loss": 0.3867,
+      "step": 7322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5563446969696972e-05,
+      "loss": 0.5557,
+      "step": 7323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5558712121212122e-05,
+      "loss": 0.3927,
+      "step": 7324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5553977272727273e-05,
+      "loss": 0.4501,
+      "step": 7325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5549242424242423e-05,
+      "loss": 0.4284,
+      "step": 7326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5544507575757577e-05,
+      "loss": 0.3917,
+      "step": 7327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5539772727272727e-05,
+      "loss": 0.345,
+      "step": 7328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.553503787878788e-05,
+      "loss": 0.3931,
+      "step": 7329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.553030303030303e-05,
+      "loss": 0.5628,
+      "step": 7330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5525568181818182e-05,
+      "loss": 0.4425,
+      "step": 7331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5520833333333336e-05,
+      "loss": 0.4593,
+      "step": 7332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5516098484848486e-05,
+      "loss": 0.5904,
+      "step": 7333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5511363636363636e-05,
+      "loss": 0.3934,
+      "step": 7334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5506628787878787e-05,
+      "loss": 0.4842,
+      "step": 7335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.550189393939394e-05,
+      "loss": 0.4462,
+      "step": 7336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.549715909090909e-05,
+      "loss": 0.3874,
+      "step": 7337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5492424242424245e-05,
+      "loss": 0.4042,
+      "step": 7338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5487689393939395e-05,
+      "loss": 0.5217,
+      "step": 7339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5482954545454545e-05,
+      "loss": 0.3504,
+      "step": 7340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.54782196969697e-05,
+      "loss": 0.3731,
+      "step": 7341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.547348484848485e-05,
+      "loss": 0.4177,
+      "step": 7342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.546875e-05,
+      "loss": 0.3851,
+      "step": 7343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.546401515151515e-05,
+      "loss": 0.3349,
+      "step": 7344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5459280303030304e-05,
+      "loss": 0.3252,
+      "step": 7345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 0.4399,
+      "step": 7346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5449810606060608e-05,
+      "loss": 0.5831,
+      "step": 7347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.544507575757576e-05,
+      "loss": 0.3938,
+      "step": 7348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.544034090909091e-05,
+      "loss": 0.3939,
+      "step": 7349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5435606060606063e-05,
+      "loss": 0.4176,
+      "step": 7350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5430871212121213e-05,
+      "loss": 0.6944,
+      "step": 7351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5426136363636363e-05,
+      "loss": 0.4842,
+      "step": 7352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5421401515151514e-05,
+      "loss": 0.4177,
+      "step": 7353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5416666666666668e-05,
+      "loss": 0.4255,
+      "step": 7354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5411931818181818e-05,
+      "loss": 0.3747,
+      "step": 7355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5407196969696972e-05,
+      "loss": 0.5268,
+      "step": 7356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5402462121212122e-05,
+      "loss": 0.4767,
+      "step": 7357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5397727272727276e-05,
+      "loss": 0.3644,
+      "step": 7358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5392992424242426e-05,
+      "loss": 0.4317,
+      "step": 7359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5388257575757577e-05,
+      "loss": 0.3598,
+      "step": 7360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5383522727272727e-05,
+      "loss": 0.38,
+      "step": 7361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5378787878787877e-05,
+      "loss": 0.4844,
+      "step": 7362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.537405303030303e-05,
+      "loss": 0.3427,
+      "step": 7363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.536931818181818e-05,
+      "loss": 0.4352,
+      "step": 7364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5364583333333335e-05,
+      "loss": 0.4375,
+      "step": 7365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5359848484848486e-05,
+      "loss": 0.4491,
+      "step": 7366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.535511363636364e-05,
+      "loss": 0.48,
+      "step": 7367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.535037878787879e-05,
+      "loss": 0.3487,
+      "step": 7368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.534564393939394e-05,
+      "loss": 0.4894,
+      "step": 7369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.534090909090909e-05,
+      "loss": 0.3312,
+      "step": 7370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.533617424242424e-05,
+      "loss": 0.4882,
+      "step": 7371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5331439393939395e-05,
+      "loss": 0.4953,
+      "step": 7372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5326704545454545e-05,
+      "loss": 0.4533,
+      "step": 7373
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.53219696969697e-05,
+      "loss": 0.4696,
+      "step": 7374
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.531723484848485e-05,
+      "loss": 0.3501,
+      "step": 7375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5312500000000003e-05,
+      "loss": 0.3875,
+      "step": 7376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5307765151515153e-05,
+      "loss": 0.3989,
+      "step": 7377
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5303030303030304e-05,
+      "loss": 0.3397,
+      "step": 7378
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5298295454545454e-05,
+      "loss": 0.4932,
+      "step": 7379
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5293560606060604e-05,
+      "loss": 0.4968,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5288825757575758e-05,
+      "loss": 0.2994,
+      "step": 7381
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.528409090909091e-05,
+      "loss": 0.4185,
+      "step": 7382
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5279356060606062e-05,
+      "loss": 0.4215,
+      "step": 7383
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5274621212121213e-05,
+      "loss": 0.3388,
+      "step": 7384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5269886363636366e-05,
+      "loss": 0.5896,
+      "step": 7385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5265151515151517e-05,
+      "loss": 0.3282,
+      "step": 7386
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5260416666666667e-05,
+      "loss": 0.4387,
+      "step": 7387
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.525568181818182e-05,
+      "loss": 0.587,
+      "step": 7388
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.525094696969697e-05,
+      "loss": 0.4378,
+      "step": 7389
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5246212121212122e-05,
+      "loss": 0.3563,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5241477272727272e-05,
+      "loss": 0.4292,
+      "step": 7391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5236742424242426e-05,
+      "loss": 0.2722,
+      "step": 7392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5232007575757576e-05,
+      "loss": 0.3965,
+      "step": 7393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5227272727272728e-05,
+      "loss": 0.4753,
+      "step": 7394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5222537878787879e-05,
+      "loss": 0.4606,
+      "step": 7395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5217803030303032e-05,
+      "loss": 0.5287,
+      "step": 7396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5213068181818183e-05,
+      "loss": 0.329,
+      "step": 7397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5208333333333333e-05,
+      "loss": 0.4978,
+      "step": 7398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5203598484848485e-05,
+      "loss": 0.4781,
+      "step": 7399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5198863636363636e-05,
+      "loss": 0.3972,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.519412878787879e-05,
+      "loss": 0.411,
+      "step": 7401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.518939393939394e-05,
+      "loss": 0.4883,
+      "step": 7402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5184659090909092e-05,
+      "loss": 0.3616,
+      "step": 7403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5179924242424242e-05,
+      "loss": 0.3465,
+      "step": 7404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5175189393939396e-05,
+      "loss": 0.3737,
+      "step": 7405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5170454545454546e-05,
+      "loss": 0.4527,
+      "step": 7406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5165719696969698e-05,
+      "loss": 0.4879,
+      "step": 7407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5160984848484849e-05,
+      "loss": 0.4165,
+      "step": 7408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5156249999999999e-05,
+      "loss": 0.5183,
+      "step": 7409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.4942,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5146780303030303e-05,
+      "loss": 0.3479,
+      "step": 7411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5142045454545455e-05,
+      "loss": 0.5076,
+      "step": 7412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5137310606060606e-05,
+      "loss": 0.3364,
+      "step": 7413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.513257575757576e-05,
+      "loss": 0.3777,
+      "step": 7414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.512784090909091e-05,
+      "loss": 0.3729,
+      "step": 7415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5123106060606062e-05,
+      "loss": 0.5554,
+      "step": 7416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5118371212121212e-05,
+      "loss": 0.374,
+      "step": 7417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5113636363636363e-05,
+      "loss": 0.437,
+      "step": 7418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5108901515151516e-05,
+      "loss": 0.5227,
+      "step": 7419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5104166666666667e-05,
+      "loss": 0.4875,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5099431818181819e-05,
+      "loss": 0.2713,
+      "step": 7421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.509469696969697e-05,
+      "loss": 0.3788,
+      "step": 7422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5089962121212123e-05,
+      "loss": 0.4048,
+      "step": 7423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5085227272727273e-05,
+      "loss": 0.4682,
+      "step": 7424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5080492424242425e-05,
+      "loss": 0.4645,
+      "step": 7425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5075757575757576e-05,
+      "loss": 0.468,
+      "step": 7426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5071022727272726e-05,
+      "loss": 0.4017,
+      "step": 7427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.506628787878788e-05,
+      "loss": 0.35,
+      "step": 7428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.506155303030303e-05,
+      "loss": 0.4081,
+      "step": 7429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5056818181818182e-05,
+      "loss": 0.5287,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5052083333333333e-05,
+      "loss": 0.5014,
+      "step": 7431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5047348484848486e-05,
+      "loss": 0.3811,
+      "step": 7432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5042613636363637e-05,
+      "loss": 0.3318,
+      "step": 7433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5037878787878789e-05,
+      "loss": 0.4662,
+      "step": 7434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.503314393939394e-05,
+      "loss": 0.3182,
+      "step": 7435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5028409090909093e-05,
+      "loss": 0.3741,
+      "step": 7436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5023674242424243e-05,
+      "loss": 0.3836,
+      "step": 7437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5018939393939394e-05,
+      "loss": 0.511,
+      "step": 7438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5014204545454546e-05,
+      "loss": 0.5766,
+      "step": 7439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5009469696969696e-05,
+      "loss": 0.3958,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.500473484848485e-05,
+      "loss": 0.3301,
+      "step": 7441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5e-05,
+      "loss": 0.4876,
+      "step": 7442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4995265151515152e-05,
+      "loss": 0.4372,
+      "step": 7443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4990530303030303e-05,
+      "loss": 0.3844,
+      "step": 7444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4985795454545457e-05,
+      "loss": 0.3923,
+      "step": 7445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4981060606060607e-05,
+      "loss": 0.3859,
+      "step": 7446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4976325757575757e-05,
+      "loss": 0.4272,
+      "step": 7447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.497159090909091e-05,
+      "loss": 0.3549,
+      "step": 7448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.496685606060606e-05,
+      "loss": 0.3298,
+      "step": 7449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4962121212121214e-05,
+      "loss": 0.4119,
+      "step": 7450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4957386363636364e-05,
+      "loss": 0.3454,
+      "step": 7451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4952651515151516e-05,
+      "loss": 0.389,
+      "step": 7452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4947916666666666e-05,
+      "loss": 0.474,
+      "step": 7453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.494318181818182e-05,
+      "loss": 0.374,
+      "step": 7454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.493844696969697e-05,
+      "loss": 0.2977,
+      "step": 7455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4933712121212121e-05,
+      "loss": 0.3011,
+      "step": 7456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4928977272727273e-05,
+      "loss": 0.3775,
+      "step": 7457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4924242424242423e-05,
+      "loss": 0.3673,
+      "step": 7458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4919507575757577e-05,
+      "loss": 0.3937,
+      "step": 7459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4914772727272727e-05,
+      "loss": 0.4344,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.491003787878788e-05,
+      "loss": 0.4226,
+      "step": 7461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.490530303030303e-05,
+      "loss": 0.4393,
+      "step": 7462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4900568181818184e-05,
+      "loss": 0.387,
+      "step": 7463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4895833333333334e-05,
+      "loss": 0.3878,
+      "step": 7464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4891098484848486e-05,
+      "loss": 0.4359,
+      "step": 7465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4886363636363636e-05,
+      "loss": 0.3999,
+      "step": 7466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4881628787878787e-05,
+      "loss": 0.359,
+      "step": 7467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.487689393939394e-05,
+      "loss": 0.5608,
+      "step": 7468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4872159090909091e-05,
+      "loss": 0.3034,
+      "step": 7469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4867424242424243e-05,
+      "loss": 0.4113,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4862689393939393e-05,
+      "loss": 0.3806,
+      "step": 7471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4857954545454547e-05,
+      "loss": 0.425,
+      "step": 7472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4853219696969698e-05,
+      "loss": 0.4099,
+      "step": 7473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.484848484848485e-05,
+      "loss": 0.5047,
+      "step": 7474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.484375e-05,
+      "loss": 0.4495,
+      "step": 7475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.483901515151515e-05,
+      "loss": 0.4482,
+      "step": 7476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4834280303030304e-05,
+      "loss": 0.3264,
+      "step": 7477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4829545454545454e-05,
+      "loss": 0.2867,
+      "step": 7478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4824810606060607e-05,
+      "loss": 0.3348,
+      "step": 7479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4820075757575757e-05,
+      "loss": 0.4763,
+      "step": 7480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.481534090909091e-05,
+      "loss": 0.3574,
+      "step": 7481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4810606060606061e-05,
+      "loss": 0.5299,
+      "step": 7482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4805871212121215e-05,
+      "loss": 0.4854,
+      "step": 7483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4801136363636363e-05,
+      "loss": 0.3945,
+      "step": 7484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4796401515151514e-05,
+      "loss": 0.4133,
+      "step": 7485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4791666666666668e-05,
+      "loss": 0.3401,
+      "step": 7486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4786931818181818e-05,
+      "loss": 0.428,
+      "step": 7487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.478219696969697e-05,
+      "loss": 0.5278,
+      "step": 7488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.477746212121212e-05,
+      "loss": 0.3391,
+      "step": 7489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4772727272727274e-05,
+      "loss": 0.3402,
+      "step": 7490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4767992424242425e-05,
+      "loss": 0.4994,
+      "step": 7491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4763257575757578e-05,
+      "loss": 0.4416,
+      "step": 7492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4758522727272727e-05,
+      "loss": 0.3675,
+      "step": 7493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.475378787878788e-05,
+      "loss": 0.3816,
+      "step": 7494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4749053030303031e-05,
+      "loss": 0.3629,
+      "step": 7495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4744318181818182e-05,
+      "loss": 0.4614,
+      "step": 7496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4739583333333334e-05,
+      "loss": 0.4009,
+      "step": 7497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4734848484848484e-05,
+      "loss": 0.4538,
+      "step": 7498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4730113636363638e-05,
+      "loss": 0.5269,
+      "step": 7499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4725378787878788e-05,
+      "loss": 0.4078,
+      "step": 7500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4720643939393942e-05,
+      "loss": 0.312,
+      "step": 7501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.471590909090909e-05,
+      "loss": 0.4035,
+      "step": 7502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4711174242424244e-05,
+      "loss": 0.3987,
+      "step": 7503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4706439393939395e-05,
+      "loss": 0.3313,
+      "step": 7504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4701704545454545e-05,
+      "loss": 0.3409,
+      "step": 7505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4696969696969697e-05,
+      "loss": 0.5148,
+      "step": 7506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4692234848484847e-05,
+      "loss": 0.396,
+      "step": 7507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4687500000000001e-05,
+      "loss": 0.4826,
+      "step": 7508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4682765151515152e-05,
+      "loss": 0.4846,
+      "step": 7509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4678030303030305e-05,
+      "loss": 0.5153,
+      "step": 7510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4673295454545454e-05,
+      "loss": 0.4604,
+      "step": 7511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4668560606060608e-05,
+      "loss": 0.3964,
+      "step": 7512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4663825757575758e-05,
+      "loss": 0.5958,
+      "step": 7513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4659090909090909e-05,
+      "loss": 0.382,
+      "step": 7514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.465435606060606e-05,
+      "loss": 0.4948,
+      "step": 7515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4649621212121211e-05,
+      "loss": 0.3394,
+      "step": 7516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4644886363636365e-05,
+      "loss": 0.4652,
+      "step": 7517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4640151515151515e-05,
+      "loss": 0.4294,
+      "step": 7518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4635416666666669e-05,
+      "loss": 0.4695,
+      "step": 7519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4630681818181818e-05,
+      "loss": 0.4186,
+      "step": 7520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4625946969696971e-05,
+      "loss": 0.4925,
+      "step": 7521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4621212121212122e-05,
+      "loss": 0.3949,
+      "step": 7522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4616477272727275e-05,
+      "loss": 0.4059,
+      "step": 7523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4611742424242424e-05,
+      "loss": 0.384,
+      "step": 7524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4607007575757575e-05,
+      "loss": 0.3876,
+      "step": 7525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4602272727272728e-05,
+      "loss": 0.4145,
+      "step": 7526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4597537878787879e-05,
+      "loss": 0.4211,
+      "step": 7527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4592803030303032e-05,
+      "loss": 0.422,
+      "step": 7528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4588068181818181e-05,
+      "loss": 0.3664,
+      "step": 7529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4583333333333335e-05,
+      "loss": 0.5033,
+      "step": 7530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4578598484848485e-05,
+      "loss": 0.5332,
+      "step": 7531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4573863636363639e-05,
+      "loss": 0.3757,
+      "step": 7532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.456912878787879e-05,
+      "loss": 0.4043,
+      "step": 7533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4564393939393938e-05,
+      "loss": 0.4121,
+      "step": 7534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4559659090909092e-05,
+      "loss": 0.3291,
+      "step": 7535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4554924242424242e-05,
+      "loss": 0.3797,
+      "step": 7536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4550189393939396e-05,
+      "loss": 0.3146,
+      "step": 7537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4545454545454545e-05,
+      "loss": 0.4253,
+      "step": 7538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4540719696969698e-05,
+      "loss": 0.4319,
+      "step": 7539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4535984848484849e-05,
+      "loss": 0.3967,
+      "step": 7540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4531250000000003e-05,
+      "loss": 0.3145,
+      "step": 7541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4526515151515153e-05,
+      "loss": 0.3395,
+      "step": 7542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4521780303030302e-05,
+      "loss": 0.4088,
+      "step": 7543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4517045454545455e-05,
+      "loss": 0.418,
+      "step": 7544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4512310606060606e-05,
+      "loss": 0.5496,
+      "step": 7545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.450757575757576e-05,
+      "loss": 0.3956,
+      "step": 7546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4502840909090908e-05,
+      "loss": 0.4145,
+      "step": 7547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4498106060606062e-05,
+      "loss": 0.4029,
+      "step": 7548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4493371212121212e-05,
+      "loss": 0.283,
+      "step": 7549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4488636363636366e-05,
+      "loss": 0.5028,
+      "step": 7550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4483901515151516e-05,
+      "loss": 0.412,
+      "step": 7551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4479166666666669e-05,
+      "loss": 0.4304,
+      "step": 7552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4474431818181819e-05,
+      "loss": 0.3561,
+      "step": 7553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.446969696969697e-05,
+      "loss": 0.3345,
+      "step": 7554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4464962121212123e-05,
+      "loss": 0.3895,
+      "step": 7555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4460227272727272e-05,
+      "loss": 0.4861,
+      "step": 7556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4455492424242425e-05,
+      "loss": 0.3125,
+      "step": 7557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4450757575757576e-05,
+      "loss": 0.4305,
+      "step": 7558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.444602272727273e-05,
+      "loss": 0.3016,
+      "step": 7559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.444128787878788e-05,
+      "loss": 0.4503,
+      "step": 7560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4436553030303032e-05,
+      "loss": 0.4636,
+      "step": 7561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4431818181818182e-05,
+      "loss": 0.3958,
+      "step": 7562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4427083333333333e-05,
+      "loss": 0.4656,
+      "step": 7563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4422348484848487e-05,
+      "loss": 0.312,
+      "step": 7564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4417613636363635e-05,
+      "loss": 0.3274,
+      "step": 7565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4412878787878789e-05,
+      "loss": 0.4315,
+      "step": 7566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.440814393939394e-05,
+      "loss": 0.4288,
+      "step": 7567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4403409090909093e-05,
+      "loss": 0.4342,
+      "step": 7568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4398674242424243e-05,
+      "loss": 0.3631,
+      "step": 7569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4393939393939396e-05,
+      "loss": 0.3741,
+      "step": 7570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4389204545454546e-05,
+      "loss": 0.4946,
+      "step": 7571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.43844696969697e-05,
+      "loss": 0.4472,
+      "step": 7572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.437973484848485e-05,
+      "loss": 0.4024,
+      "step": 7573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4374999999999999e-05,
+      "loss": 0.4281,
+      "step": 7574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4370265151515153e-05,
+      "loss": 0.4498,
+      "step": 7575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4365530303030303e-05,
+      "loss": 0.3534,
+      "step": 7576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4360795454545457e-05,
+      "loss": 0.5072,
+      "step": 7577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4356060606060607e-05,
+      "loss": 0.4119,
+      "step": 7578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4351325757575759e-05,
+      "loss": 0.233,
+      "step": 7579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.434659090909091e-05,
+      "loss": 0.2783,
+      "step": 7580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4341856060606063e-05,
+      "loss": 0.3815,
+      "step": 7581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4337121212121214e-05,
+      "loss": 0.3216,
+      "step": 7582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4332386363636362e-05,
+      "loss": 0.3184,
+      "step": 7583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4327651515151516e-05,
+      "loss": 0.5279,
+      "step": 7584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4322916666666666e-05,
+      "loss": 0.3028,
+      "step": 7585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.431818181818182e-05,
+      "loss": 0.5234,
+      "step": 7586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.431344696969697e-05,
+      "loss": 0.4376,
+      "step": 7587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4308712121212123e-05,
+      "loss": 0.3469,
+      "step": 7588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4303977272727273e-05,
+      "loss": 0.4108,
+      "step": 7589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4299242424242427e-05,
+      "loss": 0.3375,
+      "step": 7590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4294507575757577e-05,
+      "loss": 0.3993,
+      "step": 7591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4289772727272727e-05,
+      "loss": 0.4565,
+      "step": 7592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.428503787878788e-05,
+      "loss": 0.5161,
+      "step": 7593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.428030303030303e-05,
+      "loss": 0.4001,
+      "step": 7594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4275568181818184e-05,
+      "loss": 0.5425,
+      "step": 7595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4270833333333334e-05,
+      "loss": 0.4235,
+      "step": 7596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4266098484848486e-05,
+      "loss": 0.399,
+      "step": 7597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4261363636363637e-05,
+      "loss": 0.3973,
+      "step": 7598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.425662878787879e-05,
+      "loss": 0.3564,
+      "step": 7599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.425189393939394e-05,
+      "loss": 0.4725,
+      "step": 7600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4247159090909093e-05,
+      "loss": 0.4852,
+      "step": 7601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4242424242424243e-05,
+      "loss": 0.5367,
+      "step": 7602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4237689393939393e-05,
+      "loss": 0.3793,
+      "step": 7603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4232954545454547e-05,
+      "loss": 0.3447,
+      "step": 7604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4228219696969698e-05,
+      "loss": 0.4569,
+      "step": 7605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.422348484848485e-05,
+      "loss": 0.4602,
+      "step": 7606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.421875e-05,
+      "loss": 0.4204,
+      "step": 7607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4214015151515154e-05,
+      "loss": 0.3931,
+      "step": 7608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4209280303030304e-05,
+      "loss": 0.3968,
+      "step": 7609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4204545454545456e-05,
+      "loss": 0.4292,
+      "step": 7610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4199810606060607e-05,
+      "loss": 0.342,
+      "step": 7611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4195075757575757e-05,
+      "loss": 0.5643,
+      "step": 7612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.419034090909091e-05,
+      "loss": 0.3836,
+      "step": 7613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4185606060606061e-05,
+      "loss": 0.3735,
+      "step": 7614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4180871212121213e-05,
+      "loss": 0.3792,
+      "step": 7615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4176136363636364e-05,
+      "loss": 0.357,
+      "step": 7616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4171401515151517e-05,
+      "loss": 0.3758,
+      "step": 7617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4166666666666668e-05,
+      "loss": 0.4194,
+      "step": 7618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.416193181818182e-05,
+      "loss": 0.4343,
+      "step": 7619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.415719696969697e-05,
+      "loss": 0.4303,
+      "step": 7620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.415246212121212e-05,
+      "loss": 0.4631,
+      "step": 7621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4147727272727274e-05,
+      "loss": 0.4398,
+      "step": 7622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4142992424242425e-05,
+      "loss": 0.3965,
+      "step": 7623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4138257575757577e-05,
+      "loss": 0.4315,
+      "step": 7624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4133522727272727e-05,
+      "loss": 0.4166,
+      "step": 7625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.412878787878788e-05,
+      "loss": 0.4122,
+      "step": 7626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4124053030303031e-05,
+      "loss": 0.4045,
+      "step": 7627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4119318181818183e-05,
+      "loss": 0.4831,
+      "step": 7628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4114583333333334e-05,
+      "loss": 0.3489,
+      "step": 7629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4109848484848487e-05,
+      "loss": 0.5419,
+      "step": 7630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4105113636363638e-05,
+      "loss": 0.4512,
+      "step": 7631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4100378787878788e-05,
+      "loss": 0.3583,
+      "step": 7632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.409564393939394e-05,
+      "loss": 0.2947,
+      "step": 7633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.409090909090909e-05,
+      "loss": 0.4513,
+      "step": 7634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4086174242424244e-05,
+      "loss": 0.4024,
+      "step": 7635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4081439393939395e-05,
+      "loss": 0.5686,
+      "step": 7636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4076704545454547e-05,
+      "loss": 0.2451,
+      "step": 7637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4071969696969697e-05,
+      "loss": 0.4035,
+      "step": 7638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4067234848484851e-05,
+      "loss": 0.3901,
+      "step": 7639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4062500000000001e-05,
+      "loss": 0.4542,
+      "step": 7640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4057765151515152e-05,
+      "loss": 0.6454,
+      "step": 7641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4053030303030304e-05,
+      "loss": 0.4081,
+      "step": 7642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4048295454545454e-05,
+      "loss": 0.4127,
+      "step": 7643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4043560606060608e-05,
+      "loss": 0.417,
+      "step": 7644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4038825757575758e-05,
+      "loss": 0.3351,
+      "step": 7645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.403409090909091e-05,
+      "loss": 0.4361,
+      "step": 7646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.402935606060606e-05,
+      "loss": 0.3649,
+      "step": 7647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4024621212121214e-05,
+      "loss": 0.5617,
+      "step": 7648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4019886363636365e-05,
+      "loss": 0.5784,
+      "step": 7649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4015151515151515e-05,
+      "loss": 0.4137,
+      "step": 7650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4010416666666667e-05,
+      "loss": 0.4414,
+      "step": 7651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4005681818181818e-05,
+      "loss": 0.2424,
+      "step": 7652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4000946969696971e-05,
+      "loss": 0.3828,
+      "step": 7653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3996212121212122e-05,
+      "loss": 0.4058,
+      "step": 7654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3991477272727274e-05,
+      "loss": 0.4901,
+      "step": 7655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3986742424242424e-05,
+      "loss": 0.3772,
+      "step": 7656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3982007575757578e-05,
+      "loss": 0.5572,
+      "step": 7657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3977272727272728e-05,
+      "loss": 0.3869,
+      "step": 7658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.397253787878788e-05,
+      "loss": 0.5311,
+      "step": 7659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.396780303030303e-05,
+      "loss": 0.3943,
+      "step": 7660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3963068181818181e-05,
+      "loss": 0.4664,
+      "step": 7661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3958333333333335e-05,
+      "loss": 0.4405,
+      "step": 7662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3953598484848485e-05,
+      "loss": 0.4029,
+      "step": 7663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3948863636363637e-05,
+      "loss": 0.3627,
+      "step": 7664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3944128787878788e-05,
+      "loss": 0.458,
+      "step": 7665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3939393939393942e-05,
+      "loss": 0.4251,
+      "step": 7666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3934659090909092e-05,
+      "loss": 0.4832,
+      "step": 7667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3929924242424244e-05,
+      "loss": 0.4014,
+      "step": 7668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3925189393939394e-05,
+      "loss": 0.3656,
+      "step": 7669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3920454545454545e-05,
+      "loss": 0.4657,
+      "step": 7670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3915719696969698e-05,
+      "loss": 0.4543,
+      "step": 7671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3910984848484849e-05,
+      "loss": 0.4111,
+      "step": 7672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3906250000000001e-05,
+      "loss": 0.3727,
+      "step": 7673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3901515151515151e-05,
+      "loss": 0.4103,
+      "step": 7674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3896780303030305e-05,
+      "loss": 0.3004,
+      "step": 7675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3892045454545455e-05,
+      "loss": 0.4147,
+      "step": 7676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3887310606060607e-05,
+      "loss": 0.4266,
+      "step": 7677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3882575757575758e-05,
+      "loss": 0.3732,
+      "step": 7678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3877840909090908e-05,
+      "loss": 0.3178,
+      "step": 7679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3873106060606062e-05,
+      "loss": 0.3844,
+      "step": 7680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3868371212121212e-05,
+      "loss": 0.4404,
+      "step": 7681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3863636363636364e-05,
+      "loss": 0.3672,
+      "step": 7682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3858901515151515e-05,
+      "loss": 0.4989,
+      "step": 7683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3854166666666669e-05,
+      "loss": 0.4874,
+      "step": 7684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3849431818181819e-05,
+      "loss": 0.4523,
+      "step": 7685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3844696969696971e-05,
+      "loss": 0.3631,
+      "step": 7686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3839962121212121e-05,
+      "loss": 0.3215,
+      "step": 7687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3835227272727275e-05,
+      "loss": 0.3121,
+      "step": 7688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3830492424242426e-05,
+      "loss": 0.3685,
+      "step": 7689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3825757575757576e-05,
+      "loss": 0.3346,
+      "step": 7690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3821022727272728e-05,
+      "loss": 0.4437,
+      "step": 7691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3816287878787878e-05,
+      "loss": 0.3485,
+      "step": 7692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3811553030303032e-05,
+      "loss": 0.3725,
+      "step": 7693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3806818181818182e-05,
+      "loss": 0.4154,
+      "step": 7694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3802083333333335e-05,
+      "loss": 0.4312,
+      "step": 7695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3797348484848485e-05,
+      "loss": 0.276,
+      "step": 7696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3792613636363639e-05,
+      "loss": 0.5047,
+      "step": 7697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3787878787878789e-05,
+      "loss": 0.4135,
+      "step": 7698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.378314393939394e-05,
+      "loss": 0.3131,
+      "step": 7699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3778409090909091e-05,
+      "loss": 0.4729,
+      "step": 7700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3773674242424242e-05,
+      "loss": 0.3338,
+      "step": 7701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3768939393939396e-05,
+      "loss": 0.3868,
+      "step": 7702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3764204545454546e-05,
+      "loss": 0.4174,
+      "step": 7703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3759469696969698e-05,
+      "loss": 0.3313,
+      "step": 7704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3754734848484848e-05,
+      "loss": 0.3415,
+      "step": 7705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3750000000000002e-05,
+      "loss": 0.3333,
+      "step": 7706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3745265151515153e-05,
+      "loss": 0.429,
+      "step": 7707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3740530303030303e-05,
+      "loss": 0.4061,
+      "step": 7708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3735795454545455e-05,
+      "loss": 0.4423,
+      "step": 7709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3731060606060605e-05,
+      "loss": 0.3318,
+      "step": 7710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3726325757575759e-05,
+      "loss": 0.4384,
+      "step": 7711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.372159090909091e-05,
+      "loss": 0.4066,
+      "step": 7712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3716856060606062e-05,
+      "loss": 0.4692,
+      "step": 7713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3712121212121212e-05,
+      "loss": 0.3487,
+      "step": 7714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3707386363636366e-05,
+      "loss": 0.3281,
+      "step": 7715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3702651515151516e-05,
+      "loss": 0.3394,
+      "step": 7716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3697916666666668e-05,
+      "loss": 0.4222,
+      "step": 7717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3693181818181819e-05,
+      "loss": 0.461,
+      "step": 7718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3688446969696969e-05,
+      "loss": 0.4053,
+      "step": 7719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3683712121212123e-05,
+      "loss": 0.477,
+      "step": 7720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3678977272727273e-05,
+      "loss": 0.3667,
+      "step": 7721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3674242424242425e-05,
+      "loss": 0.2912,
+      "step": 7722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3669507575757575e-05,
+      "loss": 0.4664,
+      "step": 7723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.366477272727273e-05,
+      "loss": 0.4008,
+      "step": 7724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.366003787878788e-05,
+      "loss": 0.4,
+      "step": 7725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3655303030303032e-05,
+      "loss": 0.4204,
+      "step": 7726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3650568181818182e-05,
+      "loss": 0.3943,
+      "step": 7727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3645833333333332e-05,
+      "loss": 0.3987,
+      "step": 7728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3641098484848486e-05,
+      "loss": 0.4576,
+      "step": 7729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.3758,
+      "step": 7730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3631628787878789e-05,
+      "loss": 0.4012,
+      "step": 7731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3626893939393939e-05,
+      "loss": 0.3474,
+      "step": 7732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3622159090909093e-05,
+      "loss": 0.3409,
+      "step": 7733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3617424242424243e-05,
+      "loss": 0.4204,
+      "step": 7734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3612689393939395e-05,
+      "loss": 0.3136,
+      "step": 7735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3607954545454546e-05,
+      "loss": 0.381,
+      "step": 7736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.36032196969697e-05,
+      "loss": 0.4418,
+      "step": 7737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.359848484848485e-05,
+      "loss": 0.4884,
+      "step": 7738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.359375e-05,
+      "loss": 0.4309,
+      "step": 7739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3589015151515152e-05,
+      "loss": 0.3224,
+      "step": 7740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3584280303030303e-05,
+      "loss": 0.5144,
+      "step": 7741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3579545454545456e-05,
+      "loss": 0.4451,
+      "step": 7742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3574810606060607e-05,
+      "loss": 0.3381,
+      "step": 7743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3570075757575759e-05,
+      "loss": 0.4356,
+      "step": 7744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3565340909090909e-05,
+      "loss": 0.468,
+      "step": 7745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3560606060606063e-05,
+      "loss": 0.2973,
+      "step": 7746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3555871212121213e-05,
+      "loss": 0.3968,
+      "step": 7747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3551136363636364e-05,
+      "loss": 0.4714,
+      "step": 7748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3546401515151516e-05,
+      "loss": 0.4422,
+      "step": 7749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3541666666666666e-05,
+      "loss": 0.4172,
+      "step": 7750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.353693181818182e-05,
+      "loss": 0.3775,
+      "step": 7751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.353219696969697e-05,
+      "loss": 0.4011,
+      "step": 7752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3527462121212122e-05,
+      "loss": 0.3507,
+      "step": 7753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3522727272727273e-05,
+      "loss": 0.3915,
+      "step": 7754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3517992424242426e-05,
+      "loss": 0.3321,
+      "step": 7755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3513257575757577e-05,
+      "loss": 0.3085,
+      "step": 7756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3508522727272727e-05,
+      "loss": 0.3463,
+      "step": 7757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.350378787878788e-05,
+      "loss": 0.366,
+      "step": 7758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.349905303030303e-05,
+      "loss": 0.3744,
+      "step": 7759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3494318181818183e-05,
+      "loss": 0.4458,
+      "step": 7760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3489583333333334e-05,
+      "loss": 0.6526,
+      "step": 7761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3484848484848486e-05,
+      "loss": 0.3001,
+      "step": 7762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3480113636363636e-05,
+      "loss": 0.5694,
+      "step": 7763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.347537878787879e-05,
+      "loss": 0.5496,
+      "step": 7764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.347064393939394e-05,
+      "loss": 0.4264,
+      "step": 7765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3465909090909092e-05,
+      "loss": 0.4408,
+      "step": 7766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3461174242424243e-05,
+      "loss": 0.3716,
+      "step": 7767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3456439393939393e-05,
+      "loss": 0.4234,
+      "step": 7768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3451704545454547e-05,
+      "loss": 0.4987,
+      "step": 7769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3446969696969697e-05,
+      "loss": 0.3288,
+      "step": 7770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.344223484848485e-05,
+      "loss": 0.4313,
+      "step": 7771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.34375e-05,
+      "loss": 0.3399,
+      "step": 7772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3432765151515153e-05,
+      "loss": 0.5206,
+      "step": 7773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3428030303030304e-05,
+      "loss": 0.3611,
+      "step": 7774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3423295454545456e-05,
+      "loss": 0.4071,
+      "step": 7775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3418560606060606e-05,
+      "loss": 0.4682,
+      "step": 7776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3413825757575757e-05,
+      "loss": 0.4348,
+      "step": 7777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.340909090909091e-05,
+      "loss": 0.4032,
+      "step": 7778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.340435606060606e-05,
+      "loss": 0.3201,
+      "step": 7779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3399621212121213e-05,
+      "loss": 0.4417,
+      "step": 7780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3394886363636363e-05,
+      "loss": 0.3824,
+      "step": 7781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3390151515151517e-05,
+      "loss": 0.3763,
+      "step": 7782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3385416666666667e-05,
+      "loss": 0.3093,
+      "step": 7783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.338068181818182e-05,
+      "loss": 0.3267,
+      "step": 7784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.337594696969697e-05,
+      "loss": 0.4148,
+      "step": 7785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.337121212121212e-05,
+      "loss": 0.3214,
+      "step": 7786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3366477272727274e-05,
+      "loss": 0.4202,
+      "step": 7787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3361742424242424e-05,
+      "loss": 0.4103,
+      "step": 7788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3357007575757576e-05,
+      "loss": 0.4049,
+      "step": 7789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3352272727272727e-05,
+      "loss": 0.4322,
+      "step": 7790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.334753787878788e-05,
+      "loss": 0.3759,
+      "step": 7791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3342803030303031e-05,
+      "loss": 0.293,
+      "step": 7792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3338068181818183e-05,
+      "loss": 0.4164,
+      "step": 7793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.5312,
+      "step": 7794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3328598484848487e-05,
+      "loss": 0.3397,
+      "step": 7795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3323863636363637e-05,
+      "loss": 0.3675,
+      "step": 7796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3319128787878788e-05,
+      "loss": 0.4635,
+      "step": 7797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.331439393939394e-05,
+      "loss": 0.4099,
+      "step": 7798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.330965909090909e-05,
+      "loss": 0.5232,
+      "step": 7799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3304924242424244e-05,
+      "loss": 0.3618,
+      "step": 7800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3300189393939394e-05,
+      "loss": 0.4533,
+      "step": 7801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3295454545454546e-05,
+      "loss": 0.3654,
+      "step": 7802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3290719696969697e-05,
+      "loss": 0.3735,
+      "step": 7803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.328598484848485e-05,
+      "loss": 0.2973,
+      "step": 7804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3281250000000001e-05,
+      "loss": 0.3368,
+      "step": 7805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3276515151515151e-05,
+      "loss": 0.4243,
+      "step": 7806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3271780303030303e-05,
+      "loss": 0.4142,
+      "step": 7807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3267045454545454e-05,
+      "loss": 0.3097,
+      "step": 7808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3262310606060608e-05,
+      "loss": 0.3895,
+      "step": 7809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3257575757575758e-05,
+      "loss": 0.4515,
+      "step": 7810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.325284090909091e-05,
+      "loss": 0.3723,
+      "step": 7811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.324810606060606e-05,
+      "loss": 0.4604,
+      "step": 7812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3243371212121214e-05,
+      "loss": 0.3134,
+      "step": 7813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3238636363636364e-05,
+      "loss": 0.5601,
+      "step": 7814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3233901515151515e-05,
+      "loss": 0.4147,
+      "step": 7815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3229166666666667e-05,
+      "loss": 0.4368,
+      "step": 7816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3224431818181817e-05,
+      "loss": 0.4575,
+      "step": 7817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3219696969696971e-05,
+      "loss": 0.397,
+      "step": 7818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3214962121212121e-05,
+      "loss": 0.3489,
+      "step": 7819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3210227272727273e-05,
+      "loss": 0.4637,
+      "step": 7820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3205492424242424e-05,
+      "loss": 0.322,
+      "step": 7821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3200757575757578e-05,
+      "loss": 0.3982,
+      "step": 7822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3196022727272728e-05,
+      "loss": 0.3634,
+      "step": 7823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.319128787878788e-05,
+      "loss": 0.4412,
+      "step": 7824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.318655303030303e-05,
+      "loss": 0.286,
+      "step": 7825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.318181818181818e-05,
+      "loss": 0.4438,
+      "step": 7826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3177083333333335e-05,
+      "loss": 0.4568,
+      "step": 7827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3172348484848485e-05,
+      "loss": 0.3824,
+      "step": 7828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3167613636363637e-05,
+      "loss": 0.2679,
+      "step": 7829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3162878787878787e-05,
+      "loss": 0.3538,
+      "step": 7830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3158143939393941e-05,
+      "loss": 0.2599,
+      "step": 7831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3153409090909092e-05,
+      "loss": 0.4393,
+      "step": 7832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3148674242424244e-05,
+      "loss": 0.4059,
+      "step": 7833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3143939393939394e-05,
+      "loss": 0.4202,
+      "step": 7834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3139204545454544e-05,
+      "loss": 0.3041,
+      "step": 7835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3134469696969698e-05,
+      "loss": 0.4576,
+      "step": 7836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3129734848484848e-05,
+      "loss": 0.4753,
+      "step": 7837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3125e-05,
+      "loss": 0.5038,
+      "step": 7838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3120265151515151e-05,
+      "loss": 0.4335,
+      "step": 7839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3115530303030305e-05,
+      "loss": 0.3493,
+      "step": 7840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3110795454545455e-05,
+      "loss": 0.4852,
+      "step": 7841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3106060606060607e-05,
+      "loss": 0.3239,
+      "step": 7842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3101325757575757e-05,
+      "loss": 0.266,
+      "step": 7843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3096590909090908e-05,
+      "loss": 0.4846,
+      "step": 7844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3091856060606062e-05,
+      "loss": 0.3437,
+      "step": 7845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3087121212121212e-05,
+      "loss": 0.4562,
+      "step": 7846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3082386363636364e-05,
+      "loss": 0.4296,
+      "step": 7847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3077651515151514e-05,
+      "loss": 0.3729,
+      "step": 7848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3072916666666668e-05,
+      "loss": 0.4337,
+      "step": 7849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3068181818181819e-05,
+      "loss": 0.5775,
+      "step": 7850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.306344696969697e-05,
+      "loss": 0.4557,
+      "step": 7851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3058712121212121e-05,
+      "loss": 0.4583,
+      "step": 7852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3053977272727275e-05,
+      "loss": 0.4164,
+      "step": 7853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3049242424242425e-05,
+      "loss": 0.4534,
+      "step": 7854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3044507575757576e-05,
+      "loss": 0.3694,
+      "step": 7855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3039772727272728e-05,
+      "loss": 0.3541,
+      "step": 7856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3035037878787878e-05,
+      "loss": 0.3636,
+      "step": 7857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3030303030303032e-05,
+      "loss": 0.357,
+      "step": 7858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3025568181818182e-05,
+      "loss": 0.3274,
+      "step": 7859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3020833333333334e-05,
+      "loss": 0.4996,
+      "step": 7860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3016098484848485e-05,
+      "loss": 0.3613,
+      "step": 7861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3011363636363638e-05,
+      "loss": 0.4206,
+      "step": 7862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3006628787878789e-05,
+      "loss": 0.3065,
+      "step": 7863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3001893939393939e-05,
+      "loss": 0.4151,
+      "step": 7864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2997159090909091e-05,
+      "loss": 0.3667,
+      "step": 7865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2992424242424241e-05,
+      "loss": 0.4034,
+      "step": 7866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2987689393939395e-05,
+      "loss": 0.2972,
+      "step": 7867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2982954545454546e-05,
+      "loss": 0.3611,
+      "step": 7868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2978219696969698e-05,
+      "loss": 0.4118,
+      "step": 7869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2973484848484848e-05,
+      "loss": 0.3652,
+      "step": 7870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2968750000000002e-05,
+      "loss": 0.5373,
+      "step": 7871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2964015151515152e-05,
+      "loss": 0.341,
+      "step": 7872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2959280303030303e-05,
+      "loss": 0.2966,
+      "step": 7873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2954545454545455e-05,
+      "loss": 0.4951,
+      "step": 7874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2949810606060605e-05,
+      "loss": 0.3797,
+      "step": 7875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2945075757575759e-05,
+      "loss": 0.426,
+      "step": 7876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2940340909090909e-05,
+      "loss": 0.2312,
+      "step": 7877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2935606060606061e-05,
+      "loss": 0.4946,
+      "step": 7878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2930871212121212e-05,
+      "loss": 0.3919,
+      "step": 7879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2926136363636365e-05,
+      "loss": 0.3821,
+      "step": 7880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2921401515151516e-05,
+      "loss": 0.306,
+      "step": 7881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2916666666666668e-05,
+      "loss": 0.3351,
+      "step": 7882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2911931818181818e-05,
+      "loss": 0.4654,
+      "step": 7883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2907196969696969e-05,
+      "loss": 0.37,
+      "step": 7884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2902462121212122e-05,
+      "loss": 0.3529,
+      "step": 7885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2897727272727273e-05,
+      "loss": 0.4229,
+      "step": 7886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2892992424242425e-05,
+      "loss": 0.2804,
+      "step": 7887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2888257575757575e-05,
+      "loss": 0.4946,
+      "step": 7888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2883522727272729e-05,
+      "loss": 0.3034,
+      "step": 7889
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.287878787878788e-05,
+      "loss": 0.553,
+      "step": 7890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2874053030303031e-05,
+      "loss": 0.4652,
+      "step": 7891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2869318181818182e-05,
+      "loss": 0.3929,
+      "step": 7892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2864583333333332e-05,
+      "loss": 0.4976,
+      "step": 7893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2859848484848486e-05,
+      "loss": 0.3746,
+      "step": 7894
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2855113636363636e-05,
+      "loss": 0.5355,
+      "step": 7895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2850378787878788e-05,
+      "loss": 0.4155,
+      "step": 7896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2845643939393939e-05,
+      "loss": 0.3986,
+      "step": 7897
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2840909090909092e-05,
+      "loss": 0.4936,
+      "step": 7898
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2836174242424243e-05,
+      "loss": 0.2881,
+      "step": 7899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2831439393939395e-05,
+      "loss": 0.4562,
+      "step": 7900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2826704545454545e-05,
+      "loss": 0.4453,
+      "step": 7901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2821969696969699e-05,
+      "loss": 0.4297,
+      "step": 7902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.281723484848485e-05,
+      "loss": 0.4754,
+      "step": 7903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.28125e-05,
+      "loss": 0.4552,
+      "step": 7904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2807765151515152e-05,
+      "loss": 0.3266,
+      "step": 7905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2803030303030302e-05,
+      "loss": 0.3507,
+      "step": 7906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2798295454545456e-05,
+      "loss": 0.417,
+      "step": 7907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2793560606060606e-05,
+      "loss": 0.3975,
+      "step": 7908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2788825757575758e-05,
+      "loss": 0.4604,
+      "step": 7909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2784090909090909e-05,
+      "loss": 0.3187,
+      "step": 7910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2779356060606063e-05,
+      "loss": 0.3969,
+      "step": 7911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2774621212121213e-05,
+      "loss": 0.41,
+      "step": 7912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2769886363636363e-05,
+      "loss": 0.3616,
+      "step": 7913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2765151515151515e-05,
+      "loss": 0.3064,
+      "step": 7914
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2760416666666666e-05,
+      "loss": 0.3885,
+      "step": 7915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.275568181818182e-05,
+      "loss": 0.3262,
+      "step": 7916
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.275094696969697e-05,
+      "loss": 0.4805,
+      "step": 7917
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2746212121212122e-05,
+      "loss": 0.3614,
+      "step": 7918
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2741477272727272e-05,
+      "loss": 0.418,
+      "step": 7919
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2736742424242426e-05,
+      "loss": 0.3657,
+      "step": 7920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2732007575757576e-05,
+      "loss": 0.2975,
+      "step": 7921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2727272727272727e-05,
+      "loss": 0.3319,
+      "step": 7922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2722537878787879e-05,
+      "loss": 0.258,
+      "step": 7923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.271780303030303e-05,
+      "loss": 0.4178,
+      "step": 7924
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2713068181818183e-05,
+      "loss": 0.4055,
+      "step": 7925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2708333333333333e-05,
+      "loss": 0.4275,
+      "step": 7926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2703598484848485e-05,
+      "loss": 0.3772,
+      "step": 7927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2698863636363636e-05,
+      "loss": 0.4016,
+      "step": 7928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.269412878787879e-05,
+      "loss": 0.5251,
+      "step": 7929
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.268939393939394e-05,
+      "loss": 0.555,
+      "step": 7930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2684659090909094e-05,
+      "loss": 0.638,
+      "step": 7931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2679924242424242e-05,
+      "loss": 0.2909,
+      "step": 7932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2675189393939393e-05,
+      "loss": 0.2733,
+      "step": 7933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2670454545454547e-05,
+      "loss": 0.4658,
+      "step": 7934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2665719696969697e-05,
+      "loss": 0.4599,
+      "step": 7935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2660984848484849e-05,
+      "loss": 0.3316,
+      "step": 7936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.265625e-05,
+      "loss": 0.2459,
+      "step": 7937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2651515151515153e-05,
+      "loss": 0.3705,
+      "step": 7938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2646780303030303e-05,
+      "loss": 0.3937,
+      "step": 7939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2642045454545457e-05,
+      "loss": 0.4042,
+      "step": 7940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2637310606060606e-05,
+      "loss": 0.5086,
+      "step": 7941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2632575757575756e-05,
+      "loss": 0.2892,
+      "step": 7942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.262784090909091e-05,
+      "loss": 0.3372,
+      "step": 7943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.262310606060606e-05,
+      "loss": 0.4523,
+      "step": 7944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2618371212121214e-05,
+      "loss": 0.3062,
+      "step": 7945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2613636363636363e-05,
+      "loss": 0.5131,
+      "step": 7946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2608901515151517e-05,
+      "loss": 0.4335,
+      "step": 7947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2604166666666667e-05,
+      "loss": 0.3493,
+      "step": 7948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.259943181818182e-05,
+      "loss": 0.4073,
+      "step": 7949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.259469696969697e-05,
+      "loss": 0.352,
+      "step": 7950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.258996212121212e-05,
+      "loss": 0.4783,
+      "step": 7951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2585227272727274e-05,
+      "loss": 0.3137,
+      "step": 7952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2580492424242424e-05,
+      "loss": 0.3479,
+      "step": 7953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2575757575757578e-05,
+      "loss": 0.324,
+      "step": 7954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2571022727272726e-05,
+      "loss": 0.4462,
+      "step": 7955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.256628787878788e-05,
+      "loss": 0.4772,
+      "step": 7956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.256155303030303e-05,
+      "loss": 0.2715,
+      "step": 7957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2556818181818184e-05,
+      "loss": 0.362,
+      "step": 7958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2552083333333333e-05,
+      "loss": 0.3707,
+      "step": 7959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2547348484848487e-05,
+      "loss": 0.5147,
+      "step": 7960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2542613636363637e-05,
+      "loss": 0.4426,
+      "step": 7961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2537878787878787e-05,
+      "loss": 0.4239,
+      "step": 7962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2533143939393941e-05,
+      "loss": 0.3363,
+      "step": 7963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.252840909090909e-05,
+      "loss": 0.5851,
+      "step": 7964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2523674242424244e-05,
+      "loss": 0.4514,
+      "step": 7965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2518939393939394e-05,
+      "loss": 0.3341,
+      "step": 7966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2514204545454548e-05,
+      "loss": 0.4248,
+      "step": 7967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2509469696969696e-05,
+      "loss": 0.3409,
+      "step": 7968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.250473484848485e-05,
+      "loss": 0.4948,
+      "step": 7969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.25e-05,
+      "loss": 0.4587,
+      "step": 7970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2495265151515153e-05,
+      "loss": 0.4817,
+      "step": 7971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2490530303030305e-05,
+      "loss": 0.3511,
+      "step": 7972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2485795454545455e-05,
+      "loss": 0.3471,
+      "step": 7973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2481060606060607e-05,
+      "loss": 0.4091,
+      "step": 7974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2476325757575758e-05,
+      "loss": 0.4182,
+      "step": 7975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.247159090909091e-05,
+      "loss": 0.3698,
+      "step": 7976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.246685606060606e-05,
+      "loss": 0.5291,
+      "step": 7977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2462121212121212e-05,
+      "loss": 0.4241,
+      "step": 7978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2457386363636364e-05,
+      "loss": 0.3768,
+      "step": 7979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2452651515151516e-05,
+      "loss": 0.3671,
+      "step": 7980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2447916666666668e-05,
+      "loss": 0.3345,
+      "step": 7981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2443181818181819e-05,
+      "loss": 0.4451,
+      "step": 7982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.243844696969697e-05,
+      "loss": 0.3536,
+      "step": 7983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2433712121212123e-05,
+      "loss": 0.2821,
+      "step": 7984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2428977272727273e-05,
+      "loss": 0.4351,
+      "step": 7985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2424242424242424e-05,
+      "loss": 0.3993,
+      "step": 7986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2419507575757576e-05,
+      "loss": 0.3094,
+      "step": 7987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2414772727272728e-05,
+      "loss": 0.3329,
+      "step": 7988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.241003787878788e-05,
+      "loss": 0.4198,
+      "step": 7989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2405303030303032e-05,
+      "loss": 0.4345,
+      "step": 7990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2400568181818182e-05,
+      "loss": 0.4043,
+      "step": 7991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2395833333333334e-05,
+      "loss": 0.2731,
+      "step": 7992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2391098484848486e-05,
+      "loss": 0.4245,
+      "step": 7993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2386363636363638e-05,
+      "loss": 0.4846,
+      "step": 7994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2381628787878787e-05,
+      "loss": 0.4281,
+      "step": 7995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2376893939393939e-05,
+      "loss": 0.3888,
+      "step": 7996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2372159090909091e-05,
+      "loss": 0.4856,
+      "step": 7997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2367424242424243e-05,
+      "loss": 0.3528,
+      "step": 7998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2362689393939395e-05,
+      "loss": 0.3426,
+      "step": 7999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2357954545454546e-05,
+      "loss": 0.4808,
+      "step": 8000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2353219696969698e-05,
+      "loss": 0.427,
+      "step": 8001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.234848484848485e-05,
+      "loss": 0.4454,
+      "step": 8002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2343750000000002e-05,
+      "loss": 0.4506,
+      "step": 8003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2339015151515152e-05,
+      "loss": 0.4103,
+      "step": 8004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2334280303030303e-05,
+      "loss": 0.4018,
+      "step": 8005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2329545454545455e-05,
+      "loss": 0.3858,
+      "step": 8006
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2324810606060607e-05,
+      "loss": 0.4365,
+      "step": 8007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2320075757575759e-05,
+      "loss": 0.3005,
+      "step": 8008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.231534090909091e-05,
+      "loss": 0.3981,
+      "step": 8009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2310606060606061e-05,
+      "loss": 0.386,
+      "step": 8010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2305871212121213e-05,
+      "loss": 0.321,
+      "step": 8011
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2301136363636365e-05,
+      "loss": 0.4381,
+      "step": 8012
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2296401515151516e-05,
+      "loss": 0.32,
+      "step": 8013
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2291666666666666e-05,
+      "loss": 0.5238,
+      "step": 8014
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2286931818181818e-05,
+      "loss": 0.3435,
+      "step": 8015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.228219696969697e-05,
+      "loss": 0.3262,
+      "step": 8016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2277462121212122e-05,
+      "loss": 0.4673,
+      "step": 8017
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2272727272727273e-05,
+      "loss": 0.4715,
+      "step": 8018
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2267992424242425e-05,
+      "loss": 0.3767,
+      "step": 8019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2263257575757577e-05,
+      "loss": 0.3953,
+      "step": 8020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2258522727272729e-05,
+      "loss": 0.349,
+      "step": 8021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.225378787878788e-05,
+      "loss": 0.3266,
+      "step": 8022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2249053030303031e-05,
+      "loss": 0.4277,
+      "step": 8023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2244318181818182e-05,
+      "loss": 0.3929,
+      "step": 8024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2239583333333334e-05,
+      "loss": 0.4362,
+      "step": 8025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2234848484848486e-05,
+      "loss": 0.512,
+      "step": 8026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2230113636363636e-05,
+      "loss": 0.3209,
+      "step": 8027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2225378787878788e-05,
+      "loss": 0.4923,
+      "step": 8028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.222064393939394e-05,
+      "loss": 0.4258,
+      "step": 8029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2215909090909092e-05,
+      "loss": 0.309,
+      "step": 8030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2211174242424243e-05,
+      "loss": 0.4429,
+      "step": 8031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2206439393939395e-05,
+      "loss": 0.3998,
+      "step": 8032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2201704545454547e-05,
+      "loss": 0.3907,
+      "step": 8033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2196969696969697e-05,
+      "loss": 0.3895,
+      "step": 8034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.219223484848485e-05,
+      "loss": 0.3526,
+      "step": 8035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.21875e-05,
+      "loss": 0.4352,
+      "step": 8036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2182765151515152e-05,
+      "loss": 0.4349,
+      "step": 8037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2178030303030304e-05,
+      "loss": 0.4817,
+      "step": 8038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2173295454545456e-05,
+      "loss": 0.4663,
+      "step": 8039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2168560606060606e-05,
+      "loss": 0.459,
+      "step": 8040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2163825757575758e-05,
+      "loss": 0.29,
+      "step": 8041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.215909090909091e-05,
+      "loss": 0.3287,
+      "step": 8042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2154356060606061e-05,
+      "loss": 0.3158,
+      "step": 8043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2149621212121213e-05,
+      "loss": 0.387,
+      "step": 8044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2144886363636363e-05,
+      "loss": 0.4836,
+      "step": 8045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2140151515151515e-05,
+      "loss": 0.4248,
+      "step": 8046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2135416666666667e-05,
+      "loss": 0.3388,
+      "step": 8047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.213068181818182e-05,
+      "loss": 0.4001,
+      "step": 8048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.212594696969697e-05,
+      "loss": 0.3692,
+      "step": 8049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.2655,
+      "step": 8050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2116477272727274e-05,
+      "loss": 0.374,
+      "step": 8051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2111742424242426e-05,
+      "loss": 0.4031,
+      "step": 8052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2107007575757576e-05,
+      "loss": 0.4332,
+      "step": 8053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2102272727272727e-05,
+      "loss": 0.4946,
+      "step": 8054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2097537878787879e-05,
+      "loss": 0.4058,
+      "step": 8055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2092803030303031e-05,
+      "loss": 0.3544,
+      "step": 8056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2088068181818183e-05,
+      "loss": 0.3819,
+      "step": 8057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2083333333333333e-05,
+      "loss": 0.426,
+      "step": 8058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2078598484848485e-05,
+      "loss": 0.4499,
+      "step": 8059
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2073863636363638e-05,
+      "loss": 0.3869,
+      "step": 8060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.206912878787879e-05,
+      "loss": 0.3632,
+      "step": 8061
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.206439393939394e-05,
+      "loss": 0.3221,
+      "step": 8062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.205965909090909e-05,
+      "loss": 0.4375,
+      "step": 8063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2054924242424242e-05,
+      "loss": 0.4131,
+      "step": 8064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2050189393939394e-05,
+      "loss": 0.344,
+      "step": 8065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2045454545454547e-05,
+      "loss": 0.4414,
+      "step": 8066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2040719696969697e-05,
+      "loss": 0.4288,
+      "step": 8067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2035984848484849e-05,
+      "loss": 0.4272,
+      "step": 8068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2031250000000001e-05,
+      "loss": 0.4141,
+      "step": 8069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2026515151515153e-05,
+      "loss": 0.3779,
+      "step": 8070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2021780303030304e-05,
+      "loss": 0.6111,
+      "step": 8071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2017045454545454e-05,
+      "loss": 0.4317,
+      "step": 8072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2012310606060606e-05,
+      "loss": 0.3061,
+      "step": 8073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2007575757575758e-05,
+      "loss": 0.4699,
+      "step": 8074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.200284090909091e-05,
+      "loss": 0.4237,
+      "step": 8075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.199810606060606e-05,
+      "loss": 0.2875,
+      "step": 8076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1993371212121213e-05,
+      "loss": 0.4278,
+      "step": 8077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1988636363636365e-05,
+      "loss": 0.2892,
+      "step": 8078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1983901515151517e-05,
+      "loss": 0.3875,
+      "step": 8079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1979166666666667e-05,
+      "loss": 0.325,
+      "step": 8080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1974431818181819e-05,
+      "loss": 0.3298,
+      "step": 8081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.196969696969697e-05,
+      "loss": 0.5225,
+      "step": 8082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1964962121212122e-05,
+      "loss": 0.3661,
+      "step": 8083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1960227272727274e-05,
+      "loss": 0.3269,
+      "step": 8084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1955492424242424e-05,
+      "loss": 0.181,
+      "step": 8085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1950757575757576e-05,
+      "loss": 0.5114,
+      "step": 8086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1946022727272728e-05,
+      "loss": 0.5246,
+      "step": 8087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.194128787878788e-05,
+      "loss": 0.507,
+      "step": 8088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.193655303030303e-05,
+      "loss": 0.4341,
+      "step": 8089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1931818181818183e-05,
+      "loss": 0.4373,
+      "step": 8090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1927083333333335e-05,
+      "loss": 0.3976,
+      "step": 8091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1922348484848485e-05,
+      "loss": 0.3906,
+      "step": 8092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1917613636363637e-05,
+      "loss": 0.4792,
+      "step": 8093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1912878787878788e-05,
+      "loss": 0.5145,
+      "step": 8094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.190814393939394e-05,
+      "loss": 0.3391,
+      "step": 8095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1903409090909092e-05,
+      "loss": 0.3972,
+      "step": 8096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1898674242424244e-05,
+      "loss": 0.3578,
+      "step": 8097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1893939393939394e-05,
+      "loss": 0.2919,
+      "step": 8098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1889204545454546e-05,
+      "loss": 0.3311,
+      "step": 8099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1884469696969698e-05,
+      "loss": 0.3178,
+      "step": 8100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.187973484848485e-05,
+      "loss": 0.3396,
+      "step": 8101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1875e-05,
+      "loss": 0.3568,
+      "step": 8102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1870265151515151e-05,
+      "loss": 0.5739,
+      "step": 8103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1865530303030303e-05,
+      "loss": 0.4789,
+      "step": 8104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1860795454545455e-05,
+      "loss": 0.343,
+      "step": 8105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1856060606060607e-05,
+      "loss": 0.3664,
+      "step": 8106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1851325757575758e-05,
+      "loss": 0.3407,
+      "step": 8107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.184659090909091e-05,
+      "loss": 0.4279,
+      "step": 8108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1841856060606062e-05,
+      "loss": 0.4361,
+      "step": 8109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1837121212121214e-05,
+      "loss": 0.4081,
+      "step": 8110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1832386363636364e-05,
+      "loss": 0.3922,
+      "step": 8111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1827651515151515e-05,
+      "loss": 0.337,
+      "step": 8112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1822916666666667e-05,
+      "loss": 0.55,
+      "step": 8113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1818181818181819e-05,
+      "loss": 0.362,
+      "step": 8114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.181344696969697e-05,
+      "loss": 0.3435,
+      "step": 8115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1808712121212121e-05,
+      "loss": 0.4699,
+      "step": 8116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1803977272727273e-05,
+      "loss": 0.3632,
+      "step": 8117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1799242424242425e-05,
+      "loss": 0.3705,
+      "step": 8118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1794507575757577e-05,
+      "loss": 0.37,
+      "step": 8119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1789772727272728e-05,
+      "loss": 0.4411,
+      "step": 8120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1785037878787878e-05,
+      "loss": 0.2598,
+      "step": 8121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.178030303030303e-05,
+      "loss": 0.4092,
+      "step": 8122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1775568181818182e-05,
+      "loss": 0.5036,
+      "step": 8123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1770833333333334e-05,
+      "loss": 0.38,
+      "step": 8124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1766098484848485e-05,
+      "loss": 0.3752,
+      "step": 8125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1761363636363637e-05,
+      "loss": 0.4642,
+      "step": 8126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1756628787878789e-05,
+      "loss": 0.4041,
+      "step": 8127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1751893939393941e-05,
+      "loss": 0.4688,
+      "step": 8128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1747159090909091e-05,
+      "loss": 0.5415,
+      "step": 8129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1742424242424243e-05,
+      "loss": 0.4461,
+      "step": 8130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1737689393939394e-05,
+      "loss": 0.4086,
+      "step": 8131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1732954545454546e-05,
+      "loss": 0.4472,
+      "step": 8132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1728219696969698e-05,
+      "loss": 0.3391,
+      "step": 8133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1723484848484848e-05,
+      "loss": 0.4367,
+      "step": 8134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.171875e-05,
+      "loss": 0.4725,
+      "step": 8135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1714015151515152e-05,
+      "loss": 0.3902,
+      "step": 8136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1709280303030304e-05,
+      "loss": 0.3434,
+      "step": 8137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1704545454545455e-05,
+      "loss": 0.3224,
+      "step": 8138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1699810606060607e-05,
+      "loss": 0.3895,
+      "step": 8139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1695075757575757e-05,
+      "loss": 0.4217,
+      "step": 8140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.169034090909091e-05,
+      "loss": 0.4129,
+      "step": 8141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1685606060606061e-05,
+      "loss": 0.4447,
+      "step": 8142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1680871212121212e-05,
+      "loss": 0.4023,
+      "step": 8143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1676136363636364e-05,
+      "loss": 0.5124,
+      "step": 8144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1671401515151516e-05,
+      "loss": 0.4123,
+      "step": 8145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.3893,
+      "step": 8146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1661931818181818e-05,
+      "loss": 0.3174,
+      "step": 8147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.165719696969697e-05,
+      "loss": 0.2492,
+      "step": 8148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1652462121212122e-05,
+      "loss": 0.4109,
+      "step": 8149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1647727272727273e-05,
+      "loss": 0.2938,
+      "step": 8150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1642992424242425e-05,
+      "loss": 0.4377,
+      "step": 8151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1638257575757575e-05,
+      "loss": 0.3851,
+      "step": 8152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1633522727272727e-05,
+      "loss": 0.4647,
+      "step": 8153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.162878787878788e-05,
+      "loss": 0.391,
+      "step": 8154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1624053030303031e-05,
+      "loss": 0.3277,
+      "step": 8155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1619318181818182e-05,
+      "loss": 0.3407,
+      "step": 8156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1614583333333334e-05,
+      "loss": 0.2503,
+      "step": 8157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1609848484848486e-05,
+      "loss": 0.4577,
+      "step": 8158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1605113636363638e-05,
+      "loss": 0.3121,
+      "step": 8159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1600378787878788e-05,
+      "loss": 0.351,
+      "step": 8160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1595643939393939e-05,
+      "loss": 0.3745,
+      "step": 8161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.159090909090909e-05,
+      "loss": 0.2548,
+      "step": 8162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1586174242424243e-05,
+      "loss": 0.3991,
+      "step": 8163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1581439393939395e-05,
+      "loss": 0.4021,
+      "step": 8164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1576704545454545e-05,
+      "loss": 0.3488,
+      "step": 8165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1571969696969697e-05,
+      "loss": 0.3765,
+      "step": 8166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.156723484848485e-05,
+      "loss": 0.2367,
+      "step": 8167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1562500000000002e-05,
+      "loss": 0.5486,
+      "step": 8168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1557765151515152e-05,
+      "loss": 0.4278,
+      "step": 8169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1553030303030302e-05,
+      "loss": 0.3383,
+      "step": 8170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1548295454545454e-05,
+      "loss": 0.3495,
+      "step": 8171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1543560606060606e-05,
+      "loss": 0.3572,
+      "step": 8172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1538825757575758e-05,
+      "loss": 0.5853,
+      "step": 8173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1534090909090909e-05,
+      "loss": 0.334,
+      "step": 8174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1529356060606061e-05,
+      "loss": 0.4017,
+      "step": 8175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1524621212121213e-05,
+      "loss": 0.4371,
+      "step": 8176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1519886363636365e-05,
+      "loss": 0.2943,
+      "step": 8177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1515151515151517e-05,
+      "loss": 0.3888,
+      "step": 8178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1510416666666666e-05,
+      "loss": 0.331,
+      "step": 8179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1505681818181818e-05,
+      "loss": 0.4488,
+      "step": 8180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.150094696969697e-05,
+      "loss": 0.3027,
+      "step": 8181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1496212121212122e-05,
+      "loss": 0.4558,
+      "step": 8182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1491477272727272e-05,
+      "loss": 0.3097,
+      "step": 8183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1486742424242424e-05,
+      "loss": 0.3896,
+      "step": 8184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1482007575757577e-05,
+      "loss": 0.4621,
+      "step": 8185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1477272727272729e-05,
+      "loss": 0.433,
+      "step": 8186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.147253787878788e-05,
+      "loss": 0.4664,
+      "step": 8187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1467803030303031e-05,
+      "loss": 0.4745,
+      "step": 8188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1463068181818181e-05,
+      "loss": 0.4673,
+      "step": 8189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1458333333333333e-05,
+      "loss": 0.369,
+      "step": 8190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1453598484848486e-05,
+      "loss": 0.3619,
+      "step": 8191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1448863636363636e-05,
+      "loss": 0.298,
+      "step": 8192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1444128787878788e-05,
+      "loss": 0.4702,
+      "step": 8193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.143939393939394e-05,
+      "loss": 0.3692,
+      "step": 8194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1434659090909092e-05,
+      "loss": 0.4193,
+      "step": 8195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1429924242424244e-05,
+      "loss": 0.3684,
+      "step": 8196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1425189393939395e-05,
+      "loss": 0.4378,
+      "step": 8197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1420454545454547e-05,
+      "loss": 0.441,
+      "step": 8198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1415719696969697e-05,
+      "loss": 0.3021,
+      "step": 8199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1410984848484849e-05,
+      "loss": 0.4173,
+      "step": 8200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.140625e-05,
+      "loss": 0.3888,
+      "step": 8201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1401515151515151e-05,
+      "loss": 0.3583,
+      "step": 8202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1396780303030304e-05,
+      "loss": 0.579,
+      "step": 8203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1392045454545456e-05,
+      "loss": 0.3609,
+      "step": 8204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1387310606060608e-05,
+      "loss": 0.2579,
+      "step": 8205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1382575757575758e-05,
+      "loss": 0.4859,
+      "step": 8206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.137784090909091e-05,
+      "loss": 0.271,
+      "step": 8207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.137310606060606e-05,
+      "loss": 0.4039,
+      "step": 8208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1368371212121213e-05,
+      "loss": 0.2962,
+      "step": 8209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1363636363636365e-05,
+      "loss": 0.4549,
+      "step": 8210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1358901515151515e-05,
+      "loss": 0.4704,
+      "step": 8211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1354166666666667e-05,
+      "loss": 0.2932,
+      "step": 8212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1349431818181819e-05,
+      "loss": 0.5096,
+      "step": 8213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1344696969696971e-05,
+      "loss": 0.5035,
+      "step": 8214
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1339962121212122e-05,
+      "loss": 0.3043,
+      "step": 8215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1335227272727274e-05,
+      "loss": 0.3907,
+      "step": 8216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1330492424242426e-05,
+      "loss": 0.2879,
+      "step": 8217
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1325757575757576e-05,
+      "loss": 0.3543,
+      "step": 8218
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1321022727272728e-05,
+      "loss": 0.3061,
+      "step": 8219
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1316287878787879e-05,
+      "loss": 0.4591,
+      "step": 8220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.131155303030303e-05,
+      "loss": 0.4116,
+      "step": 8221
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1306818181818183e-05,
+      "loss": 0.2385,
+      "step": 8222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1302083333333335e-05,
+      "loss": 0.2853,
+      "step": 8223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1297348484848485e-05,
+      "loss": 0.4963,
+      "step": 8224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1292613636363637e-05,
+      "loss": 0.2721,
+      "step": 8225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.128787878787879e-05,
+      "loss": 0.3821,
+      "step": 8226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1283143939393941e-05,
+      "loss": 0.364,
+      "step": 8227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1278409090909092e-05,
+      "loss": 0.4102,
+      "step": 8228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1273674242424242e-05,
+      "loss": 0.3376,
+      "step": 8229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1268939393939394e-05,
+      "loss": 0.3735,
+      "step": 8230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1264204545454546e-05,
+      "loss": 0.3526,
+      "step": 8231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1259469696969698e-05,
+      "loss": 0.3104,
+      "step": 8232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1254734848484849e-05,
+      "loss": 0.3822,
+      "step": 8233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.125e-05,
+      "loss": 0.4579,
+      "step": 8234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1245265151515153e-05,
+      "loss": 0.3447,
+      "step": 8235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1240530303030305e-05,
+      "loss": 0.3552,
+      "step": 8236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1235795454545455e-05,
+      "loss": 0.3304,
+      "step": 8237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1231060606060606e-05,
+      "loss": 0.3045,
+      "step": 8238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1226325757575758e-05,
+      "loss": 0.4363,
+      "step": 8239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.122159090909091e-05,
+      "loss": 0.2615,
+      "step": 8240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1216856060606062e-05,
+      "loss": 0.3207,
+      "step": 8241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1212121212121212e-05,
+      "loss": 0.3382,
+      "step": 8242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1207386363636364e-05,
+      "loss": 0.4509,
+      "step": 8243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1202651515151516e-05,
+      "loss": 0.3473,
+      "step": 8244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1197916666666668e-05,
+      "loss": 0.5419,
+      "step": 8245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1193181818181819e-05,
+      "loss": 0.432,
+      "step": 8246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1188446969696969e-05,
+      "loss": 0.3402,
+      "step": 8247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1183712121212121e-05,
+      "loss": 0.2955,
+      "step": 8248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1178977272727273e-05,
+      "loss": 0.4172,
+      "step": 8249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1174242424242425e-05,
+      "loss": 0.3318,
+      "step": 8250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1169507575757576e-05,
+      "loss": 0.2551,
+      "step": 8251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1164772727272728e-05,
+      "loss": 0.3115,
+      "step": 8252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.116003787878788e-05,
+      "loss": 0.439,
+      "step": 8253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1155303030303032e-05,
+      "loss": 0.2911,
+      "step": 8254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1150568181818182e-05,
+      "loss": 0.3081,
+      "step": 8255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1145833333333334e-05,
+      "loss": 0.3772,
+      "step": 8256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1141098484848485e-05,
+      "loss": 0.384,
+      "step": 8257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1136363636363637e-05,
+      "loss": 0.3498,
+      "step": 8258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1131628787878789e-05,
+      "loss": 0.3943,
+      "step": 8259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.112689393939394e-05,
+      "loss": 0.334,
+      "step": 8260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1122159090909091e-05,
+      "loss": 0.357,
+      "step": 8261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1117424242424243e-05,
+      "loss": 0.3423,
+      "step": 8262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1112689393939395e-05,
+      "loss": 0.3624,
+      "step": 8263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1107954545454546e-05,
+      "loss": 0.368,
+      "step": 8264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1103219696969698e-05,
+      "loss": 0.4345,
+      "step": 8265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.109848484848485e-05,
+      "loss": 0.4503,
+      "step": 8266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.109375e-05,
+      "loss": 0.4623,
+      "step": 8267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1089015151515152e-05,
+      "loss": 0.3139,
+      "step": 8268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1084280303030303e-05,
+      "loss": 0.4243,
+      "step": 8269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1079545454545455e-05,
+      "loss": 0.4457,
+      "step": 8270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1074810606060607e-05,
+      "loss": 0.4731,
+      "step": 8271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1070075757575759e-05,
+      "loss": 0.3573,
+      "step": 8272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.106534090909091e-05,
+      "loss": 0.3553,
+      "step": 8273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1060606060606061e-05,
+      "loss": 0.3208,
+      "step": 8274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1055871212121213e-05,
+      "loss": 0.3891,
+      "step": 8275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1051136363636364e-05,
+      "loss": 0.2932,
+      "step": 8276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1046401515151516e-05,
+      "loss": 0.2831,
+      "step": 8277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1041666666666666e-05,
+      "loss": 0.4446,
+      "step": 8278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1036931818181818e-05,
+      "loss": 0.3214,
+      "step": 8279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.103219696969697e-05,
+      "loss": 0.3904,
+      "step": 8280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1027462121212122e-05,
+      "loss": 0.4046,
+      "step": 8281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1022727272727273e-05,
+      "loss": 0.3955,
+      "step": 8282
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1017992424242425e-05,
+      "loss": 0.5396,
+      "step": 8283
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1013257575757577e-05,
+      "loss": 0.3263,
+      "step": 8284
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1008522727272729e-05,
+      "loss": 0.3759,
+      "step": 8285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.100378787878788e-05,
+      "loss": 0.3434,
+      "step": 8286
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.099905303030303e-05,
+      "loss": 0.374,
+      "step": 8287
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0994318181818182e-05,
+      "loss": 0.4504,
+      "step": 8288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0989583333333334e-05,
+      "loss": 0.283,
+      "step": 8289
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0984848484848486e-05,
+      "loss": 0.3406,
+      "step": 8290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0980113636363636e-05,
+      "loss": 0.4202,
+      "step": 8291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0975378787878788e-05,
+      "loss": 0.3233,
+      "step": 8292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.097064393939394e-05,
+      "loss": 0.4676,
+      "step": 8293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0965909090909093e-05,
+      "loss": 0.3939,
+      "step": 8294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0961174242424243e-05,
+      "loss": 0.4542,
+      "step": 8295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0956439393939393e-05,
+      "loss": 0.3859,
+      "step": 8296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0951704545454545e-05,
+      "loss": 0.6974,
+      "step": 8297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0946969696969697e-05,
+      "loss": 0.3876,
+      "step": 8298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.094223484848485e-05,
+      "loss": 0.4551,
+      "step": 8299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.09375e-05,
+      "loss": 0.5007,
+      "step": 8300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0932765151515152e-05,
+      "loss": 0.448,
+      "step": 8301
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0928030303030304e-05,
+      "loss": 0.3792,
+      "step": 8302
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0923295454545456e-05,
+      "loss": 0.3557,
+      "step": 8303
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0918560606060606e-05,
+      "loss": 0.5012,
+      "step": 8304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0913825757575757e-05,
+      "loss": 0.3203,
+      "step": 8305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 0.35,
+      "step": 8306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0904356060606061e-05,
+      "loss": 0.3745,
+      "step": 8307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0899621212121213e-05,
+      "loss": 0.2902,
+      "step": 8308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0894886363636363e-05,
+      "loss": 0.3106,
+      "step": 8309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0890151515151515e-05,
+      "loss": 0.397,
+      "step": 8310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0885416666666668e-05,
+      "loss": 0.3741,
+      "step": 8311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.088068181818182e-05,
+      "loss": 0.3414,
+      "step": 8312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.087594696969697e-05,
+      "loss": 0.3368,
+      "step": 8313
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0871212121212122e-05,
+      "loss": 0.4113,
+      "step": 8314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0866477272727272e-05,
+      "loss": 0.4176,
+      "step": 8315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0861742424242424e-05,
+      "loss": 0.363,
+      "step": 8316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0857007575757577e-05,
+      "loss": 0.3482,
+      "step": 8317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0852272727272727e-05,
+      "loss": 0.477,
+      "step": 8318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0847537878787879e-05,
+      "loss": 0.3677,
+      "step": 8319
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0842803030303031e-05,
+      "loss": 0.2808,
+      "step": 8320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0838068181818183e-05,
+      "loss": 0.4502,
+      "step": 8321
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 0.4283,
+      "step": 8322
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0828598484848486e-05,
+      "loss": 0.3196,
+      "step": 8323
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0823863636363638e-05,
+      "loss": 0.3594,
+      "step": 8324
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0819128787878788e-05,
+      "loss": 0.438,
+      "step": 8325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.081439393939394e-05,
+      "loss": 0.3926,
+      "step": 8326
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.080965909090909e-05,
+      "loss": 0.3006,
+      "step": 8327
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0804924242424243e-05,
+      "loss": 0.3911,
+      "step": 8328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0800189393939395e-05,
+      "loss": 0.4263,
+      "step": 8329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0795454545454547e-05,
+      "loss": 0.3543,
+      "step": 8330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0790719696969697e-05,
+      "loss": 0.3777,
+      "step": 8331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0785984848484849e-05,
+      "loss": 0.483,
+      "step": 8332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0781250000000001e-05,
+      "loss": 0.4872,
+      "step": 8333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0776515151515152e-05,
+      "loss": 0.3727,
+      "step": 8334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0771780303030304e-05,
+      "loss": 0.4102,
+      "step": 8335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0767045454545454e-05,
+      "loss": 0.3575,
+      "step": 8336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0762310606060606e-05,
+      "loss": 0.3241,
+      "step": 8337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0757575757575758e-05,
+      "loss": 0.3646,
+      "step": 8338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.075284090909091e-05,
+      "loss": 0.4288,
+      "step": 8339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.074810606060606e-05,
+      "loss": 0.377,
+      "step": 8340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0743371212121213e-05,
+      "loss": 0.3572,
+      "step": 8341
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0738636363636365e-05,
+      "loss": 0.2837,
+      "step": 8342
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0733901515151517e-05,
+      "loss": 0.4167,
+      "step": 8343
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0729166666666667e-05,
+      "loss": 0.4508,
+      "step": 8344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0724431818181818e-05,
+      "loss": 0.4062,
+      "step": 8345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.071969696969697e-05,
+      "loss": 0.4105,
+      "step": 8346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0714962121212122e-05,
+      "loss": 0.5155,
+      "step": 8347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0710227272727274e-05,
+      "loss": 0.4801,
+      "step": 8348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0705492424242424e-05,
+      "loss": 0.4964,
+      "step": 8349
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0700757575757576e-05,
+      "loss": 0.448,
+      "step": 8350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0696022727272728e-05,
+      "loss": 0.4203,
+      "step": 8351
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.069128787878788e-05,
+      "loss": 0.5124,
+      "step": 8352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.068655303030303e-05,
+      "loss": 0.3172,
+      "step": 8353
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0681818181818181e-05,
+      "loss": 0.4956,
+      "step": 8354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0677083333333333e-05,
+      "loss": 0.3424,
+      "step": 8355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0672348484848485e-05,
+      "loss": 0.3372,
+      "step": 8356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0667613636363637e-05,
+      "loss": 0.3624,
+      "step": 8357
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0662878787878788e-05,
+      "loss": 0.3355,
+      "step": 8358
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.065814393939394e-05,
+      "loss": 0.4608,
+      "step": 8359
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0653409090909092e-05,
+      "loss": 0.4287,
+      "step": 8360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0648674242424244e-05,
+      "loss": 0.3593,
+      "step": 8361
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0643939393939394e-05,
+      "loss": 0.3688,
+      "step": 8362
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0639204545454546e-05,
+      "loss": 0.4515,
+      "step": 8363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0634469696969697e-05,
+      "loss": 0.3711,
+      "step": 8364
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0629734848484849e-05,
+      "loss": 0.2998,
+      "step": 8365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0625e-05,
+      "loss": 0.4246,
+      "step": 8366
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0620265151515151e-05,
+      "loss": 0.4217,
+      "step": 8367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0615530303030303e-05,
+      "loss": 0.3717,
+      "step": 8368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0610795454545455e-05,
+      "loss": 0.3476,
+      "step": 8369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0606060606060607e-05,
+      "loss": 0.3534,
+      "step": 8370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0601325757575758e-05,
+      "loss": 0.3609,
+      "step": 8371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.059659090909091e-05,
+      "loss": 0.2144,
+      "step": 8372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.059185606060606e-05,
+      "loss": 0.4529,
+      "step": 8373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0587121212121212e-05,
+      "loss": 0.319,
+      "step": 8374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0582386363636364e-05,
+      "loss": 0.4202,
+      "step": 8375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0577651515151515e-05,
+      "loss": 0.3319,
+      "step": 8376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0572916666666667e-05,
+      "loss": 0.5006,
+      "step": 8377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0568181818181819e-05,
+      "loss": 0.3925,
+      "step": 8378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0563446969696971e-05,
+      "loss": 0.4239,
+      "step": 8379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0558712121212121e-05,
+      "loss": 0.3288,
+      "step": 8380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0553977272727273e-05,
+      "loss": 0.4741,
+      "step": 8381
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0549242424242425e-05,
+      "loss": 0.3289,
+      "step": 8382
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0544507575757576e-05,
+      "loss": 0.2983,
+      "step": 8383
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0539772727272728e-05,
+      "loss": 0.2259,
+      "step": 8384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0535037878787878e-05,
+      "loss": 0.3362,
+      "step": 8385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.053030303030303e-05,
+      "loss": 0.2988,
+      "step": 8386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0525568181818182e-05,
+      "loss": 0.2914,
+      "step": 8387
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0520833333333334e-05,
+      "loss": 0.3904,
+      "step": 8388
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0516098484848485e-05,
+      "loss": 0.4395,
+      "step": 8389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0511363636363637e-05,
+      "loss": 0.2713,
+      "step": 8390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0506628787878789e-05,
+      "loss": 0.3447,
+      "step": 8391
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0501893939393941e-05,
+      "loss": 0.4028,
+      "step": 8392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0497159090909091e-05,
+      "loss": 0.4534,
+      "step": 8393
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0492424242424242e-05,
+      "loss": 0.4313,
+      "step": 8394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0487689393939394e-05,
+      "loss": 0.4785,
+      "step": 8395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0482954545454546e-05,
+      "loss": 0.4417,
+      "step": 8396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0478219696969698e-05,
+      "loss": 0.363,
+      "step": 8397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0473484848484848e-05,
+      "loss": 0.3713,
+      "step": 8398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.046875e-05,
+      "loss": 0.4044,
+      "step": 8399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0464015151515152e-05,
+      "loss": 0.3894,
+      "step": 8400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0459280303030304e-05,
+      "loss": 0.3686,
+      "step": 8401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0454545454545455e-05,
+      "loss": 0.3551,
+      "step": 8402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0449810606060605e-05,
+      "loss": 0.3137,
+      "step": 8403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0445075757575757e-05,
+      "loss": 0.3372,
+      "step": 8404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.044034090909091e-05,
+      "loss": 0.3003,
+      "step": 8405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0435606060606061e-05,
+      "loss": 0.3579,
+      "step": 8406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0430871212121212e-05,
+      "loss": 0.4062,
+      "step": 8407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0426136363636364e-05,
+      "loss": 0.3839,
+      "step": 8408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0421401515151516e-05,
+      "loss": 0.4872,
+      "step": 8409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 0.4511,
+      "step": 8410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.041193181818182e-05,
+      "loss": 0.4782,
+      "step": 8411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0407196969696969e-05,
+      "loss": 0.3327,
+      "step": 8412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.040246212121212e-05,
+      "loss": 0.2786,
+      "step": 8413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0397727272727273e-05,
+      "loss": 0.5727,
+      "step": 8414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0392992424242425e-05,
+      "loss": 0.3735,
+      "step": 8415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0388257575757577e-05,
+      "loss": 0.3093,
+      "step": 8416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0383522727272727e-05,
+      "loss": 0.2565,
+      "step": 8417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.037878787878788e-05,
+      "loss": 0.3243,
+      "step": 8418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0374053030303032e-05,
+      "loss": 0.4238,
+      "step": 8419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0369318181818184e-05,
+      "loss": 0.4102,
+      "step": 8420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0364583333333334e-05,
+      "loss": 0.3418,
+      "step": 8421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0359848484848484e-05,
+      "loss": 0.2959,
+      "step": 8422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0355113636363636e-05,
+      "loss": 0.5488,
+      "step": 8423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0350378787878788e-05,
+      "loss": 0.3733,
+      "step": 8424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.034564393939394e-05,
+      "loss": 0.4186,
+      "step": 8425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0340909090909091e-05,
+      "loss": 0.3462,
+      "step": 8426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0336174242424243e-05,
+      "loss": 0.4224,
+      "step": 8427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0331439393939395e-05,
+      "loss": 0.4332,
+      "step": 8428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0326704545454547e-05,
+      "loss": 0.3786,
+      "step": 8429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0321969696969698e-05,
+      "loss": 0.3722,
+      "step": 8430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.031723484848485e-05,
+      "loss": 0.4572,
+      "step": 8431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.03125e-05,
+      "loss": 0.4005,
+      "step": 8432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0307765151515152e-05,
+      "loss": 0.3864,
+      "step": 8433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0303030303030304e-05,
+      "loss": 0.4099,
+      "step": 8434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0298295454545454e-05,
+      "loss": 0.4673,
+      "step": 8435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0293560606060607e-05,
+      "loss": 0.3,
+      "step": 8436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0288825757575759e-05,
+      "loss": 0.3167,
+      "step": 8437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.028409090909091e-05,
+      "loss": 0.3976,
+      "step": 8438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0279356060606061e-05,
+      "loss": 0.3943,
+      "step": 8439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0274621212121213e-05,
+      "loss": 0.3401,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0269886363636363e-05,
+      "loss": 0.4917,
+      "step": 8441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0265151515151516e-05,
+      "loss": 0.2687,
+      "step": 8442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0260416666666668e-05,
+      "loss": 0.432,
+      "step": 8443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0255681818181818e-05,
+      "loss": 0.4094,
+      "step": 8444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.025094696969697e-05,
+      "loss": 0.3614,
+      "step": 8445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0246212121212122e-05,
+      "loss": 0.3977,
+      "step": 8446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0241477272727274e-05,
+      "loss": 0.4754,
+      "step": 8447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0236742424242425e-05,
+      "loss": 0.3468,
+      "step": 8448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0232007575757577e-05,
+      "loss": 0.2973,
+      "step": 8449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0227272727272729e-05,
+      "loss": 0.3639,
+      "step": 8450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0222537878787879e-05,
+      "loss": 0.3539,
+      "step": 8451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0217803030303031e-05,
+      "loss": 0.4587,
+      "step": 8452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0213068181818182e-05,
+      "loss": 0.3425,
+      "step": 8453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0208333333333334e-05,
+      "loss": 0.369,
+      "step": 8454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0203598484848486e-05,
+      "loss": 0.3561,
+      "step": 8455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0198863636363638e-05,
+      "loss": 0.3354,
+      "step": 8456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0194128787878788e-05,
+      "loss": 0.4674,
+      "step": 8457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.018939393939394e-05,
+      "loss": 0.3662,
+      "step": 8458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0184659090909092e-05,
+      "loss": 0.5108,
+      "step": 8459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0179924242424244e-05,
+      "loss": 0.3093,
+      "step": 8460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0175189393939395e-05,
+      "loss": 0.3974,
+      "step": 8461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0170454545454545e-05,
+      "loss": 0.5148,
+      "step": 8462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0165719696969697e-05,
+      "loss": 0.434,
+      "step": 8463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.016098484848485e-05,
+      "loss": 0.4272,
+      "step": 8464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0156250000000001e-05,
+      "loss": 0.3223,
+      "step": 8465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0151515151515152e-05,
+      "loss": 0.4354,
+      "step": 8466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0146780303030304e-05,
+      "loss": 0.4437,
+      "step": 8467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0142045454545456e-05,
+      "loss": 0.3044,
+      "step": 8468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0137310606060608e-05,
+      "loss": 0.3542,
+      "step": 8469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0132575757575758e-05,
+      "loss": 0.4161,
+      "step": 8470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0127840909090909e-05,
+      "loss": 0.4916,
+      "step": 8471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.012310606060606e-05,
+      "loss": 0.4616,
+      "step": 8472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0118371212121213e-05,
+      "loss": 0.4431,
+      "step": 8473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0113636363636365e-05,
+      "loss": 0.4449,
+      "step": 8474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0108901515151515e-05,
+      "loss": 0.3205,
+      "step": 8475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0104166666666667e-05,
+      "loss": 0.4081,
+      "step": 8476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.009943181818182e-05,
+      "loss": 0.4234,
+      "step": 8477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0094696969696971e-05,
+      "loss": 0.3132,
+      "step": 8478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0089962121212122e-05,
+      "loss": 0.3871,
+      "step": 8479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0085227272727272e-05,
+      "loss": 0.4421,
+      "step": 8480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0080492424242424e-05,
+      "loss": 0.4289,
+      "step": 8481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0075757575757576e-05,
+      "loss": 0.3604,
+      "step": 8482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0071022727272728e-05,
+      "loss": 0.2944,
+      "step": 8483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0066287878787879e-05,
+      "loss": 0.549,
+      "step": 8484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.006155303030303e-05,
+      "loss": 0.4013,
+      "step": 8485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0056818181818183e-05,
+      "loss": 0.3732,
+      "step": 8486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0052083333333335e-05,
+      "loss": 0.4664,
+      "step": 8487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0047348484848485e-05,
+      "loss": 0.3604,
+      "step": 8488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0042613636363637e-05,
+      "loss": 0.3409,
+      "step": 8489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0037878787878788e-05,
+      "loss": 0.4295,
+      "step": 8490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.003314393939394e-05,
+      "loss": 0.4084,
+      "step": 8491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0028409090909092e-05,
+      "loss": 0.3387,
+      "step": 8492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0023674242424242e-05,
+      "loss": 0.5432,
+      "step": 8493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0018939393939394e-05,
+      "loss": 0.4126,
+      "step": 8494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0014204545454546e-05,
+      "loss": 0.475,
+      "step": 8495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0009469696969698e-05,
+      "loss": 0.4167,
+      "step": 8496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0004734848484849e-05,
+      "loss": 0.3734,
+      "step": 8497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1e-05,
+      "loss": 0.4712,
+      "step": 8498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.995265151515151e-06,
+      "loss": 0.5347,
+      "step": 8499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.990530303030303e-06,
+      "loss": 0.3841,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.985795454545455e-06,
+      "loss": 0.5252,
+      "step": 8501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.981060606060606e-06,
+      "loss": 0.4656,
+      "step": 8502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.976325757575758e-06,
+      "loss": 0.3225,
+      "step": 8503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.97159090909091e-06,
+      "loss": 0.4222,
+      "step": 8504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.966856060606062e-06,
+      "loss": 0.3059,
+      "step": 8505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.962121212121212e-06,
+      "loss": 0.4837,
+      "step": 8506
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.957386363636364e-06,
+      "loss": 0.4184,
+      "step": 8507
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.952651515151516e-06,
+      "loss": 0.382,
+      "step": 8508
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.947916666666667e-06,
+      "loss": 0.3979,
+      "step": 8509
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.943181818181819e-06,
+      "loss": 0.3235,
+      "step": 8510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.93844696969697e-06,
+      "loss": 0.2849,
+      "step": 8511
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.933712121212121e-06,
+      "loss": 0.3543,
+      "step": 8512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.928977272727273e-06,
+      "loss": 0.3952,
+      "step": 8513
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.924242424242425e-06,
+      "loss": 0.4573,
+      "step": 8514
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.919507575757576e-06,
+      "loss": 0.4235,
+      "step": 8515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.914772727272728e-06,
+      "loss": 0.303,
+      "step": 8516
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.91003787878788e-06,
+      "loss": 0.4605,
+      "step": 8517
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.905303030303032e-06,
+      "loss": 0.4277,
+      "step": 8518
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.900568181818182e-06,
+      "loss": 0.3697,
+      "step": 8519
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.895833333333333e-06,
+      "loss": 0.3275,
+      "step": 8520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.891098484848485e-06,
+      "loss": 0.3586,
+      "step": 8521
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.886363636363637e-06,
+      "loss": 0.3279,
+      "step": 8522
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.881628787878789e-06,
+      "loss": 0.5423,
+      "step": 8523
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.87689393939394e-06,
+      "loss": 0.3029,
+      "step": 8524
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.872159090909091e-06,
+      "loss": 0.4941,
+      "step": 8525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.867424242424243e-06,
+      "loss": 0.3438,
+      "step": 8526
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.862689393939396e-06,
+      "loss": 0.4787,
+      "step": 8527
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.857954545454546e-06,
+      "loss": 0.3426,
+      "step": 8528
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.853219696969696e-06,
+      "loss": 0.4093,
+      "step": 8529
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.848484848484848e-06,
+      "loss": 0.3402,
+      "step": 8530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.84375e-06,
+      "loss": 0.3549,
+      "step": 8531
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839015151515152e-06,
+      "loss": 0.3414,
+      "step": 8532
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.834280303030303e-06,
+      "loss": 0.5574,
+      "step": 8533
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.829545454545455e-06,
+      "loss": 0.3572,
+      "step": 8534
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.824810606060607e-06,
+      "loss": 0.4358,
+      "step": 8535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.820075757575759e-06,
+      "loss": 0.4462,
+      "step": 8536
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.81534090909091e-06,
+      "loss": 0.4677,
+      "step": 8537
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.81060606060606e-06,
+      "loss": 0.3705,
+      "step": 8538
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.805871212121212e-06,
+      "loss": 0.3995,
+      "step": 8539
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.801136363636364e-06,
+      "loss": 0.3313,
+      "step": 8540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.796401515151516e-06,
+      "loss": 0.4169,
+      "step": 8541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.791666666666666e-06,
+      "loss": 0.348,
+      "step": 8542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.786931818181818e-06,
+      "loss": 0.432,
+      "step": 8543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.78219696969697e-06,
+      "loss": 0.2475,
+      "step": 8544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.777462121212123e-06,
+      "loss": 0.3882,
+      "step": 8545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.772727272727273e-06,
+      "loss": 0.2882,
+      "step": 8546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.767992424242425e-06,
+      "loss": 0.3629,
+      "step": 8547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.763257575757575e-06,
+      "loss": 0.4172,
+      "step": 8548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.758522727272727e-06,
+      "loss": 0.4069,
+      "step": 8549
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.75378787878788e-06,
+      "loss": 0.4309,
+      "step": 8550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.74905303030303e-06,
+      "loss": 0.3487,
+      "step": 8551
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.744318181818182e-06,
+      "loss": 0.2145,
+      "step": 8552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.739583333333334e-06,
+      "loss": 0.4679,
+      "step": 8553
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.734848484848486e-06,
+      "loss": 0.3371,
+      "step": 8554
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.730113636363636e-06,
+      "loss": 0.3638,
+      "step": 8555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.725378787878789e-06,
+      "loss": 0.3586,
+      "step": 8556
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.72064393939394e-06,
+      "loss": 0.5333,
+      "step": 8557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.715909090909091e-06,
+      "loss": 0.4038,
+      "step": 8558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.711174242424243e-06,
+      "loss": 0.4176,
+      "step": 8559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.706439393939393e-06,
+      "loss": 0.4035,
+      "step": 8560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.701704545454545e-06,
+      "loss": 0.271,
+      "step": 8561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 0.3346,
+      "step": 8562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.69223484848485e-06,
+      "loss": 0.4129,
+      "step": 8563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.6875e-06,
+      "loss": 0.3426,
+      "step": 8564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.682765151515152e-06,
+      "loss": 0.43,
+      "step": 8565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.678030303030304e-06,
+      "loss": 0.3212,
+      "step": 8566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.673295454545455e-06,
+      "loss": 0.3069,
+      "step": 8567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.668560606060607e-06,
+      "loss": 0.3363,
+      "step": 8568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.663825757575757e-06,
+      "loss": 0.3587,
+      "step": 8569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.659090909090909e-06,
+      "loss": 0.4109,
+      "step": 8570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.654356060606061e-06,
+      "loss": 0.4034,
+      "step": 8571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.649621212121213e-06,
+      "loss": 0.4164,
+      "step": 8572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.644886363636364e-06,
+      "loss": 0.4087,
+      "step": 8573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.640151515151516e-06,
+      "loss": 0.3188,
+      "step": 8574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.635416666666668e-06,
+      "loss": 0.3862,
+      "step": 8575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.63068181818182e-06,
+      "loss": 0.438,
+      "step": 8576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.62594696969697e-06,
+      "loss": 0.5004,
+      "step": 8577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.62121212121212e-06,
+      "loss": 0.3194,
+      "step": 8578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.616477272727273e-06,
+      "loss": 0.4125,
+      "step": 8579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.611742424242425e-06,
+      "loss": 0.3364,
+      "step": 8580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.607007575757577e-06,
+      "loss": 0.3363,
+      "step": 8581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.602272727272727e-06,
+      "loss": 0.5008,
+      "step": 8582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.597537878787879e-06,
+      "loss": 0.3658,
+      "step": 8583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.592803030303031e-06,
+      "loss": 0.4612,
+      "step": 8584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.588068181818183e-06,
+      "loss": 0.3113,
+      "step": 8585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.583333333333334e-06,
+      "loss": 0.3194,
+      "step": 8586
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.578598484848484e-06,
+      "loss": 0.389,
+      "step": 8587
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.573863636363636e-06,
+      "loss": 0.455,
+      "step": 8588
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.569128787878788e-06,
+      "loss": 0.4622,
+      "step": 8589
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.56439393939394e-06,
+      "loss": 0.3752,
+      "step": 8590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.55965909090909e-06,
+      "loss": 0.5109,
+      "step": 8591
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.554924242424243e-06,
+      "loss": 0.3936,
+      "step": 8592
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.550189393939395e-06,
+      "loss": 0.4193,
+      "step": 8593
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.545454545454547e-06,
+      "loss": 0.2927,
+      "step": 8594
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.540719696969697e-06,
+      "loss": 0.4767,
+      "step": 8595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.53598484848485e-06,
+      "loss": 0.3288,
+      "step": 8596
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.53125e-06,
+      "loss": 0.4127,
+      "step": 8597
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.526515151515152e-06,
+      "loss": 0.5603,
+      "step": 8598
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.521780303030304e-06,
+      "loss": 0.4531,
+      "step": 8599
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.517045454545454e-06,
+      "loss": 0.2917,
+      "step": 8600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.512310606060606e-06,
+      "loss": 0.2882,
+      "step": 8601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.507575757575758e-06,
+      "loss": 0.2942,
+      "step": 8602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.50284090909091e-06,
+      "loss": 0.3633,
+      "step": 8603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.49810606060606e-06,
+      "loss": 0.4959,
+      "step": 8604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.493371212121213e-06,
+      "loss": 0.4274,
+      "step": 8605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.488636363636363e-06,
+      "loss": 0.5684,
+      "step": 8606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.483901515151515e-06,
+      "loss": 0.2924,
+      "step": 8607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.479166666666667e-06,
+      "loss": 0.223,
+      "step": 8608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.474431818181818e-06,
+      "loss": 0.4412,
+      "step": 8609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.46969696969697e-06,
+      "loss": 0.4271,
+      "step": 8610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.464962121212122e-06,
+      "loss": 0.34,
+      "step": 8611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.460227272727274e-06,
+      "loss": 0.3609,
+      "step": 8612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.455492424242424e-06,
+      "loss": 0.3385,
+      "step": 8613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.450757575757576e-06,
+      "loss": 0.3202,
+      "step": 8614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.446022727272728e-06,
+      "loss": 0.3644,
+      "step": 8615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.441287878787879e-06,
+      "loss": 0.5013,
+      "step": 8616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.43655303030303e-06,
+      "loss": 0.4169,
+      "step": 8617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.431818181818181e-06,
+      "loss": 0.3509,
+      "step": 8618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.427083333333333e-06,
+      "loss": 0.4204,
+      "step": 8619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.422348484848485e-06,
+      "loss": 0.3614,
+      "step": 8620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.417613636363637e-06,
+      "loss": 0.2798,
+      "step": 8621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.41287878787879e-06,
+      "loss": 0.3849,
+      "step": 8622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.40814393939394e-06,
+      "loss": 0.3897,
+      "step": 8623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.403409090909092e-06,
+      "loss": 0.3058,
+      "step": 8624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.398674242424244e-06,
+      "loss": 0.2798,
+      "step": 8625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.393939393939394e-06,
+      "loss": 0.4022,
+      "step": 8626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.389204545454545e-06,
+      "loss": 0.4015,
+      "step": 8627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.384469696969697e-06,
+      "loss": 0.3627,
+      "step": 8628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.379734848484849e-06,
+      "loss": 0.5472,
+      "step": 8629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.2459,
+      "step": 8630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.370265151515153e-06,
+      "loss": 0.3282,
+      "step": 8631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.365530303030303e-06,
+      "loss": 0.5354,
+      "step": 8632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.360795454545455e-06,
+      "loss": 0.3093,
+      "step": 8633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.356060606060607e-06,
+      "loss": 0.283,
+      "step": 8634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.351325757575758e-06,
+      "loss": 0.3673,
+      "step": 8635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.346590909090908e-06,
+      "loss": 0.4751,
+      "step": 8636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.34185606060606e-06,
+      "loss": 0.2966,
+      "step": 8637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.337121212121212e-06,
+      "loss": 0.5414,
+      "step": 8638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.332386363636364e-06,
+      "loss": 0.4211,
+      "step": 8639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.327651515151516e-06,
+      "loss": 0.3717,
+      "step": 8640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.322916666666667e-06,
+      "loss": 0.2842,
+      "step": 8641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.318181818181819e-06,
+      "loss": 0.4377,
+      "step": 8642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.313446969696971e-06,
+      "loss": 0.271,
+      "step": 8643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.308712121212123e-06,
+      "loss": 0.2412,
+      "step": 8644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.303977272727272e-06,
+      "loss": 0.5022,
+      "step": 8645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.299242424242424e-06,
+      "loss": 0.3514,
+      "step": 8646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.294507575757576e-06,
+      "loss": 0.4678,
+      "step": 8647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.289772727272728e-06,
+      "loss": 0.382,
+      "step": 8648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.28503787878788e-06,
+      "loss": 0.3108,
+      "step": 8649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.28030303030303e-06,
+      "loss": 0.4647,
+      "step": 8650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.275568181818182e-06,
+      "loss": 0.4895,
+      "step": 8651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.270833333333334e-06,
+      "loss": 0.3319,
+      "step": 8652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.266098484848487e-06,
+      "loss": 0.392,
+      "step": 8653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.261363636363637e-06,
+      "loss": 0.411,
+      "step": 8654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.256628787878787e-06,
+      "loss": 0.3055,
+      "step": 8655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.25189393939394e-06,
+      "loss": 0.4416,
+      "step": 8656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.247159090909091e-06,
+      "loss": 0.3156,
+      "step": 8657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.242424242424244e-06,
+      "loss": 0.3424,
+      "step": 8658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.237689393939394e-06,
+      "loss": 0.3105,
+      "step": 8659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.232954545454546e-06,
+      "loss": 0.5042,
+      "step": 8660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.228219696969698e-06,
+      "loss": 0.4972,
+      "step": 8661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.22348484848485e-06,
+      "loss": 0.2266,
+      "step": 8662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.21875e-06,
+      "loss": 0.3963,
+      "step": 8663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.21401515151515e-06,
+      "loss": 0.4232,
+      "step": 8664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.209280303030303e-06,
+      "loss": 0.3394,
+      "step": 8665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.204545454545455e-06,
+      "loss": 0.4269,
+      "step": 8666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.199810606060607e-06,
+      "loss": 0.3946,
+      "step": 8667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.195075757575757e-06,
+      "loss": 0.3404,
+      "step": 8668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.19034090909091e-06,
+      "loss": 0.3458,
+      "step": 8669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.185606060606062e-06,
+      "loss": 0.3726,
+      "step": 8670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.180871212121214e-06,
+      "loss": 0.3379,
+      "step": 8671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.176136363636364e-06,
+      "loss": 0.4152,
+      "step": 8672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.171401515151516e-06,
+      "loss": 0.4544,
+      "step": 8673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 0.3703,
+      "step": 8674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.161931818181818e-06,
+      "loss": 0.4695,
+      "step": 8675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.15719696969697e-06,
+      "loss": 0.4461,
+      "step": 8676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.152462121212121e-06,
+      "loss": 0.3998,
+      "step": 8677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.147727272727273e-06,
+      "loss": 0.507,
+      "step": 8678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.142992424242425e-06,
+      "loss": 0.4858,
+      "step": 8679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.138257575757577e-06,
+      "loss": 0.3929,
+      "step": 8680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.133522727272728e-06,
+      "loss": 0.3529,
+      "step": 8681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.12878787878788e-06,
+      "loss": 0.2713,
+      "step": 8682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.124053030303032e-06,
+      "loss": 0.357,
+      "step": 8683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.119318181818182e-06,
+      "loss": 0.4023,
+      "step": 8684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.114583333333334e-06,
+      "loss": 0.4654,
+      "step": 8685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.109848484848484e-06,
+      "loss": 0.4055,
+      "step": 8686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.105113636363637e-06,
+      "loss": 0.3814,
+      "step": 8687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.100378787878789e-06,
+      "loss": 0.3408,
+      "step": 8688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.09564393939394e-06,
+      "loss": 0.2927,
+      "step": 8689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.2994,
+      "step": 8690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.086174242424243e-06,
+      "loss": 0.483,
+      "step": 8691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.081439393939395e-06,
+      "loss": 0.4517,
+      "step": 8692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.076704545454547e-06,
+      "loss": 0.3587,
+      "step": 8693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.071969696969698e-06,
+      "loss": 0.3212,
+      "step": 8694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.067234848484848e-06,
+      "loss": 0.4748,
+      "step": 8695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.0625e-06,
+      "loss": 0.5005,
+      "step": 8696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.057765151515152e-06,
+      "loss": 0.3613,
+      "step": 8697
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.053030303030304e-06,
+      "loss": 0.3407,
+      "step": 8698
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.048295454545455e-06,
+      "loss": 0.3151,
+      "step": 8699
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.043560606060607e-06,
+      "loss": 0.5001,
+      "step": 8700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.038825757575759e-06,
+      "loss": 0.3668,
+      "step": 8701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.03409090909091e-06,
+      "loss": 0.4912,
+      "step": 8702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.029356060606061e-06,
+      "loss": 0.4301,
+      "step": 8703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.024621212121212e-06,
+      "loss": 0.4178,
+      "step": 8704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.019886363636364e-06,
+      "loss": 0.391,
+      "step": 8705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.015151515151516e-06,
+      "loss": 0.2859,
+      "step": 8706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.010416666666668e-06,
+      "loss": 0.2821,
+      "step": 8707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.005681818181818e-06,
+      "loss": 0.4899,
+      "step": 8708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.00094696969697e-06,
+      "loss": 0.3678,
+      "step": 8709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.996212121212122e-06,
+      "loss": 0.4635,
+      "step": 8710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.991477272727274e-06,
+      "loss": 0.3236,
+      "step": 8711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.986742424242425e-06,
+      "loss": 0.4774,
+      "step": 8712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.982007575757575e-06,
+      "loss": 0.4645,
+      "step": 8713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.977272727272727e-06,
+      "loss": 0.2612,
+      "step": 8714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.97253787878788e-06,
+      "loss": 0.4722,
+      "step": 8715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.967803030303031e-06,
+      "loss": 0.3232,
+      "step": 8716
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.963068181818182e-06,
+      "loss": 0.3596,
+      "step": 8717
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.958333333333334e-06,
+      "loss": 0.3958,
+      "step": 8718
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.953598484848486e-06,
+      "loss": 0.5312,
+      "step": 8719
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.948863636363638e-06,
+      "loss": 0.2635,
+      "step": 8720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.944128787878788e-06,
+      "loss": 0.2183,
+      "step": 8721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.93939393939394e-06,
+      "loss": 0.3565,
+      "step": 8722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.93465909090909e-06,
+      "loss": 0.4562,
+      "step": 8723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.929924242424243e-06,
+      "loss": 0.3494,
+      "step": 8724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.925189393939395e-06,
+      "loss": 0.3597,
+      "step": 8725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.920454545454545e-06,
+      "loss": 0.4118,
+      "step": 8726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.915719696969697e-06,
+      "loss": 0.3816,
+      "step": 8727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.91098484848485e-06,
+      "loss": 0.3677,
+      "step": 8728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.906250000000001e-06,
+      "loss": 0.3222,
+      "step": 8729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.901515151515152e-06,
+      "loss": 0.3601,
+      "step": 8730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.896780303030304e-06,
+      "loss": 0.3348,
+      "step": 8731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.892045454545454e-06,
+      "loss": 0.3928,
+      "step": 8732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.887310606060606e-06,
+      "loss": 0.4112,
+      "step": 8733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.882575757575758e-06,
+      "loss": 0.3382,
+      "step": 8734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.877840909090909e-06,
+      "loss": 0.3198,
+      "step": 8735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.87310606060606e-06,
+      "loss": 0.4144,
+      "step": 8736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.868371212121213e-06,
+      "loss": 0.2624,
+      "step": 8737
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.863636363636365e-06,
+      "loss": 0.5103,
+      "step": 8738
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.858901515151515e-06,
+      "loss": 0.3686,
+      "step": 8739
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.854166666666667e-06,
+      "loss": 0.4845,
+      "step": 8740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.84943181818182e-06,
+      "loss": 0.2949,
+      "step": 8741
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.84469696969697e-06,
+      "loss": 0.4787,
+      "step": 8742
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.839962121212122e-06,
+      "loss": 0.3259,
+      "step": 8743
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.835227272727272e-06,
+      "loss": 0.388,
+      "step": 8744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.830492424242424e-06,
+      "loss": 0.3893,
+      "step": 8745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.825757575757576e-06,
+      "loss": 0.4216,
+      "step": 8746
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.821022727272728e-06,
+      "loss": 0.3161,
+      "step": 8747
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.816287878787879e-06,
+      "loss": 0.3268,
+      "step": 8748
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.81155303030303e-06,
+      "loss": 0.3612,
+      "step": 8749
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.806818181818183e-06,
+      "loss": 0.4305,
+      "step": 8750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.802083333333335e-06,
+      "loss": 0.4245,
+      "step": 8751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.797348484848485e-06,
+      "loss": 0.4266,
+      "step": 8752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.792613636363636e-06,
+      "loss": 0.3864,
+      "step": 8753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 0.3993,
+      "step": 8754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.78314393939394e-06,
+      "loss": 0.399,
+      "step": 8755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.778409090909092e-06,
+      "loss": 0.3931,
+      "step": 8756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.773674242424242e-06,
+      "loss": 0.3817,
+      "step": 8757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.768939393939394e-06,
+      "loss": 0.2761,
+      "step": 8758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.764204545454546e-06,
+      "loss": 0.3361,
+      "step": 8759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.759469696969698e-06,
+      "loss": 0.3539,
+      "step": 8760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.754734848484849e-06,
+      "loss": 0.3529,
+      "step": 8761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.75e-06,
+      "loss": 0.4304,
+      "step": 8762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.745265151515151e-06,
+      "loss": 0.4926,
+      "step": 8763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.740530303030303e-06,
+      "loss": 0.4636,
+      "step": 8764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.735795454545455e-06,
+      "loss": 0.4687,
+      "step": 8765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.731060606060606e-06,
+      "loss": 0.4796,
+      "step": 8766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.726325757575758e-06,
+      "loss": 0.3646,
+      "step": 8767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.72159090909091e-06,
+      "loss": 0.3588,
+      "step": 8768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.716856060606062e-06,
+      "loss": 0.446,
+      "step": 8769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.712121212121212e-06,
+      "loss": 0.2461,
+      "step": 8770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.707386363636363e-06,
+      "loss": 0.3864,
+      "step": 8771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.702651515151515e-06,
+      "loss": 0.3881,
+      "step": 8772
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.697916666666667e-06,
+      "loss": 0.2393,
+      "step": 8773
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.693181818181819e-06,
+      "loss": 0.3487,
+      "step": 8774
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.68844696969697e-06,
+      "loss": 0.3657,
+      "step": 8775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.683712121212121e-06,
+      "loss": 0.3635,
+      "step": 8776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.678977272727273e-06,
+      "loss": 0.4454,
+      "step": 8777
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.674242424242426e-06,
+      "loss": 0.2441,
+      "step": 8778
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.669507575757576e-06,
+      "loss": 0.3089,
+      "step": 8779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.664772727272728e-06,
+      "loss": 0.3463,
+      "step": 8780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.660037878787878e-06,
+      "loss": 0.5169,
+      "step": 8781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.65530303030303e-06,
+      "loss": 0.3334,
+      "step": 8782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.650568181818182e-06,
+      "loss": 0.3193,
+      "step": 8783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.645833333333333e-06,
+      "loss": 0.3583,
+      "step": 8784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.641098484848485e-06,
+      "loss": 0.2657,
+      "step": 8785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.636363636363637e-06,
+      "loss": 0.4986,
+      "step": 8786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.631628787878789e-06,
+      "loss": 0.372,
+      "step": 8787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.62689393939394e-06,
+      "loss": 0.4504,
+      "step": 8788
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.622159090909092e-06,
+      "loss": 0.4293,
+      "step": 8789
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.617424242424244e-06,
+      "loss": 0.4063,
+      "step": 8790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.612689393939394e-06,
+      "loss": 0.3287,
+      "step": 8791
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.607954545454546e-06,
+      "loss": 0.3015,
+      "step": 8792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.603219696969696e-06,
+      "loss": 0.3339,
+      "step": 8793
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.598484848484848e-06,
+      "loss": 0.315,
+      "step": 8794
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.59375e-06,
+      "loss": 0.3754,
+      "step": 8795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.589015151515153e-06,
+      "loss": 0.4282,
+      "step": 8796
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.584280303030303e-06,
+      "loss": 0.3356,
+      "step": 8797
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.579545454545455e-06,
+      "loss": 0.4339,
+      "step": 8798
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.574810606060607e-06,
+      "loss": 0.3866,
+      "step": 8799
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.570075757575757e-06,
+      "loss": 0.3542,
+      "step": 8800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.56534090909091e-06,
+      "loss": 0.528,
+      "step": 8801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.56060606060606e-06,
+      "loss": 0.5024,
+      "step": 8802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.555871212121212e-06,
+      "loss": 0.5231,
+      "step": 8803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.551136363636364e-06,
+      "loss": 0.4711,
+      "step": 8804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.546401515151516e-06,
+      "loss": 0.4406,
+      "step": 8805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.541666666666666e-06,
+      "loss": 0.3567,
+      "step": 8806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.536931818181819e-06,
+      "loss": 0.3682,
+      "step": 8807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.53219696969697e-06,
+      "loss": 0.2917,
+      "step": 8808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.527462121212123e-06,
+      "loss": 0.3876,
+      "step": 8809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.522727272727273e-06,
+      "loss": 0.3551,
+      "step": 8810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.517992424242423e-06,
+      "loss": 0.3602,
+      "step": 8811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.513257575757575e-06,
+      "loss": 0.3351,
+      "step": 8812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.508522727272728e-06,
+      "loss": 0.3923,
+      "step": 8813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.50378787878788e-06,
+      "loss": 0.4857,
+      "step": 8814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.49905303030303e-06,
+      "loss": 0.3977,
+      "step": 8815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.494318181818182e-06,
+      "loss": 0.2932,
+      "step": 8816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.489583333333334e-06,
+      "loss": 0.4385,
+      "step": 8817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 0.326,
+      "step": 8818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.480113636363637e-06,
+      "loss": 0.4355,
+      "step": 8819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.475378787878787e-06,
+      "loss": 0.3997,
+      "step": 8820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.470643939393939e-06,
+      "loss": 0.352,
+      "step": 8821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.465909090909091e-06,
+      "loss": 0.477,
+      "step": 8822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.461174242424243e-06,
+      "loss": 0.3833,
+      "step": 8823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.456439393939394e-06,
+      "loss": 0.3248,
+      "step": 8824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.451704545454546e-06,
+      "loss": 0.4445,
+      "step": 8825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.446969696969698e-06,
+      "loss": 0.3995,
+      "step": 8826
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.44223484848485e-06,
+      "loss": 0.4193,
+      "step": 8827
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.437500000000002e-06,
+      "loss": 0.4279,
+      "step": 8828
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.43276515151515e-06,
+      "loss": 0.4617,
+      "step": 8829
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.428030303030303e-06,
+      "loss": 0.3569,
+      "step": 8830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.423295454545455e-06,
+      "loss": 0.4199,
+      "step": 8831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.418560606060607e-06,
+      "loss": 0.2885,
+      "step": 8832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.413825757575757e-06,
+      "loss": 0.394,
+      "step": 8833
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.409090909090909e-06,
+      "loss": 0.3622,
+      "step": 8834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.404356060606061e-06,
+      "loss": 0.5063,
+      "step": 8835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.399621212121213e-06,
+      "loss": 0.4481,
+      "step": 8836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.394886363636365e-06,
+      "loss": 0.522,
+      "step": 8837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.390151515151516e-06,
+      "loss": 0.3914,
+      "step": 8838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.385416666666666e-06,
+      "loss": 0.3385,
+      "step": 8839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.380681818181818e-06,
+      "loss": 0.4069,
+      "step": 8840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.37594696969697e-06,
+      "loss": 0.4606,
+      "step": 8841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.37121212121212e-06,
+      "loss": 0.3208,
+      "step": 8842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.366477272727273e-06,
+      "loss": 0.3411,
+      "step": 8843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.361742424242425e-06,
+      "loss": 0.2827,
+      "step": 8844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.357007575757577e-06,
+      "loss": 0.2673,
+      "step": 8845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.352272727272729e-06,
+      "loss": 0.322,
+      "step": 8846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.34753787878788e-06,
+      "loss": 0.4246,
+      "step": 8847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.342803030303031e-06,
+      "loss": 0.3677,
+      "step": 8848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.338068181818182e-06,
+      "loss": 0.3373,
+      "step": 8849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.3142,
+      "step": 8850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.328598484848484e-06,
+      "loss": 0.3189,
+      "step": 8851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.323863636363636e-06,
+      "loss": 0.4872,
+      "step": 8852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.319128787878788e-06,
+      "loss": 0.4105,
+      "step": 8853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.31439393939394e-06,
+      "loss": 0.3026,
+      "step": 8854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.309659090909092e-06,
+      "loss": 0.278,
+      "step": 8855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.304924242424243e-06,
+      "loss": 0.577,
+      "step": 8856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.300189393939395e-06,
+      "loss": 0.4184,
+      "step": 8857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.295454545454547e-06,
+      "loss": 0.4499,
+      "step": 8858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.290719696969697e-06,
+      "loss": 0.3326,
+      "step": 8859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.285984848484848e-06,
+      "loss": 0.4144,
+      "step": 8860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.28125e-06,
+      "loss": 0.4539,
+      "step": 8861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.276515151515152e-06,
+      "loss": 0.3023,
+      "step": 8862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.271780303030304e-06,
+      "loss": 0.4582,
+      "step": 8863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.267045454545456e-06,
+      "loss": 0.3937,
+      "step": 8864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.262310606060606e-06,
+      "loss": 0.3594,
+      "step": 8865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.257575757575758e-06,
+      "loss": 0.2881,
+      "step": 8866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.25284090909091e-06,
+      "loss": 0.2699,
+      "step": 8867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.24810606060606e-06,
+      "loss": 0.3403,
+      "step": 8868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.243371212121211e-06,
+      "loss": 0.3391,
+      "step": 8869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.238636363636363e-06,
+      "loss": 0.3744,
+      "step": 8870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.233901515151515e-06,
+      "loss": 0.3552,
+      "step": 8871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.229166666666667e-06,
+      "loss": 0.3038,
+      "step": 8872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.22443181818182e-06,
+      "loss": 0.2824,
+      "step": 8873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.21969696969697e-06,
+      "loss": 0.297,
+      "step": 8874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.214962121212122e-06,
+      "loss": 0.4815,
+      "step": 8875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.210227272727274e-06,
+      "loss": 0.3875,
+      "step": 8876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.205492424242426e-06,
+      "loss": 0.3813,
+      "step": 8877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.200757575757576e-06,
+      "loss": 0.4119,
+      "step": 8878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.196022727272727e-06,
+      "loss": 0.4193,
+      "step": 8879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.191287878787879e-06,
+      "loss": 0.3659,
+      "step": 8880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.186553030303031e-06,
+      "loss": 0.4064,
+      "step": 8881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 0.6319,
+      "step": 8882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.177083333333333e-06,
+      "loss": 0.3144,
+      "step": 8883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.172348484848485e-06,
+      "loss": 0.3102,
+      "step": 8884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.167613636363637e-06,
+      "loss": 0.2801,
+      "step": 8885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.16287878787879e-06,
+      "loss": 0.3449,
+      "step": 8886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.15814393939394e-06,
+      "loss": 0.317,
+      "step": 8887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.15340909090909e-06,
+      "loss": 0.4944,
+      "step": 8888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.148674242424242e-06,
+      "loss": 0.4307,
+      "step": 8889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.143939393939394e-06,
+      "loss": 0.2919,
+      "step": 8890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.139204545454546e-06,
+      "loss": 0.3418,
+      "step": 8891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.134469696969697e-06,
+      "loss": 0.3825,
+      "step": 8892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.129734848484849e-06,
+      "loss": 0.3446,
+      "step": 8893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 0.4089,
+      "step": 8894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.120265151515153e-06,
+      "loss": 0.2979,
+      "step": 8895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.115530303030303e-06,
+      "loss": 0.4546,
+      "step": 8896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.110795454545454e-06,
+      "loss": 0.4117,
+      "step": 8897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.106060606060606e-06,
+      "loss": 0.4533,
+      "step": 8898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.101325757575758e-06,
+      "loss": 0.2692,
+      "step": 8899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.09659090909091e-06,
+      "loss": 0.3101,
+      "step": 8900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.09185606060606e-06,
+      "loss": 0.3622,
+      "step": 8901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.087121212121212e-06,
+      "loss": 0.4573,
+      "step": 8902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.082386363636365e-06,
+      "loss": 0.3713,
+      "step": 8903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.077651515151517e-06,
+      "loss": 0.3584,
+      "step": 8904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.072916666666667e-06,
+      "loss": 0.4404,
+      "step": 8905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.068181818181819e-06,
+      "loss": 0.3679,
+      "step": 8906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.06344696969697e-06,
+      "loss": 0.3104,
+      "step": 8907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.058712121212121e-06,
+      "loss": 0.3689,
+      "step": 8908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.053977272727274e-06,
+      "loss": 0.4159,
+      "step": 8909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.049242424242424e-06,
+      "loss": 0.3317,
+      "step": 8910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.044507575757576e-06,
+      "loss": 0.4869,
+      "step": 8911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.039772727272728e-06,
+      "loss": 0.3691,
+      "step": 8912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.03503787878788e-06,
+      "loss": 0.3687,
+      "step": 8913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.03030303030303e-06,
+      "loss": 0.3491,
+      "step": 8914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.025568181818183e-06,
+      "loss": 0.3715,
+      "step": 8915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.020833333333335e-06,
+      "loss": 0.4332,
+      "step": 8916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.016098484848485e-06,
+      "loss": 0.4187,
+      "step": 8917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.011363636363637e-06,
+      "loss": 0.3726,
+      "step": 8918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.006628787878787e-06,
+      "loss": 0.4043,
+      "step": 8919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.00189393939394e-06,
+      "loss": 0.3197,
+      "step": 8920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.997159090909092e-06,
+      "loss": 0.4084,
+      "step": 8921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.992424242424244e-06,
+      "loss": 0.3622,
+      "step": 8922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.987689393939394e-06,
+      "loss": 0.4277,
+      "step": 8923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.982954545454546e-06,
+      "loss": 0.3267,
+      "step": 8924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.978219696969698e-06,
+      "loss": 0.3348,
+      "step": 8925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.97348484848485e-06,
+      "loss": 0.2535,
+      "step": 8926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.96875e-06,
+      "loss": 0.3345,
+      "step": 8927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.964015151515151e-06,
+      "loss": 0.4266,
+      "step": 8928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.959280303030303e-06,
+      "loss": 0.4281,
+      "step": 8929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.954545454545455e-06,
+      "loss": 0.4311,
+      "step": 8930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.949810606060607e-06,
+      "loss": 0.3229,
+      "step": 8931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.945075757575758e-06,
+      "loss": 0.4122,
+      "step": 8932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.94034090909091e-06,
+      "loss": 0.3793,
+      "step": 8933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.935606060606062e-06,
+      "loss": 0.294,
+      "step": 8934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.930871212121214e-06,
+      "loss": 0.2665,
+      "step": 8935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.926136363636364e-06,
+      "loss": 0.3675,
+      "step": 8936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.921401515151514e-06,
+      "loss": 0.5482,
+      "step": 8937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.916666666666667e-06,
+      "loss": 0.3467,
+      "step": 8938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.911931818181819e-06,
+      "loss": 0.4379,
+      "step": 8939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.90719696969697e-06,
+      "loss": 0.3484,
+      "step": 8940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.902462121212121e-06,
+      "loss": 0.3802,
+      "step": 8941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.897727272727273e-06,
+      "loss": 0.3806,
+      "step": 8942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.892992424242425e-06,
+      "loss": 0.4834,
+      "step": 8943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.888257575757577e-06,
+      "loss": 0.3397,
+      "step": 8944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.883522727272728e-06,
+      "loss": 0.4963,
+      "step": 8945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.878787878787878e-06,
+      "loss": 0.3288,
+      "step": 8946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.87405303030303e-06,
+      "loss": 0.3089,
+      "step": 8947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.869318181818182e-06,
+      "loss": 0.3943,
+      "step": 8948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.864583333333334e-06,
+      "loss": 0.5069,
+      "step": 8949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.859848484848485e-06,
+      "loss": 0.3646,
+      "step": 8950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.855113636363637e-06,
+      "loss": 0.3784,
+      "step": 8951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.850378787878789e-06,
+      "loss": 0.3184,
+      "step": 8952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.84564393939394e-06,
+      "loss": 0.3434,
+      "step": 8953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.840909090909091e-06,
+      "loss": 0.3874,
+      "step": 8954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.836174242424243e-06,
+      "loss": 0.3163,
+      "step": 8955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.831439393939394e-06,
+      "loss": 0.3628,
+      "step": 8956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.826704545454546e-06,
+      "loss": 0.3568,
+      "step": 8957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.821969696969698e-06,
+      "loss": 0.3296,
+      "step": 8958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.817234848484848e-06,
+      "loss": 0.4113,
+      "step": 8959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.8125e-06,
+      "loss": 0.3612,
+      "step": 8960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.807765151515152e-06,
+      "loss": 0.3321,
+      "step": 8961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.803030303030304e-06,
+      "loss": 0.4,
+      "step": 8962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.798295454545455e-06,
+      "loss": 0.3681,
+      "step": 8963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.793560606060607e-06,
+      "loss": 0.3661,
+      "step": 8964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.788825757575757e-06,
+      "loss": 0.3611,
+      "step": 8965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.78409090909091e-06,
+      "loss": 0.539,
+      "step": 8966
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.779356060606061e-06,
+      "loss": 0.2447,
+      "step": 8967
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.774621212121212e-06,
+      "loss": 0.4161,
+      "step": 8968
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.769886363636364e-06,
+      "loss": 0.3997,
+      "step": 8969
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.765151515151516e-06,
+      "loss": 0.3513,
+      "step": 8970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.760416666666668e-06,
+      "loss": 0.4278,
+      "step": 8971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.755681818181818e-06,
+      "loss": 0.4281,
+      "step": 8972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.75094696969697e-06,
+      "loss": 0.4164,
+      "step": 8973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.746212121212122e-06,
+      "loss": 0.2937,
+      "step": 8974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.741477272727273e-06,
+      "loss": 0.3352,
+      "step": 8975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.736742424242425e-06,
+      "loss": 0.517,
+      "step": 8976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.732007575757575e-06,
+      "loss": 0.3406,
+      "step": 8977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.727272727272727e-06,
+      "loss": 0.2979,
+      "step": 8978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.72253787878788e-06,
+      "loss": 0.4228,
+      "step": 8979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.717803030303031e-06,
+      "loss": 0.2765,
+      "step": 8980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.713068181818182e-06,
+      "loss": 0.2964,
+      "step": 8981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.708333333333334e-06,
+      "loss": 0.4996,
+      "step": 8982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.703598484848486e-06,
+      "loss": 0.3194,
+      "step": 8983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.698863636363638e-06,
+      "loss": 0.2767,
+      "step": 8984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.694128787878788e-06,
+      "loss": 0.466,
+      "step": 8985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.689393939393939e-06,
+      "loss": 0.3318,
+      "step": 8986
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.68465909090909e-06,
+      "loss": 0.3226,
+      "step": 8987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.679924242424243e-06,
+      "loss": 0.4201,
+      "step": 8988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.675189393939395e-06,
+      "loss": 0.4408,
+      "step": 8989
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.670454545454545e-06,
+      "loss": 0.3307,
+      "step": 8990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.665719696969697e-06,
+      "loss": 0.4697,
+      "step": 8991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.66098484848485e-06,
+      "loss": 0.3933,
+      "step": 8992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.656250000000001e-06,
+      "loss": 0.3193,
+      "step": 8993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.651515151515152e-06,
+      "loss": 0.4178,
+      "step": 8994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.646780303030302e-06,
+      "loss": 0.3732,
+      "step": 8995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.642045454545454e-06,
+      "loss": 0.2756,
+      "step": 8996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.637310606060606e-06,
+      "loss": 0.2928,
+      "step": 8997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.632575757575758e-06,
+      "loss": 0.397,
+      "step": 8998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.62784090909091e-06,
+      "loss": 0.4926,
+      "step": 8999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.623106060606061e-06,
+      "loss": 0.3425,
+      "step": 9000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.618371212121213e-06,
+      "loss": 0.2634,
+      "step": 9001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.613636363636364e-06,
+      "loss": 0.3844,
+      "step": 9002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.608901515151516e-06,
+      "loss": 0.3996,
+      "step": 9003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.6041666666666666e-06,
+      "loss": 0.4777,
+      "step": 9004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.599431818181818e-06,
+      "loss": 0.3107,
+      "step": 9005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.59469696969697e-06,
+      "loss": 0.3113,
+      "step": 9006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.589962121212121e-06,
+      "loss": 0.4691,
+      "step": 9007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.585227272727273e-06,
+      "loss": 0.4829,
+      "step": 9008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.580492424242424e-06,
+      "loss": 0.3321,
+      "step": 9009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.3797,
+      "step": 9010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.571022727272728e-06,
+      "loss": 0.4209,
+      "step": 9011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.56628787878788e-06,
+      "loss": 0.3999,
+      "step": 9012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.561553030303031e-06,
+      "loss": 0.3142,
+      "step": 9013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.556818181818181e-06,
+      "loss": 0.3885,
+      "step": 9014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.552083333333333e-06,
+      "loss": 0.326,
+      "step": 9015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.547348484848485e-06,
+      "loss": 0.2756,
+      "step": 9016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.542613636363637e-06,
+      "loss": 0.3639,
+      "step": 9017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.537878787878788e-06,
+      "loss": 0.3761,
+      "step": 9018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.53314393939394e-06,
+      "loss": 0.4522,
+      "step": 9019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.528409090909091e-06,
+      "loss": 0.2724,
+      "step": 9020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.523674242424243e-06,
+      "loss": 0.3451,
+      "step": 9021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5189393939393945e-06,
+      "loss": 0.2761,
+      "step": 9022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5142045454545465e-06,
+      "loss": 0.3683,
+      "step": 9023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.509469696969697e-06,
+      "loss": 0.4498,
+      "step": 9024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.504734848484848e-06,
+      "loss": 0.2991,
+      "step": 9025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5e-06,
+      "loss": 0.2858,
+      "step": 9026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.495265151515151e-06,
+      "loss": 0.4162,
+      "step": 9027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4905303030303035e-06,
+      "loss": 0.4324,
+      "step": 9028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.485795454545455e-06,
+      "loss": 0.29,
+      "step": 9029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.481060606060607e-06,
+      "loss": 0.3901,
+      "step": 9030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.476325757575758e-06,
+      "loss": 0.4064,
+      "step": 9031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.47159090909091e-06,
+      "loss": 0.405,
+      "step": 9032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4668560606060604e-06,
+      "loss": 0.4907,
+      "step": 9033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.462121212121212e-06,
+      "loss": 0.2355,
+      "step": 9034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.457386363636364e-06,
+      "loss": 0.3789,
+      "step": 9035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.452651515151515e-06,
+      "loss": 0.3634,
+      "step": 9036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.447916666666667e-06,
+      "loss": 0.4693,
+      "step": 9037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.443181818181818e-06,
+      "loss": 0.2817,
+      "step": 9038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.43844696969697e-06,
+      "loss": 0.4675,
+      "step": 9039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4337121212121215e-06,
+      "loss": 0.4568,
+      "step": 9040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.428977272727274e-06,
+      "loss": 0.4971,
+      "step": 9041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.424242424242425e-06,
+      "loss": 0.5745,
+      "step": 9042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.419507575757575e-06,
+      "loss": 0.4416,
+      "step": 9043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.414772727272727e-06,
+      "loss": 0.3755,
+      "step": 9044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4100378787878785e-06,
+      "loss": 0.4233,
+      "step": 9045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4053030303030305e-06,
+      "loss": 0.3251,
+      "step": 9046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.400568181818182e-06,
+      "loss": 0.3701,
+      "step": 9047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.395833333333334e-06,
+      "loss": 0.2577,
+      "step": 9048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.391098484848485e-06,
+      "loss": 0.3604,
+      "step": 9049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.386363636363637e-06,
+      "loss": 0.684,
+      "step": 9050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.381628787878789e-06,
+      "loss": 0.4487,
+      "step": 9051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.37689393939394e-06,
+      "loss": 0.2791,
+      "step": 9052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.372159090909091e-06,
+      "loss": 0.3008,
+      "step": 9053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.367424242424242e-06,
+      "loss": 0.3649,
+      "step": 9054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.362689393939394e-06,
+      "loss": 0.3096,
+      "step": 9055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.357954545454545e-06,
+      "loss": 0.2757,
+      "step": 9056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.353219696969697e-06,
+      "loss": 0.3951,
+      "step": 9057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.3484848484848486e-06,
+      "loss": 0.3055,
+      "step": 9058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.343750000000001e-06,
+      "loss": 0.3508,
+      "step": 9059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.339015151515153e-06,
+      "loss": 0.4708,
+      "step": 9060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.334280303030304e-06,
+      "loss": 0.3232,
+      "step": 9061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.329545454545454e-06,
+      "loss": 0.4634,
+      "step": 9062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.3248106060606055e-06,
+      "loss": 0.2941,
+      "step": 9063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.3200757575757576e-06,
+      "loss": 0.3067,
+      "step": 9064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.315340909090909e-06,
+      "loss": 0.3628,
+      "step": 9065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.310606060606061e-06,
+      "loss": 0.3836,
+      "step": 9066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.305871212121212e-06,
+      "loss": 0.2922,
+      "step": 9067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.301136363636364e-06,
+      "loss": 0.3015,
+      "step": 9068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.296401515151516e-06,
+      "loss": 0.3269,
+      "step": 9069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.2916666666666674e-06,
+      "loss": 0.2659,
+      "step": 9070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.2869318181818195e-06,
+      "loss": 0.3414,
+      "step": 9071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.282196969696969e-06,
+      "loss": 0.3236,
+      "step": 9072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.277462121212121e-06,
+      "loss": 0.425,
+      "step": 9073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.272727272727272e-06,
+      "loss": 0.2974,
+      "step": 9074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.267992424242424e-06,
+      "loss": 0.3674,
+      "step": 9075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.2632575757575765e-06,
+      "loss": 0.2859,
+      "step": 9076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.258522727272728e-06,
+      "loss": 0.2914,
+      "step": 9077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.25378787878788e-06,
+      "loss": 0.4595,
+      "step": 9078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.249053030303031e-06,
+      "loss": 0.2384,
+      "step": 9079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.244318181818183e-06,
+      "loss": 0.2812,
+      "step": 9080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.239583333333334e-06,
+      "loss": 0.4452,
+      "step": 9081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.234848484848485e-06,
+      "loss": 0.3779,
+      "step": 9082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.230113636363636e-06,
+      "loss": 0.3924,
+      "step": 9083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.225378787878788e-06,
+      "loss": 0.3305,
+      "step": 9084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.22064393939394e-06,
+      "loss": 0.2968,
+      "step": 9085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.215909090909091e-06,
+      "loss": 0.4107,
+      "step": 9086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.211174242424243e-06,
+      "loss": 0.4986,
+      "step": 9087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.2064393939393945e-06,
+      "loss": 0.3405,
+      "step": 9088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.2017045454545466e-06,
+      "loss": 0.3191,
+      "step": 9089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.196969696969698e-06,
+      "loss": 0.3145,
+      "step": 9090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.19223484848485e-06,
+      "loss": 0.3998,
+      "step": 9091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.187499999999999e-06,
+      "loss": 0.4062,
+      "step": 9092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1827651515151514e-06,
+      "loss": 0.4388,
+      "step": 9093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1780303030303035e-06,
+      "loss": 0.2409,
+      "step": 9094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.173295454545455e-06,
+      "loss": 0.334,
+      "step": 9095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.168560606060607e-06,
+      "loss": 0.2351,
+      "step": 9096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.163825757575758e-06,
+      "loss": 0.3799,
+      "step": 9097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.15909090909091e-06,
+      "loss": 0.2593,
+      "step": 9098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.154356060606061e-06,
+      "loss": 0.4806,
+      "step": 9099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.149621212121213e-06,
+      "loss": 0.452,
+      "step": 9100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.144886363636364e-06,
+      "loss": 0.3165,
+      "step": 9101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.140151515151515e-06,
+      "loss": 0.3159,
+      "step": 9102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.135416666666667e-06,
+      "loss": 0.408,
+      "step": 9103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.130681818181818e-06,
+      "loss": 0.3998,
+      "step": 9104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.12594696969697e-06,
+      "loss": 0.3232,
+      "step": 9105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1212121212121215e-06,
+      "loss": 0.5246,
+      "step": 9106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.116477272727274e-06,
+      "loss": 0.3911,
+      "step": 9107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.111742424242425e-06,
+      "loss": 0.2836,
+      "step": 9108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.107007575757577e-06,
+      "loss": 0.4501,
+      "step": 9109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.102272727272728e-06,
+      "loss": 0.3655,
+      "step": 9110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.0975378787878785e-06,
+      "loss": 0.3772,
+      "step": 9111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.0928030303030306e-06,
+      "loss": 0.4379,
+      "step": 9112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.088068181818182e-06,
+      "loss": 0.3553,
+      "step": 9113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.083333333333334e-06,
+      "loss": 0.3509,
+      "step": 9114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.078598484848485e-06,
+      "loss": 0.3131,
+      "step": 9115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.073863636363637e-06,
+      "loss": 0.4375,
+      "step": 9116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.069128787878788e-06,
+      "loss": 0.4361,
+      "step": 9117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.06439393939394e-06,
+      "loss": 0.2738,
+      "step": 9118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.059659090909092e-06,
+      "loss": 0.3856,
+      "step": 9119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.054924242424244e-06,
+      "loss": 0.3374,
+      "step": 9120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.050189393939394e-06,
+      "loss": 0.4593,
+      "step": 9121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.045454545454545e-06,
+      "loss": 0.3663,
+      "step": 9122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.040719696969697e-06,
+      "loss": 0.3379,
+      "step": 9123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.035984848484849e-06,
+      "loss": 0.3572,
+      "step": 9124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.031250000000001e-06,
+      "loss": 0.4307,
+      "step": 9125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.026515151515152e-06,
+      "loss": 0.237,
+      "step": 9126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.021780303030304e-06,
+      "loss": 0.3786,
+      "step": 9127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.017045454545455e-06,
+      "loss": 0.3692,
+      "step": 9128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.012310606060607e-06,
+      "loss": 0.3724,
+      "step": 9129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.007575757575758e-06,
+      "loss": 0.4764,
+      "step": 9130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.002840909090909e-06,
+      "loss": 0.4565,
+      "step": 9131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.998106060606061e-06,
+      "loss": 0.3792,
+      "step": 9132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.993371212121212e-06,
+      "loss": 0.3384,
+      "step": 9133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.988636363636364e-06,
+      "loss": 0.3439,
+      "step": 9134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.983901515151515e-06,
+      "loss": 0.3354,
+      "step": 9135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.9791666666666675e-06,
+      "loss": 0.4987,
+      "step": 9136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.974431818181819e-06,
+      "loss": 0.3116,
+      "step": 9137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 0.2814,
+      "step": 9138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.964962121212122e-06,
+      "loss": 0.4267,
+      "step": 9139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.960227272727272e-06,
+      "loss": 0.3717,
+      "step": 9140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.955492424242424e-06,
+      "loss": 0.472,
+      "step": 9141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.950757575757576e-06,
+      "loss": 0.3544,
+      "step": 9142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.946022727272728e-06,
+      "loss": 0.4289,
+      "step": 9143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.941287878787879e-06,
+      "loss": 0.4202,
+      "step": 9144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.936553030303031e-06,
+      "loss": 0.3367,
+      "step": 9145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.931818181818182e-06,
+      "loss": 0.4026,
+      "step": 9146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.927083333333334e-06,
+      "loss": 0.5483,
+      "step": 9147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.9223484848484855e-06,
+      "loss": 0.2649,
+      "step": 9148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.9176136363636376e-06,
+      "loss": 0.4251,
+      "step": 9149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.912878787878788e-06,
+      "loss": 0.3537,
+      "step": 9150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.908143939393939e-06,
+      "loss": 0.3646,
+      "step": 9151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.903409090909091e-06,
+      "loss": 0.3437,
+      "step": 9152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8986742424242424e-06,
+      "loss": 0.3921,
+      "step": 9153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8939393939393945e-06,
+      "loss": 0.4349,
+      "step": 9154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.889204545454546e-06,
+      "loss": 0.3199,
+      "step": 9155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.884469696969698e-06,
+      "loss": 0.3328,
+      "step": 9156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.879734848484849e-06,
+      "loss": 0.3719,
+      "step": 9157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.875000000000001e-06,
+      "loss": 0.4593,
+      "step": 9158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8702651515151515e-06,
+      "loss": 0.4744,
+      "step": 9159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.865530303030303e-06,
+      "loss": 0.4483,
+      "step": 9160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.860795454545455e-06,
+      "loss": 0.3995,
+      "step": 9161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.856060606060606e-06,
+      "loss": 0.2533,
+      "step": 9162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.851325757575758e-06,
+      "loss": 0.3299,
+      "step": 9163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.846590909090909e-06,
+      "loss": 0.4127,
+      "step": 9164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.841856060606061e-06,
+      "loss": 0.3776,
+      "step": 9165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8371212121212125e-06,
+      "loss": 0.362,
+      "step": 9166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.832386363636365e-06,
+      "loss": 0.3786,
+      "step": 9167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.827651515151516e-06,
+      "loss": 0.3532,
+      "step": 9168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.822916666666666e-06,
+      "loss": 0.3706,
+      "step": 9169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.818181818181818e-06,
+      "loss": 0.3568,
+      "step": 9170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8134469696969695e-06,
+      "loss": 0.3808,
+      "step": 9171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8087121212121216e-06,
+      "loss": 0.4437,
+      "step": 9172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.803977272727273e-06,
+      "loss": 0.3473,
+      "step": 9173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.799242424242425e-06,
+      "loss": 0.3426,
+      "step": 9174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.794507575757576e-06,
+      "loss": 0.3026,
+      "step": 9175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.789772727272728e-06,
+      "loss": 0.4781,
+      "step": 9176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.785037878787879e-06,
+      "loss": 0.2764,
+      "step": 9177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.7803030303030314e-06,
+      "loss": 0.4363,
+      "step": 9178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.775568181818182e-06,
+      "loss": 0.3645,
+      "step": 9179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.770833333333333e-06,
+      "loss": 0.2477,
+      "step": 9180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.766098484848485e-06,
+      "loss": 0.3231,
+      "step": 9181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.761363636363636e-06,
+      "loss": 0.3295,
+      "step": 9182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.756628787878788e-06,
+      "loss": 0.463,
+      "step": 9183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.75189393939394e-06,
+      "loss": 0.399,
+      "step": 9184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.747159090909092e-06,
+      "loss": 0.5273,
+      "step": 9185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.742424242424243e-06,
+      "loss": 0.3609,
+      "step": 9186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.737689393939395e-06,
+      "loss": 0.4373,
+      "step": 9187
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.732954545454546e-06,
+      "loss": 0.4381,
+      "step": 9188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.7282196969696965e-06,
+      "loss": 0.3042,
+      "step": 9189
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.723484848484849e-06,
+      "loss": 0.2456,
+      "step": 9190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.71875e-06,
+      "loss": 0.3136,
+      "step": 9191
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.714015151515152e-06,
+      "loss": 0.3732,
+      "step": 9192
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.709280303030303e-06,
+      "loss": 0.329,
+      "step": 9193
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.704545454545455e-06,
+      "loss": 0.2361,
+      "step": 9194
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.699810606060606e-06,
+      "loss": 0.4064,
+      "step": 9195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.6950757575757585e-06,
+      "loss": 0.355,
+      "step": 9196
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.69034090909091e-06,
+      "loss": 0.4471,
+      "step": 9197
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.68560606060606e-06,
+      "loss": 0.4416,
+      "step": 9198
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.680871212121212e-06,
+      "loss": 0.4018,
+      "step": 9199
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.676136363636363e-06,
+      "loss": 0.3399,
+      "step": 9200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.6714015151515154e-06,
+      "loss": 0.3459,
+      "step": 9201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.3568,
+      "step": 9202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.661931818181819e-06,
+      "loss": 0.4399,
+      "step": 9203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.65719696969697e-06,
+      "loss": 0.3626,
+      "step": 9204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.652462121212122e-06,
+      "loss": 0.4682,
+      "step": 9205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.647727272727273e-06,
+      "loss": 0.372,
+      "step": 9206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.642992424242425e-06,
+      "loss": 0.2475,
+      "step": 9207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.638257575757576e-06,
+      "loss": 0.2229,
+      "step": 9208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.633522727272727e-06,
+      "loss": 0.3768,
+      "step": 9209
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.628787878787879e-06,
+      "loss": 0.4705,
+      "step": 9210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.62405303030303e-06,
+      "loss": 0.3724,
+      "step": 9211
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.619318181818182e-06,
+      "loss": 0.3945,
+      "step": 9212
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.6145833333333335e-06,
+      "loss": 0.5269,
+      "step": 9213
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.6098484848484855e-06,
+      "loss": 0.244,
+      "step": 9214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.605113636363637e-06,
+      "loss": 0.2938,
+      "step": 9215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.600378787878789e-06,
+      "loss": 0.3193,
+      "step": 9216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.59564393939394e-06,
+      "loss": 0.311,
+      "step": 9217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.59090909090909e-06,
+      "loss": 0.3725,
+      "step": 9218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5861742424242425e-06,
+      "loss": 0.3373,
+      "step": 9219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.581439393939394e-06,
+      "loss": 0.2619,
+      "step": 9220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.576704545454546e-06,
+      "loss": 0.4245,
+      "step": 9221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.571969696969697e-06,
+      "loss": 0.4124,
+      "step": 9222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.567234848484849e-06,
+      "loss": 0.3252,
+      "step": 9223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5625e-06,
+      "loss": 0.4757,
+      "step": 9224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.557765151515152e-06,
+      "loss": 0.3759,
+      "step": 9225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5530303030303036e-06,
+      "loss": 0.4175,
+      "step": 9226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.548295454545454e-06,
+      "loss": 0.3497,
+      "step": 9227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.543560606060606e-06,
+      "loss": 0.3257,
+      "step": 9228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.538825757575757e-06,
+      "loss": 0.2518,
+      "step": 9229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.534090909090909e-06,
+      "loss": 0.3029,
+      "step": 9230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5293560606060605e-06,
+      "loss": 0.3669,
+      "step": 9231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5246212121212126e-06,
+      "loss": 0.2823,
+      "step": 9232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.519886363636364e-06,
+      "loss": 0.3899,
+      "step": 9233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.515151515151516e-06,
+      "loss": 0.3784,
+      "step": 9234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.510416666666667e-06,
+      "loss": 0.4594,
+      "step": 9235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.505681818181819e-06,
+      "loss": 0.4844,
+      "step": 9236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.5009469696969695e-06,
+      "loss": 0.2909,
+      "step": 9237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.496212121212121e-06,
+      "loss": 0.3304,
+      "step": 9238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.491477272727273e-06,
+      "loss": 0.2943,
+      "step": 9239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.486742424242424e-06,
+      "loss": 0.4689,
+      "step": 9240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.482007575757576e-06,
+      "loss": 0.247,
+      "step": 9241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.477272727272727e-06,
+      "loss": 0.3661,
+      "step": 9242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.472537878787879e-06,
+      "loss": 0.2967,
+      "step": 9243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.467803030303031e-06,
+      "loss": 0.3468,
+      "step": 9244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.463068181818183e-06,
+      "loss": 0.4465,
+      "step": 9245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.458333333333334e-06,
+      "loss": 0.2911,
+      "step": 9246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.453598484848484e-06,
+      "loss": 0.425,
+      "step": 9247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.448863636363636e-06,
+      "loss": 0.3603,
+      "step": 9248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.4441287878787876e-06,
+      "loss": 0.3165,
+      "step": 9249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.43939393939394e-06,
+      "loss": 0.5304,
+      "step": 9250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.434659090909091e-06,
+      "loss": 0.3443,
+      "step": 9251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.429924242424243e-06,
+      "loss": 0.5284,
+      "step": 9252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.425189393939394e-06,
+      "loss": 0.3093,
+      "step": 9253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.420454545454546e-06,
+      "loss": 0.5021,
+      "step": 9254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.415719696969697e-06,
+      "loss": 0.3596,
+      "step": 9255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.4109848484848495e-06,
+      "loss": 0.3234,
+      "step": 9256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.40625e-06,
+      "loss": 0.4704,
+      "step": 9257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.401515151515151e-06,
+      "loss": 0.4096,
+      "step": 9258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.396780303030303e-06,
+      "loss": 0.3658,
+      "step": 9259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.392045454545454e-06,
+      "loss": 0.3376,
+      "step": 9260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.3873106060606064e-06,
+      "loss": 0.459,
+      "step": 9261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.382575757575758e-06,
+      "loss": 0.361,
+      "step": 9262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.37784090909091e-06,
+      "loss": 0.3882,
+      "step": 9263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.373106060606061e-06,
+      "loss": 0.3689,
+      "step": 9264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.368371212121213e-06,
+      "loss": 0.2588,
+      "step": 9265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.363636363636363e-06,
+      "loss": 0.3564,
+      "step": 9266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.358901515151515e-06,
+      "loss": 0.323,
+      "step": 9267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.354166666666667e-06,
+      "loss": 0.3443,
+      "step": 9268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.349431818181818e-06,
+      "loss": 0.4261,
+      "step": 9269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.34469696969697e-06,
+      "loss": 0.5967,
+      "step": 9270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.339962121212121e-06,
+      "loss": 0.5218,
+      "step": 9271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.335227272727273e-06,
+      "loss": 0.3533,
+      "step": 9272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.3304924242424245e-06,
+      "loss": 0.3839,
+      "step": 9273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.3257575757575765e-06,
+      "loss": 0.3311,
+      "step": 9274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.321022727272729e-06,
+      "loss": 0.4738,
+      "step": 9275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.316287878787878e-06,
+      "loss": 0.3579,
+      "step": 9276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.31155303030303e-06,
+      "loss": 0.3591,
+      "step": 9277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.306818181818181e-06,
+      "loss": 0.324,
+      "step": 9278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.3020833333333335e-06,
+      "loss": 0.3914,
+      "step": 9279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.297348484848485e-06,
+      "loss": 0.3012,
+      "step": 9280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.292613636363637e-06,
+      "loss": 0.4007,
+      "step": 9281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.287878787878789e-06,
+      "loss": 0.4282,
+      "step": 9282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.28314393939394e-06,
+      "loss": 0.3784,
+      "step": 9283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.278409090909092e-06,
+      "loss": 0.4575,
+      "step": 9284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.273674242424243e-06,
+      "loss": 0.3228,
+      "step": 9285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.268939393939394e-06,
+      "loss": 0.3866,
+      "step": 9286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.264204545454545e-06,
+      "loss": 0.2679,
+      "step": 9287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.259469696969697e-06,
+      "loss": 0.3712,
+      "step": 9288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.254734848484848e-06,
+      "loss": 0.4036,
+      "step": 9289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.25e-06,
+      "loss": 0.384,
+      "step": 9290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.245265151515152e-06,
+      "loss": 0.4186,
+      "step": 9291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.240530303030304e-06,
+      "loss": 0.4504,
+      "step": 9292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.235795454545455e-06,
+      "loss": 0.3831,
+      "step": 9293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.231060606060606e-06,
+      "loss": 0.386,
+      "step": 9294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.226325757575758e-06,
+      "loss": 0.3864,
+      "step": 9295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.221590909090909e-06,
+      "loss": 0.398,
+      "step": 9296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.216856060606061e-06,
+      "loss": 0.3608,
+      "step": 9297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.212121212121212e-06,
+      "loss": 0.4404,
+      "step": 9298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.207386363636364e-06,
+      "loss": 0.3106,
+      "step": 9299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.202651515151516e-06,
+      "loss": 0.3888,
+      "step": 9300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.197916666666667e-06,
+      "loss": 0.3798,
+      "step": 9301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.193181818181819e-06,
+      "loss": 0.4511,
+      "step": 9302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.1884469696969696e-06,
+      "loss": 0.1917,
+      "step": 9303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.183712121212122e-06,
+      "loss": 0.3289,
+      "step": 9304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.178977272727273e-06,
+      "loss": 0.3755,
+      "step": 9305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.174242424242425e-06,
+      "loss": 0.4179,
+      "step": 9306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.169507575757576e-06,
+      "loss": 0.4527,
+      "step": 9307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.164772727272727e-06,
+      "loss": 0.3149,
+      "step": 9308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.160037878787879e-06,
+      "loss": 0.2114,
+      "step": 9309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.155303030303031e-06,
+      "loss": 0.3297,
+      "step": 9310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.150568181818183e-06,
+      "loss": 0.2857,
+      "step": 9311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.145833333333333e-06,
+      "loss": 0.2766,
+      "step": 9312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.141098484848485e-06,
+      "loss": 0.2717,
+      "step": 9313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.136363636363636e-06,
+      "loss": 0.4171,
+      "step": 9314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.1316287878787884e-06,
+      "loss": 0.4806,
+      "step": 9315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.12689393939394e-06,
+      "loss": 0.2791,
+      "step": 9316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.122159090909091e-06,
+      "loss": 0.4452,
+      "step": 9317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.117424242424243e-06,
+      "loss": 0.2997,
+      "step": 9318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.112689393939394e-06,
+      "loss": 0.4238,
+      "step": 9319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.107954545454546e-06,
+      "loss": 0.2396,
+      "step": 9320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.1032196969696974e-06,
+      "loss": 0.3226,
+      "step": 9321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.098484848484849e-06,
+      "loss": 0.2865,
+      "step": 9322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.09375e-06,
+      "loss": 0.3022,
+      "step": 9323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.089015151515152e-06,
+      "loss": 0.4023,
+      "step": 9324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.084280303030303e-06,
+      "loss": 0.3534,
+      "step": 9325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.079545454545455e-06,
+      "loss": 0.248,
+      "step": 9326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0748106060606065e-06,
+      "loss": 0.3525,
+      "step": 9327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.070075757575758e-06,
+      "loss": 0.5029,
+      "step": 9328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.06534090909091e-06,
+      "loss": 0.3454,
+      "step": 9329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.3756,
+      "step": 9330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.055871212121213e-06,
+      "loss": 0.3632,
+      "step": 9331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.051136363636363e-06,
+      "loss": 0.3874,
+      "step": 9332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0464015151515155e-06,
+      "loss": 0.4646,
+      "step": 9333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.041666666666667e-06,
+      "loss": 0.4286,
+      "step": 9334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.036931818181819e-06,
+      "loss": 0.4191,
+      "step": 9335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.03219696969697e-06,
+      "loss": 0.2716,
+      "step": 9336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.027462121212121e-06,
+      "loss": 0.3479,
+      "step": 9337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.022727272727273e-06,
+      "loss": 0.4522,
+      "step": 9338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0179924242424245e-06,
+      "loss": 0.3652,
+      "step": 9339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0132575757575766e-06,
+      "loss": 0.3418,
+      "step": 9340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.008522727272727e-06,
+      "loss": 0.4856,
+      "step": 9341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.003787878787879e-06,
+      "loss": 0.3522,
+      "step": 9342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.99905303030303e-06,
+      "loss": 0.4406,
+      "step": 9343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.994318181818182e-06,
+      "loss": 0.4205,
+      "step": 9344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9895833333333335e-06,
+      "loss": 0.3281,
+      "step": 9345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.984848484848485e-06,
+      "loss": 0.3929,
+      "step": 9346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.980113636363637e-06,
+      "loss": 0.3627,
+      "step": 9347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.975378787878788e-06,
+      "loss": 0.3097,
+      "step": 9348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.97064393939394e-06,
+      "loss": 0.3876,
+      "step": 9349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.965909090909091e-06,
+      "loss": 0.3904,
+      "step": 9350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9611742424242425e-06,
+      "loss": 0.3589,
+      "step": 9351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.956439393939394e-06,
+      "loss": 0.3883,
+      "step": 9352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.951704545454546e-06,
+      "loss": 0.2073,
+      "step": 9353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.946969696969697e-06,
+      "loss": 0.427,
+      "step": 9354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.942234848484849e-06,
+      "loss": 0.5523,
+      "step": 9355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9375e-06,
+      "loss": 0.3504,
+      "step": 9356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9327651515151515e-06,
+      "loss": 0.51,
+      "step": 9357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.928030303030304e-06,
+      "loss": 0.4181,
+      "step": 9358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.923295454545455e-06,
+      "loss": 0.27,
+      "step": 9359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.918560606060607e-06,
+      "loss": 0.3514,
+      "step": 9360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.913825757575757e-06,
+      "loss": 0.409,
+      "step": 9361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.909090909090909e-06,
+      "loss": 0.3593,
+      "step": 9362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.9043560606060606e-06,
+      "loss": 0.3317,
+      "step": 9363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.899621212121213e-06,
+      "loss": 0.386,
+      "step": 9364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894886363636364e-06,
+      "loss": 0.4387,
+      "step": 9365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.890151515151515e-06,
+      "loss": 0.3804,
+      "step": 9366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.885416666666667e-06,
+      "loss": 0.2104,
+      "step": 9367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.880681818181818e-06,
+      "loss": 0.352,
+      "step": 9368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8759469696969704e-06,
+      "loss": 0.3331,
+      "step": 9369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.871212121212122e-06,
+      "loss": 0.3306,
+      "step": 9370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.866477272727273e-06,
+      "loss": 0.2665,
+      "step": 9371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.861742424242424e-06,
+      "loss": 0.3924,
+      "step": 9372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.857007575757576e-06,
+      "loss": 0.3181,
+      "step": 9373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.852272727272727e-06,
+      "loss": 0.3914,
+      "step": 9374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.847537878787879e-06,
+      "loss": 0.2646,
+      "step": 9375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.842803030303031e-06,
+      "loss": 0.3319,
+      "step": 9376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.838068181818182e-06,
+      "loss": 0.5145,
+      "step": 9377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 0.2557,
+      "step": 9378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.828598484848485e-06,
+      "loss": 0.3478,
+      "step": 9379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.823863636363636e-06,
+      "loss": 0.2616,
+      "step": 9380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.819128787878788e-06,
+      "loss": 0.4214,
+      "step": 9381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.81439393939394e-06,
+      "loss": 0.3799,
+      "step": 9382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.809659090909091e-06,
+      "loss": 0.3312,
+      "step": 9383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.804924242424243e-06,
+      "loss": 0.2963,
+      "step": 9384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.800189393939394e-06,
+      "loss": 0.3866,
+      "step": 9385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.795454545454545e-06,
+      "loss": 0.3317,
+      "step": 9386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.7907196969696975e-06,
+      "loss": 0.3681,
+      "step": 9387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.785984848484849e-06,
+      "loss": 0.3706,
+      "step": 9388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.781250000000001e-06,
+      "loss": 0.3286,
+      "step": 9389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.776515151515151e-06,
+      "loss": 0.4779,
+      "step": 9390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.771780303030303e-06,
+      "loss": 0.4172,
+      "step": 9391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7670454545454544e-06,
+      "loss": 0.3787,
+      "step": 9392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7623106060606065e-06,
+      "loss": 0.2928,
+      "step": 9393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7575757575757586e-06,
+      "loss": 0.2877,
+      "step": 9394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.752840909090909e-06,
+      "loss": 0.328,
+      "step": 9395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.748106060606061e-06,
+      "loss": 0.2794,
+      "step": 9396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.743371212121212e-06,
+      "loss": 0.3125,
+      "step": 9397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.738636363636364e-06,
+      "loss": 0.3574,
+      "step": 9398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7339015151515155e-06,
+      "loss": 0.3612,
+      "step": 9399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.729166666666667e-06,
+      "loss": 0.339,
+      "step": 9400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.724431818181818e-06,
+      "loss": 0.348,
+      "step": 9401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.71969696969697e-06,
+      "loss": 0.1713,
+      "step": 9402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.714962121212122e-06,
+      "loss": 0.3324,
+      "step": 9403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.710227272727273e-06,
+      "loss": 0.4354,
+      "step": 9404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7054924242424245e-06,
+      "loss": 0.4207,
+      "step": 9405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.700757575757576e-06,
+      "loss": 0.4543,
+      "step": 9406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.696022727272728e-06,
+      "loss": 0.3476,
+      "step": 9407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.691287878787879e-06,
+      "loss": 0.377,
+      "step": 9408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.68655303030303e-06,
+      "loss": 0.327,
+      "step": 9409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.681818181818182e-06,
+      "loss": 0.3492,
+      "step": 9410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.6770833333333335e-06,
+      "loss": 0.3856,
+      "step": 9411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.672348484848486e-06,
+      "loss": 0.2726,
+      "step": 9412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.667613636363637e-06,
+      "loss": 0.3463,
+      "step": 9413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.662878787878788e-06,
+      "loss": 0.2049,
+      "step": 9414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.658143939393939e-06,
+      "loss": 0.5773,
+      "step": 9415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.653409090909091e-06,
+      "loss": 0.4284,
+      "step": 9416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.6486742424242426e-06,
+      "loss": 0.4956,
+      "step": 9417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.643939393939395e-06,
+      "loss": 0.3994,
+      "step": 9418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.639204545454546e-06,
+      "loss": 0.5181,
+      "step": 9419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.634469696969697e-06,
+      "loss": 0.3974,
+      "step": 9420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.629734848484849e-06,
+      "loss": 0.4368,
+      "step": 9421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.625e-06,
+      "loss": 0.3952,
+      "step": 9422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.620265151515152e-06,
+      "loss": 0.4322,
+      "step": 9423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.615530303030303e-06,
+      "loss": 0.4963,
+      "step": 9424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.610795454545455e-06,
+      "loss": 0.3747,
+      "step": 9425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.606060606060606e-06,
+      "loss": 0.3747,
+      "step": 9426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.601325757575758e-06,
+      "loss": 0.3158,
+      "step": 9427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.596590909090909e-06,
+      "loss": 0.3489,
+      "step": 9428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.591856060606061e-06,
+      "loss": 0.4289,
+      "step": 9429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.587121212121213e-06,
+      "loss": 0.3749,
+      "step": 9430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.582386363636364e-06,
+      "loss": 0.4232,
+      "step": 9431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.577651515151516e-06,
+      "loss": 0.3427,
+      "step": 9432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.572916666666667e-06,
+      "loss": 0.3534,
+      "step": 9433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.568181818181818e-06,
+      "loss": 0.3184,
+      "step": 9434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.56344696969697e-06,
+      "loss": 0.417,
+      "step": 9435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.558712121212122e-06,
+      "loss": 0.4719,
+      "step": 9436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.553977272727273e-06,
+      "loss": 0.3489,
+      "step": 9437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.549242424242425e-06,
+      "loss": 0.3488,
+      "step": 9438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.544507575757576e-06,
+      "loss": 0.442,
+      "step": 9439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.539772727272727e-06,
+      "loss": 0.3517,
+      "step": 9440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5350378787878795e-06,
+      "loss": 0.3889,
+      "step": 9441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.530303030303031e-06,
+      "loss": 0.3582,
+      "step": 9442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.525568181818182e-06,
+      "loss": 0.2939,
+      "step": 9443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.520833333333333e-06,
+      "loss": 0.2749,
+      "step": 9444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.516098484848485e-06,
+      "loss": 0.3874,
+      "step": 9445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.511363636363636e-06,
+      "loss": 0.4316,
+      "step": 9446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5066287878787885e-06,
+      "loss": 0.4117,
+      "step": 9447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.50189393939394e-06,
+      "loss": 0.35,
+      "step": 9448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.497159090909091e-06,
+      "loss": 0.3957,
+      "step": 9449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.492424242424243e-06,
+      "loss": 0.4016,
+      "step": 9450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.487689393939394e-06,
+      "loss": 0.4141,
+      "step": 9451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.482954545454546e-06,
+      "loss": 0.4709,
+      "step": 9452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.478219696969697e-06,
+      "loss": 0.3121,
+      "step": 9453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.473484848484849e-06,
+      "loss": 0.3393,
+      "step": 9454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.46875e-06,
+      "loss": 0.3171,
+      "step": 9455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.464015151515152e-06,
+      "loss": 0.2231,
+      "step": 9456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.459280303030303e-06,
+      "loss": 0.3419,
+      "step": 9457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.4165,
+      "step": 9458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4498106060606065e-06,
+      "loss": 0.3434,
+      "step": 9459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.445075757575758e-06,
+      "loss": 0.3122,
+      "step": 9460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.44034090909091e-06,
+      "loss": 0.3437,
+      "step": 9461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.435606060606061e-06,
+      "loss": 0.4121,
+      "step": 9462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.430871212121212e-06,
+      "loss": 0.3333,
+      "step": 9463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4261363636363635e-06,
+      "loss": 0.3833,
+      "step": 9464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4214015151515155e-06,
+      "loss": 0.381,
+      "step": 9465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.416666666666667e-06,
+      "loss": 0.3802,
+      "step": 9466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.411931818181819e-06,
+      "loss": 0.4653,
+      "step": 9467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.40719696969697e-06,
+      "loss": 0.347,
+      "step": 9468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.402462121212121e-06,
+      "loss": 0.2983,
+      "step": 9469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.397727272727273e-06,
+      "loss": 0.4896,
+      "step": 9470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3929924242424246e-06,
+      "loss": 0.2551,
+      "step": 9471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.388257575757576e-06,
+      "loss": 0.3931,
+      "step": 9472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.383522727272727e-06,
+      "loss": 0.3722,
+      "step": 9473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.378787878787879e-06,
+      "loss": 0.2844,
+      "step": 9474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.37405303030303e-06,
+      "loss": 0.4139,
+      "step": 9475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.369318181818182e-06,
+      "loss": 0.4254,
+      "step": 9476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3645833333333336e-06,
+      "loss": 0.3938,
+      "step": 9477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.359848484848485e-06,
+      "loss": 0.3764,
+      "step": 9478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.355113636363637e-06,
+      "loss": 0.4144,
+      "step": 9479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.350378787878788e-06,
+      "loss": 0.3206,
+      "step": 9480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.34564393939394e-06,
+      "loss": 0.2866,
+      "step": 9481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3409090909090905e-06,
+      "loss": 0.3057,
+      "step": 9482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.336174242424243e-06,
+      "loss": 0.2779,
+      "step": 9483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.331439393939394e-06,
+      "loss": 0.324,
+      "step": 9484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.326704545454546e-06,
+      "loss": 0.323,
+      "step": 9485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.321969696969697e-06,
+      "loss": 0.4485,
+      "step": 9486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.317234848484848e-06,
+      "loss": 0.4879,
+      "step": 9487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3125e-06,
+      "loss": 0.3543,
+      "step": 9488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.307765151515152e-06,
+      "loss": 0.4285,
+      "step": 9489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.303030303030304e-06,
+      "loss": 0.3502,
+      "step": 9490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.298295454545455e-06,
+      "loss": 0.4612,
+      "step": 9491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.293560606060606e-06,
+      "loss": 0.5743,
+      "step": 9492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.288825757575757e-06,
+      "loss": 0.325,
+      "step": 9493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.284090909090909e-06,
+      "loss": 0.3248,
+      "step": 9494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.279356060606061e-06,
+      "loss": 0.3527,
+      "step": 9495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.274621212121213e-06,
+      "loss": 0.3528,
+      "step": 9496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.269886363636364e-06,
+      "loss": 0.3619,
+      "step": 9497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.265151515151515e-06,
+      "loss": 0.3407,
+      "step": 9498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.260416666666667e-06,
+      "loss": 0.3611,
+      "step": 9499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.255681818181818e-06,
+      "loss": 0.3011,
+      "step": 9500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2509469696969705e-06,
+      "loss": 0.2349,
+      "step": 9501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.246212121212121e-06,
+      "loss": 0.4215,
+      "step": 9502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.241477272727273e-06,
+      "loss": 0.3493,
+      "step": 9503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.236742424242424e-06,
+      "loss": 0.5611,
+      "step": 9504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.232007575757576e-06,
+      "loss": 0.4857,
+      "step": 9505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2272727272727274e-06,
+      "loss": 0.3451,
+      "step": 9506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.222537878787879e-06,
+      "loss": 0.2781,
+      "step": 9507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.217803030303031e-06,
+      "loss": 0.4047,
+      "step": 9508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.213068181818182e-06,
+      "loss": 0.2275,
+      "step": 9509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.208333333333334e-06,
+      "loss": 0.4995,
+      "step": 9510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.203598484848484e-06,
+      "loss": 0.3581,
+      "step": 9511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1988636363636364e-06,
+      "loss": 0.5115,
+      "step": 9512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1941287878787885e-06,
+      "loss": 0.4743,
+      "step": 9513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.18939393939394e-06,
+      "loss": 0.284,
+      "step": 9514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.184659090909092e-06,
+      "loss": 0.399,
+      "step": 9515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.179924242424242e-06,
+      "loss": 0.3056,
+      "step": 9516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.175189393939394e-06,
+      "loss": 0.4301,
+      "step": 9517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1704545454545455e-06,
+      "loss": 0.2717,
+      "step": 9518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1657196969696975e-06,
+      "loss": 0.2862,
+      "step": 9519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.160984848484849e-06,
+      "loss": 0.3735,
+      "step": 9520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.15625e-06,
+      "loss": 0.3829,
+      "step": 9521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.151515151515152e-06,
+      "loss": 0.4093,
+      "step": 9522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.146780303030303e-06,
+      "loss": 0.4529,
+      "step": 9523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.142045454545455e-06,
+      "loss": 0.2992,
+      "step": 9524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1373106060606065e-06,
+      "loss": 0.3938,
+      "step": 9525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.132575757575758e-06,
+      "loss": 0.3914,
+      "step": 9526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.127840909090909e-06,
+      "loss": 0.3856,
+      "step": 9527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.123106060606061e-06,
+      "loss": 0.2866,
+      "step": 9528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.118371212121212e-06,
+      "loss": 0.3468,
+      "step": 9529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.113636363636364e-06,
+      "loss": 0.2127,
+      "step": 9530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1089015151515156e-06,
+      "loss": 0.378,
+      "step": 9531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.104166666666667e-06,
+      "loss": 0.3085,
+      "step": 9532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.099431818181819e-06,
+      "loss": 0.405,
+      "step": 9533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.09469696969697e-06,
+      "loss": 0.3944,
+      "step": 9534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.089962121212122e-06,
+      "loss": 0.3706,
+      "step": 9535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0852272727272725e-06,
+      "loss": 0.3526,
+      "step": 9536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.080492424242425e-06,
+      "loss": 0.3035,
+      "step": 9537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.075757575757576e-06,
+      "loss": 0.3514,
+      "step": 9538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.071022727272728e-06,
+      "loss": 0.4393,
+      "step": 9539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.066287878787879e-06,
+      "loss": 0.3322,
+      "step": 9540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.06155303030303e-06,
+      "loss": 0.3543,
+      "step": 9541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.056818181818182e-06,
+      "loss": 0.3833,
+      "step": 9542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.052083333333334e-06,
+      "loss": 0.3479,
+      "step": 9543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.047348484848486e-06,
+      "loss": 0.3105,
+      "step": 9544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.042613636363636e-06,
+      "loss": 0.2122,
+      "step": 9545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.037878787878788e-06,
+      "loss": 0.4206,
+      "step": 9546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.033143939393939e-06,
+      "loss": 0.4799,
+      "step": 9547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.028409090909091e-06,
+      "loss": 0.3135,
+      "step": 9548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.023674242424243e-06,
+      "loss": 0.2887,
+      "step": 9549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.018939393939394e-06,
+      "loss": 0.3954,
+      "step": 9550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.014204545454546e-06,
+      "loss": 0.2887,
+      "step": 9551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.009469696969697e-06,
+      "loss": 0.4317,
+      "step": 9552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.004734848484849e-06,
+      "loss": 0.3706,
+      "step": 9553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5e-06,
+      "loss": 0.3099,
+      "step": 9554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.995265151515152e-06,
+      "loss": 0.5197,
+      "step": 9555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.990530303030303e-06,
+      "loss": 0.3536,
+      "step": 9556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.985795454545455e-06,
+      "loss": 0.4025,
+      "step": 9557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.981060606060606e-06,
+      "loss": 0.4384,
+      "step": 9558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.976325757575758e-06,
+      "loss": 0.263,
+      "step": 9559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9715909090909094e-06,
+      "loss": 0.4137,
+      "step": 9560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.966856060606061e-06,
+      "loss": 0.4626,
+      "step": 9561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.962121212121213e-06,
+      "loss": 0.2697,
+      "step": 9562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.957386363636364e-06,
+      "loss": 0.4581,
+      "step": 9563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.952651515151516e-06,
+      "loss": 0.519,
+      "step": 9564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.947916666666666e-06,
+      "loss": 0.4363,
+      "step": 9565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9431818181818184e-06,
+      "loss": 0.3054,
+      "step": 9566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.93844696969697e-06,
+      "loss": 0.3072,
+      "step": 9567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.933712121212122e-06,
+      "loss": 0.2391,
+      "step": 9568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.928977272727273e-06,
+      "loss": 0.4373,
+      "step": 9569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.924242424242424e-06,
+      "loss": 0.2747,
+      "step": 9570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.919507575757576e-06,
+      "loss": 0.3277,
+      "step": 9571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9147727272727275e-06,
+      "loss": 0.3946,
+      "step": 9572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9100378787878795e-06,
+      "loss": 0.2779,
+      "step": 9573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.90530303030303e-06,
+      "loss": 0.3181,
+      "step": 9574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.900568181818182e-06,
+      "loss": 0.2855,
+      "step": 9575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.895833333333333e-06,
+      "loss": 0.2958,
+      "step": 9576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.891098484848485e-06,
+      "loss": 0.3379,
+      "step": 9577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8863636363636365e-06,
+      "loss": 0.3739,
+      "step": 9578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.881628787878788e-06,
+      "loss": 0.4592,
+      "step": 9579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.87689393939394e-06,
+      "loss": 0.4923,
+      "step": 9580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.872159090909091e-06,
+      "loss": 0.4541,
+      "step": 9581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.867424242424243e-06,
+      "loss": 0.407,
+      "step": 9582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.862689393939394e-06,
+      "loss": 0.3462,
+      "step": 9583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8579545454545455e-06,
+      "loss": 0.5272,
+      "step": 9584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.853219696969697e-06,
+      "loss": 0.2885,
+      "step": 9585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 0.2617,
+      "step": 9586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.84375e-06,
+      "loss": 0.2948,
+      "step": 9587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.839015151515152e-06,
+      "loss": 0.2631,
+      "step": 9588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.834280303030303e-06,
+      "loss": 0.307,
+      "step": 9589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8295454545454545e-06,
+      "loss": 0.4594,
+      "step": 9590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8248106060606066e-06,
+      "loss": 0.3179,
+      "step": 9591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.820075757575758e-06,
+      "loss": 0.2796,
+      "step": 9592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.81534090909091e-06,
+      "loss": 0.3768,
+      "step": 9593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.81060606060606e-06,
+      "loss": 0.3964,
+      "step": 9594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.805871212121212e-06,
+      "loss": 0.3351,
+      "step": 9595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8011363636363635e-06,
+      "loss": 0.3572,
+      "step": 9596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.796401515151516e-06,
+      "loss": 0.3381,
+      "step": 9597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.791666666666667e-06,
+      "loss": 0.328,
+      "step": 9598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.786931818181818e-06,
+      "loss": 0.2469,
+      "step": 9599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.78219696969697e-06,
+      "loss": 0.3795,
+      "step": 9600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.777462121212121e-06,
+      "loss": 0.3863,
+      "step": 9601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.772727272727273e-06,
+      "loss": 0.4077,
+      "step": 9602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.767992424242425e-06,
+      "loss": 0.4588,
+      "step": 9603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.763257575757576e-06,
+      "loss": 0.2307,
+      "step": 9604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.758522727272727e-06,
+      "loss": 0.3927,
+      "step": 9605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.753787878787879e-06,
+      "loss": 0.4993,
+      "step": 9606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.74905303030303e-06,
+      "loss": 0.278,
+      "step": 9607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7443181818181816e-06,
+      "loss": 0.3386,
+      "step": 9608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.739583333333334e-06,
+      "loss": 0.4059,
+      "step": 9609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.734848484848485e-06,
+      "loss": 0.3672,
+      "step": 9610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.730113636363637e-06,
+      "loss": 0.3343,
+      "step": 9611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.725378787878788e-06,
+      "loss": 0.2909,
+      "step": 9612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.720643939393939e-06,
+      "loss": 0.2412,
+      "step": 9613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7159090909090906e-06,
+      "loss": 0.2399,
+      "step": 9614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.711174242424243e-06,
+      "loss": 0.3851,
+      "step": 9615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.706439393939395e-06,
+      "loss": 0.404,
+      "step": 9616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701704545454546e-06,
+      "loss": 0.3413,
+      "step": 9617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.696969696969697e-06,
+      "loss": 0.2459,
+      "step": 9618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.692234848484848e-06,
+      "loss": 0.3579,
+      "step": 9619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 0.4113,
+      "step": 9620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.682765151515152e-06,
+      "loss": 0.4839,
+      "step": 9621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.678030303030304e-06,
+      "loss": 0.6261,
+      "step": 9622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.673295454545454e-06,
+      "loss": 0.3313,
+      "step": 9623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.668560606060606e-06,
+      "loss": 0.3899,
+      "step": 9624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.663825757575758e-06,
+      "loss": 0.3557,
+      "step": 9625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6590909090909095e-06,
+      "loss": 0.2791,
+      "step": 9626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6543560606060615e-06,
+      "loss": 0.3251,
+      "step": 9627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.649621212121212e-06,
+      "loss": 0.4018,
+      "step": 9628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.644886363636364e-06,
+      "loss": 0.3682,
+      "step": 9629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.640151515151515e-06,
+      "loss": 0.386,
+      "step": 9630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.635416666666667e-06,
+      "loss": 0.4163,
+      "step": 9631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6306818181818185e-06,
+      "loss": 0.4566,
+      "step": 9632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.62594696969697e-06,
+      "loss": 0.5201,
+      "step": 9633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.621212121212122e-06,
+      "loss": 0.3408,
+      "step": 9634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.616477272727273e-06,
+      "loss": 0.3664,
+      "step": 9635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.611742424242425e-06,
+      "loss": 0.3888,
+      "step": 9636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.607007575757575e-06,
+      "loss": 0.3181,
+      "step": 9637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6022727272727275e-06,
+      "loss": 0.4711,
+      "step": 9638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.597537878787879e-06,
+      "loss": 0.3639,
+      "step": 9639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.592803030303031e-06,
+      "loss": 0.3957,
+      "step": 9640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.588068181818182e-06,
+      "loss": 0.2155,
+      "step": 9641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.583333333333333e-06,
+      "loss": 0.3475,
+      "step": 9642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.578598484848485e-06,
+      "loss": 0.4409,
+      "step": 9643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5738636363636365e-06,
+      "loss": 0.532,
+      "step": 9644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5691287878787886e-06,
+      "loss": 0.2934,
+      "step": 9645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.56439393939394e-06,
+      "loss": 0.3289,
+      "step": 9646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.559659090909091e-06,
+      "loss": 0.3743,
+      "step": 9647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.554924242424242e-06,
+      "loss": 0.3945,
+      "step": 9648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.550189393939394e-06,
+      "loss": 0.4166,
+      "step": 9649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.3433,
+      "step": 9650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.540719696969698e-06,
+      "loss": 0.356,
+      "step": 9651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.535984848484849e-06,
+      "loss": 0.3638,
+      "step": 9652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.53125e-06,
+      "loss": 0.3623,
+      "step": 9653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.526515151515152e-06,
+      "loss": 0.3168,
+      "step": 9654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.521780303030303e-06,
+      "loss": 0.4477,
+      "step": 9655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.517045454545455e-06,
+      "loss": 0.2578,
+      "step": 9656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.512310606060606e-06,
+      "loss": 0.3419,
+      "step": 9657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.507575757575758e-06,
+      "loss": 0.2996,
+      "step": 9658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.502840909090909e-06,
+      "loss": 0.4405,
+      "step": 9659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.498106060606061e-06,
+      "loss": 0.3523,
+      "step": 9660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.493371212121212e-06,
+      "loss": 0.4277,
+      "step": 9661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4886363636363636e-06,
+      "loss": 0.3988,
+      "step": 9662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.483901515151516e-06,
+      "loss": 0.5347,
+      "step": 9663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.479166666666667e-06,
+      "loss": 0.377,
+      "step": 9664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.474431818181819e-06,
+      "loss": 0.4904,
+      "step": 9665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.46969696969697e-06,
+      "loss": 0.4308,
+      "step": 9666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.464962121212121e-06,
+      "loss": 0.2958,
+      "step": 9667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4602272727272726e-06,
+      "loss": 0.4219,
+      "step": 9668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.455492424242425e-06,
+      "loss": 0.3854,
+      "step": 9669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.450757575757576e-06,
+      "loss": 0.4023,
+      "step": 9670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.446022727272727e-06,
+      "loss": 0.4308,
+      "step": 9671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.441287878787879e-06,
+      "loss": 0.3138,
+      "step": 9672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.43655303030303e-06,
+      "loss": 0.383,
+      "step": 9673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4318181818181824e-06,
+      "loss": 0.2831,
+      "step": 9674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.427083333333334e-06,
+      "loss": 0.3173,
+      "step": 9675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.422348484848485e-06,
+      "loss": 0.3407,
+      "step": 9676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.417613636363636e-06,
+      "loss": 0.3929,
+      "step": 9677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.412878787878788e-06,
+      "loss": 0.331,
+      "step": 9678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.408143939393939e-06,
+      "loss": 0.4384,
+      "step": 9679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4034090909090914e-06,
+      "loss": 0.4023,
+      "step": 9680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.398674242424243e-06,
+      "loss": 0.3231,
+      "step": 9681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.393939393939394e-06,
+      "loss": 0.331,
+      "step": 9682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.389204545454546e-06,
+      "loss": 0.2717,
+      "step": 9683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.384469696969697e-06,
+      "loss": 0.3937,
+      "step": 9684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.379734848484849e-06,
+      "loss": 0.3704,
+      "step": 9685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.375e-06,
+      "loss": 0.1927,
+      "step": 9686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.370265151515152e-06,
+      "loss": 0.373,
+      "step": 9687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.365530303030303e-06,
+      "loss": 0.3727,
+      "step": 9688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.360795454545455e-06,
+      "loss": 0.3458,
+      "step": 9689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.356060606060606e-06,
+      "loss": 0.3154,
+      "step": 9690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.351325757575757e-06,
+      "loss": 0.5192,
+      "step": 9691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3465909090909095e-06,
+      "loss": 0.3778,
+      "step": 9692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.341856060606061e-06,
+      "loss": 0.3622,
+      "step": 9693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.337121212121213e-06,
+      "loss": 0.4093,
+      "step": 9694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.332386363636364e-06,
+      "loss": 0.4924,
+      "step": 9695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.327651515151515e-06,
+      "loss": 0.2825,
+      "step": 9696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3229166666666664e-06,
+      "loss": 0.3898,
+      "step": 9697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3181818181818185e-06,
+      "loss": 0.4013,
+      "step": 9698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.31344696969697e-06,
+      "loss": 0.361,
+      "step": 9699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.308712121212122e-06,
+      "loss": 0.4713,
+      "step": 9700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.303977272727273e-06,
+      "loss": 0.4686,
+      "step": 9701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.299242424242424e-06,
+      "loss": 0.2939,
+      "step": 9702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.294507575757576e-06,
+      "loss": 0.383,
+      "step": 9703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2897727272727275e-06,
+      "loss": 0.2622,
+      "step": 9704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.285037878787879e-06,
+      "loss": 0.1934,
+      "step": 9705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.28030303030303e-06,
+      "loss": 0.2241,
+      "step": 9706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.275568181818182e-06,
+      "loss": 0.335,
+      "step": 9707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.270833333333333e-06,
+      "loss": 0.4623,
+      "step": 9708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.266098484848485e-06,
+      "loss": 0.306,
+      "step": 9709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2613636363636365e-06,
+      "loss": 0.2685,
+      "step": 9710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.256628787878788e-06,
+      "loss": 0.4362,
+      "step": 9711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.25189393939394e-06,
+      "loss": 0.353,
+      "step": 9712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.247159090909091e-06,
+      "loss": 0.3982,
+      "step": 9713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 0.4856,
+      "step": 9714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2376893939393935e-06,
+      "loss": 0.3798,
+      "step": 9715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2329545454545455e-06,
+      "loss": 0.331,
+      "step": 9716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.228219696969697e-06,
+      "loss": 0.2958,
+      "step": 9717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.223484848484849e-06,
+      "loss": 0.4762,
+      "step": 9718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.218750000000001e-06,
+      "loss": 0.4663,
+      "step": 9719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.214015151515151e-06,
+      "loss": 0.4375,
+      "step": 9720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.209280303030303e-06,
+      "loss": 0.3234,
+      "step": 9721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2045454545454546e-06,
+      "loss": 0.3642,
+      "step": 9722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.199810606060607e-06,
+      "loss": 0.4074,
+      "step": 9723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.195075757575758e-06,
+      "loss": 0.3201,
+      "step": 9724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.190340909090909e-06,
+      "loss": 0.4008,
+      "step": 9725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.18560606060606e-06,
+      "loss": 0.3854,
+      "step": 9726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.180871212121212e-06,
+      "loss": 0.2999,
+      "step": 9727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1761363636363644e-06,
+      "loss": 0.4412,
+      "step": 9728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.171401515151516e-06,
+      "loss": 0.384,
+      "step": 9729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.3836,
+      "step": 9730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.161931818181818e-06,
+      "loss": 0.296,
+      "step": 9731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.15719696969697e-06,
+      "loss": 0.367,
+      "step": 9732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.152462121212121e-06,
+      "loss": 0.4117,
+      "step": 9733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1477272727272734e-06,
+      "loss": 0.4067,
+      "step": 9734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.142992424242424e-06,
+      "loss": 0.3383,
+      "step": 9735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.138257575757576e-06,
+      "loss": 0.3792,
+      "step": 9736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.133522727272728e-06,
+      "loss": 0.3047,
+      "step": 9737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.128787878787879e-06,
+      "loss": 0.373,
+      "step": 9738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.12405303030303e-06,
+      "loss": 0.3435,
+      "step": 9739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.119318181818182e-06,
+      "loss": 0.2888,
+      "step": 9740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.114583333333334e-06,
+      "loss": 0.4209,
+      "step": 9741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.109848484848485e-06,
+      "loss": 0.5473,
+      "step": 9742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.105113636363637e-06,
+      "loss": 0.3949,
+      "step": 9743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.100378787878788e-06,
+      "loss": 0.4029,
+      "step": 9744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.095643939393939e-06,
+      "loss": 0.4353,
+      "step": 9745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0909090909090915e-06,
+      "loss": 0.3546,
+      "step": 9746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.086174242424243e-06,
+      "loss": 0.319,
+      "step": 9747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.081439393939395e-06,
+      "loss": 0.4445,
+      "step": 9748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.076704545454545e-06,
+      "loss": 0.5911,
+      "step": 9749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.071969696969697e-06,
+      "loss": 0.2701,
+      "step": 9750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0672348484848484e-06,
+      "loss": 0.4368,
+      "step": 9751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0625000000000005e-06,
+      "loss": 0.4216,
+      "step": 9752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.057765151515152e-06,
+      "loss": 0.4385,
+      "step": 9753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.053030303030303e-06,
+      "loss": 0.5199,
+      "step": 9754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.048295454545455e-06,
+      "loss": 0.4079,
+      "step": 9755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.043560606060606e-06,
+      "loss": 0.2999,
+      "step": 9756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.038825757575758e-06,
+      "loss": 0.3501,
+      "step": 9757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0340909090909095e-06,
+      "loss": 0.3929,
+      "step": 9758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.029356060606061e-06,
+      "loss": 0.3582,
+      "step": 9759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.024621212121212e-06,
+      "loss": 0.243,
+      "step": 9760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.019886363636364e-06,
+      "loss": 0.4407,
+      "step": 9761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.015151515151515e-06,
+      "loss": 0.2679,
+      "step": 9762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.010416666666667e-06,
+      "loss": 0.3809,
+      "step": 9763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0056818181818185e-06,
+      "loss": 0.3329,
+      "step": 9764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.00094696969697e-06,
+      "loss": 0.3591,
+      "step": 9765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.996212121212122e-06,
+      "loss": 0.3401,
+      "step": 9766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.991477272727273e-06,
+      "loss": 0.368,
+      "step": 9767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.986742424242425e-06,
+      "loss": 0.3531,
+      "step": 9768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9820075757575755e-06,
+      "loss": 0.4494,
+      "step": 9769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9772727272727275e-06,
+      "loss": 0.4911,
+      "step": 9770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.972537878787879e-06,
+      "loss": 0.3982,
+      "step": 9771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.967803030303031e-06,
+      "loss": 0.3067,
+      "step": 9772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.963068181818182e-06,
+      "loss": 0.3624,
+      "step": 9773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.958333333333333e-06,
+      "loss": 0.371,
+      "step": 9774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.953598484848485e-06,
+      "loss": 0.5214,
+      "step": 9775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9488636363636366e-06,
+      "loss": 0.4061,
+      "step": 9776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.944128787878789e-06,
+      "loss": 0.2551,
+      "step": 9777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.939393939393939e-06,
+      "loss": 0.3792,
+      "step": 9778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.934659090909091e-06,
+      "loss": 0.4656,
+      "step": 9779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.929924242424242e-06,
+      "loss": 0.3402,
+      "step": 9780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.925189393939394e-06,
+      "loss": 0.438,
+      "step": 9781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9204545454545456e-06,
+      "loss": 0.3089,
+      "step": 9782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.915719696969697e-06,
+      "loss": 0.3154,
+      "step": 9783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.910984848484849e-06,
+      "loss": 0.4247,
+      "step": 9784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.90625e-06,
+      "loss": 0.2983,
+      "step": 9785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.901515151515152e-06,
+      "loss": 0.2492,
+      "step": 9786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.896780303030303e-06,
+      "loss": 0.3523,
+      "step": 9787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.892045454545455e-06,
+      "loss": 0.4175,
+      "step": 9788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.887310606060606e-06,
+      "loss": 0.2592,
+      "step": 9789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.882575757575758e-06,
+      "loss": 0.3422,
+      "step": 9790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.877840909090909e-06,
+      "loss": 0.2921,
+      "step": 9791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.873106060606061e-06,
+      "loss": 0.4196,
+      "step": 9792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.868371212121212e-06,
+      "loss": 0.3699,
+      "step": 9793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.863636363636364e-06,
+      "loss": 0.3051,
+      "step": 9794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.858901515151516e-06,
+      "loss": 0.4452,
+      "step": 9795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.854166666666667e-06,
+      "loss": 0.3861,
+      "step": 9796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.849431818181819e-06,
+      "loss": 0.2898,
+      "step": 9797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.844696969696969e-06,
+      "loss": 0.4398,
+      "step": 9798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.839962121212121e-06,
+      "loss": 0.4567,
+      "step": 9799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.835227272727273e-06,
+      "loss": 0.4009,
+      "step": 9800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.830492424242425e-06,
+      "loss": 0.3042,
+      "step": 9801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.825757575757576e-06,
+      "loss": 0.29,
+      "step": 9802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.821022727272727e-06,
+      "loss": 0.2964,
+      "step": 9803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.816287878787879e-06,
+      "loss": 0.3221,
+      "step": 9804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8115530303030304e-06,
+      "loss": 0.556,
+      "step": 9805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.806818181818182e-06,
+      "loss": 0.3986,
+      "step": 9806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8020833333333333e-06,
+      "loss": 0.3511,
+      "step": 9807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.797348484848485e-06,
+      "loss": 0.3138,
+      "step": 9808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7926136363636366e-06,
+      "loss": 0.3869,
+      "step": 9809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7878787878787882e-06,
+      "loss": 0.3447,
+      "step": 9810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.78314393939394e-06,
+      "loss": 0.3141,
+      "step": 9811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7784090909090907e-06,
+      "loss": 0.2219,
+      "step": 9812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7736742424242423e-06,
+      "loss": 0.3974,
+      "step": 9813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.768939393939394e-06,
+      "loss": 0.3394,
+      "step": 9814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7642045454545456e-06,
+      "loss": 0.3653,
+      "step": 9815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7594696969696972e-06,
+      "loss": 0.3654,
+      "step": 9816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7547348484848485e-06,
+      "loss": 0.4674,
+      "step": 9817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.75e-06,
+      "loss": 0.3924,
+      "step": 9818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7452651515151517e-06,
+      "loss": 0.2536,
+      "step": 9819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7405303030303034e-06,
+      "loss": 0.3006,
+      "step": 9820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.735795454545455e-06,
+      "loss": 0.3825,
+      "step": 9821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.731060606060606e-06,
+      "loss": 0.3895,
+      "step": 9822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7263257575757575e-06,
+      "loss": 0.3335,
+      "step": 9823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.721590909090909e-06,
+      "loss": 0.453,
+      "step": 9824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7168560606060608e-06,
+      "loss": 0.3715,
+      "step": 9825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7121212121212124e-06,
+      "loss": 0.344,
+      "step": 9826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7073863636363636e-06,
+      "loss": 0.3435,
+      "step": 9827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7026515151515153e-06,
+      "loss": 0.4533,
+      "step": 9828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.697916666666667e-06,
+      "loss": 0.4062,
+      "step": 9829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6931818181818186e-06,
+      "loss": 0.431,
+      "step": 9830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.68844696969697e-06,
+      "loss": 0.452,
+      "step": 9831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.683712121212121e-06,
+      "loss": 0.4014,
+      "step": 9832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6789772727272726e-06,
+      "loss": 0.5342,
+      "step": 9833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6742424242424243e-06,
+      "loss": 0.4824,
+      "step": 9834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6695075757575763e-06,
+      "loss": 0.4621,
+      "step": 9835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.664772727272727e-06,
+      "loss": 0.3778,
+      "step": 9836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6600378787878788e-06,
+      "loss": 0.5174,
+      "step": 9837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6553030303030304e-06,
+      "loss": 0.3868,
+      "step": 9838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.650568181818182e-06,
+      "loss": 0.5344,
+      "step": 9839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6458333333333337e-06,
+      "loss": 0.4954,
+      "step": 9840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6410984848484845e-06,
+      "loss": 0.5594,
+      "step": 9841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.636363636363636e-06,
+      "loss": 0.3604,
+      "step": 9842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6316287878787882e-06,
+      "loss": 0.2963,
+      "step": 9843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.62689393939394e-06,
+      "loss": 0.2705,
+      "step": 9844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6221590909090915e-06,
+      "loss": 0.3396,
+      "step": 9845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6174242424242423e-06,
+      "loss": 0.4328,
+      "step": 9846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.612689393939394e-06,
+      "loss": 0.3449,
+      "step": 9847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6079545454545456e-06,
+      "loss": 0.3775,
+      "step": 9848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6032196969696972e-06,
+      "loss": 0.4426,
+      "step": 9849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.598484848484849e-06,
+      "loss": 0.3886,
+      "step": 9850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5937499999999997e-06,
+      "loss": 0.2803,
+      "step": 9851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5890151515151518e-06,
+      "loss": 0.4171,
+      "step": 9852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5842803030303034e-06,
+      "loss": 0.2856,
+      "step": 9853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.579545454545455e-06,
+      "loss": 0.3756,
+      "step": 9854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5748106060606067e-06,
+      "loss": 0.3101,
+      "step": 9855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5700757575757575e-06,
+      "loss": 0.2765,
+      "step": 9856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.565340909090909e-06,
+      "loss": 0.4097,
+      "step": 9857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5606060606060608e-06,
+      "loss": 0.2798,
+      "step": 9858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5558712121212124e-06,
+      "loss": 0.3246,
+      "step": 9859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.551136363636364e-06,
+      "loss": 0.3766,
+      "step": 9860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5464015151515153e-06,
+      "loss": 0.3811,
+      "step": 9861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.541666666666667e-06,
+      "loss": 0.3964,
+      "step": 9862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5369318181818186e-06,
+      "loss": 0.349,
+      "step": 9863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.53219696969697e-06,
+      "loss": 0.2672,
+      "step": 9864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.527462121212122e-06,
+      "loss": 0.5252,
+      "step": 9865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5227272727272726e-06,
+      "loss": 0.314,
+      "step": 9866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5179924242424243e-06,
+      "loss": 0.3686,
+      "step": 9867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.513257575757576e-06,
+      "loss": 0.4044,
+      "step": 9868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5085227272727276e-06,
+      "loss": 0.261,
+      "step": 9869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.503787878787879e-06,
+      "loss": 0.2969,
+      "step": 9870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4990530303030304e-06,
+      "loss": 0.3896,
+      "step": 9871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.494318181818182e-06,
+      "loss": 0.3812,
+      "step": 9872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4895833333333337e-06,
+      "loss": 0.4996,
+      "step": 9873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4848484848484854e-06,
+      "loss": 0.4593,
+      "step": 9874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.480113636363636e-06,
+      "loss": 0.3461,
+      "step": 9875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.475378787878788e-06,
+      "loss": 0.2801,
+      "step": 9876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4706439393939395e-06,
+      "loss": 0.5168,
+      "step": 9877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.465909090909091e-06,
+      "loss": 0.5571,
+      "step": 9878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4611742424242427e-06,
+      "loss": 0.4519,
+      "step": 9879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456439393939394e-06,
+      "loss": 0.3093,
+      "step": 9880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4517045454545456e-06,
+      "loss": 0.4023,
+      "step": 9881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4469696969696973e-06,
+      "loss": 0.2976,
+      "step": 9882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.442234848484849e-06,
+      "loss": 0.4198,
+      "step": 9883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4375000000000005e-06,
+      "loss": 0.4867,
+      "step": 9884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4327651515151513e-06,
+      "loss": 0.577,
+      "step": 9885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.428030303030303e-06,
+      "loss": 0.3277,
+      "step": 9886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4232954545454546e-06,
+      "loss": 0.3111,
+      "step": 9887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4185606060606063e-06,
+      "loss": 0.4055,
+      "step": 9888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.413825757575758e-06,
+      "loss": 0.4335,
+      "step": 9889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.409090909090909e-06,
+      "loss": 0.3782,
+      "step": 9890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4043560606060608e-06,
+      "loss": 0.426,
+      "step": 9891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3996212121212124e-06,
+      "loss": 0.3365,
+      "step": 9892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.394886363636364e-06,
+      "loss": 0.2911,
+      "step": 9893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3901515151515157e-06,
+      "loss": 0.2821,
+      "step": 9894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3854166666666665e-06,
+      "loss": 0.327,
+      "step": 9895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.380681818181818e-06,
+      "loss": 0.3014,
+      "step": 9896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.37594696969697e-06,
+      "loss": 0.2762,
+      "step": 9897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3712121212121214e-06,
+      "loss": 0.4054,
+      "step": 9898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.366477272727273e-06,
+      "loss": 0.374,
+      "step": 9899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3617424242424243e-06,
+      "loss": 0.4527,
+      "step": 9900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.357007575757576e-06,
+      "loss": 0.4632,
+      "step": 9901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3522727272727276e-06,
+      "loss": 0.2959,
+      "step": 9902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3475378787878792e-06,
+      "loss": 0.2984,
+      "step": 9903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.34280303030303e-06,
+      "loss": 0.452,
+      "step": 9904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3380681818181817e-06,
+      "loss": 0.3177,
+      "step": 9905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.2958,
+      "step": 9906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.328598484848485e-06,
+      "loss": 0.5836,
+      "step": 9907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3238636363636366e-06,
+      "loss": 0.3418,
+      "step": 9908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.319128787878788e-06,
+      "loss": 0.351,
+      "step": 9909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3143939393939395e-06,
+      "loss": 0.6445,
+      "step": 9910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.309659090909091e-06,
+      "loss": 0.3275,
+      "step": 9911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3049242424242428e-06,
+      "loss": 0.4547,
+      "step": 9912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3001893939393944e-06,
+      "loss": 0.5173,
+      "step": 9913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.295454545454545e-06,
+      "loss": 0.293,
+      "step": 9914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.290719696969697e-06,
+      "loss": 0.4242,
+      "step": 9915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2859848484848485e-06,
+      "loss": 0.4109,
+      "step": 9916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.28125e-06,
+      "loss": 0.3461,
+      "step": 9917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2765151515151518e-06,
+      "loss": 0.2472,
+      "step": 9918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.271780303030303e-06,
+      "loss": 0.2832,
+      "step": 9919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2670454545454546e-06,
+      "loss": 0.4165,
+      "step": 9920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2623106060606063e-06,
+      "loss": 0.3478,
+      "step": 9921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.257575757575758e-06,
+      "loss": 0.3972,
+      "step": 9922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2528409090909096e-06,
+      "loss": 0.335,
+      "step": 9923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2481060606060604e-06,
+      "loss": 0.4072,
+      "step": 9924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.243371212121212e-06,
+      "loss": 0.3562,
+      "step": 9925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2386363636363637e-06,
+      "loss": 0.3515,
+      "step": 9926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2339015151515153e-06,
+      "loss": 0.2625,
+      "step": 9927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.229166666666667e-06,
+      "loss": 0.3052,
+      "step": 9928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.224431818181818e-06,
+      "loss": 0.4666,
+      "step": 9929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.21969696969697e-06,
+      "loss": 0.3613,
+      "step": 9930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2149621212121215e-06,
+      "loss": 0.3671,
+      "step": 9931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.210227272727273e-06,
+      "loss": 0.3465,
+      "step": 9932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2054924242424247e-06,
+      "loss": 0.3243,
+      "step": 9933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2007575757575755e-06,
+      "loss": 0.343,
+      "step": 9934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.196022727272727e-06,
+      "loss": 0.332,
+      "step": 9935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.191287878787879e-06,
+      "loss": 0.4118,
+      "step": 9936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1865530303030305e-06,
+      "loss": 0.3462,
+      "step": 9937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1818181818181817e-06,
+      "loss": 0.4088,
+      "step": 9938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1770833333333333e-06,
+      "loss": 0.2776,
+      "step": 9939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.172348484848485e-06,
+      "loss": 0.3494,
+      "step": 9940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1676136363636366e-06,
+      "loss": 0.3443,
+      "step": 9941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1628787878787883e-06,
+      "loss": 0.3858,
+      "step": 9942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.158143939393939e-06,
+      "loss": 0.4701,
+      "step": 9943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1534090909090907e-06,
+      "loss": 0.4486,
+      "step": 9944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1486742424242424e-06,
+      "loss": 0.4537,
+      "step": 9945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1439393939393944e-06,
+      "loss": 0.4601,
+      "step": 9946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.139204545454546e-06,
+      "loss": 0.3125,
+      "step": 9947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.134469696969697e-06,
+      "loss": 0.3038,
+      "step": 9948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1297348484848485e-06,
+      "loss": 0.3419,
+      "step": 9949
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.125e-06,
+      "loss": 0.4601,
+      "step": 9950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.120265151515152e-06,
+      "loss": 0.3778,
+      "step": 9951
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.115530303030303e-06,
+      "loss": 0.3853,
+      "step": 9952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1107954545454547e-06,
+      "loss": 0.503,
+      "step": 9953
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.106060606060606e-06,
+      "loss": 0.3954,
+      "step": 9954
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.101325757575758e-06,
+      "loss": 0.2241,
+      "step": 9955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0965909090909096e-06,
+      "loss": 0.3204,
+      "step": 9956
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.091856060606061e-06,
+      "loss": 0.4734,
+      "step": 9957
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0871212121212125e-06,
+      "loss": 0.3457,
+      "step": 9958
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0823863636363637e-06,
+      "loss": 0.3962,
+      "step": 9959
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0776515151515153e-06,
+      "loss": 0.2937,
+      "step": 9960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0729166666666665e-06,
+      "loss": 0.3357,
+      "step": 9961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.068181818181818e-06,
+      "loss": 0.5521,
+      "step": 9962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.06344696969697e-06,
+      "loss": 0.3759,
+      "step": 9963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0587121212121215e-06,
+      "loss": 0.4268,
+      "step": 9964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.053977272727273e-06,
+      "loss": 0.4797,
+      "step": 9965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0492424242424243e-06,
+      "loss": 0.4065,
+      "step": 9966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.044507575757576e-06,
+      "loss": 0.3788,
+      "step": 9967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0397727272727276e-06,
+      "loss": 0.4003,
+      "step": 9968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.035037878787879e-06,
+      "loss": 0.2975,
+      "step": 9969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.3231,
+      "step": 9970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0255681818181817e-06,
+      "loss": 0.4469,
+      "step": 9971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0208333333333334e-06,
+      "loss": 0.2632,
+      "step": 9972
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.016098484848485e-06,
+      "loss": 0.3972,
+      "step": 9973
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0113636363636366e-06,
+      "loss": 0.3674,
+      "step": 9974
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0066287878787883e-06,
+      "loss": 0.4526,
+      "step": 9975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0018939393939395e-06,
+      "loss": 0.4951,
+      "step": 9976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.997159090909091e-06,
+      "loss": 0.2533,
+      "step": 9977
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9924242424242424e-06,
+      "loss": 0.4525,
+      "step": 9978
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.987689393939394e-06,
+      "loss": 0.3165,
+      "step": 9979
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9829545454545457e-06,
+      "loss": 0.305,
+      "step": 9980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.978219696969697e-06,
+      "loss": 0.4442,
+      "step": 9981
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9734848484848485e-06,
+      "loss": 0.185,
+      "step": 9982
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.96875e-06,
+      "loss": 0.4827,
+      "step": 9983
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.964015151515152e-06,
+      "loss": 0.4971,
+      "step": 9984
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9592803030303035e-06,
+      "loss": 0.4169,
+      "step": 9985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9545454545454547e-06,
+      "loss": 0.4183,
+      "step": 9986
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9498106060606063e-06,
+      "loss": 0.293,
+      "step": 9987
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9450757575757575e-06,
+      "loss": 0.3359,
+      "step": 9988
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.940340909090909e-06,
+      "loss": 0.5211,
+      "step": 9989
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.935606060606061e-06,
+      "loss": 0.3667,
+      "step": 9990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.930871212121212e-06,
+      "loss": 0.4546,
+      "step": 9991
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9261363636363637e-06,
+      "loss": 0.3644,
+      "step": 9992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9214015151515153e-06,
+      "loss": 0.3817,
+      "step": 9993
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.916666666666667e-06,
+      "loss": 0.2816,
+      "step": 9994
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.911931818181818e-06,
+      "loss": 0.2981,
+      "step": 9995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.90719696969697e-06,
+      "loss": 0.3614,
+      "step": 9996
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9024621212121215e-06,
+      "loss": 0.3166,
+      "step": 9997
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8977272727272727e-06,
+      "loss": 0.3899,
+      "step": 9998
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8929924242424243e-06,
+      "loss": 0.5023,
+      "step": 9999
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8882575757575756e-06,
+      "loss": 0.4396,
+      "step": 10000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8835227272727272e-06,
+      "loss": 0.3524,
+      "step": 10001
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8787878787878793e-06,
+      "loss": 0.3018,
+      "step": 10002
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8740530303030305e-06,
+      "loss": 0.2231,
+      "step": 10003
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.869318181818182e-06,
+      "loss": 0.2967,
+      "step": 10004
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8645833333333334e-06,
+      "loss": 0.2435,
+      "step": 10005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.859848484848485e-06,
+      "loss": 0.394,
+      "step": 10006
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8551136363636367e-06,
+      "loss": 0.4321,
+      "step": 10007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.850378787878788e-06,
+      "loss": 0.2159,
+      "step": 10008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8456439393939395e-06,
+      "loss": 0.2853,
+      "step": 10009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.840909090909091e-06,
+      "loss": 0.2634,
+      "step": 10010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.836174242424243e-06,
+      "loss": 0.364,
+      "step": 10011
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.831439393939394e-06,
+      "loss": 0.2208,
+      "step": 10012
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8267045454545457e-06,
+      "loss": 0.2479,
+      "step": 10013
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8219696969696973e-06,
+      "loss": 0.3573,
+      "step": 10014
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8172348484848485e-06,
+      "loss": 0.4149,
+      "step": 10015
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8125e-06,
+      "loss": 0.4422,
+      "step": 10016
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8077651515151514e-06,
+      "loss": 0.2915,
+      "step": 10017
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.803030303030303e-06,
+      "loss": 0.2926,
+      "step": 10018
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7982954545454547e-06,
+      "loss": 0.418,
+      "step": 10019
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7935606060606063e-06,
+      "loss": 0.2997,
+      "step": 10020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.788825757575758e-06,
+      "loss": 0.3421,
+      "step": 10021
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.784090909090909e-06,
+      "loss": 0.3193,
+      "step": 10022
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.779356060606061e-06,
+      "loss": 0.3336,
+      "step": 10023
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7746212121212125e-06,
+      "loss": 0.4527,
+      "step": 10024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7698863636363637e-06,
+      "loss": 0.3853,
+      "step": 10025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7651515151515153e-06,
+      "loss": 0.4123,
+      "step": 10026
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7604166666666666e-06,
+      "loss": 0.5421,
+      "step": 10027
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.755681818181818e-06,
+      "loss": 0.3727,
+      "step": 10028
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.75094696969697e-06,
+      "loss": 0.3354,
+      "step": 10029
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7462121212121215e-06,
+      "loss": 0.3817,
+      "step": 10030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.741477272727273e-06,
+      "loss": 0.3368,
+      "step": 10031
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7367424242424244e-06,
+      "loss": 0.2959,
+      "step": 10032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.732007575757576e-06,
+      "loss": 0.4501,
+      "step": 10033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 0.2511,
+      "step": 10034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.722537878787879e-06,
+      "loss": 0.403,
+      "step": 10035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7178030303030305e-06,
+      "loss": 0.4431,
+      "step": 10036
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7130681818181817e-06,
+      "loss": 0.4105,
+      "step": 10037
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7083333333333334e-06,
+      "loss": 0.3939,
+      "step": 10038
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.703598484848485e-06,
+      "loss": 0.2561,
+      "step": 10039
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6988636363636367e-06,
+      "loss": 0.2952,
+      "step": 10040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.694128787878788e-06,
+      "loss": 0.4512,
+      "step": 10041
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6893939393939395e-06,
+      "loss": 0.5199,
+      "step": 10042
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.684659090909091e-06,
+      "loss": 0.3194,
+      "step": 10043
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6799242424242424e-06,
+      "loss": 0.3019,
+      "step": 10044
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.675189393939394e-06,
+      "loss": 0.3654,
+      "step": 10045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6704545454545453e-06,
+      "loss": 0.3293,
+      "step": 10046
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.665719696969697e-06,
+      "loss": 0.3876,
+      "step": 10047
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6609848484848485e-06,
+      "loss": 0.4293,
+      "step": 10048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.65625e-06,
+      "loss": 0.3719,
+      "step": 10049
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.651515151515152e-06,
+      "loss": 0.2496,
+      "step": 10050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.646780303030303e-06,
+      "loss": 0.4973,
+      "step": 10051
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6420454545454547e-06,
+      "loss": 0.3468,
+      "step": 10052
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6373106060606063e-06,
+      "loss": 0.4132,
+      "step": 10053
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6325757575757576e-06,
+      "loss": 0.2901,
+      "step": 10054
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.627840909090909e-06,
+      "loss": 0.3697,
+      "step": 10055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6231060606060604e-06,
+      "loss": 0.3326,
+      "step": 10056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.618371212121212e-06,
+      "loss": 0.3634,
+      "step": 10057
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6136363636363637e-06,
+      "loss": 0.2699,
+      "step": 10058
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6089015151515154e-06,
+      "loss": 0.3899,
+      "step": 10059
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.604166666666667e-06,
+      "loss": 0.3436,
+      "step": 10060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5994318181818182e-06,
+      "loss": 0.4255,
+      "step": 10061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.59469696969697e-06,
+      "loss": 0.4543,
+      "step": 10062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.589962121212121e-06,
+      "loss": 0.3128,
+      "step": 10063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5852272727272727e-06,
+      "loss": 0.4425,
+      "step": 10064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5804924242424244e-06,
+      "loss": 0.3669,
+      "step": 10065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.575757575757576e-06,
+      "loss": 0.2702,
+      "step": 10066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5710227272727277e-06,
+      "loss": 0.5403,
+      "step": 10067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.566287878787879e-06,
+      "loss": 0.3567,
+      "step": 10068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5615530303030305e-06,
+      "loss": 0.4344,
+      "step": 10069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.556818181818182e-06,
+      "loss": 0.4659,
+      "step": 10070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5520833333333334e-06,
+      "loss": 0.4453,
+      "step": 10071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.547348484848485e-06,
+      "loss": 0.3688,
+      "step": 10072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5426136363636363e-06,
+      "loss": 0.4284,
+      "step": 10073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.537878787878788e-06,
+      "loss": 0.3161,
+      "step": 10074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5331439393939395e-06,
+      "loss": 0.3457,
+      "step": 10075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.528409090909091e-06,
+      "loss": 0.3597,
+      "step": 10076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.523674242424243e-06,
+      "loss": 0.4181,
+      "step": 10077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.518939393939394e-06,
+      "loss": 0.3772,
+      "step": 10078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5142045454545457e-06,
+      "loss": 0.323,
+      "step": 10079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.509469696969697e-06,
+      "loss": 0.4944,
+      "step": 10080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5047348484848486e-06,
+      "loss": 0.3874,
+      "step": 10081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5e-06,
+      "loss": 0.3782,
+      "step": 10082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4952651515151514e-06,
+      "loss": 0.3584,
+      "step": 10083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.490530303030303e-06,
+      "loss": 0.411,
+      "step": 10084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4857954545454547e-06,
+      "loss": 0.3777,
+      "step": 10085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4810606060606064e-06,
+      "loss": 0.3391,
+      "step": 10086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.476325757575758e-06,
+      "loss": 0.417,
+      "step": 10087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4715909090909092e-06,
+      "loss": 0.3551,
+      "step": 10088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.466856060606061e-06,
+      "loss": 0.2849,
+      "step": 10089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.462121212121212e-06,
+      "loss": 0.4288,
+      "step": 10090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4573863636363637e-06,
+      "loss": 0.4105,
+      "step": 10091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.452651515151515e-06,
+      "loss": 0.4303,
+      "step": 10092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4479166666666666e-06,
+      "loss": 0.3595,
+      "step": 10093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4431818181818182e-06,
+      "loss": 0.5298,
+      "step": 10094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.43844696969697e-06,
+      "loss": 0.2966,
+      "step": 10095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4337121212121215e-06,
+      "loss": 0.3355,
+      "step": 10096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4289772727272727e-06,
+      "loss": 0.3293,
+      "step": 10097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 0.4173,
+      "step": 10098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.419507575757576e-06,
+      "loss": 0.3442,
+      "step": 10099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4147727272727273e-06,
+      "loss": 0.3812,
+      "step": 10100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.410037878787879e-06,
+      "loss": 0.3667,
+      "step": 10101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.40530303030303e-06,
+      "loss": 0.2607,
+      "step": 10102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4005681818181818e-06,
+      "loss": 0.4403,
+      "step": 10103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3958333333333334e-06,
+      "loss": 0.2824,
+      "step": 10104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.391098484848485e-06,
+      "loss": 0.3054,
+      "step": 10105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3863636363636367e-06,
+      "loss": 0.2329,
+      "step": 10106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.381628787878788e-06,
+      "loss": 0.3631,
+      "step": 10107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3768939393939396e-06,
+      "loss": 0.3265,
+      "step": 10108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3721590909090908e-06,
+      "loss": 0.3925,
+      "step": 10109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3674242424242424e-06,
+      "loss": 0.3167,
+      "step": 10110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.362689393939394e-06,
+      "loss": 0.4824,
+      "step": 10111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3579545454545453e-06,
+      "loss": 0.3198,
+      "step": 10112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3532196969696974e-06,
+      "loss": 0.4666,
+      "step": 10113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3484848484848486e-06,
+      "loss": 0.2835,
+      "step": 10114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3437500000000002e-06,
+      "loss": 0.3974,
+      "step": 10115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.339015151515152e-06,
+      "loss": 0.2491,
+      "step": 10116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.334280303030303e-06,
+      "loss": 0.3801,
+      "step": 10117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3295454545454547e-06,
+      "loss": 0.4425,
+      "step": 10118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.324810606060606e-06,
+      "loss": 0.4184,
+      "step": 10119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3200757575757576e-06,
+      "loss": 0.4859,
+      "step": 10120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3153409090909092e-06,
+      "loss": 0.2559,
+      "step": 10121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.310606060606061e-06,
+      "loss": 0.3955,
+      "step": 10122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3058712121212125e-06,
+      "loss": 0.3626,
+      "step": 10123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3011363636363637e-06,
+      "loss": 0.3228,
+      "step": 10124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2964015151515154e-06,
+      "loss": 0.3566,
+      "step": 10125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2916666666666666e-06,
+      "loss": 0.4229,
+      "step": 10126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2869318181818183e-06,
+      "loss": 0.457,
+      "step": 10127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.28219696969697e-06,
+      "loss": 0.3206,
+      "step": 10128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.277462121212121e-06,
+      "loss": 0.3145,
+      "step": 10129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 0.4217,
+      "step": 10130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2679924242424244e-06,
+      "loss": 0.4704,
+      "step": 10131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.263257575757576e-06,
+      "loss": 0.4325,
+      "step": 10132
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2585227272727277e-06,
+      "loss": 0.4069,
+      "step": 10133
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.253787878787879e-06,
+      "loss": 0.2236,
+      "step": 10134
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2490530303030306e-06,
+      "loss": 0.3196,
+      "step": 10135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2443181818181818e-06,
+      "loss": 0.378,
+      "step": 10136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2395833333333334e-06,
+      "loss": 0.3035,
+      "step": 10137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.234848484848485e-06,
+      "loss": 0.2992,
+      "step": 10138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2301136363636363e-06,
+      "loss": 0.3183,
+      "step": 10139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.225378787878788e-06,
+      "loss": 0.2864,
+      "step": 10140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2206439393939396e-06,
+      "loss": 0.4468,
+      "step": 10141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2159090909090912e-06,
+      "loss": 0.3363,
+      "step": 10142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2111742424242424e-06,
+      "loss": 0.2702,
+      "step": 10143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.206439393939394e-06,
+      "loss": 0.4109,
+      "step": 10144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2017045454545457e-06,
+      "loss": 0.2965,
+      "step": 10145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.196969696969697e-06,
+      "loss": 0.3795,
+      "step": 10146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1922348484848486e-06,
+      "loss": 0.2788,
+      "step": 10147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1875e-06,
+      "loss": 0.3624,
+      "step": 10148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1827651515151515e-06,
+      "loss": 0.3333,
+      "step": 10149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.178030303030303e-06,
+      "loss": 0.3608,
+      "step": 10150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1732954545454547e-06,
+      "loss": 0.4114,
+      "step": 10151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1685606060606064e-06,
+      "loss": 0.32,
+      "step": 10152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1638257575757576e-06,
+      "loss": 0.2776,
+      "step": 10153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1590909090909092e-06,
+      "loss": 0.3038,
+      "step": 10154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.154356060606061e-06,
+      "loss": 0.5211,
+      "step": 10155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.149621212121212e-06,
+      "loss": 0.2395,
+      "step": 10156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1448863636363638e-06,
+      "loss": 0.4241,
+      "step": 10157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.140151515151515e-06,
+      "loss": 0.3085,
+      "step": 10158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1354166666666666e-06,
+      "loss": 0.2516,
+      "step": 10159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1306818181818183e-06,
+      "loss": 0.4045,
+      "step": 10160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.12594696969697e-06,
+      "loss": 0.3329,
+      "step": 10161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1212121212121216e-06,
+      "loss": 0.2906,
+      "step": 10162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1164772727272728e-06,
+      "loss": 0.2666,
+      "step": 10163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1117424242424244e-06,
+      "loss": 0.3533,
+      "step": 10164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1070075757575756e-06,
+      "loss": 0.372,
+      "step": 10165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1022727272727273e-06,
+      "loss": 0.3384,
+      "step": 10166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.097537878787879e-06,
+      "loss": 0.3348,
+      "step": 10167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.09280303030303e-06,
+      "loss": 0.4553,
+      "step": 10168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0880681818181822e-06,
+      "loss": 0.3689,
+      "step": 10169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 0.455,
+      "step": 10170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.078598484848485e-06,
+      "loss": 0.3393,
+      "step": 10171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0738636363636367e-06,
+      "loss": 0.4926,
+      "step": 10172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.069128787878788e-06,
+      "loss": 0.4213,
+      "step": 10173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0643939393939396e-06,
+      "loss": 0.3498,
+      "step": 10174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.059659090909091e-06,
+      "loss": 0.3067,
+      "step": 10175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0549242424242424e-06,
+      "loss": 0.2541,
+      "step": 10176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.050189393939394e-06,
+      "loss": 0.2716,
+      "step": 10177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0454545454545457e-06,
+      "loss": 0.2814,
+      "step": 10178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0407196969696974e-06,
+      "loss": 0.3975,
+      "step": 10179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0359848484848486e-06,
+      "loss": 0.4426,
+      "step": 10180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0312500000000002e-06,
+      "loss": 0.3313,
+      "step": 10181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0265151515151515e-06,
+      "loss": 0.3867,
+      "step": 10182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.021780303030303e-06,
+      "loss": 0.2557,
+      "step": 10183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0170454545454548e-06,
+      "loss": 0.3281,
+      "step": 10184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.012310606060606e-06,
+      "loss": 0.3748,
+      "step": 10185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0075757575757576e-06,
+      "loss": 0.4633,
+      "step": 10186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0028409090909093e-06,
+      "loss": 0.4153,
+      "step": 10187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.998106060606061e-06,
+      "loss": 0.359,
+      "step": 10188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9933712121212125e-06,
+      "loss": 0.2571,
+      "step": 10189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9886363636363638e-06,
+      "loss": 0.3145,
+      "step": 10190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9839015151515154e-06,
+      "loss": 0.3391,
+      "step": 10191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9791666666666666e-06,
+      "loss": 0.3458,
+      "step": 10192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9744318181818183e-06,
+      "loss": 0.4535,
+      "step": 10193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9696969696969695e-06,
+      "loss": 0.4445,
+      "step": 10194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.964962121212121e-06,
+      "loss": 0.5515,
+      "step": 10195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9602272727272728e-06,
+      "loss": 0.3829,
+      "step": 10196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9554924242424244e-06,
+      "loss": 0.4565,
+      "step": 10197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.950757575757576e-06,
+      "loss": 0.3559,
+      "step": 10198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9460227272727273e-06,
+      "loss": 0.4317,
+      "step": 10199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.941287878787879e-06,
+      "loss": 0.3648,
+      "step": 10200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9365530303030306e-06,
+      "loss": 0.4063,
+      "step": 10201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.931818181818182e-06,
+      "loss": 0.3239,
+      "step": 10202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9270833333333334e-06,
+      "loss": 0.3902,
+      "step": 10203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9223484848484847e-06,
+      "loss": 0.4285,
+      "step": 10204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9176136363636363e-06,
+      "loss": 0.4605,
+      "step": 10205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.912878787878788e-06,
+      "loss": 0.385,
+      "step": 10206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9081439393939396e-06,
+      "loss": 0.3386,
+      "step": 10207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.903409090909091e-06,
+      "loss": 0.3217,
+      "step": 10208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8986742424242425e-06,
+      "loss": 0.3966,
+      "step": 10209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8939393939393941e-06,
+      "loss": 0.4389,
+      "step": 10210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8892045454545453e-06,
+      "loss": 0.2636,
+      "step": 10211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.884469696969697e-06,
+      "loss": 0.3851,
+      "step": 10212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8797348484848486e-06,
+      "loss": 0.5036,
+      "step": 10213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.875e-06,
+      "loss": 0.3423,
+      "step": 10214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8702651515151517e-06,
+      "loss": 0.396,
+      "step": 10215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865530303030303e-06,
+      "loss": 0.2978,
+      "step": 10216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8607954545454546e-06,
+      "loss": 0.2692,
+      "step": 10217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8560606060606062e-06,
+      "loss": 0.3856,
+      "step": 10218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8513257575757576e-06,
+      "loss": 0.3682,
+      "step": 10219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8465909090909093e-06,
+      "loss": 0.3945,
+      "step": 10220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8418560606060605e-06,
+      "loss": 0.3782,
+      "step": 10221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8371212121212121e-06,
+      "loss": 0.3163,
+      "step": 10222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8323863636363636e-06,
+      "loss": 0.2427,
+      "step": 10223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8276515151515152e-06,
+      "loss": 0.3771,
+      "step": 10224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8229166666666669e-06,
+      "loss": 0.3873,
+      "step": 10225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.818181818181818e-06,
+      "loss": 0.3751,
+      "step": 10226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.81344696969697e-06,
+      "loss": 0.2935,
+      "step": 10227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8087121212121212e-06,
+      "loss": 0.4273,
+      "step": 10228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8039772727272728e-06,
+      "loss": 0.3844,
+      "step": 10229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7992424242424244e-06,
+      "loss": 0.4614,
+      "step": 10230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7945075757575759e-06,
+      "loss": 0.359,
+      "step": 10231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7897727272727275e-06,
+      "loss": 0.3399,
+      "step": 10232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7850378787878787e-06,
+      "loss": 0.283,
+      "step": 10233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7803030303030304e-06,
+      "loss": 0.3634,
+      "step": 10234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.775568181818182e-06,
+      "loss": 0.3409,
+      "step": 10235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7708333333333335e-06,
+      "loss": 0.4098,
+      "step": 10236
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.766098484848485e-06,
+      "loss": 0.408,
+      "step": 10237
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7613636363636363e-06,
+      "loss": 0.327,
+      "step": 10238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.756628787878788e-06,
+      "loss": 0.2885,
+      "step": 10239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7518939393939394e-06,
+      "loss": 0.4195,
+      "step": 10240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.747159090909091e-06,
+      "loss": 0.3356,
+      "step": 10241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7424242424242427e-06,
+      "loss": 0.2978,
+      "step": 10242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.737689393939394e-06,
+      "loss": 0.418,
+      "step": 10243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7329545454545456e-06,
+      "loss": 0.4739,
+      "step": 10244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.728219696969697e-06,
+      "loss": 0.355,
+      "step": 10245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7234848484848486e-06,
+      "loss": 0.3202,
+      "step": 10246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7187500000000003e-06,
+      "loss": 0.4101,
+      "step": 10247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7140151515151515e-06,
+      "loss": 0.3237,
+      "step": 10248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7092803030303031e-06,
+      "loss": 0.3092,
+      "step": 10249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7045454545454546e-06,
+      "loss": 0.3592,
+      "step": 10250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6998106060606062e-06,
+      "loss": 0.3357,
+      "step": 10251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6950757575757579e-06,
+      "loss": 0.371,
+      "step": 10252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.690340909090909e-06,
+      "loss": 0.3283,
+      "step": 10253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6856060606060607e-06,
+      "loss": 0.4181,
+      "step": 10254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6808712121212122e-06,
+      "loss": 0.3569,
+      "step": 10255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6761363636363638e-06,
+      "loss": 0.3429,
+      "step": 10256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.671401515151515e-06,
+      "loss": 0.3935,
+      "step": 10257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.4016,
+      "step": 10258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6619318181818183e-06,
+      "loss": 0.253,
+      "step": 10259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6571969696969697e-06,
+      "loss": 0.3511,
+      "step": 10260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6524621212121214e-06,
+      "loss": 0.3708,
+      "step": 10261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6477272727272726e-06,
+      "loss": 0.3126,
+      "step": 10262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6429924242424242e-06,
+      "loss": 0.4195,
+      "step": 10263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6382575757575759e-06,
+      "loss": 0.5364,
+      "step": 10264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6335227272727273e-06,
+      "loss": 0.3563,
+      "step": 10265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.628787878787879e-06,
+      "loss": 0.2985,
+      "step": 10266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6240530303030302e-06,
+      "loss": 0.2868,
+      "step": 10267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6193181818181818e-06,
+      "loss": 0.4136,
+      "step": 10268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6145833333333335e-06,
+      "loss": 0.4104,
+      "step": 10269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.609848484848485e-06,
+      "loss": 0.5594,
+      "step": 10270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6051136363636366e-06,
+      "loss": 0.4045,
+      "step": 10271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6003787878787878e-06,
+      "loss": 0.3004,
+      "step": 10272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5956439393939394e-06,
+      "loss": 0.4004,
+      "step": 10273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5909090909090908e-06,
+      "loss": 0.3669,
+      "step": 10274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5861742424242425e-06,
+      "loss": 0.3881,
+      "step": 10275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5814393939393941e-06,
+      "loss": 0.3216,
+      "step": 10276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5767045454545454e-06,
+      "loss": 0.332,
+      "step": 10277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5719696969696972e-06,
+      "loss": 0.4408,
+      "step": 10278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5672348484848484e-06,
+      "loss": 0.3975,
+      "step": 10279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5625e-06,
+      "loss": 0.2882,
+      "step": 10280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5577651515151515e-06,
+      "loss": 0.329,
+      "step": 10281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.553030303030303e-06,
+      "loss": 0.3406,
+      "step": 10282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5482954545454548e-06,
+      "loss": 0.3896,
+      "step": 10283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5435606060606062e-06,
+      "loss": 0.3565,
+      "step": 10284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5388257575757577e-06,
+      "loss": 0.3671,
+      "step": 10285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.534090909090909e-06,
+      "loss": 0.3293,
+      "step": 10286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5293560606060607e-06,
+      "loss": 0.3552,
+      "step": 10287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5246212121212122e-06,
+      "loss": 0.3136,
+      "step": 10288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5198863636363638e-06,
+      "loss": 0.3811,
+      "step": 10289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 0.4251,
+      "step": 10290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5104166666666667e-06,
+      "loss": 0.3535,
+      "step": 10291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5056818181818183e-06,
+      "loss": 0.3764,
+      "step": 10292
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5009469696969698e-06,
+      "loss": 0.2717,
+      "step": 10293
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4962121212121212e-06,
+      "loss": 0.342,
+      "step": 10294
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4914772727272728e-06,
+      "loss": 0.3614,
+      "step": 10295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4867424242424243e-06,
+      "loss": 0.3936,
+      "step": 10296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.482007575757576e-06,
+      "loss": 0.2973,
+      "step": 10297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4772727272727273e-06,
+      "loss": 0.2328,
+      "step": 10298
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4725378787878788e-06,
+      "loss": 0.3624,
+      "step": 10299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4678030303030304e-06,
+      "loss": 0.2753,
+      "step": 10300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4630681818181818e-06,
+      "loss": 0.3847,
+      "step": 10301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4583333333333335e-06,
+      "loss": 0.3722,
+      "step": 10302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.453598484848485e-06,
+      "loss": 0.355,
+      "step": 10303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4488636363636364e-06,
+      "loss": 0.3378,
+      "step": 10304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4441287878787878e-06,
+      "loss": 0.3628,
+      "step": 10305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4393939393939396e-06,
+      "loss": 0.3635,
+      "step": 10306
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.434659090909091e-06,
+      "loss": 0.2472,
+      "step": 10307
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4299242424242425e-06,
+      "loss": 0.4069,
+      "step": 10308
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.425189393939394e-06,
+      "loss": 0.3766,
+      "step": 10309
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4204545454545456e-06,
+      "loss": 0.3638,
+      "step": 10310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.415719696969697e-06,
+      "loss": 0.3593,
+      "step": 10311
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4109848484848487e-06,
+      "loss": 0.3773,
+      "step": 10312
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.40625e-06,
+      "loss": 0.4612,
+      "step": 10313
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4015151515151515e-06,
+      "loss": 0.3668,
+      "step": 10314
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3967803030303032e-06,
+      "loss": 0.3804,
+      "step": 10315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3920454545454546e-06,
+      "loss": 0.2635,
+      "step": 10316
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3873106060606062e-06,
+      "loss": 0.4081,
+      "step": 10317
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3825757575757577e-06,
+      "loss": 0.4788,
+      "step": 10318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.377840909090909e-06,
+      "loss": 0.3542,
+      "step": 10319
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3731060606060607e-06,
+      "loss": 0.4232,
+      "step": 10320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3683712121212122e-06,
+      "loss": 0.2609,
+      "step": 10321
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 0.3345,
+      "step": 10322
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3589015151515153e-06,
+      "loss": 0.3273,
+      "step": 10323
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3541666666666667e-06,
+      "loss": 0.3114,
+      "step": 10324
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3494318181818183e-06,
+      "loss": 0.3592,
+      "step": 10325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3446969696969698e-06,
+      "loss": 0.2799,
+      "step": 10326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3399621212121212e-06,
+      "loss": 0.4411,
+      "step": 10327
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3352272727272726e-06,
+      "loss": 0.3064,
+      "step": 10328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3304924242424243e-06,
+      "loss": 0.4546,
+      "step": 10329
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.325757575757576e-06,
+      "loss": 0.238,
+      "step": 10330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3210227272727273e-06,
+      "loss": 0.1797,
+      "step": 10331
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3162878787878788e-06,
+      "loss": 0.3829,
+      "step": 10332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3115530303030302e-06,
+      "loss": 0.3401,
+      "step": 10333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3068181818181819e-06,
+      "loss": 0.2883,
+      "step": 10334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3020833333333335e-06,
+      "loss": 0.5101,
+      "step": 10335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.297348484848485e-06,
+      "loss": 0.333,
+      "step": 10336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2926136363636364e-06,
+      "loss": 0.2552,
+      "step": 10337
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.287878787878788e-06,
+      "loss": 0.423,
+      "step": 10338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2831439393939394e-06,
+      "loss": 0.3075,
+      "step": 10339
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.278409090909091e-06,
+      "loss": 0.2031,
+      "step": 10340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2736742424242425e-06,
+      "loss": 0.3858,
+      "step": 10341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.268939393939394e-06,
+      "loss": 0.2696,
+      "step": 10342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2642045454545456e-06,
+      "loss": 0.3334,
+      "step": 10343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.259469696969697e-06,
+      "loss": 0.3018,
+      "step": 10344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2547348484848485e-06,
+      "loss": 0.396,
+      "step": 10345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.25e-06,
+      "loss": 0.3923,
+      "step": 10346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2452651515151515e-06,
+      "loss": 0.4382,
+      "step": 10347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2405303030303032e-06,
+      "loss": 0.2412,
+      "step": 10348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2357954545454546e-06,
+      "loss": 0.2931,
+      "step": 10349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.231060606060606e-06,
+      "loss": 0.4179,
+      "step": 10350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2263257575757575e-06,
+      "loss": 0.291,
+      "step": 10351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2215909090909091e-06,
+      "loss": 0.3005,
+      "step": 10352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2168560606060608e-06,
+      "loss": 0.3546,
+      "step": 10353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 0.3812,
+      "step": 10354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2073863636363636e-06,
+      "loss": 0.4278,
+      "step": 10355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.202651515151515e-06,
+      "loss": 0.331,
+      "step": 10356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1979166666666667e-06,
+      "loss": 0.3484,
+      "step": 10357
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1931818181818183e-06,
+      "loss": 0.3924,
+      "step": 10358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1884469696969698e-06,
+      "loss": 0.4412,
+      "step": 10359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1837121212121212e-06,
+      "loss": 0.3214,
+      "step": 10360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1789772727272726e-06,
+      "loss": 0.4308,
+      "step": 10361
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1742424242424243e-06,
+      "loss": 0.3611,
+      "step": 10362
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.169507575757576e-06,
+      "loss": 0.4762,
+      "step": 10363
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1647727272727274e-06,
+      "loss": 0.3432,
+      "step": 10364
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1600378787878788e-06,
+      "loss": 0.2912,
+      "step": 10365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1553030303030304e-06,
+      "loss": 0.448,
+      "step": 10366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1505681818181819e-06,
+      "loss": 0.3866,
+      "step": 10367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1458333333333333e-06,
+      "loss": 0.4322,
+      "step": 10368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.141098484848485e-06,
+      "loss": 0.4583,
+      "step": 10369
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1363636363636364e-06,
+      "loss": 0.3459,
+      "step": 10370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.131628787878788e-06,
+      "loss": 0.3601,
+      "step": 10371
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1268939393939395e-06,
+      "loss": 0.5054,
+      "step": 10372
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1221590909090909e-06,
+      "loss": 0.4127,
+      "step": 10373
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1174242424242425e-06,
+      "loss": 0.3373,
+      "step": 10374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.112689393939394e-06,
+      "loss": 0.4081,
+      "step": 10375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1079545454545456e-06,
+      "loss": 0.3494,
+      "step": 10376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.103219696969697e-06,
+      "loss": 0.3335,
+      "step": 10377
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0984848484848485e-06,
+      "loss": 0.2322,
+      "step": 10378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.09375e-06,
+      "loss": 0.3235,
+      "step": 10379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0890151515151515e-06,
+      "loss": 0.2862,
+      "step": 10380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0842803030303032e-06,
+      "loss": 0.3127,
+      "step": 10381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0795454545454546e-06,
+      "loss": 0.3267,
+      "step": 10382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.074810606060606e-06,
+      "loss": 0.3551,
+      "step": 10383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0700757575757575e-06,
+      "loss": 0.4485,
+      "step": 10384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0653409090909091e-06,
+      "loss": 0.2046,
+      "step": 10385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0606060606060608e-06,
+      "loss": 0.2168,
+      "step": 10386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0558712121212122e-06,
+      "loss": 0.3629,
+      "step": 10387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0511363636363636e-06,
+      "loss": 0.3399,
+      "step": 10388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.046401515151515e-06,
+      "loss": 0.3015,
+      "step": 10389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0416666666666667e-06,
+      "loss": 0.322,
+      "step": 10390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0369318181818184e-06,
+      "loss": 0.4382,
+      "step": 10391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0321969696969698e-06,
+      "loss": 0.4105,
+      "step": 10392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0274621212121212e-06,
+      "loss": 0.3054,
+      "step": 10393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0227272727272729e-06,
+      "loss": 0.3077,
+      "step": 10394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0179924242424243e-06,
+      "loss": 0.4305,
+      "step": 10395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0132575757575757e-06,
+      "loss": 0.337,
+      "step": 10396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0085227272727274e-06,
+      "loss": 0.3743,
+      "step": 10397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0037878787878788e-06,
+      "loss": 0.3238,
+      "step": 10398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.990530303030305e-07,
+      "loss": 0.2874,
+      "step": 10399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.943181818181819e-07,
+      "loss": 0.4539,
+      "step": 10400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.895833333333333e-07,
+      "loss": 0.3166,
+      "step": 10401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.848484848484847e-07,
+      "loss": 0.3321,
+      "step": 10402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.801136363636364e-07,
+      "loss": 0.3577,
+      "step": 10403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.75378787878788e-07,
+      "loss": 0.4115,
+      "step": 10404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.706439393939395e-07,
+      "loss": 0.3024,
+      "step": 10405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.65909090909091e-07,
+      "loss": 0.2892,
+      "step": 10406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.611742424242423e-07,
+      "loss": 0.2887,
+      "step": 10407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.56439393939394e-07,
+      "loss": 0.2958,
+      "step": 10408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.517045454545455e-07,
+      "loss": 0.3217,
+      "step": 10409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.469696969696971e-07,
+      "loss": 0.4328,
+      "step": 10410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.422348484848485e-07,
+      "loss": 0.3804,
+      "step": 10411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.375e-07,
+      "loss": 0.2664,
+      "step": 10412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.327651515151515e-07,
+      "loss": 0.314,
+      "step": 10413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.280303030303031e-07,
+      "loss": 0.2979,
+      "step": 10414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.232954545454546e-07,
+      "loss": 0.4766,
+      "step": 10415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.185606060606061e-07,
+      "loss": 0.3515,
+      "step": 10416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.138257575757576e-07,
+      "loss": 0.4259,
+      "step": 10417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.09090909090909e-07,
+      "loss": 0.2542,
+      "step": 10418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.043560606060606e-07,
+      "loss": 0.4358,
+      "step": 10419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.996212121212122e-07,
+      "loss": 0.2894,
+      "step": 10420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.948863636363638e-07,
+      "loss": 0.3698,
+      "step": 10421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.901515151515152e-07,
+      "loss": 0.4338,
+      "step": 10422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.854166666666667e-07,
+      "loss": 0.398,
+      "step": 10423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.806818181818182e-07,
+      "loss": 0.4116,
+      "step": 10424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.759469696969697e-07,
+      "loss": 0.3918,
+      "step": 10425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.712121212121213e-07,
+      "loss": 0.3224,
+      "step": 10426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.664772727272728e-07,
+      "loss": 0.2628,
+      "step": 10427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.617424242424243e-07,
+      "loss": 0.4584,
+      "step": 10428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.570075757575757e-07,
+      "loss": 0.3865,
+      "step": 10429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.522727272727273e-07,
+      "loss": 0.4631,
+      "step": 10430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.475378787878789e-07,
+      "loss": 0.294,
+      "step": 10431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.428030303030304e-07,
+      "loss": 0.4008,
+      "step": 10432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.380681818181819e-07,
+      "loss": 0.2775,
+      "step": 10433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 0.3043,
+      "step": 10434
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.285984848484849e-07,
+      "loss": 0.3916,
+      "step": 10435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.238636363636363e-07,
+      "loss": 0.4031,
+      "step": 10436
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.191287878787879e-07,
+      "loss": 0.415,
+      "step": 10437
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.143939393939395e-07,
+      "loss": 0.2412,
+      "step": 10438
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.096590909090909e-07,
+      "loss": 0.452,
+      "step": 10439
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.049242424242425e-07,
+      "loss": 0.3357,
+      "step": 10440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.001893939393939e-07,
+      "loss": 0.3064,
+      "step": 10441
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.954545454545454e-07,
+      "loss": 0.3285,
+      "step": 10442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.907196969696971e-07,
+      "loss": 0.3357,
+      "step": 10443
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.859848484848486e-07,
+      "loss": 0.4123,
+      "step": 10444
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.8125e-07,
+      "loss": 0.3154,
+      "step": 10445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.765151515151515e-07,
+      "loss": 0.3249,
+      "step": 10446
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.717803030303031e-07,
+      "loss": 0.4982,
+      "step": 10447
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.670454545454545e-07,
+      "loss": 0.2981,
+      "step": 10448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.623106060606061e-07,
+      "loss": 0.1878,
+      "step": 10449
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.575757575757576e-07,
+      "loss": 0.4448,
+      "step": 10450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.528409090909092e-07,
+      "loss": 0.3886,
+      "step": 10451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.481060606060606e-07,
+      "loss": 0.3048,
+      "step": 10452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.433712121212121e-07,
+      "loss": 0.3725,
+      "step": 10453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.386363636363637e-07,
+      "loss": 0.4095,
+      "step": 10454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.339015151515152e-07,
+      "loss": 0.2851,
+      "step": 10455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.291666666666667e-07,
+      "loss": 0.3917,
+      "step": 10456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.244318181818182e-07,
+      "loss": 0.3309,
+      "step": 10457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.196969696969698e-07,
+      "loss": 0.3149,
+      "step": 10458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.149621212121213e-07,
+      "loss": 0.3228,
+      "step": 10459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.102272727272728e-07,
+      "loss": 0.3714,
+      "step": 10460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.054924242424243e-07,
+      "loss": 0.5355,
+      "step": 10461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.007575757575758e-07,
+      "loss": 0.3246,
+      "step": 10462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.960227272727273e-07,
+      "loss": 0.3304,
+      "step": 10463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.912878787878788e-07,
+      "loss": 0.4089,
+      "step": 10464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.865530303030304e-07,
+      "loss": 0.5131,
+      "step": 10465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.818181818181818e-07,
+      "loss": 0.3539,
+      "step": 10466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.770833333333333e-07,
+      "loss": 0.3467,
+      "step": 10467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.723484848484849e-07,
+      "loss": 0.3451,
+      "step": 10468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.676136363636363e-07,
+      "loss": 0.3678,
+      "step": 10469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.62878787878788e-07,
+      "loss": 0.3009,
+      "step": 10470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.581439393939394e-07,
+      "loss": 0.3523,
+      "step": 10471
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.534090909090909e-07,
+      "loss": 0.269,
+      "step": 10472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.486742424242425e-07,
+      "loss": 0.4051,
+      "step": 10473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.43939393939394e-07,
+      "loss": 0.3413,
+      "step": 10474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.392045454545455e-07,
+      "loss": 0.4858,
+      "step": 10475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.34469696969697e-07,
+      "loss": 0.3523,
+      "step": 10476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.297348484848485e-07,
+      "loss": 0.3202,
+      "step": 10477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.25e-07,
+      "loss": 0.3041,
+      "step": 10478
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.202651515151516e-07,
+      "loss": 0.4594,
+      "step": 10479
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.15530303030303e-07,
+      "loss": 0.3312,
+      "step": 10480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.107954545454546e-07,
+      "loss": 0.4636,
+      "step": 10481
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 0.3004,
+      "step": 10482
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.013257575757575e-07,
+      "loss": 0.378,
+      "step": 10483
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.965909090909092e-07,
+      "loss": 0.4186,
+      "step": 10484
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.918560606060606e-07,
+      "loss": 0.3511,
+      "step": 10485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.871212121212121e-07,
+      "loss": 0.2488,
+      "step": 10486
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.823863636363637e-07,
+      "loss": 0.3573,
+      "step": 10487
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.776515151515152e-07,
+      "loss": 0.4415,
+      "step": 10488
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.729166666666667e-07,
+      "loss": 0.3312,
+      "step": 10489
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.681818181818182e-07,
+      "loss": 0.3378,
+      "step": 10490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.634469696969697e-07,
+      "loss": 0.4006,
+      "step": 10491
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.587121212121213e-07,
+      "loss": 0.222,
+      "step": 10492
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.539772727272728e-07,
+      "loss": 0.2547,
+      "step": 10493
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.492424242424242e-07,
+      "loss": 0.4973,
+      "step": 10494
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.445075757575758e-07,
+      "loss": 0.3681,
+      "step": 10495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.397727272727273e-07,
+      "loss": 0.4801,
+      "step": 10496
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.350378787878787e-07,
+      "loss": 0.4155,
+      "step": 10497
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.303030303030304e-07,
+      "loss": 0.3897,
+      "step": 10498
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.255681818181818e-07,
+      "loss": 0.3306,
+      "step": 10499
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.208333333333334e-07,
+      "loss": 0.3089,
+      "step": 10500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.160984848484849e-07,
+      "loss": 0.3124,
+      "step": 10501
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.113636363636364e-07,
+      "loss": 0.3088,
+      "step": 10502
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.066287878787879e-07,
+      "loss": 0.3232,
+      "step": 10503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.018939393939394e-07,
+      "loss": 0.3222,
+      "step": 10504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.971590909090909e-07,
+      "loss": 0.2692,
+      "step": 10505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.924242424242424e-07,
+      "loss": 0.3365,
+      "step": 10506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.87689393939394e-07,
+      "loss": 0.422,
+      "step": 10507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.829545454545455e-07,
+      "loss": 0.3349,
+      "step": 10508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.78219696969697e-07,
+      "loss": 0.3762,
+      "step": 10509
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7348484848484853e-07,
+      "loss": 0.2848,
+      "step": 10510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6875e-07,
+      "loss": 0.2641,
+      "step": 10511
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6401515151515155e-07,
+      "loss": 0.3552,
+      "step": 10512
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5928030303030303e-07,
+      "loss": 0.354,
+      "step": 10513
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.545454545454545e-07,
+      "loss": 0.4675,
+      "step": 10514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.498106060606061e-07,
+      "loss": 0.3233,
+      "step": 10515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.450757575757576e-07,
+      "loss": 0.3848,
+      "step": 10516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.403409090909091e-07,
+      "loss": 0.2515,
+      "step": 10517
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3560606060606067e-07,
+      "loss": 0.419,
+      "step": 10518
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3087121212121216e-07,
+      "loss": 0.4433,
+      "step": 10519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2613636363636364e-07,
+      "loss": 0.3668,
+      "step": 10520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.214015151515152e-07,
+      "loss": 0.3372,
+      "step": 10521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1666666666666667e-07,
+      "loss": 0.471,
+      "step": 10522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1193181818181815e-07,
+      "loss": 0.2732,
+      "step": 10523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0719696969696974e-07,
+      "loss": 0.3915,
+      "step": 10524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0246212121212123e-07,
+      "loss": 0.4274,
+      "step": 10525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.977272727272727e-07,
+      "loss": 0.2766,
+      "step": 10526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.929924242424243e-07,
+      "loss": 0.2969,
+      "step": 10527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8825757575757573e-07,
+      "loss": 0.3787,
+      "step": 10528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8352272727272727e-07,
+      "loss": 0.3745,
+      "step": 10529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.787878787878788e-07,
+      "loss": 0.3576,
+      "step": 10530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.740530303030303e-07,
+      "loss": 0.4067,
+      "step": 10531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6931818181818183e-07,
+      "loss": 0.2822,
+      "step": 10532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6458333333333337e-07,
+      "loss": 0.208,
+      "step": 10533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.598484848484849e-07,
+      "loss": 0.3243,
+      "step": 10534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.551136363636364e-07,
+      "loss": 0.3031,
+      "step": 10535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.503787878787879e-07,
+      "loss": 0.4142,
+      "step": 10536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.456439393939394e-07,
+      "loss": 0.5774,
+      "step": 10537
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.409090909090909e-07,
+      "loss": 0.2569,
+      "step": 10538
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3617424242424244e-07,
+      "loss": 0.3093,
+      "step": 10539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.31439393939394e-07,
+      "loss": 0.4212,
+      "step": 10540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2670454545454546e-07,
+      "loss": 0.401,
+      "step": 10541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.21969696969697e-07,
+      "loss": 0.296,
+      "step": 10542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.172348484848485e-07,
+      "loss": 0.456,
+      "step": 10543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.125e-07,
+      "loss": 0.2972,
+      "step": 10544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.077651515151515e-07,
+      "loss": 0.3813,
+      "step": 10545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 0.228,
+      "step": 10546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.982954545454546e-07,
+      "loss": 0.2432,
+      "step": 10547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9356060606060607e-07,
+      "loss": 0.1784,
+      "step": 10548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.888257575757576e-07,
+      "loss": 0.412,
+      "step": 10549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.840909090909091e-07,
+      "loss": 0.413,
+      "step": 10550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7935606060606063e-07,
+      "loss": 0.4045,
+      "step": 10551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.746212121212121e-07,
+      "loss": 0.3606,
+      "step": 10552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6988636363636366e-07,
+      "loss": 0.252,
+      "step": 10553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.651515151515152e-07,
+      "loss": 0.5843,
+      "step": 10554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.604166666666667e-07,
+      "loss": 0.3831,
+      "step": 10555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.556818181818182e-07,
+      "loss": 0.2536,
+      "step": 10556
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.509469696969697e-07,
+      "loss": 0.3268,
+      "step": 10557
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.462121212121212e-07,
+      "loss": 0.2557,
+      "step": 10558
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.414772727272727e-07,
+      "loss": 0.3248,
+      "step": 10559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3674242424242426e-07,
+      "loss": 0.3447,
+      "step": 10560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3200757575757577e-07,
+      "loss": 0.4054,
+      "step": 10561
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2727272727272726e-07,
+      "loss": 0.425,
+      "step": 10562
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.225378787878788e-07,
+      "loss": 0.3801,
+      "step": 10563
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1780303030303034e-07,
+      "loss": 0.3481,
+      "step": 10564
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1306818181818182e-07,
+      "loss": 0.3498,
+      "step": 10565
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0833333333333333e-07,
+      "loss": 0.2887,
+      "step": 10566
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0359848484848487e-07,
+      "loss": 0.459,
+      "step": 10567
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9886363636363636e-07,
+      "loss": 0.3628,
+      "step": 10568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9412878787878787e-07,
+      "loss": 0.3691,
+      "step": 10569
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.893939393939394e-07,
+      "loss": 0.4454,
+      "step": 10570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8465909090909092e-07,
+      "loss": 0.4496,
+      "step": 10571
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7992424242424246e-07,
+      "loss": 0.3426,
+      "step": 10572
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7518939393939394e-07,
+      "loss": 0.3262,
+      "step": 10573
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7045454545454545e-07,
+      "loss": 0.2979,
+      "step": 10574
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.65719696969697e-07,
+      "loss": 0.3851,
+      "step": 10575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.609848484848485e-07,
+      "loss": 0.3326,
+      "step": 10576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5625e-07,
+      "loss": 0.3899,
+      "step": 10577
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5151515151515152e-07,
+      "loss": 0.2775,
+      "step": 10578
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4678030303030304e-07,
+      "loss": 0.3605,
+      "step": 10579
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4204545454545455e-07,
+      "loss": 0.3316,
+      "step": 10580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3731060606060606e-07,
+      "loss": 0.2982,
+      "step": 10581
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.325757575757576e-07,
+      "loss": 0.3613,
+      "step": 10582
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.278409090909091e-07,
+      "loss": 0.311,
+      "step": 10583
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.231060606060606e-07,
+      "loss": 0.3251,
+      "step": 10584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1837121212121213e-07,
+      "loss": 0.4194,
+      "step": 10585
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1363636363636363e-07,
+      "loss": 0.3763,
+      "step": 10586
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0890151515151517e-07,
+      "loss": 0.4211,
+      "step": 10587
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0416666666666667e-07,
+      "loss": 0.3213,
+      "step": 10588
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.943181818181818e-08,
+      "loss": 0.4496,
+      "step": 10589
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.46969696969697e-08,
+      "loss": 0.3995,
+      "step": 10590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.996212121212123e-08,
+      "loss": 0.3966,
+      "step": 10591
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.522727272727273e-08,
+      "loss": 0.2745,
+      "step": 10592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.049242424242425e-08,
+      "loss": 0.2612,
+      "step": 10593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.575757575757576e-08,
+      "loss": 0.5134,
+      "step": 10594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.102272727272727e-08,
+      "loss": 0.3417,
+      "step": 10595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.62878787878788e-08,
+      "loss": 0.519,
+      "step": 10596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.15530303030303e-08,
+      "loss": 0.3192,
+      "step": 10597
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.6818181818181815e-08,
+      "loss": 0.4666,
+      "step": 10598
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.208333333333333e-08,
+      "loss": 0.3151,
+      "step": 10599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.734848484848485e-08,
+      "loss": 0.3522,
+      "step": 10600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.261363636363636e-08,
+      "loss": 0.3949,
+      "step": 10601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.787878787878788e-08,
+      "loss": 0.3288,
+      "step": 10602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.31439393939394e-08,
+      "loss": 0.3307,
+      "step": 10603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8409090909090908e-08,
+      "loss": 0.4438,
+      "step": 10604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3674242424242426e-08,
+      "loss": 0.38,
+      "step": 10605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.893939393939394e-08,
+      "loss": 0.3574,
+      "step": 10606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4204545454545454e-08,
+      "loss": 0.423,
+      "step": 10607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.46969696969697e-09,
+      "loss": 0.3434,
+      "step": 10608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.734848484848485e-09,
+      "loss": 0.3422,
+      "step": 10609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.6542,
+      "step": 10610
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8589505681918598,
+      "eval_loss": 0.3358200788497925,
+      "eval_runtime": 474.8456,
+      "eval_samples_per_second": 357.482,
+      "eval_steps_per_second": 5.587,
+      "step": 10610
+    }
+  ],
+  "max_steps": 10610,
+  "num_train_epochs": 1,
+  "total_flos": 3.145588661252904e+16,
+  "trial_name": null,
+  "trial_params": null
+}