diff --git "a/robot_action_model/trainer_state.json" "b/robot_action_model/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/robot_action_model/trainer_state.json"
@@ -0,0 +1,126016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.345547235940736,
+  "global_step": 21000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.307855626326964e-08,
+      "loss": 13.2394,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0615711252653928e-07,
+      "loss": 13.5785,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5923566878980893e-07,
+      "loss": 13.0081,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1231422505307855e-07,
+      "loss": 13.1462,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6539278131634817e-07,
+      "loss": 13.423,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1847133757961787e-07,
+      "loss": 13.3561,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.715498938428875e-07,
+      "loss": 12.9398,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.246284501061571e-07,
+      "loss": 13.0864,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.777070063694267e-07,
+      "loss": 13.0811,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.307855626326963e-07,
+      "loss": 13.042,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.83864118895966e-07,
+      "loss": 12.8726,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.369426751592357e-07,
+      "loss": 12.9285,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.900212314225053e-07,
+      "loss": 12.974,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.43099787685775e-07,
+      "loss": 12.7314,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.961783439490446e-07,
+      "loss": 12.7048,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.492569002123142e-07,
+      "loss": 12.6257,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.023354564755839e-07,
+      "loss": 12.6064,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.554140127388535e-07,
+      "loss": 12.2679,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.008492569002123e-06,
+      "loss": 12.1374,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0615711252653927e-06,
+      "loss": 12.3263,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1146496815286625e-06,
+      "loss": 11.9696,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.167728237791932e-06,
+      "loss": 11.955,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2208067940552019e-06,
+      "loss": 11.5576,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2738853503184715e-06,
+      "loss": 11.6582,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.326963906581741e-06,
+      "loss": 11.271,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3800424628450107e-06,
+      "loss": 11.1264,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4331210191082802e-06,
+      "loss": 10.9406,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.48619957537155e-06,
+      "loss": 10.7237,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5392781316348196e-06,
+      "loss": 10.6955,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5923566878980892e-06,
+      "loss": 10.6187,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6454352441613588e-06,
+      "loss": 10.3437,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6985138004246284e-06,
+      "loss": 10.1739,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7515923566878982e-06,
+      "loss": 10.1383,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8046709129511678e-06,
+      "loss": 10.2044,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8577494692144374e-06,
+      "loss": 9.7015,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.910828025477707e-06,
+      "loss": 9.2676,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.963906581740977e-06,
+      "loss": 9.0955,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.016985138004246e-06,
+      "loss": 8.7729,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.070063694267516e-06,
+      "loss": 8.7713,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1231422505307854e-06,
+      "loss": 8.4284,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.176220806794055e-06,
+      "loss": 8.1487,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.229299363057325e-06,
+      "loss": 8.0746,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2823779193205943e-06,
+      "loss": 7.6863,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.335456475583864e-06,
+      "loss": 7.5029,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.388535031847134e-06,
+      "loss": 7.3109,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4416135881104038e-06,
+      "loss": 7.0102,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.494692144373673e-06,
+      "loss": 6.9272,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.547770700636943e-06,
+      "loss": 6.5848,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6008492569002127e-06,
+      "loss": 6.7954,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.653927813163482e-06,
+      "loss": 5.9796,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.707006369426752e-06,
+      "loss": 5.8593,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7600849256900213e-06,
+      "loss": 5.4405,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.813163481953291e-06,
+      "loss": 5.3938,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8662420382165605e-06,
+      "loss": 5.3757,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9193205944798303e-06,
+      "loss": 4.9917,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9723991507431e-06,
+      "loss": 4.6684,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0254777070063695e-06,
+      "loss": 4.424,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0785562632696393e-06,
+      "loss": 4.2579,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1316348195329087e-06,
+      "loss": 4.2904,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1847133757961785e-06,
+      "loss": 4.2257,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2377919320594483e-06,
+      "loss": 4.1382,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2908704883227177e-06,
+      "loss": 3.9079,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3439490445859875e-06,
+      "loss": 3.8785,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.397027600849257e-06,
+      "loss": 3.7329,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4501061571125266e-06,
+      "loss": 3.8806,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5031847133757964e-06,
+      "loss": 3.5959,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.556263269639066e-06,
+      "loss": 3.8628,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6093418259023356e-06,
+      "loss": 3.4981,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.662420382165605e-06,
+      "loss": 3.6243,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.715498938428875e-06,
+      "loss": 3.5661,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7685774946921446e-06,
+      "loss": 3.2206,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.821656050955414e-06,
+      "loss": 3.3048,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.874734607218684e-06,
+      "loss": 3.1822,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.927813163481954e-06,
+      "loss": 3.3145,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9808917197452226e-06,
+      "loss": 3.2493,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.033970276008492e-06,
+      "loss": 3.2409,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.087048832271762e-06,
+      "loss": 3.0861,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.140127388535032e-06,
+      "loss": 2.884,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.193205944798302e-06,
+      "loss": 2.9362,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.246284501061571e-06,
+      "loss": 2.8918,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2993630573248405e-06,
+      "loss": 2.8552,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.35244161358811e-06,
+      "loss": 2.7369,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.40552016985138e-06,
+      "loss": 2.6513,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.45859872611465e-06,
+      "loss": 2.9522,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.511677282377919e-06,
+      "loss": 2.6039,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.564755838641189e-06,
+      "loss": 2.6981,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6178343949044585e-06,
+      "loss": 2.6402,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.670912951167728e-06,
+      "loss": 2.3836,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.723991507430998e-06,
+      "loss": 2.4552,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.777070063694268e-06,
+      "loss": 2.5614,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.830148619957538e-06,
+      "loss": 2.6077,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8832271762208075e-06,
+      "loss": 2.2563,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.936305732484077e-06,
+      "loss": 2.3505,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989384288747346e-06,
+      "loss": 2.3402,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.042462845010616e-06,
+      "loss": 2.0426,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.095541401273886e-06,
+      "loss": 2.3866,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.148619957537156e-06,
+      "loss": 2.1927,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2016985138004255e-06,
+      "loss": 2.1204,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2547770700636944e-06,
+      "loss": 2.2482,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.307855626326964e-06,
+      "loss": 2.4146,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.360934182590234e-06,
+      "loss": 2.3019,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.414012738853504e-06,
+      "loss": 2.3379,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.467091295116774e-06,
+      "loss": 2.2618,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.520169851380043e-06,
+      "loss": 2.0462,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.573248407643312e-06,
+      "loss": 2.2584,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.626326963906582e-06,
+      "loss": 1.9958,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.679405520169852e-06,
+      "loss": 2.1784,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.732484076433121e-06,
+      "loss": 2.0382,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.785562632696391e-06,
+      "loss": 1.976,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.838641188959661e-06,
+      "loss": 1.9094,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.89171974522293e-06,
+      "loss": 2.0613,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9447983014862e-06,
+      "loss": 1.9439,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.997876857749469e-06,
+      "loss": 1.9312,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.050955414012739e-06,
+      "loss": 2.1243,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.104033970276009e-06,
+      "loss": 1.9707,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.1571125265392786e-06,
+      "loss": 1.9748,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.210191082802548e-06,
+      "loss": 1.9749,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.263269639065817e-06,
+      "loss": 1.979,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.316348195329087e-06,
+      "loss": 1.8732,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.369426751592357e-06,
+      "loss": 1.9548,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.422505307855627e-06,
+      "loss": 1.9318,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4755838641188965e-06,
+      "loss": 1.9402,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5286624203821655e-06,
+      "loss": 1.8525,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.581740976645435e-06,
+      "loss": 1.8138,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.634819532908705e-06,
+      "loss": 1.972,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.687898089171975e-06,
+      "loss": 1.9255,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.740976645435245e-06,
+      "loss": 1.7673,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.794055201698514e-06,
+      "loss": 1.8791,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.8471337579617835e-06,
+      "loss": 1.8743,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.900212314225053e-06,
+      "loss": 2.1206,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.953290870488323e-06,
+      "loss": 1.9329,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.006369426751593e-06,
+      "loss": 1.9081,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.059447983014862e-06,
+      "loss": 2.0504,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.112526539278132e-06,
+      "loss": 1.9007,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.1656050955414014e-06,
+      "loss": 1.9854,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.218683651804671e-06,
+      "loss": 1.9312,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.271762208067941e-06,
+      "loss": 1.7299,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.32484076433121e-06,
+      "loss": 1.8772,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.37791932059448e-06,
+      "loss": 1.9035,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.43099787685775e-06,
+      "loss": 1.8971,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.484076433121019e-06,
+      "loss": 1.8346,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.537154989384289e-06,
+      "loss": 1.8914,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.590233545647558e-06,
+      "loss": 1.9371,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.643312101910828e-06,
+      "loss": 1.8659,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.696390658174098e-06,
+      "loss": 1.8345,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.749469214437368e-06,
+      "loss": 1.8419,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.802547770700637e-06,
+      "loss": 1.8226,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.855626326963907e-06,
+      "loss": 1.8649,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.908704883227177e-06,
+      "loss": 1.7414,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.961783439490445e-06,
+      "loss": 1.7607,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.014861995753715e-06,
+      "loss": 1.9958,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.067940552016985e-06,
+      "loss": 1.9977,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.121019108280255e-06,
+      "loss": 1.948,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.174097664543524e-06,
+      "loss": 1.8801,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.227176220806794e-06,
+      "loss": 1.7973,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.280254777070064e-06,
+      "loss": 1.762,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.8774,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.386411889596604e-06,
+      "loss": 1.7778,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.439490445859873e-06,
+      "loss": 1.7982,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.492569002123141e-06,
+      "loss": 1.876,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.545647558386411e-06,
+      "loss": 1.7986,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.598726114649681e-06,
+      "loss": 1.8512,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.651804670912951e-06,
+      "loss": 1.7415,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.70488322717622e-06,
+      "loss": 1.9235,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.75796178343949e-06,
+      "loss": 1.8396,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.81104033970276e-06,
+      "loss": 1.8685,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.86411889596603e-06,
+      "loss": 1.8122,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.9171974522293e-06,
+      "loss": 1.7921,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.970276008492568e-06,
+      "loss": 1.7168,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.023354564755838e-06,
+      "loss": 1.8416,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.076433121019108e-06,
+      "loss": 1.7362,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.129511677282377e-06,
+      "loss": 1.8464,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.182590233545647e-06,
+      "loss": 1.8438,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.235668789808917e-06,
+      "loss": 1.7913,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.288747346072187e-06,
+      "loss": 1.7962,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.341825902335457e-06,
+      "loss": 1.7934,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.394904458598726e-06,
+      "loss": 1.7505,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.447983014861996e-06,
+      "loss": 1.8372,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.501061571125266e-06,
+      "loss": 1.6862,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.554140127388536e-06,
+      "loss": 1.7439,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.607218683651806e-06,
+      "loss": 1.8254,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.660297239915075e-06,
+      "loss": 1.7703,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.713375796178345e-06,
+      "loss": 1.7435,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.766454352441615e-06,
+      "loss": 1.7457,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.819532908704885e-06,
+      "loss": 1.7817,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.872611464968155e-06,
+      "loss": 1.838,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.925690021231423e-06,
+      "loss": 1.7542,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978768577494693e-06,
+      "loss": 1.8811,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0031847133757962e-05,
+      "loss": 1.7568,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0084925690021232e-05,
+      "loss": 1.7377,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0138004246284502e-05,
+      "loss": 1.9874,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0191082802547772e-05,
+      "loss": 1.8294,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0244161358811042e-05,
+      "loss": 1.7966,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0297239915074311e-05,
+      "loss": 1.7603,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0350318471337581e-05,
+      "loss": 1.8469,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0403397027600851e-05,
+      "loss": 1.848,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0456475583864119e-05,
+      "loss": 1.7503,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0509554140127389e-05,
+      "loss": 1.7525,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0562632696390659e-05,
+      "loss": 1.7769,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0615711252653929e-05,
+      "loss": 1.6932,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0668789808917198e-05,
+      "loss": 1.8658,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0721868365180468e-05,
+      "loss": 1.8308,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0774946921443738e-05,
+      "loss": 1.7774,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0828025477707008e-05,
+      "loss": 1.8035,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0881104033970278e-05,
+      "loss": 1.7462,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0934182590233547e-05,
+      "loss": 1.8383,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0987261146496815e-05,
+      "loss": 1.8283,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1040339702760085e-05,
+      "loss": 1.804,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1093418259023355e-05,
+      "loss": 1.7737,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1146496815286625e-05,
+      "loss": 1.7962,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1199575371549895e-05,
+      "loss": 1.7451,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1252653927813164e-05,
+      "loss": 1.7099,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1305732484076434e-05,
+      "loss": 1.7736,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1358811040339704e-05,
+      "loss": 1.8268,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1411889596602974e-05,
+      "loss": 1.7221,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1464968152866242e-05,
+      "loss": 1.707,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1518046709129512e-05,
+      "loss": 1.7096,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1571125265392782e-05,
+      "loss": 1.7667,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1624203821656051e-05,
+      "loss": 1.8853,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1677282377919321e-05,
+      "loss": 1.7017,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1730360934182591e-05,
+      "loss": 1.7002,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.178343949044586e-05,
+      "loss": 1.7738,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.183651804670913e-05,
+      "loss": 1.7321,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.18895966029724e-05,
+      "loss": 1.7405,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.194267515923567e-05,
+      "loss": 1.7042,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1995753715498938e-05,
+      "loss": 1.757,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2048832271762208e-05,
+      "loss": 1.824,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2101910828025478e-05,
+      "loss": 1.7749,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2154989384288748e-05,
+      "loss": 1.6916,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2208067940552018e-05,
+      "loss": 1.8248,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2261146496815287e-05,
+      "loss": 1.747,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2314225053078557e-05,
+      "loss": 1.7767,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2367303609341827e-05,
+      "loss": 1.8638,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2420382165605097e-05,
+      "loss": 1.883,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2473460721868367e-05,
+      "loss": 1.695,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2526539278131635e-05,
+      "loss": 1.7912,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2579617834394904e-05,
+      "loss": 1.7668,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2632696390658174e-05,
+      "loss": 1.7254,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2685774946921444e-05,
+      "loss": 1.8065,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2738853503184714e-05,
+      "loss": 1.7407,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2791932059447984e-05,
+      "loss": 1.8269,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2845010615711253e-05,
+      "loss": 1.6442,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2898089171974523e-05,
+      "loss": 1.76,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2951167728237793e-05,
+      "loss": 1.8382,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3004246284501063e-05,
+      "loss": 1.754,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3057324840764331e-05,
+      "loss": 1.7527,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.31104033970276e-05,
+      "loss": 1.7334,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.316348195329087e-05,
+      "loss": 1.7709,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.321656050955414e-05,
+      "loss": 1.7049,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.326963906581741e-05,
+      "loss": 1.7254,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.332271762208068e-05,
+      "loss": 1.7452,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.337579617834395e-05,
+      "loss": 1.7457,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.342887473460722e-05,
+      "loss": 1.8252,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.348195329087049e-05,
+      "loss": 1.7939,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.353503184713376e-05,
+      "loss": 1.779,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3588110403397027e-05,
+      "loss": 1.6789,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3641188959660297e-05,
+      "loss": 1.8051,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3694267515923567e-05,
+      "loss": 1.7489,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3747346072186837e-05,
+      "loss": 1.7917,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3800424628450107e-05,
+      "loss": 1.7999,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3853503184713376e-05,
+      "loss": 1.7932,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3906581740976646e-05,
+      "loss": 1.7785,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3959660297239916e-05,
+      "loss": 1.8581,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4012738853503186e-05,
+      "loss": 1.6873,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4065817409766454e-05,
+      "loss": 1.7676,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4118895966029724e-05,
+      "loss": 1.7536,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4171974522292993e-05,
+      "loss": 1.7727,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4225053078556263e-05,
+      "loss": 1.8111,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4278131634819533e-05,
+      "loss": 1.7551,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4331210191082803e-05,
+      "loss": 1.7625,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4384288747346073e-05,
+      "loss": 1.7789,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4437367303609342e-05,
+      "loss": 1.7994,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4490445859872612e-05,
+      "loss": 1.7413,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4543524416135882e-05,
+      "loss": 1.8493,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.459660297239915e-05,
+      "loss": 1.7341,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.464968152866242e-05,
+      "loss": 1.6797,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.470276008492569e-05,
+      "loss": 1.7534,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.475583864118896e-05,
+      "loss": 1.6369,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.480891719745223e-05,
+      "loss": 1.7259,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.48619957537155e-05,
+      "loss": 1.8071,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4915074309978769e-05,
+      "loss": 1.6766,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4968152866242039e-05,
+      "loss": 1.7492,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5021231422505309e-05,
+      "loss": 1.8395,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5074309978768578e-05,
+      "loss": 1.7494,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5127388535031847e-05,
+      "loss": 1.8256,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5180467091295116e-05,
+      "loss": 1.7491,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5233545647558386e-05,
+      "loss": 1.7623,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5286624203821656e-05,
+      "loss": 1.7713,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5339702760084927e-05,
+      "loss": 1.7782,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5392781316348196e-05,
+      "loss": 1.7375,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5445859872611464e-05,
+      "loss": 1.8366,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5498938428874735e-05,
+      "loss": 1.7118,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5552016985138003e-05,
+      "loss": 1.8163,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5605095541401275e-05,
+      "loss": 1.8422,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5658174097664543e-05,
+      "loss": 1.7868,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5711252653927814e-05,
+      "loss": 1.7494,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5764331210191083e-05,
+      "loss": 1.78,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5817409766454354e-05,
+      "loss": 1.7398,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5870488322717622e-05,
+      "loss": 1.7319,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.592356687898089e-05,
+      "loss": 1.7439,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5976645435244162e-05,
+      "loss": 1.6106,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.602972399150743e-05,
+      "loss": 1.7283,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.60828025477707e-05,
+      "loss": 1.8201,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.613588110403397e-05,
+      "loss": 1.8263,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.618895966029724e-05,
+      "loss": 1.76,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.624203821656051e-05,
+      "loss": 1.8246,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.629511677282378e-05,
+      "loss": 1.7459,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.634819532908705e-05,
+      "loss": 1.7105,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6401273885350317e-05,
+      "loss": 1.6788,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6454352441613588e-05,
+      "loss": 1.7235,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6507430997876856e-05,
+      "loss": 1.7301,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6560509554140128e-05,
+      "loss": 1.7603,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6613588110403396e-05,
+      "loss": 1.785,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.7839,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6719745222929936e-05,
+      "loss": 1.8071,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6772823779193207e-05,
+      "loss": 1.7206,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6825902335456475e-05,
+      "loss": 1.7958,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6878980891719747e-05,
+      "loss": 1.7571,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6932059447983015e-05,
+      "loss": 1.7329,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6985138004246283e-05,
+      "loss": 1.7394,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7038216560509554e-05,
+      "loss": 1.7653,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7091295116772823e-05,
+      "loss": 1.6213,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7144373673036094e-05,
+      "loss": 1.7928,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7197452229299362e-05,
+      "loss": 1.7673,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7250530785562634e-05,
+      "loss": 1.6208,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7303609341825902e-05,
+      "loss": 1.7639,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7356687898089173e-05,
+      "loss": 1.8074,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.740976645435244e-05,
+      "loss": 1.766,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.746284501061571e-05,
+      "loss": 1.7995,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.751592356687898e-05,
+      "loss": 1.7793,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.756900212314225e-05,
+      "loss": 1.7708,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.762208067940552e-05,
+      "loss": 1.7271,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.767515923566879e-05,
+      "loss": 1.7913,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.772823779193206e-05,
+      "loss": 1.7497,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7781316348195328e-05,
+      "loss": 1.7389,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.78343949044586e-05,
+      "loss": 1.7852,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7887473460721868e-05,
+      "loss": 1.7337,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7940552016985136e-05,
+      "loss": 1.7797,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7993630573248407e-05,
+      "loss": 1.7064,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8046709129511676e-05,
+      "loss": 1.6687,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8099787685774947e-05,
+      "loss": 1.6851,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8152866242038215e-05,
+      "loss": 1.7716,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8205944798301487e-05,
+      "loss": 1.7418,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8259023354564755e-05,
+      "loss": 1.7734,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8312101910828026e-05,
+      "loss": 1.7266,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8365180467091294e-05,
+      "loss": 1.7366,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8418259023354566e-05,
+      "loss": 1.8191,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8471337579617834e-05,
+      "loss": 1.7777,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8524416135881102e-05,
+      "loss": 1.6773,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8577494692144374e-05,
+      "loss": 1.7571,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8630573248407642e-05,
+      "loss": 1.7991,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8683651804670913e-05,
+      "loss": 1.6566,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.873673036093418e-05,
+      "loss": 1.7523,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8789808917197453e-05,
+      "loss": 1.7112,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8842887473460724e-05,
+      "loss": 1.7251,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8895966029723992e-05,
+      "loss": 1.6962,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8949044585987264e-05,
+      "loss": 1.7735,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9002123142250532e-05,
+      "loss": 1.6943,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9055201698513804e-05,
+      "loss": 1.6965,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.910828025477707e-05,
+      "loss": 1.7586,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9161358811040343e-05,
+      "loss": 1.7877,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.921443736730361e-05,
+      "loss": 1.7255,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9267515923566883e-05,
+      "loss": 1.8001,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.932059447983015e-05,
+      "loss": 1.7364,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.937367303609342e-05,
+      "loss": 1.7547,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.942675159235669e-05,
+      "loss": 1.7793,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.947983014861996e-05,
+      "loss": 1.7245,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.953290870488323e-05,
+      "loss": 1.8604,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9585987261146498e-05,
+      "loss": 1.7412,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.963906581740977e-05,
+      "loss": 1.742,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9692144373673038e-05,
+      "loss": 1.7016,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.974522292993631e-05,
+      "loss": 1.8565,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9798301486199577e-05,
+      "loss": 1.7629,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9851380042462846e-05,
+      "loss": 1.7308,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9904458598726117e-05,
+      "loss": 1.7234,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957537154989385e-05,
+      "loss": 1.69,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0010615711252657e-05,
+      "loss": 1.6764,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0063694267515925e-05,
+      "loss": 1.7348,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0116772823779196e-05,
+      "loss": 1.7891,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0169851380042464e-05,
+      "loss": 1.6461,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0222929936305736e-05,
+      "loss": 1.7279,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0276008492569004e-05,
+      "loss": 1.7295,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0329087048832275e-05,
+      "loss": 1.725,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0382165605095544e-05,
+      "loss": 1.7318,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.043524416135881e-05,
+      "loss": 1.7745,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0488322717622083e-05,
+      "loss": 1.6553,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.054140127388535e-05,
+      "loss": 1.7428,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0594479830148623e-05,
+      "loss": 1.764,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.064755838641189e-05,
+      "loss": 1.7641,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0700636942675162e-05,
+      "loss": 1.72,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.075371549893843e-05,
+      "loss": 1.7725,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0806794055201702e-05,
+      "loss": 1.755,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.085987261146497e-05,
+      "loss": 1.843,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0912951167728238e-05,
+      "loss": 1.6745,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.096602972399151e-05,
+      "loss": 1.8095,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1019108280254778e-05,
+      "loss": 1.7815,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.107218683651805e-05,
+      "loss": 1.6943,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1125265392781317e-05,
+      "loss": 1.7335,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.117834394904459e-05,
+      "loss": 1.7863,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1231422505307857e-05,
+      "loss": 1.7079,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.128450106157113e-05,
+      "loss": 1.928,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1337579617834397e-05,
+      "loss": 1.7933,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1390658174097665e-05,
+      "loss": 1.7624,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1443736730360936e-05,
+      "loss": 1.7532,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1496815286624204e-05,
+      "loss": 1.7005,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1549893842887476e-05,
+      "loss": 1.7096,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1602972399150744e-05,
+      "loss": 1.7014,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1656050955414015e-05,
+      "loss": 1.7562,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1709129511677284e-05,
+      "loss": 1.7051,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1762208067940555e-05,
+      "loss": 1.7441,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1815286624203823e-05,
+      "loss": 1.7231,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1868365180467095e-05,
+      "loss": 1.6898,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921443736730363e-05,
+      "loss": 1.6999,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.197452229299363e-05,
+      "loss": 1.659,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2027600849256902e-05,
+      "loss": 1.73,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.208067940552017e-05,
+      "loss": 1.7718,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2133757961783442e-05,
+      "loss": 1.7391,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.218683651804671e-05,
+      "loss": 1.6849,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.223991507430998e-05,
+      "loss": 1.6969,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.229299363057325e-05,
+      "loss": 1.7341,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.234607218683652e-05,
+      "loss": 1.6623,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.239915074309979e-05,
+      "loss": 1.7084,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2452229299363057e-05,
+      "loss": 1.7769,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.250530785562633e-05,
+      "loss": 1.7378,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2558386411889597e-05,
+      "loss": 1.7355,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.261146496815287e-05,
+      "loss": 1.8237,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2664543524416137e-05,
+      "loss": 1.6815,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2717622080679408e-05,
+      "loss": 1.7585,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2770700636942676e-05,
+      "loss": 1.764,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2823779193205948e-05,
+      "loss": 1.7687,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2876857749469216e-05,
+      "loss": 1.6967,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2929936305732484e-05,
+      "loss": 1.658,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2983014861995755e-05,
+      "loss": 1.7407,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3036093418259024e-05,
+      "loss": 1.7635,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3089171974522295e-05,
+      "loss": 1.7269,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3142250530785563e-05,
+      "loss": 1.7232,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3195329087048835e-05,
+      "loss": 1.7296,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3248407643312103e-05,
+      "loss": 1.773,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3301486199575374e-05,
+      "loss": 1.7638,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3354564755838642e-05,
+      "loss": 1.7348,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3407643312101914e-05,
+      "loss": 1.7816,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3460721868365182e-05,
+      "loss": 1.6746,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.351380042462845e-05,
+      "loss": 1.7239,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.356687898089172e-05,
+      "loss": 1.8,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.361995753715499e-05,
+      "loss": 1.7302,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.367303609341826e-05,
+      "loss": 1.7168,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.372611464968153e-05,
+      "loss": 1.7488,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.37791932059448e-05,
+      "loss": 1.7816,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.383227176220807e-05,
+      "loss": 1.7422,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.388535031847134e-05,
+      "loss": 1.7034,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.393842887473461e-05,
+      "loss": 1.6768,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3991507430997877e-05,
+      "loss": 1.7316,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4044585987261148e-05,
+      "loss": 1.7459,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4097664543524416e-05,
+      "loss": 1.7408,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4150743099787688e-05,
+      "loss": 1.7735,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4203821656050956e-05,
+      "loss": 1.7786,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4256900212314227e-05,
+      "loss": 1.7398,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4309978768577495e-05,
+      "loss": 1.6686,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4363057324840767e-05,
+      "loss": 1.7383,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4416135881104035e-05,
+      "loss": 1.6986,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4469214437367307e-05,
+      "loss": 1.7243,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4522292993630575e-05,
+      "loss": 1.7103,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4575371549893843e-05,
+      "loss": 1.6854,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4628450106157114e-05,
+      "loss": 1.7729,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4681528662420382e-05,
+      "loss": 1.7461,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4734607218683654e-05,
+      "loss": 1.7199,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4787685774946922e-05,
+      "loss": 1.6926,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4840764331210193e-05,
+      "loss": 1.6162,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.489384288747346e-05,
+      "loss": 1.6949,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4946921443736733e-05,
+      "loss": 1.6942,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5e-05,
+      "loss": 1.7611,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.505307855626327e-05,
+      "loss": 1.7071,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.510615711252654e-05,
+      "loss": 1.7811,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.515923566878981e-05,
+      "loss": 1.6702,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.521231422505308e-05,
+      "loss": 1.785,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.526539278131635e-05,
+      "loss": 1.78,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.531847133757962e-05,
+      "loss": 1.6166,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5371549893842888e-05,
+      "loss": 1.7741,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.542462845010616e-05,
+      "loss": 1.7162,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5477707006369428e-05,
+      "loss": 1.7183,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5530785562632696e-05,
+      "loss": 1.7235,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5583864118895967e-05,
+      "loss": 1.643,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5636942675159235e-05,
+      "loss": 1.7505,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5690021231422507e-05,
+      "loss": 1.7042,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5743099787685775e-05,
+      "loss": 1.7562,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5796178343949047e-05,
+      "loss": 1.6997,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5849256900212315e-05,
+      "loss": 1.7107,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5902335456475586e-05,
+      "loss": 1.7249,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5955414012738854e-05,
+      "loss": 1.7123,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6008492569002126e-05,
+      "loss": 1.7532,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6061571125265394e-05,
+      "loss": 1.7075,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6114649681528662e-05,
+      "loss": 1.7439,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6167728237791933e-05,
+      "loss": 1.7153,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.62208067940552e-05,
+      "loss": 1.8087,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6273885350318473e-05,
+      "loss": 1.6498,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.632696390658174e-05,
+      "loss": 1.7434,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6380042462845013e-05,
+      "loss": 1.7606,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.643312101910828e-05,
+      "loss": 1.6246,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6486199575371552e-05,
+      "loss": 1.739,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.653927813163482e-05,
+      "loss": 1.7452,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.659235668789809e-05,
+      "loss": 1.7566,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.664543524416136e-05,
+      "loss": 1.7023,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6698513800424628e-05,
+      "loss": 1.6305,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.67515923566879e-05,
+      "loss": 1.7754,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6804670912951168e-05,
+      "loss": 1.6847,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.685774946921444e-05,
+      "loss": 1.7802,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6910828025477707e-05,
+      "loss": 1.7356,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.696390658174098e-05,
+      "loss": 1.7267,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7016985138004247e-05,
+      "loss": 1.7084,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.707006369426752e-05,
+      "loss": 1.7519,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7123142250530787e-05,
+      "loss": 1.7402,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7176220806794055e-05,
+      "loss": 1.7067,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7229299363057326e-05,
+      "loss": 1.6315,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7282377919320594e-05,
+      "loss": 1.6718,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7335456475583866e-05,
+      "loss": 1.6731,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7388535031847134e-05,
+      "loss": 1.6517,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7441613588110405e-05,
+      "loss": 1.7357,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7494692144373673e-05,
+      "loss": 1.7597,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7547770700636945e-05,
+      "loss": 1.6275,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7600849256900213e-05,
+      "loss": 1.7389,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.765392781316348e-05,
+      "loss": 1.6022,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7707006369426753e-05,
+      "loss": 1.7589,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.776008492569002e-05,
+      "loss": 1.6912,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7813163481953292e-05,
+      "loss": 1.6904,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.786624203821656e-05,
+      "loss": 1.746,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7919320594479832e-05,
+      "loss": 1.6886,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.79723991507431e-05,
+      "loss": 1.7287,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.802547770700637e-05,
+      "loss": 1.6906,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.807855626326964e-05,
+      "loss": 1.6573,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8131634819532908e-05,
+      "loss": 1.9117,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.818471337579618e-05,
+      "loss": 1.624,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8237791932059447e-05,
+      "loss": 1.6885,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.829087048832272e-05,
+      "loss": 1.6795,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8343949044585987e-05,
+      "loss": 1.6886,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.839702760084926e-05,
+      "loss": 1.7045,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8450106157112527e-05,
+      "loss": 1.6994,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8503184713375798e-05,
+      "loss": 1.7574,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8556263269639066e-05,
+      "loss": 1.7544,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8609341825902338e-05,
+      "loss": 1.7086,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8662420382165606e-05,
+      "loss": 1.7088,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8715498938428874e-05,
+      "loss": 1.7203,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8768577494692145e-05,
+      "loss": 1.5918,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8821656050955413e-05,
+      "loss": 1.7634,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8874734607218685e-05,
+      "loss": 1.6718,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8927813163481953e-05,
+      "loss": 1.6391,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8980891719745225e-05,
+      "loss": 1.6752,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9033970276008493e-05,
+      "loss": 1.7159,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9087048832271764e-05,
+      "loss": 1.6554,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9140127388535032e-05,
+      "loss": 1.7115,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.91932059447983e-05,
+      "loss": 1.7797,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9246284501061572e-05,
+      "loss": 1.6947,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.929936305732484e-05,
+      "loss": 1.6709,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.935244161358811e-05,
+      "loss": 1.7502,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.940552016985138e-05,
+      "loss": 1.771,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.945859872611465e-05,
+      "loss": 1.7101,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.951167728237792e-05,
+      "loss": 1.6721,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.956475583864119e-05,
+      "loss": 1.8008,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.961783439490446e-05,
+      "loss": 1.702,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9670912951167727e-05,
+      "loss": 1.7524,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9723991507431e-05,
+      "loss": 1.6954,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9777070063694267e-05,
+      "loss": 1.7373,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9830148619957538e-05,
+      "loss": 1.7402,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9883227176220806e-05,
+      "loss": 1.6732,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9936305732484078e-05,
+      "loss": 1.7241,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9989384288747346e-05,
+      "loss": 1.7487,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0042462845010617e-05,
+      "loss": 1.7186,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0095541401273885e-05,
+      "loss": 1.6738,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0148619957537157e-05,
+      "loss": 1.7535,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0201698513800425e-05,
+      "loss": 1.7228,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0254777070063693e-05,
+      "loss": 1.714,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0307855626326965e-05,
+      "loss": 1.706,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0360934182590233e-05,
+      "loss": 1.6977,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0414012738853504e-05,
+      "loss": 1.7574,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0467091295116772e-05,
+      "loss": 1.677,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0520169851380044e-05,
+      "loss": 1.6087,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.057324840764331e-05,
+      "loss": 1.7672,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.062632696390658e-05,
+      "loss": 1.7668,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0679405520169855e-05,
+      "loss": 1.6475,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.073248407643312e-05,
+      "loss": 1.688,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.078556263269639e-05,
+      "loss": 1.7369,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.083864118895966e-05,
+      "loss": 1.7315,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.089171974522293e-05,
+      "loss": 1.7317,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.09447983014862e-05,
+      "loss": 1.7168,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.099787685774947e-05,
+      "loss": 1.7728,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.105095541401274e-05,
+      "loss": 1.7395,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1104033970276007e-05,
+      "loss": 1.6962,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.115711252653928e-05,
+      "loss": 1.7178,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.121019108280255e-05,
+      "loss": 1.6498,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.126326963906582e-05,
+      "loss": 1.7007,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1316348195329086e-05,
+      "loss": 1.6954,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1369426751592354e-05,
+      "loss": 1.7347,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.142250530785563e-05,
+      "loss": 1.6871,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.14755838641189e-05,
+      "loss": 1.7315,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1528662420382165e-05,
+      "loss": 1.6435,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.158174097664543e-05,
+      "loss": 1.7452,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.163481953290871e-05,
+      "loss": 1.6703,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1687898089171976e-05,
+      "loss": 1.7151,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1740976645435244e-05,
+      "loss": 1.6608,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.179405520169851e-05,
+      "loss": 1.6194,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.184713375796178e-05,
+      "loss": 1.6481,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1900212314225055e-05,
+      "loss": 1.7341,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1953290870488323e-05,
+      "loss": 1.7129,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.200636942675159e-05,
+      "loss": 1.7071,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.205944798301486e-05,
+      "loss": 1.7723,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2112526539278135e-05,
+      "loss": 1.6893,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.21656050955414e-05,
+      "loss": 1.618,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.221868365180467e-05,
+      "loss": 1.6517,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.227176220806794e-05,
+      "loss": 1.6617,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.232484076433121e-05,
+      "loss": 1.7133,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.237791932059448e-05,
+      "loss": 1.615,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.243099787685775e-05,
+      "loss": 1.7107,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.248407643312102e-05,
+      "loss": 1.685,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2537154989384286e-05,
+      "loss": 1.7484,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.259023354564756e-05,
+      "loss": 1.6393,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.264331210191083e-05,
+      "loss": 1.6688,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.26963906581741e-05,
+      "loss": 1.6684,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2749469214437365e-05,
+      "loss": 1.656,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2802547770700634e-05,
+      "loss": 1.6195,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.285562632696391e-05,
+      "loss": 1.6412,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2908704883227177e-05,
+      "loss": 1.6654,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2961783439490445e-05,
+      "loss": 1.688,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.301486199575371e-05,
+      "loss": 1.7699,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.306794055201699e-05,
+      "loss": 1.6645,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3121019108280256e-05,
+      "loss": 1.6668,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3174097664543524e-05,
+      "loss": 1.7279,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.322717622080679e-05,
+      "loss": 1.7345,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.328025477707007e-05,
+      "loss": 1.7143,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.761,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.33864118895966e-05,
+      "loss": 1.699,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.343949044585987e-05,
+      "loss": 1.635,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.349256900212314e-05,
+      "loss": 1.7247,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3545647558386414e-05,
+      "loss": 1.6674,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.359872611464968e-05,
+      "loss": 1.7209,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.365180467091295e-05,
+      "loss": 1.7156,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.370488322717622e-05,
+      "loss": 1.668,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.375796178343949e-05,
+      "loss": 1.6838,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.381104033970276e-05,
+      "loss": 1.6752,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.386411889596603e-05,
+      "loss": 1.6031,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.39171974522293e-05,
+      "loss": 1.7016,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3970276008492566e-05,
+      "loss": 1.6352,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.402335456475584e-05,
+      "loss": 1.7107,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.407643312101911e-05,
+      "loss": 1.7099,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.412951167728238e-05,
+      "loss": 1.6599,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4182590233545645e-05,
+      "loss": 1.6673,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.423566878980892e-05,
+      "loss": 1.666,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.428874734607219e-05,
+      "loss": 1.7171,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4341825902335456e-05,
+      "loss": 1.7051,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4394904458598724e-05,
+      "loss": 1.5711,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.444798301486199e-05,
+      "loss": 1.6917,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.450106157112527e-05,
+      "loss": 1.7677,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4554140127388535e-05,
+      "loss": 1.618,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4607218683651803e-05,
+      "loss": 1.6195,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.466029723991507e-05,
+      "loss": 1.6533,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4713375796178346e-05,
+      "loss": 1.5926,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4766454352441615e-05,
+      "loss": 1.6068,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.481953290870488e-05,
+      "loss": 1.678,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.487261146496815e-05,
+      "loss": 1.6347,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.492569002123142e-05,
+      "loss": 1.6373,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4978768577494694e-05,
+      "loss": 1.6661,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.503184713375796e-05,
+      "loss": 1.6954,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.508492569002123e-05,
+      "loss": 1.6806,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.51380042462845e-05,
+      "loss": 1.7214,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.519108280254777e-05,
+      "loss": 1.6921,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.524416135881104e-05,
+      "loss": 1.6434,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.529723991507431e-05,
+      "loss": 1.666,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.535031847133758e-05,
+      "loss": 1.5945,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5403397027600845e-05,
+      "loss": 1.6778,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.545647558386412e-05,
+      "loss": 1.7031,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.550955414012739e-05,
+      "loss": 1.6798,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5562632696390657e-05,
+      "loss": 1.6707,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5615711252653925e-05,
+      "loss": 1.6282,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.56687898089172e-05,
+      "loss": 1.6953,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.572186836518047e-05,
+      "loss": 1.6288,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5774946921443736e-05,
+      "loss": 1.6373,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5828025477707004e-05,
+      "loss": 1.6108,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.588110403397027e-05,
+      "loss": 1.6638,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.593418259023355e-05,
+      "loss": 1.5918,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5987261146496815e-05,
+      "loss": 1.6317,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.604033970276008e-05,
+      "loss": 1.7241,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.609341825902335e-05,
+      "loss": 1.7456,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6146496815286626e-05,
+      "loss": 1.6568,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6199575371549894e-05,
+      "loss": 1.6791,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.625265392781316e-05,
+      "loss": 1.6846,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.630573248407643e-05,
+      "loss": 1.6788,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6358811040339705e-05,
+      "loss": 1.6797,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.641188959660297e-05,
+      "loss": 1.7185,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.646496815286624e-05,
+      "loss": 1.6563,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.651804670912951e-05,
+      "loss": 1.6662,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.657112526539278e-05,
+      "loss": 1.6534,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.662420382165605e-05,
+      "loss": 1.7525,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.667728237791932e-05,
+      "loss": 1.5653,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.673036093418259e-05,
+      "loss": 1.6767,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.678343949044586e-05,
+      "loss": 1.6519,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.683651804670913e-05,
+      "loss": 1.5879,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.68895966029724e-05,
+      "loss": 1.6358,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.694267515923567e-05,
+      "loss": 1.6195,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6995753715498936e-05,
+      "loss": 1.5995,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7048832271762204e-05,
+      "loss": 1.6749,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.710191082802548e-05,
+      "loss": 1.6506,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.715498938428875e-05,
+      "loss": 1.6787,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7208067940552015e-05,
+      "loss": 1.6947,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7261146496815283e-05,
+      "loss": 1.6927,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.731422505307856e-05,
+      "loss": 1.5945,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7367303609341826e-05,
+      "loss": 1.6209,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7420382165605095e-05,
+      "loss": 1.6782,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.747346072186836e-05,
+      "loss": 1.7465,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.752653927813164e-05,
+      "loss": 1.7504,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7579617834394906e-05,
+      "loss": 1.6585,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.763269639065818e-05,
+      "loss": 1.6572,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.768577494692145e-05,
+      "loss": 1.7366,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.773885350318472e-05,
+      "loss": 1.7074,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7791932059447985e-05,
+      "loss": 1.6489,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.784501061571126e-05,
+      "loss": 1.6636,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.789808917197453e-05,
+      "loss": 1.6607,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7951167728237796e-05,
+      "loss": 1.6365,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8004246284501064e-05,
+      "loss": 1.7146,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.805732484076434e-05,
+      "loss": 1.6575,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.811040339702761e-05,
+      "loss": 1.6855,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8163481953290875e-05,
+      "loss": 1.7229,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.821656050955414e-05,
+      "loss": 1.7244,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.826963906581741e-05,
+      "loss": 1.7057,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8322717622080686e-05,
+      "loss": 1.7036,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8375796178343954e-05,
+      "loss": 1.6633,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.842887473460722e-05,
+      "loss": 1.6632,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.848195329087049e-05,
+      "loss": 1.6657,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8535031847133766e-05,
+      "loss": 1.5941,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8588110403397034e-05,
+      "loss": 1.6687,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.86411889596603e-05,
+      "loss": 1.6804,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.869426751592357e-05,
+      "loss": 1.6402,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.874734607218684e-05,
+      "loss": 1.6608,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.880042462845011e-05,
+      "loss": 1.6169,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.885350318471338e-05,
+      "loss": 1.575,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.890658174097665e-05,
+      "loss": 1.6014,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.895966029723992e-05,
+      "loss": 1.6929,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.901273885350319e-05,
+      "loss": 1.7285,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.906581740976646e-05,
+      "loss": 1.6514,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.911889596602973e-05,
+      "loss": 1.6786,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9171974522292996e-05,
+      "loss": 1.7326,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9225053078556264e-05,
+      "loss": 1.548,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.927813163481954e-05,
+      "loss": 1.6687,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.933121019108281e-05,
+      "loss": 1.656,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9384288747346076e-05,
+      "loss": 1.5577,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9437367303609344e-05,
+      "loss": 1.773,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.949044585987262e-05,
+      "loss": 1.6667,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.954352441613589e-05,
+      "loss": 1.7739,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9596602972399155e-05,
+      "loss": 1.6042,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.964968152866242e-05,
+      "loss": 1.6756,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.970276008492569e-05,
+      "loss": 1.6155,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9755838641188966e-05,
+      "loss": 1.5683,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9808917197452234e-05,
+      "loss": 1.7452,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.98619957537155e-05,
+      "loss": 1.6494,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.991507430997877e-05,
+      "loss": 1.6514,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9968152866242045e-05,
+      "loss": 1.6452,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.002123142250531e-05,
+      "loss": 1.6043,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.007430997876858e-05,
+      "loss": 1.6459,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.012738853503185e-05,
+      "loss": 1.6508,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.018046709129512e-05,
+      "loss": 1.6801,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.023354564755839e-05,
+      "loss": 1.7403,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.028662420382166e-05,
+      "loss": 1.6588,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.033970276008493e-05,
+      "loss": 1.5769,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.03927813163482e-05,
+      "loss": 1.6742,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.044585987261147e-05,
+      "loss": 1.6325,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.049893842887474e-05,
+      "loss": 1.6096,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.055201698513801e-05,
+      "loss": 1.5629,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0605095541401276e-05,
+      "loss": 1.64,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.065817409766455e-05,
+      "loss": 1.6503,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.071125265392782e-05,
+      "loss": 1.6496,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.076433121019109e-05,
+      "loss": 1.6312,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0817409766454355e-05,
+      "loss": 1.5552,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.087048832271762e-05,
+      "loss": 1.6384,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.09235668789809e-05,
+      "loss": 1.5973,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0976645435244166e-05,
+      "loss": 1.6632,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1029723991507434e-05,
+      "loss": 1.6343,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.10828025477707e-05,
+      "loss": 1.5579,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.113588110403398e-05,
+      "loss": 1.6177,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1188959660297246e-05,
+      "loss": 1.6325,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1242038216560514e-05,
+      "loss": 1.7332,
+      "step": 777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.129511677282378e-05,
+      "loss": 1.6566,
+      "step": 778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.134819532908705e-05,
+      "loss": 1.6449,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1401273885350325e-05,
+      "loss": 1.6508,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.145435244161359e-05,
+      "loss": 1.5946,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.150743099787686e-05,
+      "loss": 1.68,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.156050955414013e-05,
+      "loss": 1.6786,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1613588110403404e-05,
+      "loss": 1.5788,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 1.6602,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.171974522292994e-05,
+      "loss": 1.6504,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.177282377919321e-05,
+      "loss": 1.6854,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.1825902335456476e-05,
+      "loss": 1.6348,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.187898089171975e-05,
+      "loss": 1.6337,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.193205944798302e-05,
+      "loss": 1.6265,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.198513800424629e-05,
+      "loss": 1.6532,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2038216560509556e-05,
+      "loss": 1.6641,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.209129511677283e-05,
+      "loss": 1.6276,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.21443736730361e-05,
+      "loss": 1.7302,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.219745222929937e-05,
+      "loss": 1.751,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2250530785562635e-05,
+      "loss": 1.5596,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.23036093418259e-05,
+      "loss": 1.653,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.235668789808918e-05,
+      "loss": 1.6084,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2409766454352446e-05,
+      "loss": 1.6591,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2462845010615714e-05,
+      "loss": 1.5998,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.251592356687898e-05,
+      "loss": 1.6234,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.256900212314226e-05,
+      "loss": 1.6553,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2622080679405525e-05,
+      "loss": 1.6825,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.267515923566879e-05,
+      "loss": 1.6316,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.272823779193206e-05,
+      "loss": 1.67,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.278131634819533e-05,
+      "loss": 1.6949,
+      "step": 806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2834394904458604e-05,
+      "loss": 1.6623,
+      "step": 807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.288747346072187e-05,
+      "loss": 1.6049,
+      "step": 808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.294055201698514e-05,
+      "loss": 1.6604,
+      "step": 809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.299363057324841e-05,
+      "loss": 1.6201,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3046709129511684e-05,
+      "loss": 1.6842,
+      "step": 811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.309978768577495e-05,
+      "loss": 1.6376,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.315286624203822e-05,
+      "loss": 1.6865,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.320594479830149e-05,
+      "loss": 1.6749,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.325902335456476e-05,
+      "loss": 1.6262,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.331210191082803e-05,
+      "loss": 1.6252,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.33651804670913e-05,
+      "loss": 1.5819,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.341825902335457e-05,
+      "loss": 1.6955,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3471337579617835e-05,
+      "loss": 1.6526,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.352441613588111e-05,
+      "loss": 1.7002,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.357749469214438e-05,
+      "loss": 1.6948,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3630573248407646e-05,
+      "loss": 1.639,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3683651804670914e-05,
+      "loss": 1.5983,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.373673036093419e-05,
+      "loss": 1.6815,
+      "step": 824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.378980891719746e-05,
+      "loss": 1.5906,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3842887473460726e-05,
+      "loss": 1.7111,
+      "step": 826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3895966029723994e-05,
+      "loss": 1.672,
+      "step": 827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.394904458598726e-05,
+      "loss": 1.617,
+      "step": 828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.400212314225054e-05,
+      "loss": 1.5561,
+      "step": 829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4055201698513805e-05,
+      "loss": 1.6724,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.410828025477707e-05,
+      "loss": 1.5813,
+      "step": 831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.416135881104034e-05,
+      "loss": 1.644,
+      "step": 832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4214437367303616e-05,
+      "loss": 1.6467,
+      "step": 833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4267515923566884e-05,
+      "loss": 1.6423,
+      "step": 834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.432059447983015e-05,
+      "loss": 1.6905,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.437367303609342e-05,
+      "loss": 1.6641,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.442675159235669e-05,
+      "loss": 1.5294,
+      "step": 837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.447983014861996e-05,
+      "loss": 1.5904,
+      "step": 838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.453290870488323e-05,
+      "loss": 1.6389,
+      "step": 839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.45859872611465e-05,
+      "loss": 1.6144,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.463906581740977e-05,
+      "loss": 1.6291,
+      "step": 841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.469214437367304e-05,
+      "loss": 1.6914,
+      "step": 842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.474522292993631e-05,
+      "loss": 1.6888,
+      "step": 843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.479830148619958e-05,
+      "loss": 1.622,
+      "step": 844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.485138004246285e-05,
+      "loss": 1.6619,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4904458598726115e-05,
+      "loss": 1.6371,
+      "step": 846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.495753715498939e-05,
+      "loss": 1.6328,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.501061571125266e-05,
+      "loss": 1.6428,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5063694267515926e-05,
+      "loss": 1.7071,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5116772823779194e-05,
+      "loss": 1.6692,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.516985138004247e-05,
+      "loss": 1.7193,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.522292993630574e-05,
+      "loss": 1.6279,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5276008492569005e-05,
+      "loss": 1.6452,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.532908704883227e-05,
+      "loss": 1.6885,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.538216560509554e-05,
+      "loss": 1.7035,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5435244161358816e-05,
+      "loss": 1.5905,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5488322717622084e-05,
+      "loss": 1.5783,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.554140127388535e-05,
+      "loss": 1.6213,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.559447983014862e-05,
+      "loss": 1.5328,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5647558386411895e-05,
+      "loss": 1.6983,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5700636942675164e-05,
+      "loss": 1.6273,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.575371549893843e-05,
+      "loss": 1.6346,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.58067940552017e-05,
+      "loss": 1.6169,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.585987261146497e-05,
+      "loss": 1.676,
+      "step": 864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.591295116772824e-05,
+      "loss": 1.6648,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.596602972399151e-05,
+      "loss": 1.7119,
+      "step": 866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.601910828025478e-05,
+      "loss": 1.6602,
+      "step": 867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.607218683651805e-05,
+      "loss": 1.5983,
+      "step": 868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.612526539278132e-05,
+      "loss": 1.632,
+      "step": 869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.617834394904459e-05,
+      "loss": 1.6303,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.623142250530786e-05,
+      "loss": 1.686,
+      "step": 871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6284501061571126e-05,
+      "loss": 1.5966,
+      "step": 872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.63375796178344e-05,
+      "loss": 1.6562,
+      "step": 873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.639065817409767e-05,
+      "loss": 1.594,
+      "step": 874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.644373673036094e-05,
+      "loss": 1.4919,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6496815286624206e-05,
+      "loss": 1.5465,
+      "step": 876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6549893842887474e-05,
+      "loss": 1.583,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.660297239915075e-05,
+      "loss": 1.5359,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.665605095541402e-05,
+      "loss": 1.6689,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6709129511677285e-05,
+      "loss": 1.7151,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.676220806794055e-05,
+      "loss": 1.6798,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.681528662420383e-05,
+      "loss": 1.6214,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6868365180467096e-05,
+      "loss": 1.5905,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6921443736730364e-05,
+      "loss": 1.6619,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.697452229299363e-05,
+      "loss": 1.7149,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.70276008492569e-05,
+      "loss": 1.5615,
+      "step": 886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7080679405520175e-05,
+      "loss": 1.6493,
+      "step": 887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.713375796178344e-05,
+      "loss": 1.6701,
+      "step": 888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.718683651804671e-05,
+      "loss": 1.6115,
+      "step": 889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.723991507430998e-05,
+      "loss": 1.6462,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7292993630573254e-05,
+      "loss": 1.6582,
+      "step": 891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.734607218683652e-05,
+      "loss": 1.6493,
+      "step": 892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.739915074309979e-05,
+      "loss": 1.6076,
+      "step": 893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.745222929936306e-05,
+      "loss": 1.6527,
+      "step": 894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.750530785562633e-05,
+      "loss": 1.617,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.75583864118896e-05,
+      "loss": 1.6289,
+      "step": 896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761146496815287e-05,
+      "loss": 1.7097,
+      "step": 897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766454352441614e-05,
+      "loss": 1.6998,
+      "step": 898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7717622080679406e-05,
+      "loss": 1.7042,
+      "step": 899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777070063694268e-05,
+      "loss": 1.5545,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782377919320595e-05,
+      "loss": 1.6985,
+      "step": 901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.787685774946922e-05,
+      "loss": 1.5871,
+      "step": 902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7929936305732485e-05,
+      "loss": 1.7256,
+      "step": 903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.798301486199575e-05,
+      "loss": 1.689,
+      "step": 904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.803609341825903e-05,
+      "loss": 1.6779,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8089171974522296e-05,
+      "loss": 1.6828,
+      "step": 906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8142250530785564e-05,
+      "loss": 1.5648,
+      "step": 907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.819532908704883e-05,
+      "loss": 1.5995,
+      "step": 908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.824840764331211e-05,
+      "loss": 1.6543,
+      "step": 909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8301486199575375e-05,
+      "loss": 1.6913,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8354564755838644e-05,
+      "loss": 1.6514,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.840764331210191e-05,
+      "loss": 1.6079,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.846072186836518e-05,
+      "loss": 1.6073,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8513800424628455e-05,
+      "loss": 1.7448,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.856687898089172e-05,
+      "loss": 1.622,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.861995753715499e-05,
+      "loss": 1.673,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.867303609341826e-05,
+      "loss": 1.5909,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8726114649681534e-05,
+      "loss": 1.6288,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.87791932059448e-05,
+      "loss": 1.6406,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.883227176220807e-05,
+      "loss": 1.5574,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.888535031847134e-05,
+      "loss": 1.6892,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.893842887473461e-05,
+      "loss": 1.6007,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.899150743099788e-05,
+      "loss": 1.547,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.904458598726115e-05,
+      "loss": 1.5884,
+      "step": 924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.909766454352442e-05,
+      "loss": 1.721,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9150743099787686e-05,
+      "loss": 1.6706,
+      "step": 926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.920382165605096e-05,
+      "loss": 1.6067,
+      "step": 927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.925690021231423e-05,
+      "loss": 1.6711,
+      "step": 928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.93099787685775e-05,
+      "loss": 1.682,
+      "step": 929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9363057324840765e-05,
+      "loss": 1.7421,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.941613588110404e-05,
+      "loss": 1.6859,
+      "step": 931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.946921443736731e-05,
+      "loss": 1.5722,
+      "step": 932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9522292993630576e-05,
+      "loss": 1.5793,
+      "step": 933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9575371549893844e-05,
+      "loss": 1.5855,
+      "step": 934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.962845010615711e-05,
+      "loss": 1.6373,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968152866242039e-05,
+      "loss": 1.6299,
+      "step": 936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9734607218683655e-05,
+      "loss": 1.6066,
+      "step": 937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.978768577494692e-05,
+      "loss": 1.6022,
+      "step": 938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.984076433121019e-05,
+      "loss": 1.5861,
+      "step": 939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9893842887473466e-05,
+      "loss": 1.6395,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9946921443736734e-05,
+      "loss": 1.5928,
+      "step": 941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5e-05,
+      "loss": 1.6496,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999999986688259e-05,
+      "loss": 1.6897,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999999946753036e-05,
+      "loss": 1.5608,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999998801943314e-05,
+      "loss": 1.7186,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999997870121453e-05,
+      "loss": 1.6578,
+      "step": 946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.99999966720648e-05,
+      "loss": 1.6882,
+      "step": 947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999995207773355e-05,
+      "loss": 1.6426,
+      "step": 948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999999347724714e-05,
+      "loss": 1.6533,
+      "step": 949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999999148048617e-05,
+      "loss": 1.6673,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999989217490474e-05,
+      "loss": 1.6079,
+      "step": 951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999998668826006e-05,
+      "loss": 1.6514,
+      "step": 952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999983892794975e-05,
+      "loss": 1.6747,
+      "step": 953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999998083109524e-05,
+      "loss": 1.6675,
+      "step": 954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999997750316089e-05,
+      "loss": 1.6473,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999973908991946e-05,
+      "loss": 1.6065,
+      "step": 956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999997004858846e-05,
+      "loss": 1.6709,
+      "step": 957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999965921950484e-05,
+      "loss": 1.6223,
+      "step": 958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999996152907804e-05,
+      "loss": 1.6117,
+      "step": 959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999995686997118e-05,
+      "loss": 1.6262,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999995194462996e-05,
+      "loss": 1.6044,
+      "step": 961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999946753054426e-05,
+      "loss": 1.6547,
+      "step": 962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999941295244644e-05,
+      "loss": 1.6825,
+      "step": 963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999935571200664e-05,
+      "loss": 1.6275,
+      "step": 964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999929580922546e-05,
+      "loss": 1.6356,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999992332441036e-05,
+      "loss": 1.5831,
+      "step": 966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999991680166416e-05,
+      "loss": 1.5974,
+      "step": 967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999991001268402e-05,
+      "loss": 1.5889,
+      "step": 968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999990295747003e-05,
+      "loss": 1.5693,
+      "step": 969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999989563602224e-05,
+      "loss": 1.7083,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999988804834075e-05,
+      "loss": 1.6652,
+      "step": 971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999988019442563e-05,
+      "loss": 1.6234,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999872074276955e-05,
+      "loss": 1.6539,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999863687894824e-05,
+      "loss": 1.6353,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999985503527932e-05,
+      "loss": 1.7224,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999846116430534e-05,
+      "loss": 1.684,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999983693134858e-05,
+      "loss": 1.5518,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999982748003353e-05,
+      "loss": 1.6447,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999817762485495e-05,
+      "loss": 1.5948,
+      "step": 979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999807778704576e-05,
+      "loss": 1.661,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999979752869089e-05,
+      "loss": 1.5777,
+      "step": 981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999978701244453e-05,
+      "loss": 1.6604,
+      "step": 982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999977622996562e-05,
+      "loss": 1.5567,
+      "step": 983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999765181254276e-05,
+      "loss": 1.6162,
+      "step": 984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999753866310604e-05,
+      "loss": 1.6566,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999974228513473e-05,
+      "loss": 1.6091,
+      "step": 986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999973043772678e-05,
+      "loss": 1.6782,
+      "step": 987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999971832408688e-05,
+      "loss": 1.5445,
+      "step": 988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999970594421516e-05,
+      "loss": 1.6519,
+      "step": 989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999693298111744e-05,
+      "loss": 1.6732,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999968038577677e-05,
+      "loss": 1.6306,
+      "step": 991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999667207210386e-05,
+      "loss": 1.5674,
+      "step": 992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999653762412716e-05,
+      "loss": 1.6281,
+      "step": 993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999640051383916e-05,
+      "loss": 1.6926,
+      "step": 994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999962607412413e-05,
+      "loss": 1.6102,
+      "step": 995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.99996118306335e-05,
+      "loss": 1.541,
+      "step": 996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999959732091218e-05,
+      "loss": 1.6947,
+      "step": 997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999958254496033e-05,
+      "loss": 1.6895,
+      "step": 998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.99995675027781e-05,
+      "loss": 1.6594,
+      "step": 999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999955219436565e-05,
+      "loss": 1.7354,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999953661972315e-05,
+      "loss": 1.6363,
+      "step": 1001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999952077885077e-05,
+      "loss": 1.6459,
+      "step": 1002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999504671748666e-05,
+      "loss": 1.6498,
+      "step": 1003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999948829841701e-05,
+      "loss": 1.5847,
+      "step": 1004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999947165885599e-05,
+      "loss": 1.7089,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999454753065764e-05,
+      "loss": 1.5749,
+      "step": 1006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999437581046525e-05,
+      "loss": 1.6296,
+      "step": 1007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999942014279846e-05,
+      "loss": 1.6629,
+      "step": 1008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999940243832175e-05,
+      "loss": 1.5811,
+      "step": 1009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999938446761657e-05,
+      "loss": 1.6635,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999936623068313e-05,
+      "loss": 1.6341,
+      "step": 1011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999347727521616e-05,
+      "loss": 1.6314,
+      "step": 1012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999932895813223e-05,
+      "loss": 1.565,
+      "step": 1013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999309922515155e-05,
+      "loss": 1.5913,
+      "step": 1014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999929062067062e-05,
+      "loss": 1.5819,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999271052598816e-05,
+      "loss": 1.6286,
+      "step": 1016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999925121829995e-05,
+      "loss": 1.6307,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999923111777424e-05,
+      "loss": 1.6262,
+      "step": 1018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999210751021886e-05,
+      "loss": 1.5698,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999919011804313e-05,
+      "loss": 1.7123,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999916921883816e-05,
+      "loss": 1.6878,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999914805340722e-05,
+      "loss": 1.5606,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999912662175053e-05,
+      "loss": 1.6193,
+      "step": 1023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999910492386833e-05,
+      "loss": 1.5977,
+      "step": 1024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999082959760826e-05,
+      "loss": 1.6263,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999060729428275e-05,
+      "loss": 1.5559,
+      "step": 1026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999038232870896e-05,
+      "loss": 1.5983,
+      "step": 1027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999901547008895e-05,
+      "loss": 1.5759,
+      "step": 1028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999899244108266e-05,
+      "loss": 1.5514,
+      "step": 1029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999896914585227e-05,
+      "loss": 1.6188,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999894558439805e-05,
+      "loss": 1.6395,
+      "step": 1031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999892175672023e-05,
+      "loss": 1.6279,
+      "step": 1032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999889766281907e-05,
+      "loss": 1.6735,
+      "step": 1033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9998873302694826e-05,
+      "loss": 1.6564,
+      "step": 1034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9998848676347756e-05,
+      "loss": 1.5908,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999882378377813e-05,
+      "loss": 1.571,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99987986249862e-05,
+      "loss": 1.6665,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999877319997225e-05,
+      "loss": 1.599,
+      "step": 1038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999874750873654e-05,
+      "loss": 1.6727,
+      "step": 1039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999872155127935e-05,
+      "loss": 1.5708,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999869532760095e-05,
+      "loss": 1.6338,
+      "step": 1041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999866883770162e-05,
+      "loss": 1.6357,
+      "step": 1042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999864208158164e-05,
+      "loss": 1.5489,
+      "step": 1043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99986150592413e-05,
+      "loss": 1.5578,
+      "step": 1044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999858777068089e-05,
+      "loss": 1.6962,
+      "step": 1045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998560215900706e-05,
+      "loss": 1.6064,
+      "step": 1046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998532394901024e-05,
+      "loss": 1.6674,
+      "step": 1047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998504307682147e-05,
+      "loss": 1.5959,
+      "step": 1048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999847595424438e-05,
+      "loss": 1.641,
+      "step": 1049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999844733458802e-05,
+      "loss": 1.6225,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999841844871337e-05,
+      "loss": 1.628,
+      "step": 1051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999838929662074e-05,
+      "loss": 1.5731,
+      "step": 1052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998359878310445e-05,
+      "loss": 1.681,
+      "step": 1053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998330193782796e-05,
+      "loss": 1.5954,
+      "step": 1054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99983002430381e-05,
+      "loss": 1.5389,
+      "step": 1055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998270026076685e-05,
+      "loss": 1.5232,
+      "step": 1056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999823954289888e-05,
+      "loss": 1.6373,
+      "step": 1057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999820879350499e-05,
+      "loss": 1.6521,
+      "step": 1058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999817777789536e-05,
+      "loss": 1.6187,
+      "step": 1059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999814649607031e-05,
+      "loss": 1.5383,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999811494803017e-05,
+      "loss": 1.6322,
+      "step": 1061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99980831337753e-05,
+      "loss": 1.6159,
+      "step": 1062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998051053306003e-05,
+      "loss": 1.5783,
+      "step": 1063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998018706622654e-05,
+      "loss": 1.6267,
+      "step": 1064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999798609372557e-05,
+      "loss": 1.5875,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999795321461512e-05,
+      "loss": 1.6695,
+      "step": 1066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999792006929164e-05,
+      "loss": 1.621,
+      "step": 1067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997886657755486e-05,
+      "loss": 1.6694,
+      "step": 1068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999785298000702e-05,
+      "loss": 1.6315,
+      "step": 1069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997819036046595e-05,
+      "loss": 1.5939,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999778482587457e-05,
+      "loss": 1.5834,
+      "step": 1071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997750349491324e-05,
+      "loss": 1.6603,
+      "step": 1072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99977156068972e-05,
+      "loss": 1.5668,
+      "step": 1073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999768059809259e-05,
+      "loss": 1.6648,
+      "step": 1074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997645323077855e-05,
+      "loss": 1.6295,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999760978185337e-05,
+      "loss": 1.6549,
+      "step": 1076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999757397441952e-05,
+      "loss": 1.6004,
+      "step": 1077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999753790077669e-05,
+      "loss": 1.6048,
+      "step": 1078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997501560925255e-05,
+      "loss": 1.5022,
+      "step": 1079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99974649548656e-05,
+      "loss": 1.553,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999742808259812e-05,
+      "loss": 1.6843,
+      "step": 1081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997390944123214e-05,
+      "loss": 1.5793,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999735353944126e-05,
+      "loss": 1.7546,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999731586855268e-05,
+      "loss": 1.6006,
+      "step": 1084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999727793145785e-05,
+      "loss": 1.6024,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999723972815719e-05,
+      "loss": 1.7511,
+      "step": 1086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997201258651105e-05,
+      "loss": 1.6299,
+      "step": 1087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999716252294001e-05,
+      "loss": 1.5952,
+      "step": 1088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99971235210243e-05,
+      "loss": 1.5775,
+      "step": 1089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99970842529044e-05,
+      "loss": 1.6345,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999704471858073e-05,
+      "loss": 1.6713,
+      "step": 1091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999700491805371e-05,
+      "loss": 1.5788,
+      "step": 1092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999696485132377e-05,
+      "loss": 1.6198,
+      "step": 1093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999692451839133e-05,
+      "loss": 1.5681,
+      "step": 1094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9996883919256806e-05,
+      "loss": 1.6197,
+      "step": 1095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999684305392066e-05,
+      "loss": 1.5995,
+      "step": 1096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999680192238331e-05,
+      "loss": 1.6146,
+      "step": 1097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999676052464519e-05,
+      "loss": 1.6732,
+      "step": 1098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999671886070674e-05,
+      "loss": 1.6541,
+      "step": 1099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999667693056842e-05,
+      "loss": 1.6044,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996634734230664e-05,
+      "loss": 1.5225,
+      "step": 1101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999659227169392e-05,
+      "loss": 1.6052,
+      "step": 1102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999654954295865e-05,
+      "loss": 1.6567,
+      "step": 1103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.99965065480253e-05,
+      "loss": 1.5832,
+      "step": 1104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996463286894335e-05,
+      "loss": 1.5614,
+      "step": 1105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996419759566204e-05,
+      "loss": 1.6605,
+      "step": 1106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999637596604139e-05,
+      "loss": 1.5666,
+      "step": 1107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999633190632034e-05,
+      "loss": 1.6734,
+      "step": 1108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999628758040353e-05,
+      "loss": 1.5461,
+      "step": 1109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999624298829143e-05,
+      "loss": 1.6312,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996198129984526e-05,
+      "loss": 1.6459,
+      "step": 1111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996153005483284e-05,
+      "loss": 1.6297,
+      "step": 1112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999610761478818e-05,
+      "loss": 1.6803,
+      "step": 1113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999606195789972e-05,
+      "loss": 1.5429,
+      "step": 1114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996016034818364e-05,
+      "loss": 1.5696,
+      "step": 1115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995969845544625e-05,
+      "loss": 1.636,
+      "step": 1116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999592339007897e-05,
+      "loss": 1.6106,
+      "step": 1117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999587666842191e-05,
+      "loss": 1.5904,
+      "step": 1118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995829680573944e-05,
+      "loss": 1.583,
+      "step": 1119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999578242653556e-05,
+      "loss": 1.6345,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999573490630727e-05,
+      "loss": 1.5639,
+      "step": 1121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995687119889576e-05,
+      "loss": 1.6425,
+      "step": 1122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995639067283e-05,
+      "loss": 1.5586,
+      "step": 1123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995590748488035e-05,
+      "loss": 1.6422,
+      "step": 1124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.99955421635052e-05,
+      "loss": 1.5998,
+      "step": 1125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999549331233503e-05,
+      "loss": 1.6615,
+      "step": 1126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999544419497802e-05,
+      "loss": 1.6211,
+      "step": 1127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.99953948114347e-05,
+      "loss": 1.6153,
+      "step": 1128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999534516170561e-05,
+      "loss": 1.6644,
+      "step": 1129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995295245791274e-05,
+      "loss": 1.5111,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999524506369221e-05,
+      "loss": 1.6353,
+      "step": 1131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999519461540896e-05,
+      "loss": 1.6166,
+      "step": 1132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995143900942064e-05,
+      "loss": 1.5671,
+      "step": 1133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995092920292064e-05,
+      "loss": 1.6156,
+      "step": 1134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995041673459495e-05,
+      "loss": 1.6298,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999499016044491e-05,
+      "loss": 1.5876,
+      "step": 1136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994938381248854e-05,
+      "loss": 1.5248,
+      "step": 1137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994886335871874e-05,
+      "loss": 1.5703,
+      "step": 1138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999483402431453e-05,
+      "loss": 1.5588,
+      "step": 1139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994781446577385e-05,
+      "loss": 1.6364,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999472860266099e-05,
+      "loss": 1.5515,
+      "step": 1141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999467549256591e-05,
+      "loss": 1.5938,
+      "step": 1142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999462211629271e-05,
+      "loss": 1.6405,
+      "step": 1143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994568473841964e-05,
+      "loss": 1.5685,
+      "step": 1144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994514565214226e-05,
+      "loss": 1.6342,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999446039041009e-05,
+      "loss": 1.6262,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999440594943012e-05,
+      "loss": 1.6683,
+      "step": 1147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999435124227491e-05,
+      "loss": 1.5899,
+      "step": 1148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999429626894503e-05,
+      "loss": 1.626,
+      "step": 1149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999424102944107e-05,
+      "loss": 1.6026,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999418552376362e-05,
+      "loss": 1.5562,
+      "step": 1151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994129751913264e-05,
+      "loss": 1.6389,
+      "step": 1152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.99940737138906e-05,
+      "loss": 1.6244,
+      "step": 1153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994017409696226e-05,
+      "loss": 1.6773,
+      "step": 1154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999396083933075e-05,
+      "loss": 1.5495,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999390400279475e-05,
+      "loss": 1.7222,
+      "step": 1156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999384690008885e-05,
+      "loss": 1.6356,
+      "step": 1157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9993789531213656e-05,
+      "loss": 1.5318,
+      "step": 1158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999373189616978e-05,
+      "loss": 1.5819,
+      "step": 1159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9993673994957834e-05,
+      "loss": 1.5838,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999361582757843e-05,
+      "loss": 1.6922,
+      "step": 1161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993557394032196e-05,
+      "loss": 1.5868,
+      "step": 1162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993498694319754e-05,
+      "loss": 1.6193,
+      "step": 1163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993439728441716e-05,
+      "loss": 1.6935,
+      "step": 1164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999338049639872e-05,
+      "loss": 1.7002,
+      "step": 1165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993320998191394e-05,
+      "loss": 1.6387,
+      "step": 1166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999326123382038e-05,
+      "loss": 1.534,
+      "step": 1167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99932012032863e-05,
+      "loss": 1.5505,
+      "step": 1168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999314090658981e-05,
+      "loss": 1.6548,
+      "step": 1169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999308034373153e-05,
+      "loss": 1.6073,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993019514712126e-05,
+      "loss": 1.7097,
+      "step": 1171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9992958419532247e-05,
+      "loss": 1.5011,
+      "step": 1172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9992897058192524e-05,
+      "loss": 1.601,
+      "step": 1173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999283543069362e-05,
+      "loss": 1.6228,
+      "step": 1174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9992773537036195e-05,
+      "loss": 1.5894,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99927113772209e-05,
+      "loss": 1.5596,
+      "step": 1176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99926489512484e-05,
+      "loss": 1.5736,
+      "step": 1177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999258625911938e-05,
+      "loss": 1.5897,
+      "step": 1178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9992523300834474e-05,
+      "loss": 1.6323,
+      "step": 1179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999246007639437e-05,
+      "loss": 1.7217,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999239658579974e-05,
+      "loss": 1.4924,
+      "step": 1181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999233282905127e-05,
+      "loss": 1.6405,
+      "step": 1182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999226880614962e-05,
+      "loss": 1.6633,
+      "step": 1183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999220451709548e-05,
+      "loss": 1.5926,
+      "step": 1184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999213996188953e-05,
+      "loss": 1.5635,
+      "step": 1185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9992075140532466e-05,
+      "loss": 1.5985,
+      "step": 1186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999201005302499e-05,
+      "loss": 1.5394,
+      "step": 1187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9991944699367754e-05,
+      "loss": 1.5914,
+      "step": 1188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99918790795615e-05,
+      "loss": 1.6612,
+      "step": 1189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99918131936069e-05,
+      "loss": 1.6192,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9991747041504655e-05,
+      "loss": 1.7092,
+      "step": 1191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9991680623255485e-05,
+      "loss": 1.6522,
+      "step": 1192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9991613938860094e-05,
+      "loss": 1.517,
+      "step": 1193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999154698831918e-05,
+      "loss": 1.5238,
+      "step": 1194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999147977163346e-05,
+      "loss": 1.6851,
+      "step": 1195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9991412288803666e-05,
+      "loss": 1.6009,
+      "step": 1196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9991344539830486e-05,
+      "loss": 1.6052,
+      "step": 1197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999127652471467e-05,
+      "loss": 1.6008,
+      "step": 1198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999120824345693e-05,
+      "loss": 1.5627,
+      "step": 1199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9991139696058e-05,
+      "loss": 1.6124,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99910708825186e-05,
+      "loss": 1.5857,
+      "step": 1201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999100180283946e-05,
+      "loss": 1.6319,
+      "step": 1202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999093245702133e-05,
+      "loss": 1.5621,
+      "step": 1203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999086284506494e-05,
+      "loss": 1.6334,
+      "step": 1204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9990792966971045e-05,
+      "loss": 1.6033,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999072282274036e-05,
+      "loss": 1.6982,
+      "step": 1206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9990652412373665e-05,
+      "loss": 1.5569,
+      "step": 1207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999058173587168e-05,
+      "loss": 1.57,
+      "step": 1208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999051079323518e-05,
+      "loss": 1.5979,
+      "step": 1209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999043958446491e-05,
+      "loss": 1.6519,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999036810956163e-05,
+      "loss": 1.6625,
+      "step": 1211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99902963685261e-05,
+      "loss": 1.5818,
+      "step": 1212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999022436135909e-05,
+      "loss": 1.5845,
+      "step": 1213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999015208806136e-05,
+      "loss": 1.5544,
+      "step": 1214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9990079548633686e-05,
+      "loss": 1.5887,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999000674307683e-05,
+      "loss": 1.5933,
+      "step": 1216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9989933671391576e-05,
+      "loss": 1.5923,
+      "step": 1217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9989860333578696e-05,
+      "loss": 1.6532,
+      "step": 1218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.998978672963899e-05,
+      "loss": 1.5238,
+      "step": 1219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9989712859573215e-05,
+      "loss": 1.6016,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9989638723382174e-05,
+      "loss": 1.6493,
+      "step": 1221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.998956432106664e-05,
+      "loss": 1.5693,
+      "step": 1222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.998948965262743e-05,
+      "loss": 1.612,
+      "step": 1223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.998941471806533e-05,
+      "loss": 1.6732,
+      "step": 1224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998933951738112e-05,
+      "loss": 1.6226,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9989264050575627e-05,
+      "loss": 1.5863,
+      "step": 1226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998918831764964e-05,
+      "loss": 1.5684,
+      "step": 1227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998911231860397e-05,
+      "loss": 1.549,
+      "step": 1228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998903605343942e-05,
+      "loss": 1.6421,
+      "step": 1229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9988959522156814e-05,
+      "loss": 1.6148,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998888272475696e-05,
+      "loss": 1.5608,
+      "step": 1231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998880566124067e-05,
+      "loss": 1.6176,
+      "step": 1232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998872833160878e-05,
+      "loss": 1.5552,
+      "step": 1233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.99886507358621e-05,
+      "loss": 1.598,
+      "step": 1234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9988572874001454e-05,
+      "loss": 1.5962,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998849474602768e-05,
+      "loss": 1.6732,
+      "step": 1236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9988416351941615e-05,
+      "loss": 1.5698,
+      "step": 1237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998833769174408e-05,
+      "loss": 1.563,
+      "step": 1238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998825876543593e-05,
+      "loss": 1.6091,
+      "step": 1239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9988179573017985e-05,
+      "loss": 1.5528,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9988100114491096e-05,
+      "loss": 1.6399,
+      "step": 1241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9988020389856114e-05,
+      "loss": 1.5469,
+      "step": 1242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9987940399113894e-05,
+      "loss": 1.6065,
+      "step": 1243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998786014226527e-05,
+      "loss": 1.683,
+      "step": 1244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9987779619311106e-05,
+      "loss": 1.6682,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998769883025226e-05,
+      "loss": 1.5985,
+      "step": 1246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.99876177750896e-05,
+      "loss": 1.5341,
+      "step": 1247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998753645382398e-05,
+      "loss": 1.5715,
+      "step": 1248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998745486645625e-05,
+      "loss": 1.601,
+      "step": 1249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9987373012987314e-05,
+      "loss": 1.5323,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998729089341802e-05,
+      "loss": 1.6123,
+      "step": 1251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998720850774925e-05,
+      "loss": 1.6463,
+      "step": 1252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9987125855981874e-05,
+      "loss": 1.5015,
+      "step": 1253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998704293811678e-05,
+      "loss": 1.6104,
+      "step": 1254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998695975415485e-05,
+      "loss": 1.5546,
+      "step": 1255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9986876304096966e-05,
+      "loss": 1.6368,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998679258794402e-05,
+      "loss": 1.6148,
+      "step": 1257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9986708605696905e-05,
+      "loss": 1.5793,
+      "step": 1258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998662435735652e-05,
+      "loss": 1.7033,
+      "step": 1259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998653984292374e-05,
+      "loss": 1.5479,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998645506239949e-05,
+      "loss": 1.542,
+      "step": 1261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998637001578466e-05,
+      "loss": 1.6403,
+      "step": 1262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998628470308017e-05,
+      "loss": 1.5576,
+      "step": 1263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.99861991242869e-05,
+      "loss": 1.5767,
+      "step": 1264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9986113279405786e-05,
+      "loss": 1.5549,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998602716843773e-05,
+      "loss": 1.6682,
+      "step": 1266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998594079138367e-05,
+      "loss": 1.6455,
+      "step": 1267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9985854148244494e-05,
+      "loss": 1.5145,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9985767239021146e-05,
+      "loss": 1.5891,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998568006371455e-05,
+      "loss": 1.6187,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998559262232562e-05,
+      "loss": 1.6104,
+      "step": 1271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998550491485531e-05,
+      "loss": 1.6524,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998541694130453e-05,
+      "loss": 1.5435,
+      "step": 1273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998532870167424e-05,
+      "loss": 1.6014,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998524019596536e-05,
+      "loss": 1.6035,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998515142417884e-05,
+      "loss": 1.5596,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998506238631563e-05,
+      "loss": 1.6439,
+      "step": 1277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998497308237668e-05,
+      "loss": 1.6366,
+      "step": 1278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998488351236293e-05,
+      "loss": 1.5285,
+      "step": 1279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998479367627534e-05,
+      "loss": 1.6112,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9984703574114863e-05,
+      "loss": 1.5378,
+      "step": 1281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998461320588246e-05,
+      "loss": 1.4846,
+      "step": 1282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.99845225715791e-05,
+      "loss": 1.5814,
+      "step": 1283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9984431671205747e-05,
+      "loss": 1.6221,
+      "step": 1284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998434050476336e-05,
+      "loss": 1.5925,
+      "step": 1285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998424907225291e-05,
+      "loss": 1.6229,
+      "step": 1286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998415737367538e-05,
+      "loss": 1.5622,
+      "step": 1287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998406540903174e-05,
+      "loss": 1.5941,
+      "step": 1288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998397317832298e-05,
+      "loss": 1.5723,
+      "step": 1289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9983880681550065e-05,
+      "loss": 1.5894,
+      "step": 1290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998378791871399e-05,
+      "loss": 1.6195,
+      "step": 1291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9983694889815745e-05,
+      "loss": 1.5589,
+      "step": 1292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9983601594856314e-05,
+      "loss": 1.6174,
+      "step": 1293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.99835080338367e-05,
+      "loss": 1.5964,
+      "step": 1294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998341420675788e-05,
+      "loss": 1.5243,
+      "step": 1295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998332011362088e-05,
+      "loss": 1.5378,
+      "step": 1296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998322575442669e-05,
+      "loss": 1.627,
+      "step": 1297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998313112917631e-05,
+      "loss": 1.5654,
+      "step": 1298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998303623787074e-05,
+      "loss": 1.5811,
+      "step": 1299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9982941080511015e-05,
+      "loss": 1.5704,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998284565709813e-05,
+      "loss": 1.6062,
+      "step": 1301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9982749967633104e-05,
+      "loss": 1.6045,
+      "step": 1302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998265401211696e-05,
+      "loss": 1.6317,
+      "step": 1303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9982557790550724e-05,
+      "loss": 1.5313,
+      "step": 1304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.99824613029354e-05,
+      "loss": 1.5834,
+      "step": 1305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9982364549272046e-05,
+      "loss": 1.5358,
+      "step": 1306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998226752956167e-05,
+      "loss": 1.6296,
+      "step": 1307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998217024380532e-05,
+      "loss": 1.6665,
+      "step": 1308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998207269200401e-05,
+      "loss": 1.6588,
+      "step": 1309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.99819748741588e-05,
+      "loss": 1.5905,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998187679027072e-05,
+      "loss": 1.5906,
+      "step": 1311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998177844034083e-05,
+      "loss": 1.6717,
+      "step": 1312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998167982437015e-05,
+      "loss": 1.5995,
+      "step": 1313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9981580942359764e-05,
+      "loss": 1.5129,
+      "step": 1314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.99814817943107e-05,
+      "loss": 1.6252,
+      "step": 1315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9981382380224016e-05,
+      "loss": 1.634,
+      "step": 1316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9981282700100784e-05,
+      "loss": 1.6123,
+      "step": 1317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9981182753942054e-05,
+      "loss": 1.6246,
+      "step": 1318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998108254174889e-05,
+      "loss": 1.5902,
+      "step": 1319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9980982063522374e-05,
+      "loss": 1.5207,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998088131926356e-05,
+      "loss": 1.6447,
+      "step": 1321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998078030897352e-05,
+      "loss": 1.6123,
+      "step": 1322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998067903265334e-05,
+      "loss": 1.6169,
+      "step": 1323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9980577490304104e-05,
+      "loss": 1.5057,
+      "step": 1324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998047568192688e-05,
+      "loss": 1.6145,
+      "step": 1325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9980373607522746e-05,
+      "loss": 1.5117,
+      "step": 1326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9980271267092806e-05,
+      "loss": 1.5782,
+      "step": 1327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998016866063814e-05,
+      "loss": 1.613,
+      "step": 1328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9980065788159845e-05,
+      "loss": 1.5676,
+      "step": 1329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997996264965902e-05,
+      "loss": 1.5935,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997985924513675e-05,
+      "loss": 1.5859,
+      "step": 1331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9979755574594153e-05,
+      "loss": 1.5363,
+      "step": 1332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997965163803232e-05,
+      "loss": 1.585,
+      "step": 1333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997954743545236e-05,
+      "loss": 1.5596,
+      "step": 1334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997944296685539e-05,
+      "loss": 1.5492,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997933823224252e-05,
+      "loss": 1.6442,
+      "step": 1336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997923323161486e-05,
+      "loss": 1.6706,
+      "step": 1337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997912796497353e-05,
+      "loss": 1.6211,
+      "step": 1338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997902243231966e-05,
+      "loss": 1.5939,
+      "step": 1339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997891663365435e-05,
+      "loss": 1.5975,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9978810568978754e-05,
+      "loss": 1.5957,
+      "step": 1341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997870423829399e-05,
+      "loss": 1.5249,
+      "step": 1342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997859764160119e-05,
+      "loss": 1.6746,
+      "step": 1343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997849077890149e-05,
+      "loss": 1.5031,
+      "step": 1344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9978383650196025e-05,
+      "loss": 1.6232,
+      "step": 1345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9978276255485937e-05,
+      "loss": 1.6124,
+      "step": 1346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997816859477238e-05,
+      "loss": 1.6144,
+      "step": 1347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9978060668056484e-05,
+      "loss": 1.6711,
+      "step": 1348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997795247533942e-05,
+      "loss": 1.5285,
+      "step": 1349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997784401662231e-05,
+      "loss": 1.5074,
+      "step": 1350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9977735291906334e-05,
+      "loss": 1.605,
+      "step": 1351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9977626301192636e-05,
+      "loss": 1.5539,
+      "step": 1352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997751704448239e-05,
+      "loss": 1.5838,
+      "step": 1353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997740752177675e-05,
+      "loss": 1.6221,
+      "step": 1354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997729773307688e-05,
+      "loss": 1.5671,
+      "step": 1355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9977187678383955e-05,
+      "loss": 1.6057,
+      "step": 1356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997707735769915e-05,
+      "loss": 1.5788,
+      "step": 1357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9976966771023626e-05,
+      "loss": 1.6197,
+      "step": 1358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997685591835858e-05,
+      "loss": 1.6018,
+      "step": 1359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997674479970518e-05,
+      "loss": 1.5736,
+      "step": 1360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.99766334150646e-05,
+      "loss": 1.543,
+      "step": 1361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997652176443804e-05,
+      "loss": 1.5866,
+      "step": 1362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.99764098478267e-05,
+      "loss": 1.614,
+      "step": 1363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9976297665231755e-05,
+      "loss": 1.5349,
+      "step": 1364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9976185216654396e-05,
+      "loss": 1.5439,
+      "step": 1365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9976072502095836e-05,
+      "loss": 1.6392,
+      "step": 1366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9975959521557256e-05,
+      "loss": 1.5394,
+      "step": 1367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997584627503988e-05,
+      "loss": 1.6814,
+      "step": 1368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9975732762544905e-05,
+      "loss": 1.6371,
+      "step": 1369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997561898407353e-05,
+      "loss": 1.5917,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997550493962699e-05,
+      "loss": 1.4918,
+      "step": 1371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9975390629206474e-05,
+      "loss": 1.5907,
+      "step": 1372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997527605281321e-05,
+      "loss": 1.5495,
+      "step": 1373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997516121044843e-05,
+      "loss": 1.5909,
+      "step": 1374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997504610211334e-05,
+      "loss": 1.5852,
+      "step": 1375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997493072780917e-05,
+      "loss": 1.5967,
+      "step": 1376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997481508753715e-05,
+      "loss": 1.6089,
+      "step": 1377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997469918129851e-05,
+      "loss": 1.4776,
+      "step": 1378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997458300909449e-05,
+      "loss": 1.6026,
+      "step": 1379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997446657092632e-05,
+      "loss": 1.5414,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997434986679525e-05,
+      "loss": 1.624,
+      "step": 1381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997423289670251e-05,
+      "loss": 1.607,
+      "step": 1382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997411566064936e-05,
+      "loss": 1.5079,
+      "step": 1383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997399815863703e-05,
+      "loss": 1.5604,
+      "step": 1384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9973880390666786e-05,
+      "loss": 1.5528,
+      "step": 1385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997376235673988e-05,
+      "loss": 1.5641,
+      "step": 1386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997364405685756e-05,
+      "loss": 1.5646,
+      "step": 1387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9973525491021103e-05,
+      "loss": 1.6356,
+      "step": 1388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997340665923175e-05,
+      "loss": 1.5415,
+      "step": 1389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997328756149079e-05,
+      "loss": 1.5461,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997316819779947e-05,
+      "loss": 1.6126,
+      "step": 1391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9973048568159074e-05,
+      "loss": 1.6397,
+      "step": 1392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9972928672570866e-05,
+      "loss": 1.6278,
+      "step": 1393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997280851103614e-05,
+      "loss": 1.6078,
+      "step": 1394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997268808355616e-05,
+      "loss": 1.6455,
+      "step": 1395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997256739013221e-05,
+      "loss": 1.5928,
+      "step": 1396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997244643076558e-05,
+      "loss": 1.6446,
+      "step": 1397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997232520545756e-05,
+      "loss": 1.607,
+      "step": 1398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997220371420943e-05,
+      "loss": 1.6353,
+      "step": 1399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9972081957022497e-05,
+      "loss": 1.5877,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997195993389805e-05,
+      "loss": 1.5467,
+      "step": 1401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.99718376448374e-05,
+      "loss": 1.5352,
+      "step": 1402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9971715089841826e-05,
+      "loss": 1.6382,
+      "step": 1403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997159226891265e-05,
+      "loss": 1.6328,
+      "step": 1404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997146918205118e-05,
+      "loss": 1.6789,
+      "step": 1405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9971345829258723e-05,
+      "loss": 1.5754,
+      "step": 1406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997122221053659e-05,
+      "loss": 1.6615,
+      "step": 1407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997109832588611e-05,
+      "loss": 1.6994,
+      "step": 1408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9970974175308595e-05,
+      "loss": 1.5404,
+      "step": 1409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997084975880535e-05,
+      "loss": 1.5964,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997072507637772e-05,
+      "loss": 1.6341,
+      "step": 1411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997060012802704e-05,
+      "loss": 1.6252,
+      "step": 1412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9970474913754614e-05,
+      "loss": 1.5648,
+      "step": 1413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.99703494335618e-05,
+      "loss": 1.6418,
+      "step": 1414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997022368744992e-05,
+      "loss": 1.5887,
+      "step": 1415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997009767542032e-05,
+      "loss": 1.6233,
+      "step": 1416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9969971397474336e-05,
+      "loss": 1.605,
+      "step": 1417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996984485361332e-05,
+      "loss": 1.5995,
+      "step": 1418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996971804383861e-05,
+      "loss": 1.5545,
+      "step": 1419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996959096815157e-05,
+      "loss": 1.5845,
+      "step": 1420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996946362655354e-05,
+      "loss": 1.5611,
+      "step": 1421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996933601904589e-05,
+      "loss": 1.6341,
+      "step": 1422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996920814562996e-05,
+      "loss": 1.6001,
+      "step": 1423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9969080006307124e-05,
+      "loss": 1.5878,
+      "step": 1424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996895160107875e-05,
+      "loss": 1.6013,
+      "step": 1425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.99688229299462e-05,
+      "loss": 1.5466,
+      "step": 1426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996869399291084e-05,
+      "loss": 1.6014,
+      "step": 1427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9968564789974046e-05,
+      "loss": 1.6418,
+      "step": 1428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9968435321137204e-05,
+      "loss": 1.6236,
+      "step": 1429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996830558640168e-05,
+      "loss": 1.6153,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996817558576886e-05,
+      "loss": 1.4568,
+      "step": 1431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996804531924013e-05,
+      "loss": 1.5487,
+      "step": 1432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996791478681688e-05,
+      "loss": 1.663,
+      "step": 1433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996778398850049e-05,
+      "loss": 1.6363,
+      "step": 1434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9967652924292355e-05,
+      "loss": 1.5676,
+      "step": 1435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996752159419389e-05,
+      "loss": 1.5738,
+      "step": 1436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9967389998206457e-05,
+      "loss": 1.5297,
+      "step": 1437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9967258136331495e-05,
+      "loss": 1.5625,
+      "step": 1438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9967126008570384e-05,
+      "loss": 1.6757,
+      "step": 1439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996699361492454e-05,
+      "loss": 1.5897,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996686095539538e-05,
+      "loss": 1.5471,
+      "step": 1441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.99667280299843e-05,
+      "loss": 1.5899,
+      "step": 1442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996659483869273e-05,
+      "loss": 1.5831,
+      "step": 1443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996646138152207e-05,
+      "loss": 1.59,
+      "step": 1444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996632765847377e-05,
+      "loss": 1.5871,
+      "step": 1445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9966193669549225e-05,
+      "loss": 1.5835,
+      "step": 1446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9966059414749876e-05,
+      "loss": 1.59,
+      "step": 1447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996592489407715e-05,
+      "loss": 1.5496,
+      "step": 1448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9965790107532494e-05,
+      "loss": 1.5596,
+      "step": 1449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996565505511732e-05,
+      "loss": 1.5891,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996551973683308e-05,
+      "loss": 1.6479,
+      "step": 1451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9965384152681216e-05,
+      "loss": 1.4931,
+      "step": 1452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996524830266316e-05,
+      "loss": 1.587,
+      "step": 1453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9965112186780374e-05,
+      "loss": 1.5587,
+      "step": 1454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996497580503429e-05,
+      "loss": 1.5409,
+      "step": 1455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996483915742637e-05,
+      "loss": 1.5674,
+      "step": 1456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9964702243958084e-05,
+      "loss": 1.6137,
+      "step": 1457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996456506463086e-05,
+      "loss": 1.5948,
+      "step": 1458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9964427619446185e-05,
+      "loss": 1.6226,
+      "step": 1459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.99642899084055e-05,
+      "loss": 1.6522,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996415193151029e-05,
+      "loss": 1.5382,
+      "step": 1461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996401368876202e-05,
+      "loss": 1.5297,
+      "step": 1462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996387518016217e-05,
+      "loss": 1.6098,
+      "step": 1463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996373640571219e-05,
+      "loss": 1.5696,
+      "step": 1464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996359736541359e-05,
+      "loss": 1.6559,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996345805926781e-05,
+      "loss": 1.6141,
+      "step": 1466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996331848727637e-05,
+      "loss": 1.52,
+      "step": 1467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996317864944074e-05,
+      "loss": 1.5948,
+      "step": 1468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9963038545762416e-05,
+      "loss": 1.5472,
+      "step": 1469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996289817624289e-05,
+      "loss": 1.5787,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9962757540883654e-05,
+      "loss": 1.6151,
+      "step": 1471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.99626166396862e-05,
+      "loss": 1.5837,
+      "step": 1472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9962475472652035e-05,
+      "loss": 1.5537,
+      "step": 1473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9962334039782665e-05,
+      "loss": 1.4961,
+      "step": 1474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996219234107959e-05,
+      "loss": 1.5734,
+      "step": 1475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996205037654432e-05,
+      "loss": 1.6559,
+      "step": 1476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996190814617837e-05,
+      "loss": 1.6164,
+      "step": 1477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996176564998325e-05,
+      "loss": 1.4802,
+      "step": 1478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996162288796049e-05,
+      "loss": 1.6826,
+      "step": 1479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996147986011159e-05,
+      "loss": 1.5429,
+      "step": 1480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996133656643809e-05,
+      "loss": 1.6439,
+      "step": 1481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996119300694151e-05,
+      "loss": 1.6654,
+      "step": 1482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996104918162338e-05,
+      "loss": 1.5422,
+      "step": 1483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996090509048522e-05,
+      "loss": 1.6221,
+      "step": 1484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9960760733528576e-05,
+      "loss": 1.5944,
+      "step": 1485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9960616110755e-05,
+      "loss": 1.6133,
+      "step": 1486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9960471222166006e-05,
+      "loss": 1.5431,
+      "step": 1487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996032606776314e-05,
+      "loss": 1.6096,
+      "step": 1488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996018064754797e-05,
+      "loss": 1.6349,
+      "step": 1489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9960034961522016e-05,
+      "loss": 1.5386,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995988900968686e-05,
+      "loss": 1.606,
+      "step": 1491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9959742792044026e-05,
+      "loss": 1.5594,
+      "step": 1492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995959630859509e-05,
+      "loss": 1.5453,
+      "step": 1493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9959449559341604e-05,
+      "loss": 1.598,
+      "step": 1494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995930254428514e-05,
+      "loss": 1.537,
+      "step": 1495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995915526342725e-05,
+      "loss": 1.5215,
+      "step": 1496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9959007716769505e-05,
+      "loss": 1.5409,
+      "step": 1497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995885990431349e-05,
+      "loss": 1.5933,
+      "step": 1498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995871182606077e-05,
+      "loss": 1.5762,
+      "step": 1499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995856348201291e-05,
+      "loss": 1.5785,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9958414872171514e-05,
+      "loss": 1.5821,
+      "step": 1501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995826599653814e-05,
+      "loss": 1.6081,
+      "step": 1502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.99581168551144e-05,
+      "loss": 1.5752,
+      "step": 1503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995796744790186e-05,
+      "loss": 1.5794,
+      "step": 1504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9957817774902126e-05,
+      "loss": 1.5044,
+      "step": 1505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995766783611678e-05,
+      "loss": 1.5604,
+      "step": 1506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9957517631547424e-05,
+      "loss": 1.5973,
+      "step": 1507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9957367161195656e-05,
+      "loss": 1.4991,
+      "step": 1508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995721642506308e-05,
+      "loss": 1.545,
+      "step": 1509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995706542315131e-05,
+      "loss": 1.5324,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9956914155461935e-05,
+      "loss": 1.583,
+      "step": 1511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995676262199658e-05,
+      "loss": 1.5482,
+      "step": 1512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9956610822756843e-05,
+      "loss": 1.5441,
+      "step": 1513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9956458757744365e-05,
+      "loss": 1.6172,
+      "step": 1514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995630642696075e-05,
+      "loss": 1.6468,
+      "step": 1515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995615383040763e-05,
+      "loss": 1.6056,
+      "step": 1516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995600096808661e-05,
+      "loss": 1.6096,
+      "step": 1517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995584783999934e-05,
+      "loss": 1.5591,
+      "step": 1518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9955694446147436e-05,
+      "loss": 1.6283,
+      "step": 1519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995554078653254e-05,
+      "loss": 1.5679,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995538686115628e-05,
+      "loss": 1.5956,
+      "step": 1521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.99552326700203e-05,
+      "loss": 1.5847,
+      "step": 1522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995507821312625e-05,
+      "loss": 1.5905,
+      "step": 1523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995492349047577e-05,
+      "loss": 1.5215,
+      "step": 1524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9954768502070496e-05,
+      "loss": 1.6338,
+      "step": 1525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.99546132479121e-05,
+      "loss": 1.5194,
+      "step": 1526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995445772800221e-05,
+      "loss": 1.5093,
+      "step": 1527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9954301942342504e-05,
+      "loss": 1.62,
+      "step": 1528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995414589093463e-05,
+      "loss": 1.5658,
+      "step": 1529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995398957378025e-05,
+      "loss": 1.68,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995383299088103e-05,
+      "loss": 1.6146,
+      "step": 1531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995367614223865e-05,
+      "loss": 1.6113,
+      "step": 1532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9953519027854754e-05,
+      "loss": 1.5803,
+      "step": 1533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9953361647731037e-05,
+      "loss": 1.5521,
+      "step": 1534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995320400186917e-05,
+      "loss": 1.5905,
+      "step": 1535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995304609027083e-05,
+      "loss": 1.5903,
+      "step": 1536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995288791293769e-05,
+      "loss": 1.6543,
+      "step": 1537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995272946987145e-05,
+      "loss": 1.6263,
+      "step": 1538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995257076107379e-05,
+      "loss": 1.5687,
+      "step": 1539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9952411786546395e-05,
+      "loss": 1.567,
+      "step": 1540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995225254629097e-05,
+      "loss": 1.5735,
+      "step": 1541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99520930403092e-05,
+      "loss": 1.5993,
+      "step": 1542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995193326860279e-05,
+      "loss": 1.4458,
+      "step": 1543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995177323117344e-05,
+      "loss": 1.5634,
+      "step": 1544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9951612928022854e-05,
+      "loss": 1.6608,
+      "step": 1545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995145235915274e-05,
+      "loss": 1.6298,
+      "step": 1546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99512915245648e-05,
+      "loss": 1.6441,
+      "step": 1547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9951130424260764e-05,
+      "loss": 1.541,
+      "step": 1548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995096905824233e-05,
+      "loss": 1.5465,
+      "step": 1549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9950807426511225e-05,
+      "loss": 1.5513,
+      "step": 1550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9950645529069165e-05,
+      "loss": 1.531,
+      "step": 1551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995048336591789e-05,
+      "loss": 1.6093,
+      "step": 1552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99503209370591e-05,
+      "loss": 1.6099,
+      "step": 1553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995015824249455e-05,
+      "loss": 1.628,
+      "step": 1554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994999528222596e-05,
+      "loss": 1.5411,
+      "step": 1555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9949832056255065e-05,
+      "loss": 1.5661,
+      "step": 1556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994966856458361e-05,
+      "loss": 1.5127,
+      "step": 1557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994950480721333e-05,
+      "loss": 1.5995,
+      "step": 1558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994934078414597e-05,
+      "loss": 1.5666,
+      "step": 1559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9949176495383276e-05,
+      "loss": 1.5375,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9949011940927006e-05,
+      "loss": 1.6444,
+      "step": 1561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9948847120778896e-05,
+      "loss": 1.6275,
+      "step": 1562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994868203494072e-05,
+      "loss": 1.6109,
+      "step": 1563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9948516683414224e-05,
+      "loss": 1.5342,
+      "step": 1564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994835106620117e-05,
+      "loss": 1.5827,
+      "step": 1565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994818518330333e-05,
+      "loss": 1.5873,
+      "step": 1566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994801903472246e-05,
+      "loss": 1.5908,
+      "step": 1567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9947852620460326e-05,
+      "loss": 1.5568,
+      "step": 1568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9947685940518715e-05,
+      "loss": 1.5424,
+      "step": 1569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99475189948994e-05,
+      "loss": 1.5654,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994735178360415e-05,
+      "loss": 1.5563,
+      "step": 1571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994718430663474e-05,
+      "loss": 1.5737,
+      "step": 1572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9947016563992976e-05,
+      "loss": 1.5917,
+      "step": 1573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994684855568062e-05,
+      "loss": 1.6063,
+      "step": 1574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994668028169949e-05,
+      "loss": 1.5323,
+      "step": 1575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994651174205134e-05,
+      "loss": 1.6348,
+      "step": 1576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9946342936738e-05,
+      "loss": 1.6397,
+      "step": 1577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9946173865761245e-05,
+      "loss": 1.5384,
+      "step": 1578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9946004529122885e-05,
+      "loss": 1.506,
+      "step": 1579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994583492682472e-05,
+      "loss": 1.6094,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994566505886856e-05,
+      "loss": 1.5542,
+      "step": 1581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994549492525622e-05,
+      "loss": 1.5944,
+      "step": 1582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994532452598949e-05,
+      "loss": 1.5473,
+      "step": 1583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994515386107021e-05,
+      "loss": 1.5554,
+      "step": 1584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9944982930500185e-05,
+      "loss": 1.5718,
+      "step": 1585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994481173428123e-05,
+      "loss": 1.5928,
+      "step": 1586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994464027241518e-05,
+      "loss": 1.6273,
+      "step": 1587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9944468544903846e-05,
+      "loss": 1.5383,
+      "step": 1588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994429655174907e-05,
+      "loss": 1.5892,
+      "step": 1589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994412429295269e-05,
+      "loss": 1.5277,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994395176851652e-05,
+      "loss": 1.6086,
+      "step": 1591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9943778978442405e-05,
+      "loss": 1.5081,
+      "step": 1592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9943605922732195e-05,
+      "loss": 1.5789,
+      "step": 1593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9943432601387715e-05,
+      "loss": 1.5667,
+      "step": 1594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994325901441084e-05,
+      "loss": 1.5078,
+      "step": 1595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994308516180338e-05,
+      "loss": 1.6467,
+      "step": 1596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994291104356722e-05,
+      "loss": 1.6152,
+      "step": 1597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9942736659704195e-05,
+      "loss": 1.5965,
+      "step": 1598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994256201021617e-05,
+      "loss": 1.5866,
+      "step": 1599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9942387095105004e-05,
+      "loss": 1.626,
+      "step": 1600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9942211914372556e-05,
+      "loss": 1.5536,
+      "step": 1601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994203646802069e-05,
+      "loss": 1.6201,
+      "step": 1602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994186075605128e-05,
+      "loss": 1.5251,
+      "step": 1603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9941684778466206e-05,
+      "loss": 1.499,
+      "step": 1604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994150853526732e-05,
+      "loss": 1.6022,
+      "step": 1605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994133202645651e-05,
+      "loss": 1.5863,
+      "step": 1606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994115525203566e-05,
+      "loss": 1.6115,
+      "step": 1607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994097821200665e-05,
+      "loss": 1.6167,
+      "step": 1608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994080090637137e-05,
+      "loss": 1.624,
+      "step": 1609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9940623335131686e-05,
+      "loss": 1.6694,
+      "step": 1610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994044549828951e-05,
+      "loss": 1.551,
+      "step": 1611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994026739584674e-05,
+      "loss": 1.5576,
+      "step": 1612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994008902780526e-05,
+      "loss": 1.627,
+      "step": 1613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993991039416697e-05,
+      "loss": 1.5467,
+      "step": 1614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9939731494933776e-05,
+      "loss": 1.5397,
+      "step": 1615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9939552330107585e-05,
+      "loss": 1.5972,
+      "step": 1616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9939372899690304e-05,
+      "loss": 1.5021,
+      "step": 1617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993919320368384e-05,
+      "loss": 1.5827,
+      "step": 1618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9939013242090114e-05,
+      "loss": 1.5978,
+      "step": 1619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993883301491103e-05,
+      "loss": 1.6192,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993865252214852e-05,
+      "loss": 1.5773,
+      "step": 1621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.99384717638045e-05,
+      "loss": 1.5751,
+      "step": 1622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.99382907398809e-05,
+      "loss": 1.592,
+      "step": 1623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993810945037964e-05,
+      "loss": 1.6641,
+      "step": 1624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993792789530266e-05,
+      "loss": 1.5728,
+      "step": 1625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993774607465188e-05,
+      "loss": 1.5928,
+      "step": 1626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993756398842925e-05,
+      "loss": 1.5747,
+      "step": 1627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993738163663671e-05,
+      "loss": 1.6022,
+      "step": 1628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9937199019276184e-05,
+      "loss": 1.5327,
+      "step": 1629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993701613634963e-05,
+      "loss": 1.5142,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993683298785899e-05,
+      "loss": 1.5659,
+      "step": 1631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993664957380623e-05,
+      "loss": 1.5375,
+      "step": 1632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9936465894193285e-05,
+      "loss": 1.5902,
+      "step": 1633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993628194902211e-05,
+      "loss": 1.5757,
+      "step": 1634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993609773829468e-05,
+      "loss": 1.5472,
+      "step": 1635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9935913262012945e-05,
+      "loss": 1.554,
+      "step": 1636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9935728520178884e-05,
+      "loss": 1.5,
+      "step": 1637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993554351279444e-05,
+      "loss": 1.5165,
+      "step": 1638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.99353582398616e-05,
+      "loss": 1.5723,
+      "step": 1639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993517270138233e-05,
+      "loss": 1.6063,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9934986897358614e-05,
+      "loss": 1.5801,
+      "step": 1641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9934800827792426e-05,
+      "loss": 1.6071,
+      "step": 1642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993461449268575e-05,
+      "loss": 1.5933,
+      "step": 1643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993442789204056e-05,
+      "loss": 1.5858,
+      "step": 1644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993424102585885e-05,
+      "loss": 1.548,
+      "step": 1645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993405389414262e-05,
+      "loss": 1.5757,
+      "step": 1646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993386649689385e-05,
+      "loss": 1.6398,
+      "step": 1647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993367883411454e-05,
+      "loss": 1.5163,
+      "step": 1648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993349090580669e-05,
+      "loss": 1.4922,
+      "step": 1649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9933302711972294e-05,
+      "loss": 1.5768,
+      "step": 1650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993311425261337e-05,
+      "loss": 1.6556,
+      "step": 1651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9932925527731903e-05,
+      "loss": 1.5638,
+      "step": 1652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993273653732993e-05,
+      "loss": 1.5374,
+      "step": 1653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9932547281409436e-05,
+      "loss": 1.5662,
+      "step": 1654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993235775997246e-05,
+      "loss": 1.5516,
+      "step": 1655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993216797302101e-05,
+      "loss": 1.6056,
+      "step": 1656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993197792055711e-05,
+      "loss": 1.5755,
+      "step": 1657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993178760258278e-05,
+      "loss": 1.672,
+      "step": 1658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993159701910004e-05,
+      "loss": 1.5877,
+      "step": 1659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993140617011094e-05,
+      "loss": 1.6058,
+      "step": 1660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.99312150556175e-05,
+      "loss": 1.5786,
+      "step": 1661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9931023675621746e-05,
+      "loss": 1.5194,
+      "step": 1662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993083203012573e-05,
+      "loss": 1.5602,
+      "step": 1663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9930640119131486e-05,
+      "loss": 1.5448,
+      "step": 1664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.993044794264107e-05,
+      "loss": 1.6776,
+      "step": 1665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.993025550065651e-05,
+      "loss": 1.5241,
+      "step": 1666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.993006279317987e-05,
+      "loss": 1.6127,
+      "step": 1667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9929869820213194e-05,
+      "loss": 1.6335,
+      "step": 1668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992967658175854e-05,
+      "loss": 1.5563,
+      "step": 1669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9929483077817964e-05,
+      "loss": 1.5567,
+      "step": 1670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9929289308393535e-05,
+      "loss": 1.5104,
+      "step": 1671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9929095273487303e-05,
+      "loss": 1.5553,
+      "step": 1672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992890097310134e-05,
+      "loss": 1.57,
+      "step": 1673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9928706407237725e-05,
+      "loss": 1.5715,
+      "step": 1674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992851157589851e-05,
+      "loss": 1.55,
+      "step": 1675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992831647908579e-05,
+      "loss": 1.4816,
+      "step": 1676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9928121116801627e-05,
+      "loss": 1.641,
+      "step": 1677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992792548904811e-05,
+      "loss": 1.6208,
+      "step": 1678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992772959582733e-05,
+      "loss": 1.5712,
+      "step": 1679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992753343714135e-05,
+      "loss": 1.5972,
+      "step": 1680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992733701299228e-05,
+      "loss": 1.5222,
+      "step": 1681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9927140323382205e-05,
+      "loss": 1.5775,
+      "step": 1682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992694336831322e-05,
+      "loss": 1.5518,
+      "step": 1683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992674614778742e-05,
+      "loss": 1.518,
+      "step": 1684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.99265486618069e-05,
+      "loss": 1.4474,
+      "step": 1685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992635091037378e-05,
+      "loss": 1.5236,
+      "step": 1686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9926152893490155e-05,
+      "loss": 1.6394,
+      "step": 1687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9925954611158134e-05,
+      "loss": 1.5407,
+      "step": 1688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9925756063379825e-05,
+      "loss": 1.6057,
+      "step": 1689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992555725015735e-05,
+      "loss": 1.5933,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992535817149282e-05,
+      "loss": 1.5014,
+      "step": 1691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992515882738836e-05,
+      "loss": 1.5658,
+      "step": 1692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992495921784609e-05,
+      "loss": 1.598,
+      "step": 1693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9924759342868144e-05,
+      "loss": 1.6056,
+      "step": 1694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992455920245663e-05,
+      "loss": 1.5578,
+      "step": 1695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.99243587966137e-05,
+      "loss": 1.6493,
+      "step": 1696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9924158125341487e-05,
+      "loss": 1.6237,
+      "step": 1697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992395718864211e-05,
+      "loss": 1.6115,
+      "step": 1698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992375598651773e-05,
+      "loss": 1.6192,
+      "step": 1699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992355451897047e-05,
+      "loss": 1.5902,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9923352786002495e-05,
+      "loss": 1.5126,
+      "step": 1701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9923150787615944e-05,
+      "loss": 1.6219,
+      "step": 1702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992294852381296e-05,
+      "loss": 1.6127,
+      "step": 1703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9922745994595714e-05,
+      "loss": 1.4968,
+      "step": 1704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992254319996635e-05,
+      "loss": 1.6156,
+      "step": 1705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9922340139927035e-05,
+      "loss": 1.4843,
+      "step": 1706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9922136814479926e-05,
+      "loss": 1.5695,
+      "step": 1707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992193322362719e-05,
+      "loss": 1.5525,
+      "step": 1708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9921729367370996e-05,
+      "loss": 1.6223,
+      "step": 1709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992152524571352e-05,
+      "loss": 1.5591,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9921320858656917e-05,
+      "loss": 1.5153,
+      "step": 1711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992111620620339e-05,
+      "loss": 1.5442,
+      "step": 1712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9920911288355096e-05,
+      "loss": 1.6123,
+      "step": 1713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9920706105114235e-05,
+      "loss": 1.5414,
+      "step": 1714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9920500656482976e-05,
+      "loss": 1.5105,
+      "step": 1715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992029494246352e-05,
+      "loss": 1.6019,
+      "step": 1716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9920088963058054e-05,
+      "loss": 1.632,
+      "step": 1717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991988271826876e-05,
+      "loss": 1.6229,
+      "step": 1718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991967620809785e-05,
+      "loss": 1.6511,
+      "step": 1719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991946943254752e-05,
+      "loss": 1.5524,
+      "step": 1720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991926239161997e-05,
+      "loss": 1.6662,
+      "step": 1721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.99190550853174e-05,
+      "loss": 1.5812,
+      "step": 1722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991884751364203e-05,
+      "loss": 1.6107,
+      "step": 1723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9918639676596054e-05,
+      "loss": 1.5487,
+      "step": 1724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.99184315741817e-05,
+      "loss": 1.5926,
+      "step": 1725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991822320640117e-05,
+      "loss": 1.6449,
+      "step": 1726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9918014573256697e-05,
+      "loss": 1.4919,
+      "step": 1727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.99178056747505e-05,
+      "loss": 1.544,
+      "step": 1728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991759651088479e-05,
+      "loss": 1.5918,
+      "step": 1729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991738708166182e-05,
+      "loss": 1.5973,
+      "step": 1730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991717738708379e-05,
+      "loss": 1.5972,
+      "step": 1731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9916967427152955e-05,
+      "loss": 1.6276,
+      "step": 1732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9916757201871545e-05,
+      "loss": 1.6011,
+      "step": 1733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.99165467112418e-05,
+      "loss": 1.543,
+      "step": 1734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991633595526596e-05,
+      "loss": 1.5599,
+      "step": 1735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991612493394626e-05,
+      "loss": 1.5785,
+      "step": 1736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991591364728496e-05,
+      "loss": 1.5943,
+      "step": 1737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9915702095284314e-05,
+      "loss": 1.6055,
+      "step": 1738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991549027794656e-05,
+      "loss": 1.603,
+      "step": 1739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9915278195273965e-05,
+      "loss": 1.5718,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9915065847268785e-05,
+      "loss": 1.5487,
+      "step": 1741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991485323393328e-05,
+      "loss": 1.5947,
+      "step": 1742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.99146403552697e-05,
+      "loss": 1.5383,
+      "step": 1743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9914427211280346e-05,
+      "loss": 1.5804,
+      "step": 1744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991421380196746e-05,
+      "loss": 1.5571,
+      "step": 1745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991400012733332e-05,
+      "loss": 1.5741,
+      "step": 1746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9913786187380207e-05,
+      "loss": 1.5593,
+      "step": 1747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.99135719821104e-05,
+      "loss": 1.6089,
+      "step": 1748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991335751152617e-05,
+      "loss": 1.5537,
+      "step": 1749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9913142775629814e-05,
+      "loss": 1.6154,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991292777442361e-05,
+      "loss": 1.6174,
+      "step": 1751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9912712507909856e-05,
+      "loss": 1.5881,
+      "step": 1752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991249697609083e-05,
+      "loss": 1.5273,
+      "step": 1753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9912281178968834e-05,
+      "loss": 1.5973,
+      "step": 1754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991206511654617e-05,
+      "loss": 1.5697,
+      "step": 1755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9911848788825145e-05,
+      "loss": 1.5783,
+      "step": 1756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991163219580804e-05,
+      "loss": 1.5127,
+      "step": 1757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9911415337497194e-05,
+      "loss": 1.5417,
+      "step": 1758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991119821389489e-05,
+      "loss": 1.5549,
+      "step": 1759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991098082500345e-05,
+      "loss": 1.5518,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991076317082519e-05,
+      "loss": 1.53,
+      "step": 1761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991054525136242e-05,
+      "loss": 1.6367,
+      "step": 1762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991032706661747e-05,
+      "loss": 1.604,
+      "step": 1763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9910108616592664e-05,
+      "loss": 1.569,
+      "step": 1764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990988990129032e-05,
+      "loss": 1.5744,
+      "step": 1765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990967092071277e-05,
+      "loss": 1.5526,
+      "step": 1766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9909451674862354e-05,
+      "loss": 1.5759,
+      "step": 1767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990923216374139e-05,
+      "loss": 1.555,
+      "step": 1768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990901238735224e-05,
+      "loss": 1.6006,
+      "step": 1769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990879234569722e-05,
+      "loss": 1.6084,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990857203877869e-05,
+      "loss": 1.5577,
+      "step": 1771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990835146659899e-05,
+      "loss": 1.5775,
+      "step": 1772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990813062916047e-05,
+      "loss": 1.5441,
+      "step": 1773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990790952646547e-05,
+      "loss": 1.5368,
+      "step": 1774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9907688158516365e-05,
+      "loss": 1.5506,
+      "step": 1775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9907466525315505e-05,
+      "loss": 1.5449,
+      "step": 1776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990724462686524e-05,
+      "loss": 1.5946,
+      "step": 1777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9907022463167945e-05,
+      "loss": 1.5671,
+      "step": 1778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990680003422599e-05,
+      "loss": 1.6369,
+      "step": 1779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990657734004173e-05,
+      "loss": 1.6216,
+      "step": 1780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9906354380617536e-05,
+      "loss": 1.5824,
+      "step": 1781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990613115595579e-05,
+      "loss": 1.5501,
+      "step": 1782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9905907666058875e-05,
+      "loss": 1.5763,
+      "step": 1783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990568391092916e-05,
+      "loss": 1.6227,
+      "step": 1784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990545989056904e-05,
+      "loss": 1.5576,
+      "step": 1785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990523560498088e-05,
+      "loss": 1.6115,
+      "step": 1786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990501105416709e-05,
+      "loss": 1.6053,
+      "step": 1787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990478623813005e-05,
+      "loss": 1.6048,
+      "step": 1788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990456115687215e-05,
+      "loss": 1.6046,
+      "step": 1789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990433581039581e-05,
+      "loss": 1.6226,
+      "step": 1790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.99041101987034e-05,
+      "loss": 1.5574,
+      "step": 1791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9903884321797335e-05,
+      "loss": 1.5949,
+      "step": 1792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990365817968003e-05,
+      "loss": 1.5922,
+      "step": 1793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990343177235388e-05,
+      "loss": 1.4946,
+      "step": 1794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.99032050998213e-05,
+      "loss": 1.6104,
+      "step": 1795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990297816208471e-05,
+      "loss": 1.5938,
+      "step": 1796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9902750959146515e-05,
+      "loss": 1.5456,
+      "step": 1797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9902523491009145e-05,
+      "loss": 1.5705,
+      "step": 1798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9902295757675015e-05,
+      "loss": 1.5584,
+      "step": 1799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990206775914655e-05,
+      "loss": 1.6138,
+      "step": 1800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9901839495426193e-05,
+      "loss": 1.5169,
+      "step": 1801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990161096651636e-05,
+      "loss": 1.5718,
+      "step": 1802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9901382172419486e-05,
+      "loss": 1.5379,
+      "step": 1803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990115311313801e-05,
+      "loss": 1.5985,
+      "step": 1804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9900923788674374e-05,
+      "loss": 1.5783,
+      "step": 1805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9900694199031016e-05,
+      "loss": 1.5487,
+      "step": 1806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990046434421039e-05,
+      "loss": 1.5858,
+      "step": 1807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990023422421492e-05,
+      "loss": 1.55,
+      "step": 1808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990000383904708e-05,
+      "loss": 1.5693,
+      "step": 1809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989977318870932e-05,
+      "loss": 1.527,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989954227320409e-05,
+      "loss": 1.6597,
+      "step": 1811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989931109253385e-05,
+      "loss": 1.6115,
+      "step": 1812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9899079646701064e-05,
+      "loss": 1.6447,
+      "step": 1813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9898847935708196e-05,
+      "loss": 1.5669,
+      "step": 1814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9898615959557715e-05,
+      "loss": 1.5005,
+      "step": 1815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989838371825209e-05,
+      "loss": 1.515,
+      "step": 1816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.98981512117938e-05,
+      "loss": 1.5581,
+      "step": 1817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989791844018531e-05,
+      "loss": 1.534,
+      "step": 1818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.98976854034291e-05,
+      "loss": 1.5842,
+      "step": 1819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989745210152766e-05,
+      "loss": 1.6236,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989721853448348e-05,
+      "loss": 1.5282,
+      "step": 1821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989698470229902e-05,
+      "loss": 1.5674,
+      "step": 1822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9896750604976806e-05,
+      "loss": 1.572,
+      "step": 1823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9896516242519305e-05,
+      "loss": 1.5687,
+      "step": 1824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989628161492902e-05,
+      "loss": 1.5666,
+      "step": 1825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989604672220846e-05,
+      "loss": 1.5155,
+      "step": 1826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9895811564360103e-05,
+      "loss": 1.5591,
+      "step": 1827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989557614138647e-05,
+      "loss": 1.5881,
+      "step": 1828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9895340453290074e-05,
+      "loss": 1.6097,
+      "step": 1829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9895104500073406e-05,
+      "loss": 1.5175,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9894868281738996e-05,
+      "loss": 1.5688,
+      "step": 1831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989463179828936e-05,
+      "loss": 1.6016,
+      "step": 1832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9894395049726995e-05,
+      "loss": 1.5806,
+      "step": 1833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9894158036054445e-05,
+      "loss": 1.5302,
+      "step": 1834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989392075727423e-05,
+      "loss": 1.5932,
+      "step": 1835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9893683213388856e-05,
+      "loss": 1.6397,
+      "step": 1836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989344540440089e-05,
+      "loss": 1.5038,
+      "step": 1837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989320733031282e-05,
+      "loss": 1.5173,
+      "step": 1838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989296899112722e-05,
+      "loss": 1.5576,
+      "step": 1839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9892730386846614e-05,
+      "loss": 1.6143,
+      "step": 1840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989249151747354e-05,
+      "loss": 1.5702,
+      "step": 1841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9892252383010544e-05,
+      "loss": 1.571,
+      "step": 1842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989201298346017e-05,
+      "loss": 1.5899,
+      "step": 1843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989177331882497e-05,
+      "loss": 1.6332,
+      "step": 1844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.98915333891075e-05,
+      "loss": 1.5762,
+      "step": 1845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989129319431031e-05,
+      "loss": 1.6087,
+      "step": 1846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989105273443595e-05,
+      "loss": 1.5419,
+      "step": 1847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9890812009487006e-05,
+      "loss": 1.5385,
+      "step": 1848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989057101946602e-05,
+      "loss": 1.5068,
+      "step": 1849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9890329764375554e-05,
+      "loss": 1.5015,
+      "step": 1850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9890088244218205e-05,
+      "loss": 1.5503,
+      "step": 1851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988984645899651e-05,
+      "loss": 1.5297,
+      "step": 1852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988960440871308e-05,
+      "loss": 1.6086,
+      "step": 1853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9889362093370454e-05,
+      "loss": 1.5262,
+      "step": 1854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988911951297124e-05,
+      "loss": 1.5445,
+      "step": 1855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988887666751801e-05,
+      "loss": 1.5855,
+      "step": 1856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9888633557013365e-05,
+      "loss": 1.5462,
+      "step": 1857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9888390181459876e-05,
+      "loss": 1.5219,
+      "step": 1858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988814654086015e-05,
+      "loss": 1.5952,
+      "step": 1859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988790263521676e-05,
+      "loss": 1.5205,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988765846453232e-05,
+      "loss": 1.5575,
+      "step": 1861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988741402880943e-05,
+      "loss": 1.5284,
+      "step": 1862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988716932805069e-05,
+      "loss": 1.5562,
+      "step": 1863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988692436225871e-05,
+      "loss": 1.5355,
+      "step": 1864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988667913143609e-05,
+      "loss": 1.6123,
+      "step": 1865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9886433635585444e-05,
+      "loss": 1.6079,
+      "step": 1866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.98861878747094e-05,
+      "loss": 1.5506,
+      "step": 1867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9885941848810546e-05,
+      "loss": 1.5863,
+      "step": 1868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988569555789153e-05,
+      "loss": 1.5487,
+      "step": 1869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988544900195496e-05,
+      "loss": 1.578,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9885202181003474e-05,
+      "loss": 1.5895,
+      "step": 1871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988495509503969e-05,
+      "loss": 1.5093,
+      "step": 1872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9884707744066246e-05,
+      "loss": 1.5679,
+      "step": 1873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9884460128085765e-05,
+      "loss": 1.5054,
+      "step": 1874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9884212247100895e-05,
+      "loss": 1.6426,
+      "step": 1875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9883964101114274e-05,
+      "loss": 1.6153,
+      "step": 1876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988371569012854e-05,
+      "loss": 1.6016,
+      "step": 1877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988346701414634e-05,
+      "loss": 1.6128,
+      "step": 1878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9883218073170324e-05,
+      "loss": 1.6249,
+      "step": 1879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988296886720314e-05,
+      "loss": 1.5633,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9882719396247455e-05,
+      "loss": 1.584,
+      "step": 1881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988246966030591e-05,
+      "loss": 1.5326,
+      "step": 1882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988221965938117e-05,
+      "loss": 1.5697,
+      "step": 1883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988196939347589e-05,
+      "loss": 1.5698,
+      "step": 1884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988171886259275e-05,
+      "loss": 1.5213,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9881468066734405e-05,
+      "loss": 1.6296,
+      "step": 1886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9881217005903534e-05,
+      "loss": 1.6003,
+      "step": 1887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.98809656801028e-05,
+      "loss": 1.6009,
+      "step": 1888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.98807140893349e-05,
+      "loss": 1.5629,
+      "step": 1889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988046223360249e-05,
+      "loss": 1.5951,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988021011290826e-05,
+      "loss": 1.5202,
+      "step": 1891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987995772725491e-05,
+      "loss": 1.4981,
+      "step": 1892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9879705076645114e-05,
+      "loss": 1.523,
+      "step": 1893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987945216108156e-05,
+      "loss": 1.5759,
+      "step": 1894
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987919898056694e-05,
+      "loss": 1.575,
+      "step": 1895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987894553510396e-05,
+      "loss": 1.5239,
+      "step": 1896
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987869182469531e-05,
+      "loss": 1.5469,
+      "step": 1897
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987843784934371e-05,
+      "loss": 1.6668,
+      "step": 1898
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987818360905184e-05,
+      "loss": 1.464,
+      "step": 1899
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987792910382242e-05,
+      "loss": 1.5851,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987767433365815e-05,
+      "loss": 1.5928,
+      "step": 1901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987741929856177e-05,
+      "loss": 1.5679,
+      "step": 1902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987716399853597e-05,
+      "loss": 1.5692,
+      "step": 1903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987690843358347e-05,
+      "loss": 1.5474,
+      "step": 1904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9876652603707004e-05,
+      "loss": 1.5309,
+      "step": 1905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987639650890929e-05,
+      "loss": 1.5635,
+      "step": 1906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9876140149193054e-05,
+      "loss": 1.6348,
+      "step": 1907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987588352456103e-05,
+      "loss": 1.5378,
+      "step": 1908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987562663501595e-05,
+      "loss": 1.5119,
+      "step": 1909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987536948056055e-05,
+      "loss": 1.5915,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987511206119757e-05,
+      "loss": 1.4871,
+      "step": 1911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9874854376929736e-05,
+      "loss": 1.612,
+      "step": 1912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987459642775981e-05,
+      "loss": 1.5581,
+      "step": 1913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9874338213690544e-05,
+      "loss": 1.5773,
+      "step": 1914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9874079734724666e-05,
+      "loss": 1.5728,
+      "step": 1915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987382099086495e-05,
+      "loss": 1.5796,
+      "step": 1916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987356198211413e-05,
+      "loss": 1.5565,
+      "step": 1917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987330270847498e-05,
+      "loss": 1.5474,
+      "step": 1918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987304316995026e-05,
+      "loss": 1.5721,
+      "step": 1919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987278336654273e-05,
+      "loss": 1.5643,
+      "step": 1920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987252329825516e-05,
+      "loss": 1.6115,
+      "step": 1921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9872262965090315e-05,
+      "loss": 1.6183,
+      "step": 1922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987200236705097e-05,
+      "loss": 1.5371,
+      "step": 1923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.98717415041399e-05,
+      "loss": 1.5765,
+      "step": 1924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987148037635988e-05,
+      "loss": 1.5549,
+      "step": 1925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9871218983713694e-05,
+      "loss": 1.4963,
+      "step": 1926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987095732620414e-05,
+      "loss": 1.6348,
+      "step": 1927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987069540383397e-05,
+      "loss": 1.5913,
+      "step": 1928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9870433216606e-05,
+      "loss": 1.5423,
+      "step": 1929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9870170764523014e-05,
+      "loss": 1.5137,
+      "step": 1930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986990804758782e-05,
+      "loss": 1.5256,
+      "step": 1931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9869645065803195e-05,
+      "loss": 1.5343,
+      "step": 1932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986938181917195e-05,
+      "loss": 1.529,
+      "step": 1933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9869118307696884e-05,
+      "loss": 1.5781,
+      "step": 1934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986885453138081e-05,
+      "loss": 1.5503,
+      "step": 1935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986859049022653e-05,
+      "loss": 1.6009,
+      "step": 1936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9868326184236865e-05,
+      "loss": 1.5365,
+      "step": 1937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986806161341462e-05,
+      "loss": 1.499,
+      "step": 1938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986779677776262e-05,
+      "loss": 1.5145,
+      "step": 1939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986753167728368e-05,
+      "loss": 1.5695,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986726631198062e-05,
+      "loss": 1.6001,
+      "step": 1941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986700068185628e-05,
+      "loss": 1.5938,
+      "step": 1942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986673478691347e-05,
+      "loss": 1.5853,
+      "step": 1943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986646862715504e-05,
+      "loss": 1.5075,
+      "step": 1944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986620220258381e-05,
+      "loss": 1.6068,
+      "step": 1945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986593551320263e-05,
+      "loss": 1.529,
+      "step": 1946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986566855901433e-05,
+      "loss": 1.549,
+      "step": 1947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9865401340021754e-05,
+      "loss": 1.5568,
+      "step": 1948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9865133856227756e-05,
+      "loss": 1.5586,
+      "step": 1949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9864866107635165e-05,
+      "loss": 1.619,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9864598094246854e-05,
+      "loss": 1.6364,
+      "step": 1951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986432981606567e-05,
+      "loss": 1.4709,
+      "step": 1952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986406127309447e-05,
+      "loss": 1.5687,
+      "step": 1953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986379246533611e-05,
+      "loss": 1.4914,
+      "step": 1954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9863523392793454e-05,
+      "loss": 1.5519,
+      "step": 1955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986325405546938e-05,
+      "loss": 1.6434,
+      "step": 1956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986298445336674e-05,
+      "loss": 1.5343,
+      "step": 1957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.98627145864884e-05,
+      "loss": 1.568,
+      "step": 1958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986244445483725e-05,
+      "loss": 1.618,
+      "step": 1959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986217405841616e-05,
+      "loss": 1.5692,
+      "step": 1960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986190339722801e-05,
+      "loss": 1.5703,
+      "step": 1961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9861632471275686e-05,
+      "loss": 1.5563,
+      "step": 1962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9861361280562076e-05,
+      "loss": 1.5529,
+      "step": 1963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986108982509005e-05,
+      "loss": 1.5505,
+      "step": 1964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9860818104862514e-05,
+      "loss": 1.6082,
+      "step": 1965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986054611988237e-05,
+      "loss": 1.5741,
+      "step": 1966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986027387015249e-05,
+      "loss": 1.5838,
+      "step": 1967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986000135567579e-05,
+      "loss": 1.6063,
+      "step": 1968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.985972857645517e-05,
+      "loss": 1.6125,
+      "step": 1969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.985945553249353e-05,
+      "loss": 1.662,
+      "step": 1970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9859182223793785e-05,
+      "loss": 1.5918,
+      "step": 1971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9858908650358836e-05,
+      "loss": 1.5905,
+      "step": 1972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9858634812191606e-05,
+      "loss": 1.5702,
+      "step": 1973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.985836070929501e-05,
+      "loss": 1.5547,
+      "step": 1974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.985808634167196e-05,
+      "loss": 1.5677,
+      "step": 1975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9857811709325386e-05,
+      "loss": 1.5612,
+      "step": 1976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9857536812258206e-05,
+      "loss": 1.653,
+      "step": 1977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9857261650473356e-05,
+      "loss": 1.6093,
+      "step": 1978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985698622397375e-05,
+      "loss": 1.5353,
+      "step": 1979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985671053276234e-05,
+      "loss": 1.5199,
+      "step": 1980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9856434576842047e-05,
+      "loss": 1.5084,
+      "step": 1981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985615835621582e-05,
+      "loss": 1.6298,
+      "step": 1982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.98558818708866e-05,
+      "loss": 1.5344,
+      "step": 1983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985560512085732e-05,
+      "loss": 1.5277,
+      "step": 1984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985532810613095e-05,
+      "loss": 1.5651,
+      "step": 1985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985505082671042e-05,
+      "loss": 1.4719,
+      "step": 1986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9854773282598674e-05,
+      "loss": 1.6055,
+      "step": 1987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.98544954737987e-05,
+      "loss": 1.6445,
+      "step": 1988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985421740031343e-05,
+      "loss": 1.5269,
+      "step": 1989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985393906214584e-05,
+      "loss": 1.6496,
+      "step": 1990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985366045929888e-05,
+      "loss": 1.5464,
+      "step": 1991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985338159177553e-05,
+      "loss": 1.6411,
+      "step": 1992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9853102459578757e-05,
+      "loss": 1.5482,
+      "step": 1993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985282306271153e-05,
+      "loss": 1.5049,
+      "step": 1994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985254340117683e-05,
+      "loss": 1.5627,
+      "step": 1995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9852263474977624e-05,
+      "loss": 1.5557,
+      "step": 1996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.98519832841169e-05,
+      "loss": 1.606,
+      "step": 1997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985170282859765e-05,
+      "loss": 1.5925,
+      "step": 1998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985142210842285e-05,
+      "loss": 1.4961,
+      "step": 1999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985114112359548e-05,
+      "loss": 1.5754,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985085987411856e-05,
+      "loss": 1.5619,
+      "step": 2001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985057835999506e-05,
+      "loss": 1.4947,
+      "step": 2002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985029658122799e-05,
+      "loss": 1.5248,
+      "step": 2003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9850014537820355e-05,
+      "loss": 1.5788,
+      "step": 2004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984973222977515e-05,
+      "loss": 1.5114,
+      "step": 2005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9849449657095375e-05,
+      "loss": 1.5827,
+      "step": 2006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984916681978406e-05,
+      "loss": 1.5926,
+      "step": 2007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984888371784419e-05,
+      "loss": 1.5169,
+      "step": 2008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984860035127881e-05,
+      "loss": 1.596,
+      "step": 2009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984831672009091e-05,
+      "loss": 1.5233,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984803282428353e-05,
+      "loss": 1.5459,
+      "step": 2011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984774866385968e-05,
+      "loss": 1.5893,
+      "step": 2012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984746423882239e-05,
+      "loss": 1.5934,
+      "step": 2013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.98471795491747e-05,
+      "loss": 1.6183,
+      "step": 2014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984689459491963e-05,
+      "loss": 1.6519,
+      "step": 2015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9846609376060215e-05,
+      "loss": 1.603,
+      "step": 2016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.98463238925995e-05,
+      "loss": 1.5204,
+      "step": 2017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9846038144540515e-05,
+      "loss": 1.6336,
+      "step": 2018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984575213188631e-05,
+      "loss": 1.5589,
+      "step": 2019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9845465854639927e-05,
+      "loss": 1.5417,
+      "step": 2020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984517931280442e-05,
+      "loss": 1.5692,
+      "step": 2021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984489250638283e-05,
+      "loss": 1.5102,
+      "step": 2022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9844605435378236e-05,
+      "loss": 1.5028,
+      "step": 2023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9844318099793663e-05,
+      "loss": 1.5577,
+      "step": 2024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984403049963219e-05,
+      "loss": 1.5171,
+      "step": 2025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984374263489687e-05,
+      "loss": 1.5714,
+      "step": 2026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9843454505590773e-05,
+      "loss": 1.5368,
+      "step": 2027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9843166111716976e-05,
+      "loss": 1.5285,
+      "step": 2028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984287745327854e-05,
+      "loss": 1.5181,
+      "step": 2029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984258853027854e-05,
+      "loss": 1.5585,
+      "step": 2030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9842299342720053e-05,
+      "loss": 1.501,
+      "step": 2031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9842009890606165e-05,
+      "loss": 1.5616,
+      "step": 2032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984172017393995e-05,
+      "loss": 1.5461,
+      "step": 2033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.98414301927245e-05,
+      "loss": 1.6413,
+      "step": 2034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.98411399469629e-05,
+      "loss": 1.5896,
+      "step": 2035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9840849436658244e-05,
+      "loss": 1.5915,
+      "step": 2036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984055866181361e-05,
+      "loss": 1.515,
+      "step": 2037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984026762243212e-05,
+      "loss": 1.5333,
+      "step": 2038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.983997631851686e-05,
+      "loss": 1.5279,
+      "step": 2039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9839684750070924e-05,
+      "loss": 1.5562,
+      "step": 2040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983939291709743e-05,
+      "loss": 1.5838,
+      "step": 2041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983910081959948e-05,
+      "loss": 1.6348,
+      "step": 2042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983880845758019e-05,
+      "loss": 1.5741,
+      "step": 2043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9838515831042675e-05,
+      "loss": 1.5715,
+      "step": 2044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983822293999004e-05,
+      "loss": 1.5374,
+      "step": 2045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983792978442541e-05,
+      "loss": 1.5116,
+      "step": 2046
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983763636435191e-05,
+      "loss": 1.4818,
+      "step": 2047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983734267977266e-05,
+      "loss": 1.5729,
+      "step": 2048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9837048730690785e-05,
+      "loss": 1.6604,
+      "step": 2049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983675451710943e-05,
+      "loss": 1.5007,
+      "step": 2050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9836460039031705e-05,
+      "loss": 1.5523,
+      "step": 2051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983616529646077e-05,
+      "loss": 1.5843,
+      "step": 2052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9835870289399756e-05,
+      "loss": 1.5606,
+      "step": 2053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9835575017851795e-05,
+      "loss": 1.5641,
+      "step": 2054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983527948182004e-05,
+      "loss": 1.5327,
+      "step": 2055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983498368130764e-05,
+      "loss": 1.5814,
+      "step": 2056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9834687616317724e-05,
+      "loss": 1.5604,
+      "step": 2057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9834391286853475e-05,
+      "loss": 1.4442,
+      "step": 2058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983409469291804e-05,
+      "loss": 1.5977,
+      "step": 2059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9833797834514575e-05,
+      "loss": 1.6483,
+      "step": 2060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983350071164623e-05,
+      "loss": 1.6006,
+      "step": 2061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983320332431618e-05,
+      "loss": 1.5645,
+      "step": 2062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9832905672527594e-05,
+      "loss": 1.6629,
+      "step": 2063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983260775628365e-05,
+      "loss": 1.5357,
+      "step": 2064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9832309575587496e-05,
+      "loss": 1.5527,
+      "step": 2065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9832011130442325e-05,
+      "loss": 1.5168,
+      "step": 2066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983171242085131e-05,
+      "loss": 1.5781,
+      "step": 2067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983141344681763e-05,
+      "loss": 1.5493,
+      "step": 2068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983111420834448e-05,
+      "loss": 1.5955,
+      "step": 2069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9830814705435035e-05,
+      "loss": 1.5487,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9830514938092494e-05,
+      "loss": 1.5851,
+      "step": 2071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983021490632004e-05,
+      "loss": 1.5355,
+      "step": 2072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982991461012087e-05,
+      "loss": 1.5625,
+      "step": 2073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9829614049498186e-05,
+      "loss": 1.5986,
+      "step": 2074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982931322445519e-05,
+      "loss": 1.5332,
+      "step": 2075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9829012134995076e-05,
+      "loss": 1.5747,
+      "step": 2076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982871078112106e-05,
+      "loss": 1.5615,
+      "step": 2077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982840916283634e-05,
+      "loss": 1.6279,
+      "step": 2078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982810728014415e-05,
+      "loss": 1.5825,
+      "step": 2079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9827805133047684e-05,
+      "loss": 1.5529,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982750272155017e-05,
+      "loss": 1.5619,
+      "step": 2081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982720004565482e-05,
+      "loss": 1.5475,
+      "step": 2082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982689710536487e-05,
+      "loss": 1.6112,
+      "step": 2083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982659390068353e-05,
+      "loss": 1.5047,
+      "step": 2084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9826290431614034e-05,
+      "loss": 1.5379,
+      "step": 2085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982598669815963e-05,
+      "loss": 1.6242,
+      "step": 2086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982568270032353e-05,
+      "loss": 1.5514,
+      "step": 2087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982537843810899e-05,
+      "loss": 1.5641,
+      "step": 2088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982507391151924e-05,
+      "loss": 1.5833,
+      "step": 2089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9824769120557515e-05,
+      "loss": 1.5998,
+      "step": 2090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982446406522708e-05,
+      "loss": 1.544,
+      "step": 2091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982415874553117e-05,
+      "loss": 1.5121,
+      "step": 2092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982385316147304e-05,
+      "loss": 1.5531,
+      "step": 2093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982354731305595e-05,
+      "loss": 1.5793,
+      "step": 2094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982324120028314e-05,
+      "loss": 1.5814,
+      "step": 2095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9822934823157894e-05,
+      "loss": 1.5527,
+      "step": 2096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982262818168346e-05,
+      "loss": 1.5731,
+      "step": 2097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.98223212758631e-05,
+      "loss": 1.6286,
+      "step": 2098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.98220141057001e-05,
+      "loss": 1.5567,
+      "step": 2099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982170667119771e-05,
+      "loss": 1.6187,
+      "step": 2100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9821398972359215e-05,
+      "loss": 1.578,
+      "step": 2101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9821091009187894e-05,
+      "loss": 1.619,
+      "step": 2102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.982078278168701e-05,
+      "loss": 1.5404,
+      "step": 2103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9820474289859876e-05,
+      "loss": 1.6558,
+      "step": 2104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.982016553370975e-05,
+      "loss": 1.5744,
+      "step": 2105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9819856513239936e-05,
+      "loss": 1.467,
+      "step": 2106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981954722845371e-05,
+      "loss": 1.5807,
+      "step": 2107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981923767935438e-05,
+      "loss": 1.5156,
+      "step": 2108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981892786594524e-05,
+      "loss": 1.6139,
+      "step": 2109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981861778822958e-05,
+      "loss": 1.6066,
+      "step": 2110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981830744621071e-05,
+      "loss": 1.5765,
+      "step": 2111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9817996839891937e-05,
+      "loss": 1.5422,
+      "step": 2112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981768596927656e-05,
+      "loss": 1.5788,
+      "step": 2113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981737483436789e-05,
+      "loss": 1.5347,
+      "step": 2114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9817063435169255e-05,
+      "loss": 1.5776,
+      "step": 2115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981675177168396e-05,
+      "loss": 1.5941,
+      "step": 2116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981643984391532e-05,
+      "loss": 1.4957,
+      "step": 2117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981612765186667e-05,
+      "loss": 1.5352,
+      "step": 2118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9815815195541316e-05,
+      "loss": 1.5262,
+      "step": 2119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9815502474942605e-05,
+      "loss": 1.5361,
+      "step": 2120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9815189490073856e-05,
+      "loss": 1.6029,
+      "step": 2121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9814876240938404e-05,
+      "loss": 1.5916,
+      "step": 2122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981456272753958e-05,
+      "loss": 1.5916,
+      "step": 2123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981424894988074e-05,
+      "loss": 1.5503,
+      "step": 2124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9813934907965205e-05,
+      "loss": 1.5912,
+      "step": 2125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981362060179633e-05,
+      "loss": 1.5366,
+      "step": 2126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981330603137746e-05,
+      "loss": 1.5257,
+      "step": 2127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9812991196711944e-05,
+      "loss": 1.5718,
+      "step": 2128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9812676097803146e-05,
+      "loss": 1.5765,
+      "step": 2129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.98123607346544e-05,
+      "loss": 1.5392,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981204510726908e-05,
+      "loss": 1.537,
+      "step": 2131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9811729215650545e-05,
+      "loss": 1.5081,
+      "step": 2132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981141305980216e-05,
+      "loss": 1.5267,
+      "step": 2133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981109663972728e-05,
+      "loss": 1.5234,
+      "step": 2134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981077995542929e-05,
+      "loss": 1.6201,
+      "step": 2135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981046300691156e-05,
+      "loss": 1.5041,
+      "step": 2136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981014579417746e-05,
+      "loss": 1.5255,
+      "step": 2137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9809828317230355e-05,
+      "loss": 1.5311,
+      "step": 2138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980951057607366e-05,
+      "loss": 1.5295,
+      "step": 2139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980919257071073e-05,
+      "loss": 1.5798,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9808874301144965e-05,
+      "loss": 1.6021,
+      "step": 2141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980855576737975e-05,
+      "loss": 1.6465,
+      "step": 2142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9808236969418475e-05,
+      "loss": 1.5339,
+      "step": 2143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980791790726454e-05,
+      "loss": 1.5604,
+      "step": 2144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980759858092135e-05,
+      "loss": 1.5262,
+      "step": 2145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9807278990392284e-05,
+      "loss": 1.5754,
+      "step": 2146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9806959135680753e-05,
+      "loss": 1.4919,
+      "step": 2147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980663901679018e-05,
+      "loss": 1.5449,
+      "step": 2148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9806318633723956e-05,
+      "loss": 1.5202,
+      "step": 2149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.98059979864855e-05,
+      "loss": 1.5625,
+      "step": 2150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980567707507822e-05,
+      "loss": 1.6237,
+      "step": 2151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980535589950555e-05,
+      "loss": 1.5526,
+      "step": 2152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9805034459770895e-05,
+      "loss": 1.5381,
+      "step": 2153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9804712755877674e-05,
+      "loss": 1.6294,
+      "step": 2154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980439078782933e-05,
+      "loss": 1.4889,
+      "step": 2155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980406855562928e-05,
+      "loss": 1.501,
+      "step": 2156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980374605928097e-05,
+      "loss": 1.5512,
+      "step": 2157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980342329878781e-05,
+      "loss": 1.5811,
+      "step": 2158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980310027415325e-05,
+      "loss": 1.5606,
+      "step": 2159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980277698538074e-05,
+      "loss": 1.5329,
+      "step": 2160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9802453432473706e-05,
+      "loss": 1.6675,
+      "step": 2161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.98021296154356e-05,
+      "loss": 1.5697,
+      "step": 2162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9801805534269875e-05,
+      "loss": 1.6119,
+      "step": 2163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9801481188979985e-05,
+      "loss": 1.541,
+      "step": 2164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9801156579569365e-05,
+      "loss": 1.5301,
+      "step": 2165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.980083170604149e-05,
+      "loss": 1.6203,
+      "step": 2166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9800506568399814e-05,
+      "loss": 1.5653,
+      "step": 2167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9800181166647796e-05,
+      "loss": 1.4551,
+      "step": 2168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979985550078891e-05,
+      "loss": 1.6159,
+      "step": 2169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979952957082662e-05,
+      "loss": 1.5435,
+      "step": 2170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979920337676439e-05,
+      "loss": 1.5904,
+      "step": 2171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979887691860571e-05,
+      "loss": 1.5149,
+      "step": 2172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9798550196354046e-05,
+      "loss": 1.5433,
+      "step": 2173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979822321001287e-05,
+      "loss": 1.5874,
+      "step": 2174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9797895959585675e-05,
+      "loss": 1.5248,
+      "step": 2175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979756844507594e-05,
+      "loss": 1.4982,
+      "step": 2176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9797240666487164e-05,
+      "loss": 1.5879,
+      "step": 2177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979691262382282e-05,
+      "loss": 1.6005,
+      "step": 2178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979658431708642e-05,
+      "loss": 1.5834,
+      "step": 2179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979625574628145e-05,
+      "loss": 1.5948,
+      "step": 2180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979592691141141e-05,
+      "loss": 1.6069,
+      "step": 2181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9795597812479797e-05,
+      "loss": 1.6394,
+      "step": 2182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979526844949013e-05,
+      "loss": 1.568,
+      "step": 2183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9794938822445895e-05,
+      "loss": 1.5794,
+      "step": 2184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9794608931350626e-05,
+      "loss": 1.603,
+      "step": 2185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979427877620782e-05,
+      "loss": 1.5549,
+      "step": 2186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9793948357021e-05,
+      "loss": 1.5625,
+      "step": 2187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979361767379368e-05,
+      "loss": 1.5378,
+      "step": 2188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979328672652939e-05,
+      "loss": 1.5873,
+      "step": 2189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979295551523164e-05,
+      "loss": 1.5616,
+      "step": 2190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979262403990398e-05,
+      "loss": 1.5227,
+      "step": 2191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9792292300549915e-05,
+      "loss": 1.5845,
+      "step": 2192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979196029717299e-05,
+      "loss": 1.5493,
+      "step": 2193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979162802977674e-05,
+      "loss": 1.5208,
+      "step": 2194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.97912954983647e-05,
+      "loss": 1.5741,
+      "step": 2195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979096270294042e-05,
+      "loss": 1.6113,
+      "step": 2196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9790629643507434e-05,
+      "loss": 1.5156,
+      "step": 2197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9790296320069294e-05,
+      "loss": 1.5501,
+      "step": 2198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9789962732629545e-05,
+      "loss": 1.5468,
+      "step": 2199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978962888119175e-05,
+      "loss": 1.5929,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9789294765759456e-05,
+      "loss": 1.6066,
+      "step": 2201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9788960386336224e-05,
+      "loss": 1.6345,
+      "step": 2202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978862574292561e-05,
+      "loss": 1.5936,
+      "step": 2203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978829083553118e-05,
+      "loss": 1.5669,
+      "step": 2204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.97879556641565e-05,
+      "loss": 1.5878,
+      "step": 2205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978762022880515e-05,
+      "loss": 1.6045,
+      "step": 2206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9787284529480695e-05,
+      "loss": 1.5627,
+      "step": 2207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.97869485661867e-05,
+      "loss": 1.5818,
+      "step": 2208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978661233892675e-05,
+      "loss": 1.5547,
+      "step": 2209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9786275847704434e-05,
+      "loss": 1.6066,
+      "step": 2210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978593909252333e-05,
+      "loss": 1.6055,
+      "step": 2211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9785602073387014e-05,
+      "loss": 1.537,
+      "step": 2212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978526479029909e-05,
+      "loss": 1.5192,
+      "step": 2213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978492724326314e-05,
+      "loss": 1.5433,
+      "step": 2214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9784589432282764e-05,
+      "loss": 1.5702,
+      "step": 2215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978425135736156e-05,
+      "loss": 1.5726,
+      "step": 2216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978391301850312e-05,
+      "loss": 1.6179,
+      "step": 2217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9783574415711056e-05,
+      "loss": 1.5798,
+      "step": 2218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978323554898897e-05,
+      "loss": 1.541,
+      "step": 2219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9782896418340475e-05,
+      "loss": 1.5649,
+      "step": 2220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978255702376917e-05,
+      "loss": 1.577,
+      "step": 2221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978221736527869e-05,
+      "loss": 1.5516,
+      "step": 2222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9781877442872636e-05,
+      "loss": 1.6066,
+      "step": 2223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978153725655463e-05,
+      "loss": 1.5246,
+      "step": 2224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9781196806328304e-05,
+      "loss": 1.5728,
+      "step": 2225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978085609219727e-05,
+      "loss": 1.5311,
+      "step": 2226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978051511416517e-05,
+      "loss": 1.5716,
+      "step": 2227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978017387223563e-05,
+      "loss": 1.6053,
+      "step": 2228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977983236641227e-05,
+      "loss": 1.4961,
+      "step": 2229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977949059669875e-05,
+      "loss": 1.5274,
+      "step": 2230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.97791485630987e-05,
+      "loss": 1.6004,
+      "step": 2231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977880626561575e-05,
+      "loss": 1.6664,
+      "step": 2232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977846370425357e-05,
+      "loss": 1.5544,
+      "step": 2233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977812087901579e-05,
+      "loss": 1.5444,
+      "step": 2234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9777777789906064e-05,
+      "loss": 1.5685,
+      "step": 2235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977743443692805e-05,
+      "loss": 1.57,
+      "step": 2236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.97770908200854e-05,
+      "loss": 1.514,
+      "step": 2237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977674693938178e-05,
+      "loss": 1.5519,
+      "step": 2238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977640279482084e-05,
+      "loss": 1.5785,
+      "step": 2239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977605838640626e-05,
+      "loss": 1.6035,
+      "step": 2240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977571371414169e-05,
+      "loss": 1.5322,
+      "step": 2241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9775368778030815e-05,
+      "loss": 1.5791,
+      "step": 2242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977502357807731e-05,
+      "loss": 1.5705,
+      "step": 2243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9774678114284846e-05,
+      "loss": 1.5949,
+      "step": 2244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.97743323866571e-05,
+      "loss": 1.5604,
+      "step": 2245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9773986395197746e-05,
+      "loss": 1.5824,
+      "step": 2246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977364013991048e-05,
+      "loss": 1.5218,
+      "step": 2247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9773293620798986e-05,
+      "loss": 1.4997,
+      "step": 2248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9772946837866965e-05,
+      "loss": 1.5633,
+      "step": 2249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977259979111809e-05,
+      "loss": 1.5885,
+      "step": 2250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977225248055607e-05,
+      "loss": 1.6475,
+      "step": 2251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977190490618461e-05,
+      "loss": 1.6014,
+      "step": 2252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977155706800739e-05,
+      "loss": 1.5143,
+      "step": 2253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9771208966028124e-05,
+      "loss": 1.5299,
+      "step": 2254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9770860600250534e-05,
+      "loss": 1.5545,
+      "step": 2255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9770511970678315e-05,
+      "loss": 1.5182,
+      "step": 2256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9770163077315174e-05,
+      "loss": 1.5697,
+      "step": 2257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9769813920164845e-05,
+      "loss": 1.5593,
+      "step": 2258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976946449923102e-05,
+      "loss": 1.5347,
+      "step": 2259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976911481451745e-05,
+      "loss": 1.5319,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976876486602784e-05,
+      "loss": 1.5379,
+      "step": 2261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976841465376592e-05,
+      "loss": 1.6384,
+      "step": 2262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976806417773543e-05,
+      "loss": 1.5832,
+      "step": 2263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9767713437940086e-05,
+      "loss": 1.5259,
+      "step": 2264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976736243438363e-05,
+      "loss": 1.5633,
+      "step": 2265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.97670111670698e-05,
+      "loss": 1.5624,
+      "step": 2266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9766659636002346e-05,
+      "loss": 1.5618,
+      "step": 2267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976630784118499e-05,
+      "loss": 1.5594,
+      "step": 2268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.97659557826215e-05,
+      "loss": 1.5563,
+      "step": 2269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9765603460315615e-05,
+      "loss": 1.5125,
+      "step": 2270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9765250874271093e-05,
+      "loss": 1.5754,
+      "step": 2271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976489802449168e-05,
+      "loss": 1.6237,
+      "step": 2272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976454491098114e-05,
+      "loss": 1.5908,
+      "step": 2273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9764191533743225e-05,
+      "loss": 1.5288,
+      "step": 2274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976383789278171e-05,
+      "loss": 1.6018,
+      "step": 2275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9763483988100354e-05,
+      "loss": 1.5085,
+      "step": 2276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976312981970294e-05,
+      "loss": 1.5573,
+      "step": 2277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9762775387593206e-05,
+      "loss": 1.5374,
+      "step": 2278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9762420691774964e-05,
+      "loss": 1.5695,
+      "step": 2279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9762065732251973e-05,
+      "loss": 1.5521,
+      "step": 2280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9761710509028016e-05,
+      "loss": 1.5399,
+      "step": 2281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976135502210687e-05,
+      "loss": 1.5697,
+      "step": 2282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9760999271492326e-05,
+      "loss": 1.559,
+      "step": 2283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9760643257188175e-05,
+      "loss": 1.5965,
+      "step": 2284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976028697919821e-05,
+      "loss": 1.5412,
+      "step": 2285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9759930437526215e-05,
+      "loss": 1.5816,
+      "step": 2286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9759573632175996e-05,
+      "loss": 1.5549,
+      "step": 2287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9759216563151346e-05,
+      "loss": 1.527,
+      "step": 2288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9758859230456066e-05,
+      "loss": 1.5579,
+      "step": 2289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.975850163409398e-05,
+      "loss": 1.556,
+      "step": 2290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.975814377406888e-05,
+      "loss": 1.5864,
+      "step": 2291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975778565038457e-05,
+      "loss": 1.601,
+      "step": 2292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975742726304489e-05,
+      "loss": 1.6148,
+      "step": 2293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9757068612053625e-05,
+      "loss": 1.6011,
+      "step": 2294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9756709697414615e-05,
+      "loss": 1.596,
+      "step": 2295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975635051913168e-05,
+      "loss": 1.5,
+      "step": 2296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9755991077208644e-05,
+      "loss": 1.59,
+      "step": 2297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9755631371649326e-05,
+      "loss": 1.6227,
+      "step": 2298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975527140245756e-05,
+      "loss": 1.5291,
+      "step": 2299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975491116963719e-05,
+      "loss": 1.543,
+      "step": 2300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975455067319205e-05,
+      "loss": 1.5407,
+      "step": 2301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975418991312596e-05,
+      "loss": 1.5277,
+      "step": 2302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9753828889442784e-05,
+      "loss": 1.5042,
+      "step": 2303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975346760214636e-05,
+      "loss": 1.5444,
+      "step": 2304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9753106051240526e-05,
+      "loss": 1.5057,
+      "step": 2305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975274423672916e-05,
+      "loss": 1.5557,
+      "step": 2306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975238215861607e-05,
+      "loss": 1.5247,
+      "step": 2307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975201981690516e-05,
+      "loss": 1.5848,
+      "step": 2308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975165721160026e-05,
+      "loss": 1.5947,
+      "step": 2309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9751294342705224e-05,
+      "loss": 1.5469,
+      "step": 2310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975093121022394e-05,
+      "loss": 1.5252,
+      "step": 2311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9750567814160263e-05,
+      "loss": 1.5535,
+      "step": 2312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9750204154518076e-05,
+      "loss": 1.5677,
+      "step": 2313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974984023130123e-05,
+      "loss": 1.6056,
+      "step": 2314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974947604451362e-05,
+      "loss": 1.5993,
+      "step": 2315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9749111594159106e-05,
+      "loss": 1.5487,
+      "step": 2316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974874688024158e-05,
+      "loss": 1.5674,
+      "step": 2317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974838190276493e-05,
+      "loss": 1.4691,
+      "step": 2318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974801666173304e-05,
+      "loss": 1.5744,
+      "step": 2319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974765115714979e-05,
+      "loss": 1.5562,
+      "step": 2320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974728538901908e-05,
+      "loss": 1.5064,
+      "step": 2321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974691935734481e-05,
+      "loss": 1.506,
+      "step": 2322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974655306213088e-05,
+      "loss": 1.5138,
+      "step": 2323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9746186503381174e-05,
+      "loss": 1.5907,
+      "step": 2324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9745819681099606e-05,
+      "loss": 1.5527,
+      "step": 2325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974545259529008e-05,
+      "loss": 1.5682,
+      "step": 2326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974508524595651e-05,
+      "loss": 1.533,
+      "step": 2327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974471763310281e-05,
+      "loss": 1.5138,
+      "step": 2328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9744349756732884e-05,
+      "loss": 1.5811,
+      "step": 2329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974398161685066e-05,
+      "loss": 1.5772,
+      "step": 2330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974361321346005e-05,
+      "loss": 1.6348,
+      "step": 2331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974324454656498e-05,
+      "loss": 1.5257,
+      "step": 2332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974287561616938e-05,
+      "loss": 1.536,
+      "step": 2333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974250642227718e-05,
+      "loss": 1.6019,
+      "step": 2334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974213696489231e-05,
+      "loss": 1.5726,
+      "step": 2335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974176724401869e-05,
+      "loss": 1.5412,
+      "step": 2336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9741397259660284e-05,
+      "loss": 1.6188,
+      "step": 2337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9741027011821006e-05,
+      "loss": 1.5739,
+      "step": 2338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9740656500504826e-05,
+      "loss": 1.57,
+      "step": 2339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9740285725715665e-05,
+      "loss": 1.5506,
+      "step": 2340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9739914687457476e-05,
+      "loss": 1.5475,
+      "step": 2341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973954338573422e-05,
+      "loss": 1.5796,
+      "step": 2342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9739171820549846e-05,
+      "loss": 1.6305,
+      "step": 2343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9738799991908313e-05,
+      "loss": 1.5718,
+      "step": 2344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973842789981358e-05,
+      "loss": 1.4912,
+      "step": 2345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9738055544269614e-05,
+      "loss": 1.4889,
+      "step": 2346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9737682925280374e-05,
+      "loss": 1.5925,
+      "step": 2347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9737310042849814e-05,
+      "loss": 1.5695,
+      "step": 2348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973693689698194e-05,
+      "loss": 1.5226,
+      "step": 2349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9736563487680696e-05,
+      "loss": 1.5153,
+      "step": 2350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973618981495007e-05,
+      "loss": 1.5796,
+      "step": 2351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973581587879405e-05,
+      "loss": 1.6146,
+      "step": 2352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9735441679216595e-05,
+      "loss": 1.5235,
+      "step": 2353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973506721622171e-05,
+      "loss": 1.55,
+      "step": 2354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973469248981337e-05,
+      "loss": 1.5789,
+      "step": 2355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9734317499995575e-05,
+      "loss": 1.5204,
+      "step": 2356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973394224677231e-05,
+      "loss": 1.5137,
+      "step": 2357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973356673014759e-05,
+      "loss": 1.6133,
+      "step": 2358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9733190950125394e-05,
+      "loss": 1.5347,
+      "step": 2359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973281490670973e-05,
+      "loss": 1.4959,
+      "step": 2360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9732438599904596e-05,
+      "loss": 1.5702,
+      "step": 2361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973206202971402e-05,
+      "loss": 1.6179,
+      "step": 2362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973168519614199e-05,
+      "loss": 1.5719,
+      "step": 2363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973130809919252e-05,
+      "loss": 1.6253,
+      "step": 2364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973093073886965e-05,
+      "loss": 1.5916,
+      "step": 2365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973055311517737e-05,
+      "loss": 1.5303,
+      "step": 2366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973017522811973e-05,
+      "loss": 1.5843,
+      "step": 2367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9729797077700715e-05,
+      "loss": 1.5515,
+      "step": 2368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972941866392439e-05,
+      "loss": 1.5278,
+      "step": 2369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9729039986794764e-05,
+      "loss": 1.5744,
+      "step": 2370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972866104631588e-05,
+      "loss": 1.5728,
+      "step": 2371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9728281842491764e-05,
+      "loss": 1.5114,
+      "step": 2372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972790237532646e-05,
+      "loss": 1.5326,
+      "step": 2373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972752264482401e-05,
+      "loss": 1.5462,
+      "step": 2374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9727142650988454e-05,
+      "loss": 1.4865,
+      "step": 2375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9726762393823845e-05,
+      "loss": 1.5383,
+      "step": 2376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972638187333423e-05,
+      "loss": 1.5754,
+      "step": 2377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972600108952366e-05,
+      "loss": 1.5539,
+      "step": 2378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972562004239618e-05,
+      "loss": 1.6265,
+      "step": 2379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972523873195586e-05,
+      "loss": 1.5829,
+      "step": 2380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972485715820676e-05,
+      "loss": 1.5877,
+      "step": 2381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9724475321152955e-05,
+      "loss": 1.5125,
+      "step": 2382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9724093220798485e-05,
+      "loss": 1.5812,
+      "step": 2383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972371085714743e-05,
+      "loss": 1.5539,
+      "step": 2384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972332823020387e-05,
+      "loss": 1.4987,
+      "step": 2385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972294533997188e-05,
+      "loss": 1.4867,
+      "step": 2386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9722562186455526e-05,
+      "loss": 1.4632,
+      "step": 2387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972217876965889e-05,
+      "loss": 1.5985,
+      "step": 2388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972179508958606e-05,
+      "loss": 1.5479,
+      "step": 2389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972141114624113e-05,
+      "loss": 1.5622,
+      "step": 2390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972102693962818e-05,
+      "loss": 1.5619,
+      "step": 2391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9720642469751294e-05,
+      "loss": 1.5367,
+      "step": 2392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9720257736614574e-05,
+      "loss": 1.5426,
+      "step": 2393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9719872740222114e-05,
+      "loss": 1.5705,
+      "step": 2394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971948748057803e-05,
+      "loss": 1.5929,
+      "step": 2395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.97191019576864e-05,
+      "loss": 1.5687,
+      "step": 2396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971871617155135e-05,
+      "loss": 1.6174,
+      "step": 2397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971833012217698e-05,
+      "loss": 1.5633,
+      "step": 2398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971794380956739e-05,
+      "loss": 1.6029,
+      "step": 2399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9717557233726716e-05,
+      "loss": 1.6021,
+      "step": 2400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971717039465906e-05,
+      "loss": 1.5304,
+      "step": 2401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9716783292368545e-05,
+      "loss": 1.516,
+      "step": 2402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9716395926859294e-05,
+      "loss": 1.5916,
+      "step": 2403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971600829813543e-05,
+      "loss": 1.5187,
+      "step": 2404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971562040620109e-05,
+      "loss": 1.5636,
+      "step": 2405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.97152322510604e-05,
+      "loss": 1.4987,
+      "step": 2406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971484383271748e-05,
+      "loss": 1.6297,
+      "step": 2407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971445515117648e-05,
+      "loss": 1.5553,
+      "step": 2408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971406620644154e-05,
+      "loss": 1.495,
+      "step": 2409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.97136769985168e-05,
+      "loss": 1.527,
+      "step": 2410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971328752740641e-05,
+      "loss": 1.5539,
+      "step": 2411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.97128977931145e-05,
+      "loss": 1.5233,
+      "step": 2412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9712507795645235e-05,
+      "loss": 1.519,
+      "step": 2413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971211753500277e-05,
+      "loss": 1.648,
+      "step": 2414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971172701119125e-05,
+      "loss": 1.4483,
+      "step": 2415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9711336224214844e-05,
+      "loss": 1.5173,
+      "step": 2416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.971094517407771e-05,
+      "loss": 1.4766,
+      "step": 2417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.971055386078401e-05,
+      "loss": 1.5622,
+      "step": 2418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9710162284337915e-05,
+      "loss": 1.6351,
+      "step": 2419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9709770444743595e-05,
+      "loss": 1.5147,
+      "step": 2420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970937834200522e-05,
+      "loss": 1.4544,
+      "step": 2421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970898597612696e-05,
+      "loss": 1.5996,
+      "step": 2422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970859334711301e-05,
+      "loss": 1.5778,
+      "step": 2423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9708200454967545e-05,
+      "loss": 1.5379,
+      "step": 2424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970780729969474e-05,
+      "loss": 1.5129,
+      "step": 2425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970741388129879e-05,
+      "loss": 1.5441,
+      "step": 2426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970702019978388e-05,
+      "loss": 1.4994,
+      "step": 2427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.97066262551542e-05,
+      "loss": 1.537,
+      "step": 2428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970623204741396e-05,
+      "loss": 1.5853,
+      "step": 2429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970583757656734e-05,
+      "loss": 1.5781,
+      "step": 2430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970544284261855e-05,
+      "loss": 1.6284,
+      "step": 2431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.97050478455718e-05,
+      "loss": 1.57,
+      "step": 2432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970465258543129e-05,
+      "loss": 1.5607,
+      "step": 2433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970425706220122e-05,
+      "loss": 1.5319,
+      "step": 2434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970386127588582e-05,
+      "loss": 1.5521,
+      "step": 2435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9703465226489285e-05,
+      "loss": 1.5651,
+      "step": 2436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9703068914015846e-05,
+      "loss": 1.5501,
+      "step": 2437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970267233846972e-05,
+      "loss": 1.5918,
+      "step": 2438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970227549985514e-05,
+      "loss": 1.5633,
+      "step": 2439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9701878398176317e-05,
+      "loss": 1.6304,
+      "step": 2440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970148103343748e-05,
+      "loss": 1.5487,
+      "step": 2441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970108340564288e-05,
+      "loss": 1.5392,
+      "step": 2442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9700685514796724e-05,
+      "loss": 1.5716,
+      "step": 2443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970028736090328e-05,
+      "loss": 1.53,
+      "step": 2444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9699888943966756e-05,
+      "loss": 1.513,
+      "step": 2445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969949026399141e-05,
+      "loss": 1.5443,
+      "step": 2446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.96990913209815e-05,
+      "loss": 1.543,
+      "step": 2447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969869211494125e-05,
+      "loss": 1.5335,
+      "step": 2448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969829264587493e-05,
+      "loss": 1.5239,
+      "step": 2449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969789291378679e-05,
+      "loss": 1.582,
+      "step": 2450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969749291868108e-05,
+      "loss": 1.5446,
+      "step": 2451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969709266056207e-05,
+      "loss": 1.6165,
+      "step": 2452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9696692139434016e-05,
+      "loss": 1.6323,
+      "step": 2453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9696291355301186e-05,
+      "loss": 1.5448,
+      "step": 2454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9695890308167844e-05,
+      "loss": 1.5296,
+      "step": 2455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969548899803826e-05,
+      "loss": 1.5383,
+      "step": 2456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9695087424916716e-05,
+      "loss": 1.5138,
+      "step": 2457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9694685588807475e-05,
+      "loss": 1.512,
+      "step": 2458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969428348971484e-05,
+      "loss": 1.5474,
+      "step": 2459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969388112764307e-05,
+      "loss": 1.5933,
+      "step": 2460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969347850259646e-05,
+      "loss": 1.5619,
+      "step": 2461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969307561457929e-05,
+      "loss": 1.5431,
+      "step": 2462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969267246359586e-05,
+      "loss": 1.5352,
+      "step": 2463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969226904965046e-05,
+      "loss": 1.5853,
+      "step": 2464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969186537274738e-05,
+      "loss": 1.4808,
+      "step": 2465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9691461432890936e-05,
+      "loss": 1.5063,
+      "step": 2466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969105723008541e-05,
+      "loss": 1.6042,
+      "step": 2467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969065276433512e-05,
+      "loss": 1.5179,
+      "step": 2468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969024803564436e-05,
+      "loss": 1.6146,
+      "step": 2469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.968984304401746e-05,
+      "loss": 1.5643,
+      "step": 2470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9689437789458715e-05,
+      "loss": 1.5147,
+      "step": 2471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9689032271972446e-05,
+      "loss": 1.5407,
+      "step": 2472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9688626491562976e-05,
+      "loss": 1.655,
+      "step": 2473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9688220448234614e-05,
+      "loss": 1.5907,
+      "step": 2474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.96878141419917e-05,
+      "loss": 1.5579,
+      "step": 2475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.968740757283855e-05,
+      "loss": 1.5588,
+      "step": 2476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.96870007407795e-05,
+      "loss": 1.5329,
+      "step": 2477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.968659364581888e-05,
+      "loss": 1.5864,
+      "step": 2478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.968618628796102e-05,
+      "loss": 1.6384,
+      "step": 2479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968577866721027e-05,
+      "loss": 1.5041,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9685370783570964e-05,
+      "loss": 1.5768,
+      "step": 2481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968496263704744e-05,
+      "loss": 1.5925,
+      "step": 2482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968455422764406e-05,
+      "loss": 1.5578,
+      "step": 2483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968414555536516e-05,
+      "loss": 1.5674,
+      "step": 2484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968373662021509e-05,
+      "loss": 1.5565,
+      "step": 2485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9683327422198226e-05,
+      "loss": 1.5433,
+      "step": 2486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968291796131889e-05,
+      "loss": 1.5814,
+      "step": 2487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968250823758148e-05,
+      "loss": 1.5145,
+      "step": 2488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968209825099033e-05,
+      "loss": 1.6273,
+      "step": 2489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968168800154982e-05,
+      "loss": 1.561,
+      "step": 2490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968127748926433e-05,
+      "loss": 1.652,
+      "step": 2491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968086671413821e-05,
+      "loss": 1.5985,
+      "step": 2492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968045567617584e-05,
+      "loss": 1.5698,
+      "step": 2493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968004437538161e-05,
+      "loss": 1.5443,
+      "step": 2494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967963281175988e-05,
+      "loss": 1.5584,
+      "step": 2495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967922098531506e-05,
+      "loss": 1.4725,
+      "step": 2496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.96788088960515e-05,
+      "loss": 1.5832,
+      "step": 2497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967839654397362e-05,
+      "loss": 1.5933,
+      "step": 2498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9677983929085796e-05,
+      "loss": 1.6271,
+      "step": 2499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9677571051392424e-05,
+      "loss": 1.4832,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967715791089791e-05,
+      "loss": 1.5286,
+      "step": 2501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9676744507606644e-05,
+      "loss": 1.5,
+      "step": 2502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9676330841523024e-05,
+      "loss": 1.5472,
+      "step": 2503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9675916912651465e-05,
+      "loss": 1.5264,
+      "step": 2504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967550272099638e-05,
+      "loss": 1.6138,
+      "step": 2505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967508826656216e-05,
+      "loss": 1.5308,
+      "step": 2506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967467354935323e-05,
+      "loss": 1.5925,
+      "step": 2507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9674258569374014e-05,
+      "loss": 1.5197,
+      "step": 2508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9673843326628925e-05,
+      "loss": 1.5871,
+      "step": 2509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9673427821122384e-05,
+      "loss": 1.5383,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9673012052858816e-05,
+      "loss": 1.5972,
+      "step": 2511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967259602184264e-05,
+      "loss": 1.5739,
+      "step": 2512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967217972807831e-05,
+      "loss": 1.576,
+      "step": 2513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967176317157024e-05,
+      "loss": 1.5978,
+      "step": 2514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967134635232286e-05,
+      "loss": 1.5498,
+      "step": 2515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9670929270340626e-05,
+      "loss": 1.6174,
+      "step": 2516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9670511925627984e-05,
+      "loss": 1.5547,
+      "step": 2517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967009431818935e-05,
+      "loss": 1.5645,
+      "step": 2518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.96696764480292e-05,
+      "loss": 1.5641,
+      "step": 2519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966925831515198e-05,
+      "loss": 1.6047,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966883991956212e-05,
+      "loss": 1.5778,
+      "step": 2521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.96684212612641e-05,
+      "loss": 1.4686,
+      "step": 2522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966800234026237e-05,
+      "loss": 1.6092,
+      "step": 2523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9667583156561395e-05,
+      "loss": 1.5682,
+      "step": 2524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9667163710165626e-05,
+      "loss": 1.5702,
+      "step": 2525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9666744001079544e-05,
+      "loss": 1.554,
+      "step": 2526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966632402930761e-05,
+      "loss": 1.6021,
+      "step": 2527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9665903794854305e-05,
+      "loss": 1.6117,
+      "step": 2528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.96654832977241e-05,
+      "loss": 1.5103,
+      "step": 2529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966506253792148e-05,
+      "loss": 1.4199,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9664641515450907e-05,
+      "loss": 1.5194,
+      "step": 2531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9664220230316884e-05,
+      "loss": 1.523,
+      "step": 2532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966379868252388e-05,
+      "loss": 1.6364,
+      "step": 2533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.96633768720764e-05,
+      "loss": 1.5907,
+      "step": 2534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966295479897893e-05,
+      "loss": 1.5555,
+      "step": 2535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9662532463235965e-05,
+      "loss": 1.5571,
+      "step": 2536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9662109864852004e-05,
+      "loss": 1.5513,
+      "step": 2537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9661687003831534e-05,
+      "loss": 1.6139,
+      "step": 2538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9661263880179085e-05,
+      "loss": 1.4961,
+      "step": 2539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966084049389914e-05,
+      "loss": 1.481,
+      "step": 2540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966041684499621e-05,
+      "loss": 1.5234,
+      "step": 2541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.965999293347482e-05,
+      "loss": 1.5703,
+      "step": 2542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965956875933948e-05,
+      "loss": 1.5409,
+      "step": 2543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9659144322594694e-05,
+      "loss": 1.4945,
+      "step": 2544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9658719623245e-05,
+      "loss": 1.5664,
+      "step": 2545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9658294661294914e-05,
+      "loss": 1.5415,
+      "step": 2546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965786943674896e-05,
+      "loss": 1.5202,
+      "step": 2547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965744394961166e-05,
+      "loss": 1.6548,
+      "step": 2548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965701819988755e-05,
+      "loss": 1.533,
+      "step": 2549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965659218758117e-05,
+      "loss": 1.5178,
+      "step": 2550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9656165912697054e-05,
+      "loss": 1.5425,
+      "step": 2551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965573937523974e-05,
+      "loss": 1.5734,
+      "step": 2552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965531257521378e-05,
+      "loss": 1.6209,
+      "step": 2553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965488551262369e-05,
+      "loss": 1.5594,
+      "step": 2554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965445818747405e-05,
+      "loss": 1.5348,
+      "step": 2555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.96540305997694e-05,
+      "loss": 1.5527,
+      "step": 2556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965360274951428e-05,
+      "loss": 1.5726,
+      "step": 2557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9653174636713274e-05,
+      "loss": 1.5786,
+      "step": 2558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9652746261370914e-05,
+      "loss": 1.5417,
+      "step": 2559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9652317623491776e-05,
+      "loss": 1.5667,
+      "step": 2560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965188872308042e-05,
+      "loss": 1.4981,
+      "step": 2561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965145956014142e-05,
+      "loss": 1.4808,
+      "step": 2562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965103013467934e-05,
+      "loss": 1.5615,
+      "step": 2563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965060044669876e-05,
+      "loss": 1.5158,
+      "step": 2564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965017049620424e-05,
+      "loss": 1.5879,
+      "step": 2565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964974028320038e-05,
+      "loss": 1.523,
+      "step": 2566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964930980769175e-05,
+      "loss": 1.5423,
+      "step": 2567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964887906968293e-05,
+      "loss": 1.6413,
+      "step": 2568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964844806917852e-05,
+      "loss": 1.6304,
+      "step": 2569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9648016806183096e-05,
+      "loss": 1.5604,
+      "step": 2570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964758528070126e-05,
+      "loss": 1.5975,
+      "step": 2571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.96471534927376e-05,
+      "loss": 1.5532,
+      "step": 2572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964672144229672e-05,
+      "loss": 1.5373,
+      "step": 2573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964628912938323e-05,
+      "loss": 1.5441,
+      "step": 2574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9645856554001714e-05,
+      "loss": 1.4878,
+      "step": 2575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964542371615679e-05,
+      "loss": 1.5622,
+      "step": 2576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9644990615853074e-05,
+      "loss": 1.6068,
+      "step": 2577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964455725309516e-05,
+      "loss": 1.5557,
+      "step": 2578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964412362788767e-05,
+      "loss": 1.6232,
+      "step": 2579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9643689740235235e-05,
+      "loss": 1.5487,
+      "step": 2580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964325559014247e-05,
+      "loss": 1.5562,
+      "step": 2581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964282117761398e-05,
+      "loss": 1.5575,
+      "step": 2582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9642386502654416e-05,
+      "loss": 1.5477,
+      "step": 2583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964195156526839e-05,
+      "loss": 1.6159,
+      "step": 2584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9641516365460543e-05,
+      "loss": 1.5627,
+      "step": 2585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964108090323551e-05,
+      "loss": 1.5153,
+      "step": 2586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9640645178597924e-05,
+      "loss": 1.5309,
+      "step": 2587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9640209191552426e-05,
+      "loss": 1.5298,
+      "step": 2588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9639772942103656e-05,
+      "loss": 1.495,
+      "step": 2589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963933643025627e-05,
+      "loss": 1.5791,
+      "step": 2590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.96388996560149e-05,
+      "loss": 1.5697,
+      "step": 2591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9638462619384215e-05,
+      "loss": 1.5524,
+      "step": 2592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9638025320368863e-05,
+      "loss": 1.5575,
+      "step": 2593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963758775897349e-05,
+      "loss": 1.5024,
+      "step": 2594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9637149935202775e-05,
+      "loss": 1.5181,
+      "step": 2595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963671184906137e-05,
+      "loss": 1.5389,
+      "step": 2596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9636273500553934e-05,
+      "loss": 1.5332,
+      "step": 2597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963583488968515e-05,
+      "loss": 1.5409,
+      "step": 2598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963539601645968e-05,
+      "loss": 1.5508,
+      "step": 2599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9634956880882194e-05,
+      "loss": 1.5349,
+      "step": 2600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963451748295738e-05,
+      "loss": 1.5793,
+      "step": 2601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963407782268991e-05,
+      "loss": 1.4815,
+      "step": 2602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9633637900084464e-05,
+      "loss": 1.5187,
+      "step": 2603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9633197715145726e-05,
+      "loss": 1.5687,
+      "step": 2604
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.963275726787839e-05,
+      "loss": 1.5557,
+      "step": 2605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.963231655828715e-05,
+      "loss": 1.572,
+      "step": 2606
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9631875586376684e-05,
+      "loss": 1.522,
+      "step": 2607
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96314343521517e-05,
+      "loss": 1.6156,
+      "step": 2608
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96309928556169e-05,
+      "loss": 1.9898,
+      "step": 2609
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9630551096776975e-05,
+      "loss": 1.5677,
+      "step": 2610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.963010907563663e-05,
+      "loss": 1.5572,
+      "step": 2611
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962966679220059e-05,
+      "loss": 1.5062,
+      "step": 2612
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962922424647354e-05,
+      "loss": 1.5807,
+      "step": 2613
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962878143846021e-05,
+      "loss": 1.4979,
+      "step": 2614
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962833836816531e-05,
+      "loss": 1.5184,
+      "step": 2615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962789503559355e-05,
+      "loss": 1.5698,
+      "step": 2616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9627451440749675e-05,
+      "loss": 1.5461,
+      "step": 2617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9627007583638386e-05,
+      "loss": 1.5436,
+      "step": 2618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962656346426441e-05,
+      "loss": 1.5841,
+      "step": 2619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96261190826325e-05,
+      "loss": 1.5434,
+      "step": 2620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962567443874736e-05,
+      "loss": 1.5966,
+      "step": 2621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962522953261374e-05,
+      "loss": 1.6105,
+      "step": 2622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962478436423638e-05,
+      "loss": 1.5888,
+      "step": 2623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962433893362002e-05,
+      "loss": 1.4936,
+      "step": 2624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962389324076939e-05,
+      "loss": 1.5052,
+      "step": 2625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962344728568925e-05,
+      "loss": 1.5164,
+      "step": 2626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962300106838435e-05,
+      "loss": 1.5235,
+      "step": 2627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9622554588859435e-05,
+      "loss": 1.5549,
+      "step": 2628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9622107847119256e-05,
+      "loss": 1.5468,
+      "step": 2629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962166084316859e-05,
+      "loss": 1.5491,
+      "step": 2630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962121357701217e-05,
+      "loss": 1.5473,
+      "step": 2631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962076604865479e-05,
+      "loss": 1.5462,
+      "step": 2632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9620318258101184e-05,
+      "loss": 1.4351,
+      "step": 2633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9619870205356145e-05,
+      "loss": 1.5033,
+      "step": 2634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961942189042443e-05,
+      "loss": 1.5975,
+      "step": 2635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9618973313310826e-05,
+      "loss": 1.5166,
+      "step": 2636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9618524474020086e-05,
+      "loss": 1.6133,
+      "step": 2637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961807537255703e-05,
+      "loss": 1.5156,
+      "step": 2638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96176260089264e-05,
+      "loss": 1.5955,
+      "step": 2639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9617176383133e-05,
+      "loss": 1.5573,
+      "step": 2640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961672649518162e-05,
+      "loss": 1.5318,
+      "step": 2641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961627634507705e-05,
+      "loss": 1.5419,
+      "step": 2642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9615825932824085e-05,
+      "loss": 1.557,
+      "step": 2643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961537525842751e-05,
+      "loss": 1.5804,
+      "step": 2644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961492432189214e-05,
+      "loss": 1.4831,
+      "step": 2645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961447312322277e-05,
+      "loss": 1.5137,
+      "step": 2646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96140216624242e-05,
+      "loss": 1.535,
+      "step": 2647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961356993950125e-05,
+      "loss": 1.5917,
+      "step": 2648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961311795445871e-05,
+      "loss": 1.5328,
+      "step": 2649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961266570730142e-05,
+      "loss": 1.5181,
+      "step": 2650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961221319803418e-05,
+      "loss": 1.5667,
+      "step": 2651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96117604266618e-05,
+      "loss": 1.5008,
+      "step": 2652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961130739318913e-05,
+      "loss": 1.5905,
+      "step": 2653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961085409762097e-05,
+      "loss": 1.5479,
+      "step": 2654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961040053996216e-05,
+      "loss": 1.5456,
+      "step": 2655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960994672021752e-05,
+      "loss": 1.5646,
+      "step": 2656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96094926383919e-05,
+      "loss": 1.6162,
+      "step": 2657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960903829449011e-05,
+      "loss": 1.5791,
+      "step": 2658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9608583688517004e-05,
+      "loss": 1.5594,
+      "step": 2659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960812882047743e-05,
+      "loss": 1.6097,
+      "step": 2660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960767369037622e-05,
+      "loss": 1.578,
+      "step": 2661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960721829821823e-05,
+      "loss": 1.5521,
+      "step": 2662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9606762644008295e-05,
+      "loss": 1.5031,
+      "step": 2663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9606306727751284e-05,
+      "loss": 1.5926,
+      "step": 2664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960585054945204e-05,
+      "loss": 1.4943,
+      "step": 2665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960539410911543e-05,
+      "loss": 1.562,
+      "step": 2666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960493740674631e-05,
+      "loss": 1.5786,
+      "step": 2667
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960448044234954e-05,
+      "loss": 1.4926,
+      "step": 2668
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9604023215929994e-05,
+      "loss": 1.543,
+      "step": 2669
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960356572749254e-05,
+      "loss": 1.5389,
+      "step": 2670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9603107977042054e-05,
+      "loss": 1.5764,
+      "step": 2671
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.96026499645834e-05,
+      "loss": 1.5505,
+      "step": 2672
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9602191690121455e-05,
+      "loss": 1.5723,
+      "step": 2673
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9601733153661114e-05,
+      "loss": 1.5754,
+      "step": 2674
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960127435520724e-05,
+      "loss": 1.583,
+      "step": 2675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960081529476474e-05,
+      "loss": 1.5466,
+      "step": 2676
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9600355972338486e-05,
+      "loss": 1.5879,
+      "step": 2677
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959989638793338e-05,
+      "loss": 1.4751,
+      "step": 2678
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959943654155431e-05,
+      "loss": 1.5272,
+      "step": 2679
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9598976433206176e-05,
+      "loss": 1.5319,
+      "step": 2680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9598516062893884e-05,
+      "loss": 1.5359,
+      "step": 2681
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959805543062233e-05,
+      "loss": 1.5993,
+      "step": 2682
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9597594536396405e-05,
+      "loss": 1.5625,
+      "step": 2683
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959713338022104e-05,
+      "loss": 1.528,
+      "step": 2684
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959667196210114e-05,
+      "loss": 1.5205,
+      "step": 2685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9596210282041614e-05,
+      "loss": 1.5824,
+      "step": 2686
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959574834004739e-05,
+      "loss": 1.4889,
+      "step": 2687
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959528613612336e-05,
+      "loss": 1.5181,
+      "step": 2688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959482367027448e-05,
+      "loss": 1.4784,
+      "step": 2689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959436094250566e-05,
+      "loss": 1.5873,
+      "step": 2690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9593897952821824e-05,
+      "loss": 1.5498,
+      "step": 2691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9593434701227906e-05,
+      "loss": 1.5412,
+      "step": 2692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959297118772883e-05,
+      "loss": 1.5277,
+      "step": 2693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9592507412329556e-05,
+      "loss": 1.5251,
+      "step": 2694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9592043375035005e-05,
+      "loss": 1.5796,
+      "step": 2695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959157907585012e-05,
+      "loss": 1.572,
+      "step": 2696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959111451477985e-05,
+      "loss": 1.572,
+      "step": 2697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959064969182913e-05,
+      "loss": 1.5558,
+      "step": 2698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9590184607002934e-05,
+      "loss": 1.5137,
+      "step": 2699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958971926030619e-05,
+      "loss": 1.4932,
+      "step": 2700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958925365174387e-05,
+      "loss": 1.508,
+      "step": 2701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9588787781320924e-05,
+      "loss": 1.5536,
+      "step": 2702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958832164904232e-05,
+      "loss": 1.5977,
+      "step": 2703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958785525491302e-05,
+      "loss": 1.5845,
+      "step": 2704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9587388598937986e-05,
+      "loss": 1.5349,
+      "step": 2705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958692168112219e-05,
+      "loss": 1.5321,
+      "step": 2706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958645450147061e-05,
+      "loss": 1.6263,
+      "step": 2707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958598705998821e-05,
+      "loss": 1.5038,
+      "step": 2708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958551935667998e-05,
+      "loss": 1.5199,
+      "step": 2709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9585051391550893e-05,
+      "loss": 1.6587,
+      "step": 2710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958458316460593e-05,
+      "loss": 1.5739,
+      "step": 2711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9584114675850086e-05,
+      "loss": 1.5425,
+      "step": 2712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958364592528835e-05,
+      "loss": 1.5129,
+      "step": 2713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.95831769129257e-05,
+      "loss": 1.5298,
+      "step": 2714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9582707638767144e-05,
+      "loss": 1.5609,
+      "step": 2715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958223810281768e-05,
+      "loss": 1.5676,
+      "step": 2716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958176830508231e-05,
+      "loss": 1.5633,
+      "step": 2717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958129824556602e-05,
+      "loss": 1.5589,
+      "step": 2718
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958082792427383e-05,
+      "loss": 1.5212,
+      "step": 2719
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9580357341210756e-05,
+      "loss": 1.5166,
+      "step": 2720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957988649638179e-05,
+      "loss": 1.5192,
+      "step": 2721
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9579415389791954e-05,
+      "loss": 1.5581,
+      "step": 2722
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957894402144627e-05,
+      "loss": 1.5956,
+      "step": 2723
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957847239134976e-05,
+      "loss": 1.5537,
+      "step": 2724
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957800049950743e-05,
+      "loss": 1.4813,
+      "step": 2725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9577528345924326e-05,
+      "loss": 1.5077,
+      "step": 2726
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957705593060546e-05,
+      "loss": 1.5355,
+      "step": 2727
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9576583253555866e-05,
+      "loss": 1.5326,
+      "step": 2728
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9576110314780586e-05,
+      "loss": 1.5275,
+      "step": 2729
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9575637114284655e-05,
+      "loss": 1.4572,
+      "step": 2730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9575163652073107e-05,
+      "loss": 1.5225,
+      "step": 2731
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957468992815098e-05,
+      "loss": 1.5187,
+      "step": 2732
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957421594252334e-05,
+      "loss": 1.5396,
+      "step": 2733
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9573741695195205e-05,
+      "loss": 1.5388,
+      "step": 2734
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9573267186171647e-05,
+      "loss": 1.6019,
+      "step": 2735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9572792415457706e-05,
+      "loss": 1.5495,
+      "step": 2736
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957231738305845e-05,
+      "loss": 1.5354,
+      "step": 2737
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9571842088978934e-05,
+      "loss": 1.5702,
+      "step": 2738
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957136653322421e-05,
+      "loss": 1.5243,
+      "step": 2739
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957089071579936e-05,
+      "loss": 1.5311,
+      "step": 2740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9570414636709436e-05,
+      "loss": 1.5881,
+      "step": 2741
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9569938295959504e-05,
+      "loss": 1.5562,
+      "step": 2742
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956946169355466e-05,
+      "loss": 1.5653,
+      "step": 2743
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956898482949996e-05,
+      "loss": 1.6301,
+      "step": 2744
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9568507703800486e-05,
+      "loss": 1.5785,
+      "step": 2745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956803031646132e-05,
+      "loss": 1.5625,
+      "step": 2746
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9567552667487555e-05,
+      "loss": 1.583,
+      "step": 2747
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9567074756884265e-05,
+      "loss": 1.5199,
+      "step": 2748
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9566596584656546e-05,
+      "loss": 1.5666,
+      "step": 2749
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956611815080948e-05,
+      "loss": 1.62,
+      "step": 2750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9565639455348174e-05,
+      "loss": 1.5584,
+      "step": 2751
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956516049827772e-05,
+      "loss": 1.5843,
+      "step": 2752
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956468127960322e-05,
+      "loss": 1.5184,
+      "step": 2753
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956420179932979e-05,
+      "loss": 1.5217,
+      "step": 2754
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956372205746251e-05,
+      "loss": 1.5731,
+      "step": 2755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956324205400651e-05,
+      "loss": 1.5542,
+      "step": 2756
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956276178896689e-05,
+      "loss": 1.5565,
+      "step": 2757
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956228126234878e-05,
+      "loss": 1.5996,
+      "step": 2758
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956180047415728e-05,
+      "loss": 1.551,
+      "step": 2759
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956131942439751e-05,
+      "loss": 1.5794,
+      "step": 2760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956083811307461e-05,
+      "loss": 1.5513,
+      "step": 2761
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9560356540193684e-05,
+      "loss": 1.5562,
+      "step": 2762
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955987470575988e-05,
+      "loss": 1.5223,
+      "step": 2763
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9559392609778324e-05,
+      "loss": 1.5166,
+      "step": 2764
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955891025225414e-05,
+      "loss": 1.6125,
+      "step": 2765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9558427633192483e-05,
+      "loss": 1.5174,
+      "step": 2766
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955794475259847e-05,
+      "loss": 1.5378,
+      "step": 2767
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955746161047726e-05,
+      "loss": 1.575,
+      "step": 2768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9556978206833994e-05,
+      "loss": 1.5589,
+      "step": 2769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955649454167381e-05,
+      "loss": 1.6569,
+      "step": 2770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955601061500188e-05,
+      "loss": 1.5379,
+      "step": 2771
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955552642682334e-05,
+      "loss": 1.5488,
+      "step": 2772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955504197714335e-05,
+      "loss": 1.4933,
+      "step": 2773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955455726596707e-05,
+      "loss": 1.55,
+      "step": 2774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955407229329966e-05,
+      "loss": 1.5602,
+      "step": 2775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9553587059146295e-05,
+      "loss": 1.5749,
+      "step": 2776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955310156351213e-05,
+      "loss": 1.4847,
+      "step": 2777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955261580640234e-05,
+      "loss": 1.5596,
+      "step": 2778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.95521297878221e-05,
+      "loss": 1.5583,
+      "step": 2779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9551643507776586e-05,
+      "loss": 1.4634,
+      "step": 2780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9551156966270975e-05,
+      "loss": 1.5184,
+      "step": 2781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955067016331044e-05,
+      "loss": 1.5765,
+      "step": 2782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955018309890017e-05,
+      "loss": 1.5672,
+      "step": 2783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954969577304537e-05,
+      "loss": 1.5747,
+      "step": 2784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.95492081857512e-05,
+      "loss": 1.54,
+      "step": 2785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954872033702287e-05,
+      "loss": 1.5523,
+      "step": 2786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954823222686558e-05,
+      "loss": 1.516,
+      "step": 2787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954774385528451e-05,
+      "loss": 1.5871,
+      "step": 2788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954725522228488e-05,
+      "loss": 1.5523,
+      "step": 2789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954676632787188e-05,
+      "loss": 1.5654,
+      "step": 2790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954627717205073e-05,
+      "loss": 1.4933,
+      "step": 2791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954578775482662e-05,
+      "loss": 1.5301,
+      "step": 2792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954529807620477e-05,
+      "loss": 1.5443,
+      "step": 2793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954480813619041e-05,
+      "loss": 1.4714,
+      "step": 2794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954431793478874e-05,
+      "loss": 1.5539,
+      "step": 2795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9543827472004975e-05,
+      "loss": 1.5881,
+      "step": 2796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954333674784436e-05,
+      "loss": 1.533,
+      "step": 2797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954284576231211e-05,
+      "loss": 1.5576,
+      "step": 2798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9542354515413446e-05,
+      "loss": 1.5285,
+      "step": 2799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954186300715361e-05,
+      "loss": 1.4932,
+      "step": 2800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9541371237537834e-05,
+      "loss": 1.5428,
+      "step": 2801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954087920657135e-05,
+      "loss": 1.5604,
+      "step": 2802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9540386914259404e-05,
+      "loss": 1.5778,
+      "step": 2803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953989436060723e-05,
+      "loss": 1.5999,
+      "step": 2804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953940154562009e-05,
+      "loss": 1.5428,
+      "step": 2805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9538908469303215e-05,
+      "loss": 1.5448,
+      "step": 2806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953841513166187e-05,
+      "loss": 1.5178,
+      "step": 2807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9537921532701294e-05,
+      "loss": 1.5065,
+      "step": 2808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953742767242675e-05,
+      "loss": 1.5176,
+      "step": 2809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.95369335508435e-05,
+      "loss": 1.5435,
+      "step": 2810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9536439167956803e-05,
+      "loss": 1.5687,
+      "step": 2811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953594452377194e-05,
+      "loss": 1.5902,
+      "step": 2812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953544961829415e-05,
+      "loss": 1.4671,
+      "step": 2813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9534954451528714e-05,
+      "loss": 1.5462,
+      "step": 2814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953445902348092e-05,
+      "loss": 1.6021,
+      "step": 2815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953396333415602e-05,
+      "loss": 1.5718,
+      "step": 2816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953346738355932e-05,
+      "loss": 1.5837,
+      "step": 2817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9532971171696074e-05,
+      "loss": 1.5737,
+      "step": 2818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953247469857159e-05,
+      "loss": 1.5275,
+      "step": 2819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953197796419113e-05,
+      "loss": 1.4904,
+      "step": 2820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9531480968560014e-05,
+      "loss": 1.5622,
+      "step": 2821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953098371168351e-05,
+      "loss": 1.6126,
+      "step": 2822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953048619356693e-05,
+      "loss": 1.5005,
+      "step": 2823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952998841421555e-05,
+      "loss": 1.5127,
+      "step": 2824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.95294903736347e-05,
+      "loss": 1.5433,
+      "step": 2825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9528992071829666e-05,
+      "loss": 1.5462,
+      "step": 2826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9528493508805765e-05,
+      "loss": 1.6175,
+      "step": 2827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952799468456829e-05,
+      "loss": 1.5726,
+      "step": 2828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952749559912256e-05,
+      "loss": 1.5498,
+      "step": 2829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.95269962524739e-05,
+      "loss": 1.6356,
+      "step": 2830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952649664462762e-05,
+      "loss": 1.5545,
+      "step": 2831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952599677558904e-05,
+      "loss": 1.5375,
+      "step": 2832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9525496645363484e-05,
+      "loss": 1.5542,
+      "step": 2833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952499625395628e-05,
+      "loss": 1.6095,
+      "step": 2834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9524495601372755e-05,
+      "loss": 1.5803,
+      "step": 2835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952399468761824e-05,
+      "loss": 1.5389,
+      "step": 2836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9523493512698074e-05,
+      "loss": 1.5794,
+      "step": 2837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9522992076617584e-05,
+      "loss": 1.536,
+      "step": 2838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9522490379382125e-05,
+      "loss": 1.5632,
+      "step": 2839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952198842099702e-05,
+      "loss": 1.5267,
+      "step": 2840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952148620146764e-05,
+      "loss": 1.5415,
+      "step": 2841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9520983720799304e-05,
+      "loss": 1.5799,
+      "step": 2842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9520480978997394e-05,
+      "loss": 1.4915,
+      "step": 2843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951997797606724e-05,
+      "loss": 1.5529,
+      "step": 2844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.95194747120142e-05,
+      "loss": 1.5729,
+      "step": 2845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9518971186843656e-05,
+      "loss": 1.5627,
+      "step": 2846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951846740056094e-05,
+      "loss": 1.5977,
+      "step": 2847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9517963353171446e-05,
+      "loss": 1.5736,
+      "step": 2848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9517459044680516e-05,
+      "loss": 1.5715,
+      "step": 2849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951695447509354e-05,
+      "loss": 1.5036,
+      "step": 2850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9516449644415876e-05,
+      "loss": 1.5308,
+      "step": 2851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951594455265291e-05,
+      "loss": 1.6149,
+      "step": 2852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951543919981002e-05,
+      "loss": 1.5171,
+      "step": 2853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951493358589259e-05,
+      "loss": 1.5119,
+      "step": 2854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9514427710906e-05,
+      "loss": 1.5072,
+      "step": 2855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951392157485563e-05,
+      "loss": 1.4671,
+      "step": 2856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.951341517774688e-05,
+      "loss": 1.5109,
+      "step": 2857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.951290851958514e-05,
+      "loss": 1.5534,
+      "step": 2858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9512401600375814e-05,
+      "loss": 1.5487,
+      "step": 2859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9511894420124296e-05,
+      "loss": 1.5306,
+      "step": 2860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.951138697883597e-05,
+      "loss": 1.5116,
+      "step": 2861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.951087927651627e-05,
+      "loss": 1.5776,
+      "step": 2862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9510371313170564e-05,
+      "loss": 1.5402,
+      "step": 2863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950986308880431e-05,
+      "loss": 1.5391,
+      "step": 2864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9509354603422875e-05,
+      "loss": 1.6019,
+      "step": 2865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9508845857031696e-05,
+      "loss": 1.5996,
+      "step": 2866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950833684963619e-05,
+      "loss": 1.5612,
+      "step": 2867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950782758124178e-05,
+      "loss": 1.5448,
+      "step": 2868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950731805185389e-05,
+      "loss": 1.5195,
+      "step": 2869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950680826147793e-05,
+      "loss": 1.53,
+      "step": 2870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950629821011935e-05,
+      "loss": 1.4981,
+      "step": 2871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950578789778357e-05,
+      "loss": 1.5659,
+      "step": 2872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950527732447602e-05,
+      "loss": 1.5939,
+      "step": 2873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9504766490202145e-05,
+      "loss": 1.5482,
+      "step": 2874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9504255394967394e-05,
+      "loss": 1.6113,
+      "step": 2875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950374403877719e-05,
+      "loss": 1.5358,
+      "step": 2876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9503232421636994e-05,
+      "loss": 1.5635,
+      "step": 2877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950272054355225e-05,
+      "loss": 1.5137,
+      "step": 2878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9502208404528404e-05,
+      "loss": 1.5225,
+      "step": 2879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950169600457092e-05,
+      "loss": 1.5746,
+      "step": 2880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9501183343685244e-05,
+      "loss": 1.5046,
+      "step": 2881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950067042187684e-05,
+      "loss": 1.5293,
+      "step": 2882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950015723915117e-05,
+      "loss": 1.5731,
+      "step": 2883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9499643795513693e-05,
+      "loss": 1.5037,
+      "step": 2884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.94991300909699e-05,
+      "loss": 1.5254,
+      "step": 2885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949861612552523e-05,
+      "loss": 1.6104,
+      "step": 2886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949810189918518e-05,
+      "loss": 1.5801,
+      "step": 2887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949758741195522e-05,
+      "loss": 1.5754,
+      "step": 2888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949707266384083e-05,
+      "loss": 1.6393,
+      "step": 2889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949655765484747e-05,
+      "loss": 1.5706,
+      "step": 2890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949604238498066e-05,
+      "loss": 1.513,
+      "step": 2891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9495526854245865e-05,
+      "loss": 1.605,
+      "step": 2892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9495011062648586e-05,
+      "loss": 1.557,
+      "step": 2893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949449501019431e-05,
+      "loss": 1.5493,
+      "step": 2894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949397869688853e-05,
+      "loss": 1.4779,
+      "step": 2895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949346212273675e-05,
+      "loss": 1.64,
+      "step": 2896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9492945287744456e-05,
+      "loss": 1.5334,
+      "step": 2897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949242819191718e-05,
+      "loss": 1.5825,
+      "step": 2898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.94919108352604e-05,
+      "loss": 1.5464,
+      "step": 2899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949139321777965e-05,
+      "loss": 1.4857,
+      "step": 2900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949087533948043e-05,
+      "loss": 1.5457,
+      "step": 2901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949035720036825e-05,
+      "loss": 1.5114,
+      "step": 2902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948983880044864e-05,
+      "loss": 1.6188,
+      "step": 2903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948932013972712e-05,
+      "loss": 1.5156,
+      "step": 2904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.94888012182092e-05,
+      "loss": 1.5622,
+      "step": 2905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948828203590041e-05,
+      "loss": 1.5233,
+      "step": 2906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9487762592806295e-05,
+      "loss": 1.4922,
+      "step": 2907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948724288893237e-05,
+      "loss": 1.5371,
+      "step": 2908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948672292428418e-05,
+      "loss": 1.5282,
+      "step": 2909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9486202698867247e-05,
+      "loss": 1.5306,
+      "step": 2910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948568221268713e-05,
+      "loss": 1.5005,
+      "step": 2911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948516146574936e-05,
+      "loss": 1.5785,
+      "step": 2912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948464045805949e-05,
+      "loss": 1.5819,
+      "step": 2913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948411918962306e-05,
+      "loss": 1.5387,
+      "step": 2914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9483597660445616e-05,
+      "loss": 1.5483,
+      "step": 2915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948307587053274e-05,
+      "loss": 1.5581,
+      "step": 2916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948255381988996e-05,
+      "loss": 1.5889,
+      "step": 2917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948203150852284e-05,
+      "loss": 1.5181,
+      "step": 2918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.948150893643695e-05,
+      "loss": 1.5236,
+      "step": 2919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.948098610363786e-05,
+      "loss": 1.5114,
+      "step": 2920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.948046301013113e-05,
+      "loss": 1.5477,
+      "step": 2921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9479939655922325e-05,
+      "loss": 1.5894,
+      "step": 2922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9479416041017025e-05,
+      "loss": 1.5789,
+      "step": 2923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947889216542081e-05,
+      "loss": 1.5571,
+      "step": 2924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947836802913926e-05,
+      "loss": 1.4854,
+      "step": 2925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947784363217794e-05,
+      "loss": 1.5584,
+      "step": 2926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947731897454245e-05,
+      "loss": 1.5409,
+      "step": 2927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947679405623838e-05,
+      "loss": 1.4893,
+      "step": 2928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9476268877271304e-05,
+      "loss": 1.5599,
+      "step": 2929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9475743437646835e-05,
+      "loss": 1.5674,
+      "step": 2930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947521773737055e-05,
+      "loss": 1.53,
+      "step": 2931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9474691776448057e-05,
+      "loss": 1.5202,
+      "step": 2932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9474165554884954e-05,
+      "loss": 1.4593,
+      "step": 2933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947363907268685e-05,
+      "loss": 1.5964,
+      "step": 2934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947311232985935e-05,
+      "loss": 1.5259,
+      "step": 2935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947258532640805e-05,
+      "loss": 1.5593,
+      "step": 2936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947205806233859e-05,
+      "loss": 1.5344,
+      "step": 2937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947153053765656e-05,
+      "loss": 1.5098,
+      "step": 2938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947100275236759e-05,
+      "loss": 1.4828,
+      "step": 2939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947047470647729e-05,
+      "loss": 1.5335,
+      "step": 2940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9469946399991304e-05,
+      "loss": 1.516,
+      "step": 2941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946941783291523e-05,
+      "loss": 1.5536,
+      "step": 2942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946888900525473e-05,
+      "loss": 1.5339,
+      "step": 2943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.94683599170154e-05,
+      "loss": 1.5445,
+      "step": 2944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.94678305682029e-05,
+      "loss": 1.5917,
+      "step": 2945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946730095882286e-05,
+      "loss": 1.5192,
+      "step": 2946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9466771088880915e-05,
+      "loss": 1.567,
+      "step": 2947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946624095838271e-05,
+      "loss": 1.5689,
+      "step": 2948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9465710567333903e-05,
+      "loss": 1.4878,
+      "step": 2949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9465179915740124e-05,
+      "loss": 1.5065,
+      "step": 2950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9464649003607034e-05,
+      "loss": 1.5941,
+      "step": 2951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946411783094028e-05,
+      "loss": 1.5514,
+      "step": 2952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946358639774553e-05,
+      "loss": 1.5052,
+      "step": 2953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9463054704028434e-05,
+      "loss": 1.5319,
+      "step": 2954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946252274979466e-05,
+      "loss": 1.5815,
+      "step": 2955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946199053504988e-05,
+      "loss": 1.5636,
+      "step": 2956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946145805979973e-05,
+      "loss": 1.5843,
+      "step": 2957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946092532404991e-05,
+      "loss": 1.6244,
+      "step": 2958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946039232780609e-05,
+      "loss": 1.4971,
+      "step": 2959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9459859071073943e-05,
+      "loss": 1.5078,
+      "step": 2960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945932555385915e-05,
+      "loss": 1.549,
+      "step": 2961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9458791776167376e-05,
+      "loss": 1.486,
+      "step": 2962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945825773800433e-05,
+      "loss": 1.5448,
+      "step": 2963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945772343937569e-05,
+      "loss": 1.5524,
+      "step": 2964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945718888028714e-05,
+      "loss": 1.5422,
+      "step": 2965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9456654060744376e-05,
+      "loss": 1.6579,
+      "step": 2966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9456118980753095e-05,
+      "loss": 1.6528,
+      "step": 2967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945558364031899e-05,
+      "loss": 1.515,
+      "step": 2968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945504803944777e-05,
+      "loss": 1.6287,
+      "step": 2969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945451217814514e-05,
+      "loss": 1.5269,
+      "step": 2970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.94539760564168e-05,
+      "loss": 1.6467,
+      "step": 2971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945343967426846e-05,
+      "loss": 1.5311,
+      "step": 2972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9452903031705836e-05,
+      "loss": 1.5243,
+      "step": 2973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945236612873464e-05,
+      "loss": 1.6227,
+      "step": 2974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945182896536058e-05,
+      "loss": 1.6099,
+      "step": 2975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.94512915415894e-05,
+      "loss": 1.5785,
+      "step": 2976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.94507538574268e-05,
+      "loss": 1.5677,
+      "step": 2977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9450215912878514e-05,
+      "loss": 1.5143,
+      "step": 2978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.944967770795028e-05,
+      "loss": 1.5047,
+      "step": 2979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.944913924264782e-05,
+      "loss": 1.6481,
+      "step": 2980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9448600516976876e-05,
+      "loss": 1.5485,
+      "step": 2981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9448061530943174e-05,
+      "loss": 1.528,
+      "step": 2982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944752228455246e-05,
+      "loss": 1.5925,
+      "step": 2983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944698277781048e-05,
+      "loss": 1.5921,
+      "step": 2984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944644301072297e-05,
+      "loss": 1.5596,
+      "step": 2985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9445902983295686e-05,
+      "loss": 1.4956,
+      "step": 2986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944536269553438e-05,
+      "loss": 1.4772,
+      "step": 2987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944482214744479e-05,
+      "loss": 1.5282,
+      "step": 2988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9444281339032694e-05,
+      "loss": 1.605,
+      "step": 2989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9443740270303845e-05,
+      "loss": 1.5466,
+      "step": 2990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9443198941264e-05,
+      "loss": 1.4836,
+      "step": 2991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944265735191892e-05,
+      "loss": 1.5233,
+      "step": 2992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944211550227439e-05,
+      "loss": 1.5384,
+      "step": 2993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944157339233616e-05,
+      "loss": 1.6087,
+      "step": 2994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944103102211002e-05,
+      "loss": 1.4491,
+      "step": 2995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944048839160173e-05,
+      "loss": 1.4919,
+      "step": 2996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9439945500817084e-05,
+      "loss": 1.6125,
+      "step": 2997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943940234976185e-05,
+      "loss": 1.6213,
+      "step": 2998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9438858938441826e-05,
+      "loss": 1.4899,
+      "step": 2999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943831526686278e-05,
+      "loss": 1.5334,
+      "step": 3000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9437771335030516e-05,
+      "loss": 1.5995,
+      "step": 3001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943722714295083e-05,
+      "loss": 1.5641,
+      "step": 3002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943668269062951e-05,
+      "loss": 1.5112,
+      "step": 3003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943613797807235e-05,
+      "loss": 1.5915,
+      "step": 3004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9435593005285166e-05,
+      "loss": 1.5394,
+      "step": 3005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9435047772273744e-05,
+      "loss": 1.516,
+      "step": 3006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9434502279043906e-05,
+      "loss": 1.5799,
+      "step": 3007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943395652560144e-05,
+      "loss": 1.592,
+      "step": 3008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943341051195218e-05,
+      "loss": 1.4963,
+      "step": 3009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943286423810193e-05,
+      "loss": 1.5319,
+      "step": 3010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943231770405651e-05,
+      "loss": 1.5599,
+      "step": 3011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943177090982174e-05,
+      "loss": 1.5448,
+      "step": 3012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943122385540344e-05,
+      "loss": 1.5381,
+      "step": 3013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943067654080744e-05,
+      "loss": 1.5269,
+      "step": 3014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943012896603957e-05,
+      "loss": 1.5871,
+      "step": 3015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942958113110565e-05,
+      "loss": 1.4961,
+      "step": 3016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942903303601153e-05,
+      "loss": 1.5335,
+      "step": 3017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942848468076304e-05,
+      "loss": 1.5373,
+      "step": 3018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942793606536602e-05,
+      "loss": 1.63,
+      "step": 3019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942738718982631e-05,
+      "loss": 1.471,
+      "step": 3020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942683805414975e-05,
+      "loss": 1.492,
+      "step": 3021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.94262886583422e-05,
+      "loss": 1.5731,
+      "step": 3022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942573900240951e-05,
+      "loss": 1.5588,
+      "step": 3023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942518908635752e-05,
+      "loss": 1.5161,
+      "step": 3024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.94246389101921e-05,
+      "loss": 1.5322,
+      "step": 3025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942408847391911e-05,
+      "loss": 1.5565,
+      "step": 3026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942353777754439e-05,
+      "loss": 1.5584,
+      "step": 3027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942298682107383e-05,
+      "loss": 1.4992,
+      "step": 3028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9422435604513285e-05,
+      "loss": 1.5078,
+      "step": 3029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9421884127868634e-05,
+      "loss": 1.5763,
+      "step": 3030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942133239114574e-05,
+      "loss": 1.4943,
+      "step": 3031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9420780394350485e-05,
+      "loss": 1.5389,
+      "step": 3032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942022813748874e-05,
+      "loss": 1.5179,
+      "step": 3033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.94196756205664e-05,
+      "loss": 1.5251,
+      "step": 3034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941912284358933e-05,
+      "loss": 1.5807,
+      "step": 3035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941856980656343e-05,
+      "loss": 1.5371,
+      "step": 3036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941801650949459e-05,
+      "loss": 1.5275,
+      "step": 3037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9417462952388696e-05,
+      "loss": 1.4925,
+      "step": 3038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941690913525164e-05,
+      "loss": 1.568,
+      "step": 3039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941635505808934e-05,
+      "loss": 1.5073,
+      "step": 3040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941580072090768e-05,
+      "loss": 1.5537,
+      "step": 3041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941524612371256e-05,
+      "loss": 1.634,
+      "step": 3042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9414691266509886e-05,
+      "loss": 1.5747,
+      "step": 3043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941413614930558e-05,
+      "loss": 1.5309,
+      "step": 3044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.941358077210554e-05,
+      "loss": 1.5385,
+      "step": 3045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.941302513491569e-05,
+      "loss": 1.5391,
+      "step": 3046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9412469237741945e-05,
+      "loss": 1.5316,
+      "step": 3047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.941191308059022e-05,
+      "loss": 1.5949,
+      "step": 3048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9411356663466446e-05,
+      "loss": 1.4953,
+      "step": 3049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9410799986376544e-05,
+      "loss": 1.5632,
+      "step": 3050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.941024304932644e-05,
+      "loss": 1.5233,
+      "step": 3051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9409685852322066e-05,
+      "loss": 1.5643,
+      "step": 3052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940912839536936e-05,
+      "loss": 1.5946,
+      "step": 3053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9408570678474256e-05,
+      "loss": 1.5506,
+      "step": 3054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9408012701642694e-05,
+      "loss": 1.589,
+      "step": 3055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9407454464880614e-05,
+      "loss": 1.5275,
+      "step": 3056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940689596819396e-05,
+      "loss": 1.5417,
+      "step": 3057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9406337211588684e-05,
+      "loss": 1.4852,
+      "step": 3058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940577819507073e-05,
+      "loss": 1.555,
+      "step": 3059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940521891864606e-05,
+      "loss": 1.5592,
+      "step": 3060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9404659382320626e-05,
+      "loss": 1.5072,
+      "step": 3061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940409958610038e-05,
+      "loss": 1.5047,
+      "step": 3062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.94035395299913e-05,
+      "loss": 1.682,
+      "step": 3063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9402979213999334e-05,
+      "loss": 1.5703,
+      "step": 3064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9402418638130454e-05,
+      "loss": 1.5519,
+      "step": 3065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9401857802390626e-05,
+      "loss": 1.5856,
+      "step": 3066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940129670678584e-05,
+      "loss": 1.5277,
+      "step": 3067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9400735351322045e-05,
+      "loss": 1.5817,
+      "step": 3068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940017373600524e-05,
+      "loss": 1.5239,
+      "step": 3069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939961186084139e-05,
+      "loss": 1.4945,
+      "step": 3070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.93990497258365e-05,
+      "loss": 1.5252,
+      "step": 3071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939848733099654e-05,
+      "loss": 1.5508,
+      "step": 3072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939792467632749e-05,
+      "loss": 1.5497,
+      "step": 3073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939736176183537e-05,
+      "loss": 1.5711,
+      "step": 3074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939679858752615e-05,
+      "loss": 1.6273,
+      "step": 3075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939623515340585e-05,
+      "loss": 1.5247,
+      "step": 3076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939567145948044e-05,
+      "loss": 1.61,
+      "step": 3077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939510750575594e-05,
+      "loss": 1.5923,
+      "step": 3078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939454329223837e-05,
+      "loss": 1.5042,
+      "step": 3079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9393978818933715e-05,
+      "loss": 1.5107,
+      "step": 3080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9393414085848e-05,
+      "loss": 1.5334,
+      "step": 3081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939284909298722e-05,
+      "loss": 1.4847,
+      "step": 3082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9392283840357425e-05,
+      "loss": 1.5309,
+      "step": 3083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939171832796461e-05,
+      "loss": 1.5612,
+      "step": 3084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.93911525558148e-05,
+      "loss": 1.6147,
+      "step": 3085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939058652391403e-05,
+      "loss": 1.5409,
+      "step": 3086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939002023226832e-05,
+      "loss": 1.5173,
+      "step": 3087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.93894536808837e-05,
+      "loss": 1.4253,
+      "step": 3088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9388886869766206e-05,
+      "loss": 1.5747,
+      "step": 3089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938831979892187e-05,
+      "loss": 1.5654,
+      "step": 3090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9387752468356735e-05,
+      "loss": 1.4834,
+      "step": 3091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9387184878076844e-05,
+      "loss": 1.5111,
+      "step": 3092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938661702808824e-05,
+      "loss": 1.5716,
+      "step": 3093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938604891839697e-05,
+      "loss": 1.5483,
+      "step": 3094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938548054900909e-05,
+      "loss": 1.4801,
+      "step": 3095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9384911919930634e-05,
+      "loss": 1.6065,
+      "step": 3096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938434303116768e-05,
+      "loss": 1.5907,
+      "step": 3097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938377388272628e-05,
+      "loss": 1.5405,
+      "step": 3098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9383204474612485e-05,
+      "loss": 1.5041,
+      "step": 3099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938263480683237e-05,
+      "loss": 1.4689,
+      "step": 3100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9382064879391995e-05,
+      "loss": 1.492,
+      "step": 3101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938149469229743e-05,
+      "loss": 1.5776,
+      "step": 3102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938092424555475e-05,
+      "loss": 1.5322,
+      "step": 3103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938035353917003e-05,
+      "loss": 1.5305,
+      "step": 3104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9379782573149345e-05,
+      "loss": 1.577,
+      "step": 3105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9379211347498777e-05,
+      "loss": 1.5705,
+      "step": 3106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.937863986222441e-05,
+      "loss": 1.5228,
+      "step": 3107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937806811733233e-05,
+      "loss": 1.6008,
+      "step": 3108
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9377496112828626e-05,
+      "loss": 1.5101,
+      "step": 3109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9376923848719384e-05,
+      "loss": 1.6022,
+      "step": 3110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9376351325010706e-05,
+      "loss": 1.5526,
+      "step": 3111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937577854170868e-05,
+      "loss": 1.5655,
+      "step": 3112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937520549881942e-05,
+      "loss": 1.5208,
+      "step": 3113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9374632196349014e-05,
+      "loss": 1.5347,
+      "step": 3114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9374058634303574e-05,
+      "loss": 1.5218,
+      "step": 3115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937348481268921e-05,
+      "loss": 1.5298,
+      "step": 3116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9372910731512036e-05,
+      "loss": 1.5246,
+      "step": 3117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937233639077815e-05,
+      "loss": 1.5145,
+      "step": 3118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937176179049368e-05,
+      "loss": 1.5807,
+      "step": 3119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9371186930664746e-05,
+      "loss": 1.5191,
+      "step": 3120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937061181129746e-05,
+      "loss": 1.5337,
+      "step": 3121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937003643239796e-05,
+      "loss": 1.5448,
+      "step": 3122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936946079397237e-05,
+      "loss": 1.514,
+      "step": 3123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9368884896026814e-05,
+      "loss": 1.5651,
+      "step": 3124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936830873856743e-05,
+      "loss": 1.5356,
+      "step": 3125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936773232160035e-05,
+      "loss": 1.5117,
+      "step": 3126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936715564513172e-05,
+      "loss": 1.572,
+      "step": 3127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936657870916767e-05,
+      "loss": 1.457,
+      "step": 3128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936600151371435e-05,
+      "loss": 1.5536,
+      "step": 3129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936542405877791e-05,
+      "loss": 1.5415,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93648463443645e-05,
+      "loss": 1.5711,
+      "step": 3131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936426837048026e-05,
+      "loss": 1.4845,
+      "step": 3132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936369013713136e-05,
+      "loss": 1.5702,
+      "step": 3133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936311164432395e-05,
+      "loss": 1.5781,
+      "step": 3134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9362532892064195e-05,
+      "loss": 1.5711,
+      "step": 3135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936195388035825e-05,
+      "loss": 1.6061,
+      "step": 3136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936137460921229e-05,
+      "loss": 1.5296,
+      "step": 3137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9360795078632484e-05,
+      "loss": 1.6034,
+      "step": 3138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936021528862499e-05,
+      "loss": 1.5647,
+      "step": 3139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9359635239196e-05,
+      "loss": 1.6123,
+      "step": 3140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9359054930351676e-05,
+      "loss": 1.5085,
+      "step": 3141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935847436209822e-05,
+      "loss": 1.4588,
+      "step": 3142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935789353444179e-05,
+      "loss": 1.5319,
+      "step": 3143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935731244738858e-05,
+      "loss": 1.5721,
+      "step": 3144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9356731100944785e-05,
+      "loss": 1.555,
+      "step": 3145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9356149495116596e-05,
+      "loss": 1.5208,
+      "step": 3146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9355567629910195e-05,
+      "loss": 1.5457,
+      "step": 3147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935498550533178e-05,
+      "loss": 1.6217,
+      "step": 3148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935440312138757e-05,
+      "loss": 1.5643,
+      "step": 3149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935382047808375e-05,
+      "loss": 1.5597,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9353237575426515e-05,
+      "loss": 1.5127,
+      "step": 3151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93526544134221e-05,
+      "loss": 1.6064,
+      "step": 3152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935207099207669e-05,
+      "loss": 1.52,
+      "step": 3153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935148731139651e-05,
+      "loss": 1.5316,
+      "step": 3154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935090337138777e-05,
+      "loss": 1.4985,
+      "step": 3155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9350319172056704e-05,
+      "loss": 1.4912,
+      "step": 3156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9349734713409515e-05,
+      "loss": 1.5837,
+      "step": 3157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934914999545244e-05,
+      "loss": 1.5553,
+      "step": 3158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9348565018191694e-05,
+      "loss": 1.5827,
+      "step": 3159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934797978163351e-05,
+      "loss": 1.5697,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934739428578413e-05,
+      "loss": 1.5343,
+      "step": 3161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9346808530649785e-05,
+      "loss": 1.5822,
+      "step": 3162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9346222516236704e-05,
+      "loss": 1.5962,
+      "step": 3163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934563624255114e-05,
+      "loss": 1.5734,
+      "step": 3164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934504970959933e-05,
+      "loss": 1.5505,
+      "step": 3165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9344462917387515e-05,
+      "loss": 1.5039,
+      "step": 3166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934387586592195e-05,
+      "loss": 1.6201,
+      "step": 3167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9343288555208894e-05,
+      "loss": 1.5734,
+      "step": 3168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9342700985254593e-05,
+      "loss": 1.512,
+      "step": 3169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93421131560653e-05,
+      "loss": 1.4844,
+      "step": 3170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.934152506764728e-05,
+      "loss": 1.5876,
+      "step": 3171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93409367200068e-05,
+      "loss": 1.5075,
+      "step": 3172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.934034811315012e-05,
+      "loss": 1.5026,
+      "step": 3173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933975924708351e-05,
+      "loss": 1.5563,
+      "step": 3174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933917012181324e-05,
+      "loss": 1.6165,
+      "step": 3175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933858073734558e-05,
+      "loss": 1.4744,
+      "step": 3176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933799109368683e-05,
+      "loss": 1.4831,
+      "step": 3177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9337401190843234e-05,
+      "loss": 1.5067,
+      "step": 3178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93368110288211e-05,
+      "loss": 1.5534,
+      "step": 3179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933622060762669e-05,
+      "loss": 1.5524,
+      "step": 3180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933562992726632e-05,
+      "loss": 1.5438,
+      "step": 3181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933503898774626e-05,
+      "loss": 1.5913,
+      "step": 3182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9334447789072815e-05,
+      "loss": 1.5376,
+      "step": 3183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9333856331252256e-05,
+      "loss": 1.556,
+      "step": 3184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933326461429092e-05,
+      "loss": 1.4886,
+      "step": 3185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933267263819508e-05,
+      "loss": 1.5861,
+      "step": 3186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933208040297105e-05,
+      "loss": 1.5684,
+      "step": 3187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933148790862514e-05,
+      "loss": 1.4678,
+      "step": 3188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933089515516366e-05,
+      "loss": 1.4561,
+      "step": 3189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933030214259291e-05,
+      "loss": 1.5365,
+      "step": 3190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9329708870919214e-05,
+      "loss": 1.5915,
+      "step": 3191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932911534014889e-05,
+      "loss": 1.5259,
+      "step": 3192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932852155028826e-05,
+      "loss": 1.5527,
+      "step": 3193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932792750134364e-05,
+      "loss": 1.4826,
+      "step": 3194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932733319332137e-05,
+      "loss": 1.5042,
+      "step": 3195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932673862622776e-05,
+      "loss": 1.5804,
+      "step": 3196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932614380006917e-05,
+      "loss": 1.6123,
+      "step": 3197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93255487148519e-05,
+      "loss": 1.5335,
+      "step": 3198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932495337058232e-05,
+      "loss": 1.5566,
+      "step": 3199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9324357767266744e-05,
+      "loss": 1.591,
+      "step": 3200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9323761904911525e-05,
+      "loss": 1.5785,
+      "step": 3201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932316578352302e-05,
+      "loss": 1.5614,
+      "step": 3202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9322569403107557e-05,
+      "loss": 1.5316,
+      "step": 3203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93219727636715e-05,
+      "loss": 1.5249,
+      "step": 3204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9321375865221194e-05,
+      "loss": 1.5256,
+      "step": 3205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932077870776301e-05,
+      "loss": 1.5495,
+      "step": 3206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932018129130329e-05,
+      "loss": 1.4862,
+      "step": 3207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9319583615848404e-05,
+      "loss": 1.575,
+      "step": 3208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931898568140472e-05,
+      "loss": 1.5194,
+      "step": 3209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93183874879786e-05,
+      "loss": 1.4727,
+      "step": 3210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931778903557642e-05,
+      "loss": 1.5576,
+      "step": 3211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931719032420455e-05,
+      "loss": 1.606,
+      "step": 3212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931659135386937e-05,
+      "loss": 1.4697,
+      "step": 3213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931599212457725e-05,
+      "loss": 1.5524,
+      "step": 3214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9315392636334576e-05,
+      "loss": 1.5697,
+      "step": 3215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9314792889147746e-05,
+      "loss": 1.506,
+      "step": 3216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9314192883023114e-05,
+      "loss": 1.5832,
+      "step": 3217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93135926179671e-05,
+      "loss": 1.4943,
+      "step": 3218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931299209398609e-05,
+      "loss": 1.5181,
+      "step": 3219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9312391311086464e-05,
+      "loss": 1.5677,
+      "step": 3220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931179026927464e-05,
+      "loss": 1.5065,
+      "step": 3221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9311188968557e-05,
+      "loss": 1.5946,
+      "step": 3222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9310587408939965e-05,
+      "loss": 1.5283,
+      "step": 3223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.930998559042993e-05,
+      "loss": 1.5734,
+      "step": 3224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9309383513033314e-05,
+      "loss": 1.4858,
+      "step": 3225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.930878117675651e-05,
+      "loss": 1.5241,
+      "step": 3226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9308178581605954e-05,
+      "loss": 1.5176,
+      "step": 3227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9307575727588054e-05,
+      "loss": 1.5879,
+      "step": 3228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.930697261470922e-05,
+      "loss": 1.5383,
+      "step": 3229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9306369242975895e-05,
+      "loss": 1.5239,
+      "step": 3230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9305765612394496e-05,
+      "loss": 1.5884,
+      "step": 3231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9305161722971443e-05,
+      "loss": 1.5493,
+      "step": 3232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9304557574713174e-05,
+      "loss": 1.508,
+      "step": 3233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9303953167626125e-05,
+      "loss": 1.5389,
+      "step": 3234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930334850171673e-05,
+      "loss": 1.5324,
+      "step": 3235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9302743576991426e-05,
+      "loss": 1.5356,
+      "step": 3236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9302138393456664e-05,
+      "loss": 1.5086,
+      "step": 3237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930153295111888e-05,
+      "loss": 1.6035,
+      "step": 3238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930092724998452e-05,
+      "loss": 1.5526,
+      "step": 3239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930032129006004e-05,
+      "loss": 1.5079,
+      "step": 3240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9299715071351886e-05,
+      "loss": 1.5472,
+      "step": 3241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929910859386653e-05,
+      "loss": 1.6065,
+      "step": 3242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9298501857610415e-05,
+      "loss": 1.4963,
+      "step": 3243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929789486259e-05,
+      "loss": 1.5744,
+      "step": 3244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929728760881176e-05,
+      "loss": 1.5539,
+      "step": 3245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9296680096282164e-05,
+      "loss": 1.5637,
+      "step": 3246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929607232500767e-05,
+      "loss": 1.5693,
+      "step": 3247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9295464294994766e-05,
+      "loss": 1.5028,
+      "step": 3248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92948560062499e-05,
+      "loss": 1.5038,
+      "step": 3249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929424745877959e-05,
+      "loss": 1.5127,
+      "step": 3250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9293638652590275e-05,
+      "loss": 1.5521,
+      "step": 3251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9293029587688464e-05,
+      "loss": 1.5788,
+      "step": 3252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929242026408064e-05,
+      "loss": 1.5494,
+      "step": 3253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929181068177328e-05,
+      "loss": 1.5894,
+      "step": 3254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9291200840772886e-05,
+      "loss": 1.4835,
+      "step": 3255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9290590741085956e-05,
+      "loss": 1.5099,
+      "step": 3256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928998038271898e-05,
+      "loss": 1.5438,
+      "step": 3257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928936976567846e-05,
+      "loss": 1.5629,
+      "step": 3258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92887588899709e-05,
+      "loss": 1.5545,
+      "step": 3259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9288147755602797e-05,
+      "loss": 1.528,
+      "step": 3260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928753636258067e-05,
+      "loss": 1.554,
+      "step": 3261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928692471091103e-05,
+      "loss": 1.6149,
+      "step": 3262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928631280060038e-05,
+      "loss": 1.5253,
+      "step": 3263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928570063165525e-05,
+      "loss": 1.5462,
+      "step": 3264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928508820408215e-05,
+      "loss": 1.5628,
+      "step": 3265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92844755178876e-05,
+      "loss": 1.5413,
+      "step": 3266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928386257307813e-05,
+      "loss": 1.5331,
+      "step": 3267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928324936966027e-05,
+      "loss": 1.5317,
+      "step": 3268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928263590764055e-05,
+      "loss": 1.5262,
+      "step": 3269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928202218702549e-05,
+      "loss": 1.4823,
+      "step": 3270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928140820782164e-05,
+      "loss": 1.5044,
+      "step": 3271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9280793970035547e-05,
+      "loss": 1.5615,
+      "step": 3272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9280179473673725e-05,
+      "loss": 1.5212,
+      "step": 3273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927956471874273e-05,
+      "loss": 1.5176,
+      "step": 3274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927894970524912e-05,
+      "loss": 1.4966,
+      "step": 3275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9278334433199434e-05,
+      "loss": 1.5288,
+      "step": 3276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9277718902600226e-05,
+      "loss": 1.5407,
+      "step": 3277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927710311345805e-05,
+      "loss": 1.5596,
+      "step": 3278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927648706577947e-05,
+      "loss": 1.5859,
+      "step": 3279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927587075957103e-05,
+      "loss": 1.5842,
+      "step": 3280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9275254194839316e-05,
+      "loss": 1.5288,
+      "step": 3281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927463737159088e-05,
+      "loss": 1.5083,
+      "step": 3282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927402028983229e-05,
+      "loss": 1.5016,
+      "step": 3283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927340294957012e-05,
+      "loss": 1.5469,
+      "step": 3284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9272785350810946e-05,
+      "loss": 1.5672,
+      "step": 3285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927216749356135e-05,
+      "loss": 1.5225,
+      "step": 3286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9271549377827894e-05,
+      "loss": 1.5324,
+      "step": 3287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9270931003617184e-05,
+      "loss": 1.5908,
+      "step": 3288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9270312370935786e-05,
+      "loss": 1.6198,
+      "step": 3289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926969347979031e-05,
+      "loss": 1.5907,
+      "step": 3290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926907433018732e-05,
+      "loss": 1.5921,
+      "step": 3291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926845492213343e-05,
+      "loss": 1.5155,
+      "step": 3292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926783525563522e-05,
+      "loss": 1.4963,
+      "step": 3293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9267215330699304e-05,
+      "loss": 1.5269,
+      "step": 3294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926659514733228e-05,
+      "loss": 1.5114,
+      "step": 3295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926597470554074e-05,
+      "loss": 1.6582,
+      "step": 3296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926535400533132e-05,
+      "loss": 1.5723,
+      "step": 3297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92647330467106e-05,
+      "loss": 1.5674,
+      "step": 3298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92641118296852e-05,
+      "loss": 1.5788,
+      "step": 3299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9263490354261745e-05,
+      "loss": 1.5514,
+      "step": 3300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926286862044685e-05,
+      "loss": 1.625,
+      "step": 3301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926224662824713e-05,
+      "loss": 1.5451,
+      "step": 3302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926162437766922e-05,
+      "loss": 1.5627,
+      "step": 3303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926100186871974e-05,
+      "loss": 1.5384,
+      "step": 3304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926037910140531e-05,
+      "loss": 1.5534,
+      "step": 3305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925975607573258e-05,
+      "loss": 1.5513,
+      "step": 3306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9259132791708175e-05,
+      "loss": 1.5731,
+      "step": 3307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9258509249338724e-05,
+      "loss": 1.529,
+      "step": 3308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925788544863089e-05,
+      "loss": 1.4688,
+      "step": 3309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92572613895913e-05,
+      "loss": 1.4945,
+      "step": 3310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9256637072226594e-05,
+      "loss": 1.492,
+      "step": 3311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925601249654345e-05,
+      "loss": 1.5124,
+      "step": 3312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925538766254848e-05,
+      "loss": 1.5085,
+      "step": 3313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925476257024836e-05,
+      "loss": 1.5173,
+      "step": 3314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925413721964975e-05,
+      "loss": 1.5592,
+      "step": 3315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925351161075931e-05,
+      "loss": 1.507,
+      "step": 3316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925288574358368e-05,
+      "loss": 1.4984,
+      "step": 3317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925225961812955e-05,
+      "loss": 1.6012,
+      "step": 3318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925163323440358e-05,
+      "loss": 1.52,
+      "step": 3319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925100659241243e-05,
+      "loss": 1.5968,
+      "step": 3320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925037969216279e-05,
+      "loss": 1.4461,
+      "step": 3321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924975253366133e-05,
+      "loss": 1.5614,
+      "step": 3322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924912511691473e-05,
+      "loss": 1.5441,
+      "step": 3323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924849744192966e-05,
+      "loss": 1.6478,
+      "step": 3324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924786950871283e-05,
+      "loss": 1.5563,
+      "step": 3325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92472413172709e-05,
+      "loss": 1.6322,
+      "step": 3326
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924661286761057e-05,
+      "loss": 1.4868,
+      "step": 3327
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9245984159738534e-05,
+      "loss": 1.4828,
+      "step": 3328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924535519366149e-05,
+      "loss": 1.6361,
+      "step": 3329
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924472596938613e-05,
+      "loss": 1.5874,
+      "step": 3330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924409648691916e-05,
+      "loss": 1.5594,
+      "step": 3331
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924346674626728e-05,
+      "loss": 1.5415,
+      "step": 3332
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92428367474372e-05,
+      "loss": 1.5247,
+      "step": 3333
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9242206490435625e-05,
+      "loss": 1.5037,
+      "step": 3334
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9241575975269275e-05,
+      "loss": 1.5267,
+      "step": 3335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9240945201944846e-05,
+      "loss": 1.5939,
+      "step": 3336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924031417046907e-05,
+      "loss": 1.547,
+      "step": 3337
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923968288084867e-05,
+      "loss": 1.52,
+      "step": 3338
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923905133309036e-05,
+      "loss": 1.5458,
+      "step": 3339
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9238419527200873e-05,
+      "loss": 1.5132,
+      "step": 3340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923778746318693e-05,
+      "loss": 1.4404,
+      "step": 3341
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923715514105527e-05,
+      "loss": 1.5262,
+      "step": 3342
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923652256081262e-05,
+      "loss": 1.6297,
+      "step": 3343
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9235889722465725e-05,
+      "loss": 1.599,
+      "step": 3344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923525662602131e-05,
+      "loss": 1.5557,
+      "step": 3345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923462327148612e-05,
+      "loss": 1.5073,
+      "step": 3346
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923398965886692e-05,
+      "loss": 1.5552,
+      "step": 3347
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923335578817043e-05,
+      "loss": 1.4857,
+      "step": 3348
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9232721659403424e-05,
+      "loss": 1.5343,
+      "step": 3349
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923208727257264e-05,
+      "loss": 1.4774,
+      "step": 3350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9231452627684834e-05,
+      "loss": 1.506,
+      "step": 3351
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923081772474678e-05,
+      "loss": 1.5103,
+      "step": 3352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923018256376521e-05,
+      "loss": 1.5023,
+      "step": 3353
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.922954714474692e-05,
+      "loss": 1.5645,
+      "step": 3354
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.922891146769867e-05,
+      "loss": 1.5062,
+      "step": 3355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.922827553262721e-05,
+      "loss": 1.5674,
+      "step": 3356
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.922763933953932e-05,
+      "loss": 1.5088,
+      "step": 3357
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.922700288844179e-05,
+      "loss": 1.4331,
+      "step": 3358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9226366179341385e-05,
+      "loss": 1.571,
+      "step": 3359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922572921224489e-05,
+      "loss": 1.5384,
+      "step": 3360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922509198715909e-05,
+      "loss": 1.6045,
+      "step": 3361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9224454504090764e-05,
+      "loss": 1.6408,
+      "step": 3362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.92238167630467e-05,
+      "loss": 1.4928,
+      "step": 3363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922317876403369e-05,
+      "loss": 1.5474,
+      "step": 3364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922254050705855e-05,
+      "loss": 1.5715,
+      "step": 3365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922190199212805e-05,
+      "loss": 1.5085,
+      "step": 3366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9221263219249e-05,
+      "loss": 1.5037,
+      "step": 3367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.92206241884282e-05,
+      "loss": 1.507,
+      "step": 3368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9219984899672456e-05,
+      "loss": 1.5129,
+      "step": 3369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9219345352988575e-05,
+      "loss": 1.5298,
+      "step": 3370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9218705548383385e-05,
+      "loss": 1.6235,
+      "step": 3371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921806548586367e-05,
+      "loss": 1.5369,
+      "step": 3372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921742516543626e-05,
+      "loss": 1.5625,
+      "step": 3373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921678458710798e-05,
+      "loss": 1.5526,
+      "step": 3374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921614375088564e-05,
+      "loss": 1.5098,
+      "step": 3375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921550265677608e-05,
+      "loss": 1.4622,
+      "step": 3376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921486130478612e-05,
+      "loss": 1.4994,
+      "step": 3377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921421969492258e-05,
+      "loss": 1.5166,
+      "step": 3378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.92135778271923e-05,
+      "loss": 1.5304,
+      "step": 3379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921293570160213e-05,
+      "loss": 1.535,
+      "step": 3380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921229331815888e-05,
+      "loss": 1.5261,
+      "step": 3381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9211650676869414e-05,
+      "loss": 1.6126,
+      "step": 3382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9211007777740556e-05,
+      "loss": 1.4762,
+      "step": 3383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921036462077918e-05,
+      "loss": 1.5283,
+      "step": 3384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920972120599211e-05,
+      "loss": 1.612,
+      "step": 3385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920907753338621e-05,
+      "loss": 1.4819,
+      "step": 3386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920843360296833e-05,
+      "loss": 1.4483,
+      "step": 3387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920778941474533e-05,
+      "loss": 1.5046,
+      "step": 3388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9207144968724076e-05,
+      "loss": 1.5072,
+      "step": 3389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9206500264911415e-05,
+      "loss": 1.4483,
+      "step": 3390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920585530331423e-05,
+      "loss": 1.5073,
+      "step": 3391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9205210083939374e-05,
+      "loss": 1.5892,
+      "step": 3392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920456460679373e-05,
+      "loss": 1.5374,
+      "step": 3393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920391887188417e-05,
+      "loss": 1.5423,
+      "step": 3394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9203272879217564e-05,
+      "loss": 1.5308,
+      "step": 3395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.92026266288008e-05,
+      "loss": 1.6107,
+      "step": 3396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9201980120640754e-05,
+      "loss": 1.4393,
+      "step": 3397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920133335474432e-05,
+      "loss": 1.5793,
+      "step": 3398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920068633111836e-05,
+      "loss": 1.5435,
+      "step": 3399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9200039049769796e-05,
+      "loss": 1.5313,
+      "step": 3400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9199391510705516e-05,
+      "loss": 1.4945,
+      "step": 3401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.91987437139324e-05,
+      "loss": 1.589,
+      "step": 3402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919809565945736e-05,
+      "loss": 1.4881,
+      "step": 3403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9197447347287275e-05,
+      "loss": 1.479,
+      "step": 3404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919679877742908e-05,
+      "loss": 1.4608,
+      "step": 3405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919614994988967e-05,
+      "loss": 1.4741,
+      "step": 3406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9195500864675946e-05,
+      "loss": 1.5188,
+      "step": 3407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919485152179484e-05,
+      "loss": 1.4735,
+      "step": 3408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919420192125324e-05,
+      "loss": 1.5625,
+      "step": 3409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919355206305808e-05,
+      "loss": 1.5125,
+      "step": 3410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919290194721629e-05,
+      "loss": 1.6034,
+      "step": 3411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919225157373477e-05,
+      "loss": 1.5791,
+      "step": 3412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9191600942620465e-05,
+      "loss": 1.5412,
+      "step": 3413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9190950053880295e-05,
+      "loss": 1.5335,
+      "step": 3414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919029890752119e-05,
+      "loss": 1.5023,
+      "step": 3415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.91896475035501e-05,
+      "loss": 1.4815,
+      "step": 3416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.918899584197394e-05,
+      "loss": 1.5173,
+      "step": 3417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9188343922799654e-05,
+      "loss": 1.5485,
+      "step": 3418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.91876917460342e-05,
+      "loss": 1.5425,
+      "step": 3419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9187039311684504e-05,
+      "loss": 1.5384,
+      "step": 3420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918638661975754e-05,
+      "loss": 1.5319,
+      "step": 3421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918573367026022e-05,
+      "loss": 1.4564,
+      "step": 3422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918508046319954e-05,
+      "loss": 1.6089,
+      "step": 3423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9184426998582436e-05,
+      "loss": 1.4982,
+      "step": 3424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9183773276415854e-05,
+      "loss": 1.5503,
+      "step": 3425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918311929670678e-05,
+      "loss": 1.5054,
+      "step": 3426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9182465059462156e-05,
+      "loss": 1.5584,
+      "step": 3427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9181810564688966e-05,
+      "loss": 1.5595,
+      "step": 3428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918115581239418e-05,
+      "loss": 1.5015,
+      "step": 3429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918050080258476e-05,
+      "loss": 1.5267,
+      "step": 3430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917984553526769e-05,
+      "loss": 1.5426,
+      "step": 3431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917919001044994e-05,
+      "loss": 1.5684,
+      "step": 3432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.91785342281385e-05,
+      "loss": 1.5381,
+      "step": 3433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9177878188340346e-05,
+      "loss": 1.5814,
+      "step": 3434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917722189106248e-05,
+      "loss": 1.5304,
+      "step": 3435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917656533631187e-05,
+      "loss": 1.4631,
+      "step": 3436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917590852409551e-05,
+      "loss": 1.5413,
+      "step": 3437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9175251454420406e-05,
+      "loss": 1.5506,
+      "step": 3438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917459412729355e-05,
+      "loss": 1.5275,
+      "step": 3439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917393654272195e-05,
+      "loss": 1.4969,
+      "step": 3440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.91732787007126e-05,
+      "loss": 1.5723,
+      "step": 3441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917262060127251e-05,
+      "loss": 1.5217,
+      "step": 3442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917196224440868e-05,
+      "loss": 1.5122,
+      "step": 3443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917130363012813e-05,
+      "loss": 1.5356,
+      "step": 3444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9170644758437875e-05,
+      "loss": 1.5356,
+      "step": 3445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916998562934493e-05,
+      "loss": 1.5568,
+      "step": 3446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.91693262428563e-05,
+      "loss": 1.5573,
+      "step": 3447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916866659897903e-05,
+      "loss": 1.486,
+      "step": 3448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9168006697720125e-05,
+      "loss": 1.5049,
+      "step": 3449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916734653908662e-05,
+      "loss": 1.5322,
+      "step": 3450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9166686123085556e-05,
+      "loss": 1.5798,
+      "step": 3451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916602544972395e-05,
+      "loss": 1.5855,
+      "step": 3452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916536451900885e-05,
+      "loss": 1.533,
+      "step": 3453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9164703330947294e-05,
+      "loss": 1.4865,
+      "step": 3454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9164041885546307e-05,
+      "loss": 1.5885,
+      "step": 3455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916338018281295e-05,
+      "loss": 1.542,
+      "step": 3456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9162718222754255e-05,
+      "loss": 1.52,
+      "step": 3457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916205600537729e-05,
+      "loss": 1.5832,
+      "step": 3458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.91613935306891e-05,
+      "loss": 1.5547,
+      "step": 3459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916073079869673e-05,
+      "loss": 1.5356,
+      "step": 3460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916006780940725e-05,
+      "loss": 1.5171,
+      "step": 3461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915940456282772e-05,
+      "loss": 1.5783,
+      "step": 3462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915874105896519e-05,
+      "loss": 1.5773,
+      "step": 3463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9158077297826735e-05,
+      "loss": 1.5609,
+      "step": 3464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9157413279419425e-05,
+      "loss": 1.5033,
+      "step": 3465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915674900375033e-05,
+      "loss": 1.5369,
+      "step": 3466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915608447082652e-05,
+      "loss": 1.5828,
+      "step": 3467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915541968065509e-05,
+      "loss": 1.5776,
+      "step": 3468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915475463324309e-05,
+      "loss": 1.4852,
+      "step": 3469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915408932859763e-05,
+      "loss": 1.5244,
+      "step": 3470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915342376672577e-05,
+      "loss": 1.5239,
+      "step": 3471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915275794763462e-05,
+      "loss": 1.5402,
+      "step": 3472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915209187133126e-05,
+      "loss": 1.6081,
+      "step": 3473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9151425537822784e-05,
+      "loss": 1.5529,
+      "step": 3474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9150758947116296e-05,
+      "loss": 1.498,
+      "step": 3475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915009209921888e-05,
+      "loss": 1.5186,
+      "step": 3476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.914942499413765e-05,
+      "loss": 1.5313,
+      "step": 3477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9148757631879696e-05,
+      "loss": 1.5418,
+      "step": 3478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9148090012452134e-05,
+      "loss": 1.5272,
+      "step": 3479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.914742213586208e-05,
+      "loss": 1.5469,
+      "step": 3480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.914675400211664e-05,
+      "loss": 1.4826,
+      "step": 3481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.914608561122293e-05,
+      "loss": 1.5008,
+      "step": 3482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.914541696318807e-05,
+      "loss": 1.5101,
+      "step": 3483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914474805801918e-05,
+      "loss": 1.4445,
+      "step": 3484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9144078895723365e-05,
+      "loss": 1.4878,
+      "step": 3485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914340947630779e-05,
+      "loss": 1.5212,
+      "step": 3486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914273979977955e-05,
+      "loss": 1.5343,
+      "step": 3487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914206986614579e-05,
+      "loss": 1.4775,
+      "step": 3488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914139967541364e-05,
+      "loss": 1.5422,
+      "step": 3489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9140729227590246e-05,
+      "loss": 1.5081,
+      "step": 3490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914005852268274e-05,
+      "loss": 1.4535,
+      "step": 3491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913938756069826e-05,
+      "loss": 1.4565,
+      "step": 3492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913871634164397e-05,
+      "loss": 1.4896,
+      "step": 3493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9138044865527e-05,
+      "loss": 1.5285,
+      "step": 3494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.91373731323545e-05,
+      "loss": 1.4751,
+      "step": 3495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913670114213364e-05,
+      "loss": 1.4396,
+      "step": 3496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913602889487157e-05,
+      "loss": 1.5817,
+      "step": 3497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913535639057544e-05,
+      "loss": 1.5483,
+      "step": 3498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913468362925241e-05,
+      "loss": 1.627,
+      "step": 3499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913401061090966e-05,
+      "loss": 1.4071,
+      "step": 3500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9133337335554346e-05,
+      "loss": 1.4924,
+      "step": 3501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913266380319364e-05,
+      "loss": 1.6407,
+      "step": 3502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913199001383472e-05,
+      "loss": 1.5736,
+      "step": 3503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913131596748476e-05,
+      "loss": 1.5745,
+      "step": 3504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913064166415093e-05,
+      "loss": 1.5022,
+      "step": 3505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9129967103840414e-05,
+      "loss": 1.5531,
+      "step": 3506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.91292922865604e-05,
+      "loss": 1.535,
+      "step": 3507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9128617212318074e-05,
+      "loss": 1.5233,
+      "step": 3508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912794188112063e-05,
+      "loss": 1.4761,
+      "step": 3509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912726629297524e-05,
+      "loss": 1.5236,
+      "step": 3510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912659044788912e-05,
+      "loss": 1.5457,
+      "step": 3511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912591434586946e-05,
+      "loss": 1.5179,
+      "step": 3512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912523798692346e-05,
+      "loss": 1.5516,
+      "step": 3513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912456137105832e-05,
+      "loss": 1.5246,
+      "step": 3514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912388449828124e-05,
+      "loss": 1.5552,
+      "step": 3515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912320736859944e-05,
+      "loss": 1.6052,
+      "step": 3516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912252998202014e-05,
+      "loss": 1.5368,
+      "step": 3517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912185233855053e-05,
+      "loss": 1.4419,
+      "step": 3518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9121174438197834e-05,
+      "loss": 1.5285,
+      "step": 3519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9120496280969286e-05,
+      "loss": 1.4937,
+      "step": 3520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911981786687209e-05,
+      "loss": 1.4883,
+      "step": 3521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911913919591347e-05,
+      "loss": 1.5597,
+      "step": 3522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911846026810068e-05,
+      "loss": 1.4826,
+      "step": 3523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9117781083440905e-05,
+      "loss": 1.451,
+      "step": 3524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911710164194142e-05,
+      "loss": 1.5635,
+      "step": 3525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911642194360944e-05,
+      "loss": 1.5793,
+      "step": 3526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911574198845221e-05,
+      "loss": 1.5038,
+      "step": 3527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9115061776476966e-05,
+      "loss": 1.5381,
+      "step": 3528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911438130769096e-05,
+      "loss": 1.4762,
+      "step": 3529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911370058210143e-05,
+      "loss": 1.516,
+      "step": 3530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9113019599715624e-05,
+      "loss": 1.5093,
+      "step": 3531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911233836054081e-05,
+      "loss": 1.5037,
+      "step": 3532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911165686458422e-05,
+      "loss": 1.5702,
+      "step": 3533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911097511185313e-05,
+      "loss": 1.5169,
+      "step": 3534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9110293102354794e-05,
+      "loss": 1.6195,
+      "step": 3535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9109610836096473e-05,
+      "loss": 1.5609,
+      "step": 3536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910892831308543e-05,
+      "loss": 1.5077,
+      "step": 3537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910824553332894e-05,
+      "loss": 1.4558,
+      "step": 3538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910756249683427e-05,
+      "loss": 1.4668,
+      "step": 3539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9106879203608694e-05,
+      "loss": 1.4867,
+      "step": 3540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910619565365949e-05,
+      "loss": 1.6364,
+      "step": 3541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9105511846993946e-05,
+      "loss": 1.5658,
+      "step": 3542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910482778361932e-05,
+      "loss": 1.5007,
+      "step": 3543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910414346354293e-05,
+      "loss": 1.5755,
+      "step": 3544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9103458886772033e-05,
+      "loss": 1.5286,
+      "step": 3545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910277405331394e-05,
+      "loss": 1.5584,
+      "step": 3546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9102088963175936e-05,
+      "loss": 1.4797,
+      "step": 3547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9101403616365315e-05,
+      "loss": 1.5579,
+      "step": 3548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.910071801288938e-05,
+      "loss": 1.5496,
+      "step": 3549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.910003215275543e-05,
+      "loss": 1.5431,
+      "step": 3550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9099346035970764e-05,
+      "loss": 1.5327,
+      "step": 3551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909865966254269e-05,
+      "loss": 1.5407,
+      "step": 3552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909797303247853e-05,
+      "loss": 1.5658,
+      "step": 3553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9097286145785595e-05,
+      "loss": 1.5438,
+      "step": 3554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909659900247118e-05,
+      "loss": 1.5815,
+      "step": 3555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9095911602542615e-05,
+      "loss": 1.5632,
+      "step": 3556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909522394600722e-05,
+      "loss": 1.526,
+      "step": 3557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9094536032872326e-05,
+      "loss": 1.4515,
+      "step": 3558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909384786314525e-05,
+      "loss": 1.5732,
+      "step": 3559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9093159436833316e-05,
+      "loss": 1.4987,
+      "step": 3560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9092470753943866e-05,
+      "loss": 1.53,
+      "step": 3561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909178181448423e-05,
+      "loss": 1.4792,
+      "step": 3562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909109261846174e-05,
+      "loss": 1.5168,
+      "step": 3563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909040316588374e-05,
+      "loss": 1.4813,
+      "step": 3564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908971345675758e-05,
+      "loss": 1.486,
+      "step": 3565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908902349109059e-05,
+      "loss": 1.4875,
+      "step": 3566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908833326889012e-05,
+      "loss": 1.5088,
+      "step": 3567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908764279016354e-05,
+      "loss": 1.4977,
+      "step": 3568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908695205491817e-05,
+      "loss": 1.5703,
+      "step": 3569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9086261063161395e-05,
+      "loss": 1.5921,
+      "step": 3570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9085569814900554e-05,
+      "loss": 1.5264,
+      "step": 3571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9084878310143026e-05,
+      "loss": 1.5073,
+      "step": 3572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908418654889617e-05,
+      "loss": 1.5233,
+      "step": 3573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908349453116734e-05,
+      "loss": 1.6053,
+      "step": 3574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9082802256963913e-05,
+      "loss": 1.5371,
+      "step": 3575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908210972629327e-05,
+      "loss": 1.5112,
+      "step": 3576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908141693916278e-05,
+      "loss": 1.547,
+      "step": 3577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9080723895579816e-05,
+      "loss": 1.5636,
+      "step": 3578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9080030595551765e-05,
+      "loss": 1.5768,
+      "step": 3579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907933703908601e-05,
+      "loss": 1.5443,
+      "step": 3580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907864322618992e-05,
+      "loss": 1.5518,
+      "step": 3581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907794915687091e-05,
+      "loss": 1.554,
+      "step": 3582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907725483113636e-05,
+      "loss": 1.5697,
+      "step": 3583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9076560248993664e-05,
+      "loss": 1.5334,
+      "step": 3584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907586541045022e-05,
+      "loss": 1.5524,
+      "step": 3585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907517031551342e-05,
+      "loss": 1.5532,
+      "step": 3586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907447496419067e-05,
+      "loss": 1.5763,
+      "step": 3587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907377935648938e-05,
+      "loss": 1.5687,
+      "step": 3588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907308349241696e-05,
+      "loss": 1.5075,
+      "step": 3589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907238737198081e-05,
+      "loss": 1.5633,
+      "step": 3590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907169099518836e-05,
+      "loss": 1.5519,
+      "step": 3591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907099436204701e-05,
+      "loss": 1.4784,
+      "step": 3592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9070297472564176e-05,
+      "loss": 1.5081,
+      "step": 3593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906960032674729e-05,
+      "loss": 1.5361,
+      "step": 3594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9068902924603777e-05,
+      "loss": 1.5127,
+      "step": 3595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9068205266141065e-05,
+      "loss": 1.5282,
+      "step": 3596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906750735136657e-05,
+      "loss": 1.5213,
+      "step": 3597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9066809180287745e-05,
+      "loss": 1.5578,
+      "step": 3598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9066110752911996e-05,
+      "loss": 1.4959,
+      "step": 3599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.90654120692468e-05,
+      "loss": 1.5269,
+      "step": 3600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906471312929955e-05,
+      "loss": 1.518,
+      "step": 3601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906401393307774e-05,
+      "loss": 1.4965,
+      "step": 3602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906331448058878e-05,
+      "loss": 1.5217,
+      "step": 3603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906261477184013e-05,
+      "loss": 1.5037,
+      "step": 3604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906191480683924e-05,
+      "loss": 1.5015,
+      "step": 3605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9061214585593574e-05,
+      "loss": 1.5508,
+      "step": 3606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906051410811057e-05,
+      "loss": 1.5519,
+      "step": 3607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.90598133743977e-05,
+      "loss": 1.4463,
+      "step": 3608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.905911238446243e-05,
+      "loss": 1.5931,
+      "step": 3609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905841113831222e-05,
+      "loss": 1.5731,
+      "step": 3610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9057709635954526e-05,
+      "loss": 1.5576,
+      "step": 3611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905700787739684e-05,
+      "loss": 1.4689,
+      "step": 3612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905630586264663e-05,
+      "loss": 1.5083,
+      "step": 3613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9055603591711365e-05,
+      "loss": 1.5374,
+      "step": 3614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905490106459853e-05,
+      "loss": 1.5783,
+      "step": 3615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905419828131559e-05,
+      "loss": 1.5194,
+      "step": 3616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905349524187005e-05,
+      "loss": 1.5881,
+      "step": 3617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9052791946269384e-05,
+      "loss": 1.5259,
+      "step": 3618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905208839452109e-05,
+      "loss": 1.596,
+      "step": 3619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9051384586632656e-05,
+      "loss": 1.5996,
+      "step": 3620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9050680522611576e-05,
+      "loss": 1.5202,
+      "step": 3621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904997620246535e-05,
+      "loss": 1.5249,
+      "step": 3622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904927162620148e-05,
+      "loss": 1.5275,
+      "step": 3623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904856679382747e-05,
+      "loss": 1.5439,
+      "step": 3624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9047861705350825e-05,
+      "loss": 1.5007,
+      "step": 3625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9047156360779054e-05,
+      "loss": 1.5148,
+      "step": 3626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904645076011966e-05,
+      "loss": 1.5177,
+      "step": 3627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9045744903380165e-05,
+      "loss": 1.5874,
+      "step": 3628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904503879056809e-05,
+      "loss": 1.5011,
+      "step": 3629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904433242169095e-05,
+      "loss": 1.5446,
+      "step": 3630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9043625796756264e-05,
+      "loss": 1.5475,
+      "step": 3631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904291891577156e-05,
+      "loss": 1.5788,
+      "step": 3632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904221177874437e-05,
+      "loss": 1.5381,
+      "step": 3633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904150438568222e-05,
+      "loss": 1.4889,
+      "step": 3634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904079673659264e-05,
+      "loss": 1.5122,
+      "step": 3635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9040088831483176e-05,
+      "loss": 1.5029,
+      "step": 3636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903938067036136e-05,
+      "loss": 1.5249,
+      "step": 3637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903867225323473e-05,
+      "loss": 1.5259,
+      "step": 3638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903796358011084e-05,
+      "loss": 1.5146,
+      "step": 3639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9037254650997234e-05,
+      "loss": 1.481,
+      "step": 3640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9036545465901454e-05,
+      "loss": 1.4909,
+      "step": 3641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903583602483106e-05,
+      "loss": 1.5145,
+      "step": 3642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9035126327793604e-05,
+      "loss": 1.487,
+      "step": 3643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9034416374796646e-05,
+      "loss": 1.5736,
+      "step": 3644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903370616584774e-05,
+      "loss": 1.5568,
+      "step": 3645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9032995700954454e-05,
+      "loss": 1.5689,
+      "step": 3646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9032284980124366e-05,
+      "loss": 1.4702,
+      "step": 3647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903157400336502e-05,
+      "loss": 1.599,
+      "step": 3648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903086277068401e-05,
+      "loss": 1.5005,
+      "step": 3649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9030151282088896e-05,
+      "loss": 1.4967,
+      "step": 3650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902943953758726e-05,
+      "loss": 1.5596,
+      "step": 3651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902872753718668e-05,
+      "loss": 1.5575,
+      "step": 3652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9028015280894746e-05,
+      "loss": 1.5407,
+      "step": 3653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902730276871904e-05,
+      "loss": 1.5049,
+      "step": 3654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902659000066714e-05,
+      "loss": 1.4945,
+      "step": 3655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902587697674664e-05,
+      "loss": 1.5104,
+      "step": 3656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902516369696515e-05,
+      "loss": 1.5731,
+      "step": 3657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902445016133024e-05,
+      "loss": 1.5798,
+      "step": 3658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9023736369849525e-05,
+      "loss": 1.5604,
+      "step": 3659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.90230223225306e-05,
+      "loss": 1.499,
+      "step": 3660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902230801938108e-05,
+      "loss": 1.528,
+      "step": 3661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902159346040855e-05,
+      "loss": 1.5479,
+      "step": 3662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902087864562065e-05,
+      "loss": 1.5635,
+      "step": 3663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9020163575024966e-05,
+      "loss": 1.5122,
+      "step": 3664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9019448248629125e-05,
+      "loss": 1.5026,
+      "step": 3665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9018732666440745e-05,
+      "loss": 1.499,
+      "step": 3666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9018016828467436e-05,
+      "loss": 1.5628,
+      "step": 3667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.901730073471684e-05,
+      "loss": 1.625,
+      "step": 3668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9016584385196565e-05,
+      "loss": 1.5666,
+      "step": 3669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9015867779914256e-05,
+      "loss": 1.5856,
+      "step": 3670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.901515091887753e-05,
+      "loss": 1.527,
+      "step": 3671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.901443380209402e-05,
+      "loss": 1.57,
+      "step": 3672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9013716429571375e-05,
+      "loss": 1.5477,
+      "step": 3673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901299880131723e-05,
+      "loss": 1.6076,
+      "step": 3674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9012280917339233e-05,
+      "loss": 1.5762,
+      "step": 3675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901156277764502e-05,
+      "loss": 1.5275,
+      "step": 3676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901084438224223e-05,
+      "loss": 1.5246,
+      "step": 3677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9010125731138534e-05,
+      "loss": 1.5127,
+      "step": 3678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9009406824341573e-05,
+      "loss": 1.5055,
+      "step": 3679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9008687661859004e-05,
+      "loss": 1.5688,
+      "step": 3680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900796824369849e-05,
+      "loss": 1.5653,
+      "step": 3681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900724856986769e-05,
+      "loss": 1.5505,
+      "step": 3682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900652864037427e-05,
+      "loss": 1.5671,
+      "step": 3683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900580845522589e-05,
+      "loss": 1.5549,
+      "step": 3684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9005088014430226e-05,
+      "loss": 1.4974,
+      "step": 3685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900436731799495e-05,
+      "loss": 1.5356,
+      "step": 3686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900364636592774e-05,
+      "loss": 1.6123,
+      "step": 3687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900292515823626e-05,
+      "loss": 1.4766,
+      "step": 3688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.90022036949282e-05,
+      "loss": 1.5166,
+      "step": 3689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900148197601124e-05,
+      "loss": 1.4969,
+      "step": 3690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9000760001493086e-05,
+      "loss": 1.5326,
+      "step": 3691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900003777138139e-05,
+      "loss": 1.4714,
+      "step": 3692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8999315285683875e-05,
+      "loss": 1.508,
+      "step": 3693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8998592544408214e-05,
+      "loss": 1.5877,
+      "step": 3694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8997869547562113e-05,
+      "loss": 1.5791,
+      "step": 3695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899714629515328e-05,
+      "loss": 1.4924,
+      "step": 3696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.89964227871894e-05,
+      "loss": 1.4684,
+      "step": 3697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899569902367819e-05,
+      "loss": 1.5729,
+      "step": 3698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8994975004627354e-05,
+      "loss": 1.4771,
+      "step": 3699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8994250730044595e-05,
+      "loss": 1.5278,
+      "step": 3700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899352619993764e-05,
+      "loss": 1.596,
+      "step": 3701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899280141431419e-05,
+      "loss": 1.5552,
+      "step": 3702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899207637318198e-05,
+      "loss": 1.5262,
+      "step": 3703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899135107654872e-05,
+      "loss": 1.4888,
+      "step": 3704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8990625524422126e-05,
+      "loss": 1.4795,
+      "step": 3705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8989899716809953e-05,
+      "loss": 1.5199,
+      "step": 3706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.89891736537199e-05,
+      "loss": 1.584,
+      "step": 3707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898844733515972e-05,
+      "loss": 1.5526,
+      "step": 3708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898772076113713e-05,
+      "loss": 1.5049,
+      "step": 3709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8986993931659885e-05,
+      "loss": 1.4777,
+      "step": 3710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8986266846735716e-05,
+      "loss": 1.4601,
+      "step": 3711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898553950637237e-05,
+      "loss": 1.5967,
+      "step": 3712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8984811910577585e-05,
+      "loss": 1.5301,
+      "step": 3713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8984084059359114e-05,
+      "loss": 1.5218,
+      "step": 3714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8983355952724715e-05,
+      "loss": 1.5768,
+      "step": 3715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898262759068213e-05,
+      "loss": 1.5508,
+      "step": 3716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898189897323913e-05,
+      "loss": 1.5547,
+      "step": 3717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898117010040345e-05,
+      "loss": 1.4857,
+      "step": 3718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8980440972182884e-05,
+      "loss": 1.6054,
+      "step": 3719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897971158858517e-05,
+      "loss": 1.4764,
+      "step": 3720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8978981949618084e-05,
+      "loss": 1.5772,
+      "step": 3721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897825205528941e-05,
+      "loss": 1.5488,
+      "step": 3722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.89775219056069e-05,
+      "loss": 1.5646,
+      "step": 3723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897679150057834e-05,
+      "loss": 1.5472,
+      "step": 3724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897606084021151e-05,
+      "loss": 1.5029,
+      "step": 3725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897532992451419e-05,
+      "loss": 1.5233,
+      "step": 3726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897459875349416e-05,
+      "loss": 1.5334,
+      "step": 3727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8973867327159203e-05,
+      "loss": 1.5514,
+      "step": 3728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8973135645517124e-05,
+      "loss": 1.5116,
+      "step": 3729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8972403708575695e-05,
+      "loss": 1.4754,
+      "step": 3730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897167151634273e-05,
+      "loss": 1.6024,
+      "step": 3731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897093906882602e-05,
+      "loss": 1.5443,
+      "step": 3732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8970206366033346e-05,
+      "loss": 1.4927,
+      "step": 3733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.896947340797253e-05,
+      "loss": 1.5171,
+      "step": 3734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8968740194651384e-05,
+      "loss": 1.5589,
+      "step": 3735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8968006726077705e-05,
+      "loss": 1.4964,
+      "step": 3736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.89672730022593e-05,
+      "loss": 1.5396,
+      "step": 3737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896653902320399e-05,
+      "loss": 1.541,
+      "step": 3738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896580478891959e-05,
+      "loss": 1.4969,
+      "step": 3739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8965070299413924e-05,
+      "loss": 1.5552,
+      "step": 3740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896433555469481e-05,
+      "loss": 1.4842,
+      "step": 3741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896360055477006e-05,
+      "loss": 1.576,
+      "step": 3742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896286529964752e-05,
+      "loss": 1.4826,
+      "step": 3743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896212978933501e-05,
+      "loss": 1.5063,
+      "step": 3744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896139402384037e-05,
+      "loss": 1.4447,
+      "step": 3745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8960658003171426e-05,
+      "loss": 1.4608,
+      "step": 3746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895992172733602e-05,
+      "loss": 1.5682,
+      "step": 3747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8959185196342e-05,
+      "loss": 1.4439,
+      "step": 3748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.89584484101972e-05,
+      "loss": 1.5207,
+      "step": 3749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8957711368909476e-05,
+      "loss": 1.5742,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895697407248666e-05,
+      "loss": 1.6318,
+      "step": 3751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895623652093663e-05,
+      "loss": 1.564,
+      "step": 3752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895549871426721e-05,
+      "loss": 1.5671,
+      "step": 3753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8954760652486276e-05,
+      "loss": 1.5283,
+      "step": 3754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895402233560169e-05,
+      "loss": 1.5057,
+      "step": 3755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895328376362131e-05,
+      "loss": 1.9413,
+      "step": 3756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895254493655299e-05,
+      "loss": 1.5426,
+      "step": 3757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895180585440462e-05,
+      "loss": 1.5291,
+      "step": 3758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8951066517184054e-05,
+      "loss": 1.494,
+      "step": 3759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895032692489917e-05,
+      "loss": 1.4373,
+      "step": 3760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894958707755785e-05,
+      "loss": 1.537,
+      "step": 3761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894884697516796e-05,
+      "loss": 1.4955,
+      "step": 3762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894810661773739e-05,
+      "loss": 1.4699,
+      "step": 3763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894736600527402e-05,
+      "loss": 1.529,
+      "step": 3764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894662513778575e-05,
+      "loss": 1.4813,
+      "step": 3765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894588401528046e-05,
+      "loss": 1.4899,
+      "step": 3766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894514263776604e-05,
+      "loss": 1.5233,
+      "step": 3767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8944401005250386e-05,
+      "loss": 1.4976,
+      "step": 3768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.89436591177414e-05,
+      "loss": 1.5991,
+      "step": 3769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894291697524699e-05,
+      "loss": 1.5964,
+      "step": 3770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8942174577775033e-05,
+      "loss": 1.5226,
+      "step": 3771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894143192533347e-05,
+      "loss": 1.5244,
+      "step": 3772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8940689017930184e-05,
+      "loss": 1.5658,
+      "step": 3773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893994585557309e-05,
+      "loss": 1.4845,
+      "step": 3774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893920243827011e-05,
+      "loss": 1.5226,
+      "step": 3775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893845876602916e-05,
+      "loss": 1.5409,
+      "step": 3776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893771483885816e-05,
+      "loss": 1.5251,
+      "step": 3777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893697065676502e-05,
+      "loss": 1.5404,
+      "step": 3778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893622621975769e-05,
+      "loss": 1.5757,
+      "step": 3779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893548152784407e-05,
+      "loss": 1.5132,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893473658103211e-05,
+      "loss": 1.4901,
+      "step": 3781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893399137932974e-05,
+      "loss": 1.4383,
+      "step": 3782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893324592274488e-05,
+      "loss": 1.5187,
+      "step": 3783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893250021128549e-05,
+      "loss": 1.5495,
+      "step": 3784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.89317542449595e-05,
+      "loss": 1.4907,
+      "step": 3785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893100802377486e-05,
+      "loss": 1.4455,
+      "step": 3786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893026154773951e-05,
+      "loss": 1.5607,
+      "step": 3787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892951481686141e-05,
+      "loss": 1.5197,
+      "step": 3788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892876783114849e-05,
+      "loss": 1.5448,
+      "step": 3789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8928020590608736e-05,
+      "loss": 1.592,
+      "step": 3790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892727309525008e-05,
+      "loss": 1.488,
+      "step": 3791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.89265253450805e-05,
+      "loss": 1.5946,
+      "step": 3792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892577734010794e-05,
+      "loss": 1.5457,
+      "step": 3793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8925029080340387e-05,
+      "loss": 1.5646,
+      "step": 3794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8924280565785795e-05,
+      "loss": 1.5015,
+      "step": 3795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8923531796452136e-05,
+      "loss": 1.5846,
+      "step": 3796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.89227827723474e-05,
+      "loss": 1.5275,
+      "step": 3797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.892203349347954e-05,
+      "loss": 1.512,
+      "step": 3798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.892128395985655e-05,
+      "loss": 1.5496,
+      "step": 3799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.892053417148641e-05,
+      "loss": 1.557,
+      "step": 3800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.89197841283771e-05,
+      "loss": 1.5052,
+      "step": 3801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.891903383053662e-05,
+      "loss": 1.4925,
+      "step": 3802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8918283277972946e-05,
+      "loss": 1.5101,
+      "step": 3803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.891753247069407e-05,
+      "loss": 1.5472,
+      "step": 3804
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8916781408708e-05,
+      "loss": 1.5252,
+      "step": 3805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.891603009202272e-05,
+      "loss": 1.6042,
+      "step": 3806
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.891527852064626e-05,
+      "loss": 1.5399,
+      "step": 3807
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8914526694586585e-05,
+      "loss": 1.5218,
+      "step": 3808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8913774613851724e-05,
+      "loss": 1.5111,
+      "step": 3809
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8913022278449685e-05,
+      "loss": 1.5182,
+      "step": 3810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8912269688388465e-05,
+      "loss": 1.4613,
+      "step": 3811
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8911516843676106e-05,
+      "loss": 1.5348,
+      "step": 3812
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.89107637443206e-05,
+      "loss": 1.5042,
+      "step": 3813
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.891001039032998e-05,
+      "loss": 1.4671,
+      "step": 3814
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890925678171226e-05,
+      "loss": 1.5498,
+      "step": 3815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890850291847549e-05,
+      "loss": 1.6016,
+      "step": 3816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8907748800627655e-05,
+      "loss": 1.5426,
+      "step": 3817
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890699442817682e-05,
+      "loss": 1.5985,
+      "step": 3818
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890623980113101e-05,
+      "loss": 1.6034,
+      "step": 3819
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890548491949826e-05,
+      "loss": 1.4862,
+      "step": 3820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890472978328661e-05,
+      "loss": 1.505,
+      "step": 3821
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890397439250409e-05,
+      "loss": 1.5529,
+      "step": 3822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890321874715877e-05,
+      "loss": 1.5656,
+      "step": 3823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8902462847258676e-05,
+      "loss": 1.4989,
+      "step": 3824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890170669281186e-05,
+      "loss": 1.4657,
+      "step": 3825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890095028382638e-05,
+      "loss": 1.5234,
+      "step": 3826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.89001936203103e-05,
+      "loss": 1.4915,
+      "step": 3827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889943670227166e-05,
+      "loss": 1.5166,
+      "step": 3828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8898679529718524e-05,
+      "loss": 1.4709,
+      "step": 3829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889792210265897e-05,
+      "loss": 1.543,
+      "step": 3830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889716442110105e-05,
+      "loss": 1.5259,
+      "step": 3831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889640648505284e-05,
+      "loss": 1.5303,
+      "step": 3832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889564829452241e-05,
+      "loss": 1.5933,
+      "step": 3833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8894889849517825e-05,
+      "loss": 1.4928,
+      "step": 3834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889413115004717e-05,
+      "loss": 1.5049,
+      "step": 3835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889337219611854e-05,
+      "loss": 1.5876,
+      "step": 3836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889261298773998e-05,
+      "loss": 1.5618,
+      "step": 3837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8891853524919616e-05,
+      "loss": 1.5241,
+      "step": 3838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889109380766551e-05,
+      "loss": 1.5356,
+      "step": 3839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889033383598576e-05,
+      "loss": 1.5374,
+      "step": 3840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8889573609888455e-05,
+      "loss": 1.492,
+      "step": 3841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.88888131293817e-05,
+      "loss": 1.5487,
+      "step": 3842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888805239447358e-05,
+      "loss": 1.5065,
+      "step": 3843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888729140517222e-05,
+      "loss": 1.5798,
+      "step": 3844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.88865301614857e-05,
+      "loss": 1.509,
+      "step": 3845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8885768663422135e-05,
+      "loss": 1.5275,
+      "step": 3846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8885006910989634e-05,
+      "loss": 1.5825,
+      "step": 3847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888424490419631e-05,
+      "loss": 1.5107,
+      "step": 3848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888348264305028e-05,
+      "loss": 1.4799,
+      "step": 3849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888272012755966e-05,
+      "loss": 1.479,
+      "step": 3850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8881957357732574e-05,
+      "loss": 1.5342,
+      "step": 3851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888119433357713e-05,
+      "loss": 1.4829,
+      "step": 3852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888043105510148e-05,
+      "loss": 1.4683,
+      "step": 3853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8879667522313735e-05,
+      "loss": 1.535,
+      "step": 3854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8878903735222015e-05,
+      "loss": 1.4865,
+      "step": 3855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.887813969383448e-05,
+      "loss": 1.5467,
+      "step": 3856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.887737539815925e-05,
+      "loss": 1.5882,
+      "step": 3857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.887661084820446e-05,
+      "loss": 1.5941,
+      "step": 3858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.887584604397828e-05,
+      "loss": 1.5026,
+      "step": 3859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.887508098548882e-05,
+      "loss": 1.5426,
+      "step": 3860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8874315672744245e-05,
+      "loss": 1.5721,
+      "step": 3861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.887355010575271e-05,
+      "loss": 1.5534,
+      "step": 3862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.887278428452235e-05,
+      "loss": 1.4777,
+      "step": 3863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.887201820906133e-05,
+      "loss": 1.5052,
+      "step": 3864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8871251879377815e-05,
+      "loss": 1.5081,
+      "step": 3865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.887048529547996e-05,
+      "loss": 1.5112,
+      "step": 3866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886971845737593e-05,
+      "loss": 1.5454,
+      "step": 3867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886895136507389e-05,
+      "loss": 1.4842,
+      "step": 3868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886818401858201e-05,
+      "loss": 1.5584,
+      "step": 3869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8867416417908454e-05,
+      "loss": 1.4878,
+      "step": 3870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8866648563061406e-05,
+      "loss": 1.5029,
+      "step": 3871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886588045404904e-05,
+      "loss": 1.5628,
+      "step": 3872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886511209087954e-05,
+      "loss": 1.5412,
+      "step": 3873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886434347356108e-05,
+      "loss": 1.5404,
+      "step": 3874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886357460210186e-05,
+      "loss": 1.515,
+      "step": 3875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8862805476510056e-05,
+      "loss": 1.4847,
+      "step": 3876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886203609679385e-05,
+      "loss": 1.5081,
+      "step": 3877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8861266462961456e-05,
+      "loss": 1.4751,
+      "step": 3878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8860496575021056e-05,
+      "loss": 1.5042,
+      "step": 3879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8859726432980854e-05,
+      "loss": 1.4967,
+      "step": 3880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.885895603684906e-05,
+      "loss": 1.6007,
+      "step": 3881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.885818538663386e-05,
+      "loss": 1.6216,
+      "step": 3882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.885741448234348e-05,
+      "loss": 1.492,
+      "step": 3883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8856643323986104e-05,
+      "loss": 1.5594,
+      "step": 3884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8855871911569975e-05,
+      "loss": 1.5343,
+      "step": 3885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.885510024510329e-05,
+      "loss": 1.4914,
+      "step": 3886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.885432832459427e-05,
+      "loss": 1.5658,
+      "step": 3887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.885355615005113e-05,
+      "loss": 1.4899,
+      "step": 3888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.88527837214821e-05,
+      "loss": 1.5067,
+      "step": 3889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.885201103889541e-05,
+      "loss": 1.4762,
+      "step": 3890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.885123810229928e-05,
+      "loss": 1.5532,
+      "step": 3891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8850464911701945e-05,
+      "loss": 1.5339,
+      "step": 3892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8849691467111636e-05,
+      "loss": 1.5524,
+      "step": 3893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8848917768536594e-05,
+      "loss": 1.5542,
+      "step": 3894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884814381598506e-05,
+      "loss": 1.494,
+      "step": 3895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884736960946527e-05,
+      "loss": 1.4797,
+      "step": 3896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884659514898547e-05,
+      "loss": 1.5002,
+      "step": 3897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884582043455391e-05,
+      "loss": 1.509,
+      "step": 3898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8845045466178854e-05,
+      "loss": 1.5226,
+      "step": 3899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8844270243868525e-05,
+      "loss": 1.5021,
+      "step": 3900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8843494767631196e-05,
+      "loss": 1.5342,
+      "step": 3901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884271903747513e-05,
+      "loss": 1.4839,
+      "step": 3902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8841943053408566e-05,
+      "loss": 1.4673,
+      "step": 3903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.88411668154398e-05,
+      "loss": 1.499,
+      "step": 3904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884039032357707e-05,
+      "loss": 1.4642,
+      "step": 3905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8839613577828666e-05,
+      "loss": 1.5308,
+      "step": 3906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883883657820284e-05,
+      "loss": 1.4486,
+      "step": 3907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883805932470789e-05,
+      "loss": 1.5535,
+      "step": 3908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8837281817352075e-05,
+      "loss": 1.5373,
+      "step": 3909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883650405614368e-05,
+      "loss": 1.5137,
+      "step": 3910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883572604109099e-05,
+      "loss": 1.5236,
+      "step": 3911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883494777220228e-05,
+      "loss": 1.5861,
+      "step": 3912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883416924948586e-05,
+      "loss": 1.4795,
+      "step": 3913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883339047295e-05,
+      "loss": 1.4852,
+      "step": 3914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8832611442603007e-05,
+      "loss": 1.6292,
+      "step": 3915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883183215845316e-05,
+      "loss": 1.5487,
+      "step": 3916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8831052620508774e-05,
+      "loss": 1.5871,
+      "step": 3917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883027282877815e-05,
+      "loss": 1.5169,
+      "step": 3918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.882949278326959e-05,
+      "loss": 1.5112,
+      "step": 3919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.882871248399139e-05,
+      "loss": 1.562,
+      "step": 3920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8827931930951875e-05,
+      "loss": 1.5482,
+      "step": 3921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8827151124159355e-05,
+      "loss": 1.5509,
+      "step": 3922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.882637006362214e-05,
+      "loss": 1.5251,
+      "step": 3923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8825588749348537e-05,
+      "loss": 1.5965,
+      "step": 3924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.882480718134689e-05,
+      "loss": 1.5957,
+      "step": 3925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.882402535962551e-05,
+      "loss": 1.5558,
+      "step": 3926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.882324328419272e-05,
+      "loss": 1.5239,
+      "step": 3927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8822460955056854e-05,
+      "loss": 1.5824,
+      "step": 3928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8821678372226244e-05,
+      "loss": 1.5697,
+      "step": 3929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.882089553570922e-05,
+      "loss": 1.5153,
+      "step": 3930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.882011244551412e-05,
+      "loss": 1.5418,
+      "step": 3931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881932910164929e-05,
+      "loss": 1.5457,
+      "step": 3932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881854550412306e-05,
+      "loss": 1.5781,
+      "step": 3933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8817761652943776e-05,
+      "loss": 1.5047,
+      "step": 3934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.88169775481198e-05,
+      "loss": 1.5015,
+      "step": 3935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881619318965946e-05,
+      "loss": 1.542,
+      "step": 3936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8815408577571135e-05,
+      "loss": 1.4874,
+      "step": 3937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881462371186317e-05,
+      "loss": 1.5763,
+      "step": 3938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881383859254391e-05,
+      "loss": 1.5594,
+      "step": 3939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881305321962173e-05,
+      "loss": 1.5054,
+      "step": 3940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881226759310499e-05,
+      "loss": 1.5564,
+      "step": 3941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881148171300206e-05,
+      "loss": 1.5246,
+      "step": 3942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.88106955793213e-05,
+      "loss": 1.6084,
+      "step": 3943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.88099091920711e-05,
+      "loss": 1.5837,
+      "step": 3944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880912255125981e-05,
+      "loss": 1.5579,
+      "step": 3945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880833565689582e-05,
+      "loss": 1.4906,
+      "step": 3946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8807548508987513e-05,
+      "loss": 1.5155,
+      "step": 3947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8806761107543276e-05,
+      "loss": 1.5169,
+      "step": 3948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880597345257147e-05,
+      "loss": 1.5744,
+      "step": 3949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880518554408051e-05,
+      "loss": 1.4331,
+      "step": 3950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880439738207878e-05,
+      "loss": 1.542,
+      "step": 3951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880360896657466e-05,
+      "loss": 1.5596,
+      "step": 3952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880282029757657e-05,
+      "loss": 1.5555,
+      "step": 3953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8802031375092884e-05,
+      "loss": 1.5591,
+      "step": 3954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8801242199132016e-05,
+      "loss": 1.5184,
+      "step": 3955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880045276970237e-05,
+      "loss": 1.5851,
+      "step": 3956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879966308681235e-05,
+      "loss": 1.4842,
+      "step": 3957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879887315047037e-05,
+      "loss": 1.5355,
+      "step": 3958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879808296068484e-05,
+      "loss": 1.5625,
+      "step": 3959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879729251746417e-05,
+      "loss": 1.5876,
+      "step": 3960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8796501820816785e-05,
+      "loss": 1.5068,
+      "step": 3961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.87957108707511e-05,
+      "loss": 1.5026,
+      "step": 3962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879491966727555e-05,
+      "loss": 1.5304,
+      "step": 3963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879412821039854e-05,
+      "loss": 1.4899,
+      "step": 3964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8793336500128514e-05,
+      "loss": 1.5477,
+      "step": 3965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8792544536473905e-05,
+      "loss": 1.4964,
+      "step": 3966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879175231944314e-05,
+      "loss": 1.5646,
+      "step": 3967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879095984904466e-05,
+      "loss": 1.5612,
+      "step": 3968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879016712528689e-05,
+      "loss": 1.4902,
+      "step": 3969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878937414817829e-05,
+      "loss": 1.5195,
+      "step": 3970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.87885809177273e-05,
+      "loss": 1.6009,
+      "step": 3971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878778743394237e-05,
+      "loss": 1.6139,
+      "step": 3972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8786993696831935e-05,
+      "loss": 1.5495,
+      "step": 3973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878619970640446e-05,
+      "loss": 1.5828,
+      "step": 3974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8785405462668406e-05,
+      "loss": 1.6001,
+      "step": 3975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878461096563222e-05,
+      "loss": 1.4653,
+      "step": 3976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8783816215304366e-05,
+      "loss": 1.4507,
+      "step": 3977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8783021211693315e-05,
+      "loss": 1.548,
+      "step": 3978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878222595480751e-05,
+      "loss": 1.5243,
+      "step": 3979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878143044465546e-05,
+      "loss": 1.5549,
+      "step": 3980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.87806346812456e-05,
+      "loss": 1.4784,
+      "step": 3981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.877983866458642e-05,
+      "loss": 1.5203,
+      "step": 3982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.87790423946864e-05,
+      "loss": 1.5814,
+      "step": 3983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.877824587155401e-05,
+      "loss": 1.5519,
+      "step": 3984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.877744909519775e-05,
+      "loss": 1.4888,
+      "step": 3985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.877665206562607e-05,
+      "loss": 1.5711,
+      "step": 3986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.87758547828475e-05,
+      "loss": 1.5267,
+      "step": 3987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8775057246870506e-05,
+      "loss": 1.4779,
+      "step": 3988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.877425945770358e-05,
+      "loss": 1.521,
+      "step": 3989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.877346141535523e-05,
+      "loss": 1.4797,
+      "step": 3990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.877266311983395e-05,
+      "loss": 1.568,
+      "step": 3991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.877186457114824e-05,
+      "loss": 1.5164,
+      "step": 3992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8771065769306604e-05,
+      "loss": 1.5744,
+      "step": 3993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8770266714317545e-05,
+      "loss": 1.5396,
+      "step": 3994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8769467406189586e-05,
+      "loss": 1.5744,
+      "step": 3995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8768667844931224e-05,
+      "loss": 1.5661,
+      "step": 3996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876786803055098e-05,
+      "loss": 1.5155,
+      "step": 3997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876706796305737e-05,
+      "loss": 1.5671,
+      "step": 3998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8766267642458916e-05,
+      "loss": 1.5632,
+      "step": 3999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876546706876415e-05,
+      "loss": 1.5239,
+      "step": 4000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876466624198157e-05,
+      "loss": 1.5474,
+      "step": 4001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8763865162119734e-05,
+      "loss": 1.5078,
+      "step": 4002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876306382918716e-05,
+      "loss": 1.5075,
+      "step": 4003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8762262243192383e-05,
+      "loss": 1.4699,
+      "step": 4004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8761460404143944e-05,
+      "loss": 1.5314,
+      "step": 4005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876065831205037e-05,
+      "loss": 1.5378,
+      "step": 4006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875985596692021e-05,
+      "loss": 1.5495,
+      "step": 4007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875905336876201e-05,
+      "loss": 1.522,
+      "step": 4008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875825051758433e-05,
+      "loss": 1.5067,
+      "step": 4009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875744741339569e-05,
+      "loss": 1.5026,
+      "step": 4010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875664405620466e-05,
+      "loss": 1.488,
+      "step": 4011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8755840446019796e-05,
+      "loss": 1.526,
+      "step": 4012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8755036582849654e-05,
+      "loss": 1.4969,
+      "step": 4013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8754232466702796e-05,
+      "loss": 1.5164,
+      "step": 4014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875342809758778e-05,
+      "loss": 1.5015,
+      "step": 4015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8752623475513185e-05,
+      "loss": 1.4676,
+      "step": 4016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875181860048757e-05,
+      "loss": 1.5571,
+      "step": 4017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875101347251949e-05,
+      "loss": 1.5274,
+      "step": 4018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875020809161755e-05,
+      "loss": 1.5752,
+      "step": 4019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874940245779031e-05,
+      "loss": 1.4937,
+      "step": 4020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874859657104635e-05,
+      "loss": 1.5977,
+      "step": 4021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874779043139426e-05,
+      "loss": 1.485,
+      "step": 4022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8746984038842616e-05,
+      "loss": 1.4777,
+      "step": 4023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874617739340001e-05,
+      "loss": 1.5399,
+      "step": 4024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874537049507503e-05,
+      "loss": 1.5394,
+      "step": 4025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8744563343876274e-05,
+      "loss": 1.5567,
+      "step": 4026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874375593981233e-05,
+      "loss": 1.4922,
+      "step": 4027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.87429482828918e-05,
+      "loss": 1.5562,
+      "step": 4028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874214037312329e-05,
+      "loss": 1.5306,
+      "step": 4029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8741332210515404e-05,
+      "loss": 1.4714,
+      "step": 4030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8740523795076734e-05,
+      "loss": 1.5142,
+      "step": 4031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.87397151268159e-05,
+      "loss": 1.589,
+      "step": 4032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873890620574151e-05,
+      "loss": 1.5156,
+      "step": 4033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8738097031862186e-05,
+      "loss": 1.5282,
+      "step": 4034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873728760518654e-05,
+      "loss": 1.4683,
+      "step": 4035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873647792572319e-05,
+      "loss": 1.5127,
+      "step": 4036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873566799348076e-05,
+      "loss": 1.5197,
+      "step": 4037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8734857808467885e-05,
+      "loss": 1.5519,
+      "step": 4038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8734047370693175e-05,
+      "loss": 1.5565,
+      "step": 4039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8733236680165276e-05,
+      "loss": 1.5272,
+      "step": 4040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873242573689281e-05,
+      "loss": 1.5635,
+      "step": 4041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873161454088442e-05,
+      "loss": 1.465,
+      "step": 4042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873080309214874e-05,
+      "loss": 1.4719,
+      "step": 4043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.872999139069442e-05,
+      "loss": 1.4572,
+      "step": 4044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8729179436530095e-05,
+      "loss": 1.55,
+      "step": 4045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.872836722966442e-05,
+      "loss": 1.5511,
+      "step": 4046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.872755477010603e-05,
+      "loss": 1.5396,
+      "step": 4047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8726742057863596e-05,
+      "loss": 1.5389,
+      "step": 4048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.872592909294576e-05,
+      "loss": 1.5859,
+      "step": 4049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8725115875361185e-05,
+      "loss": 1.5239,
+      "step": 4050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.872430240511853e-05,
+      "loss": 1.5199,
+      "step": 4051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8723488682226454e-05,
+      "loss": 1.4943,
+      "step": 4052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8722674706693624e-05,
+      "loss": 1.4429,
+      "step": 4053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8721860478528725e-05,
+      "loss": 1.4871,
+      "step": 4054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.87210459977404e-05,
+      "loss": 1.5039,
+      "step": 4055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8720231264337346e-05,
+      "loss": 1.4855,
+      "step": 4056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871941627832822e-05,
+      "loss": 1.536,
+      "step": 4057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8718601039721726e-05,
+      "loss": 1.6058,
+      "step": 4058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871778554852652e-05,
+      "loss": 1.5112,
+      "step": 4059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8716969804751295e-05,
+      "loss": 1.5379,
+      "step": 4060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871615380840474e-05,
+      "loss": 1.4569,
+      "step": 4061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871533755949556e-05,
+      "loss": 1.5506,
+      "step": 4062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8714521058032416e-05,
+      "loss": 1.5316,
+      "step": 4063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871370430402403e-05,
+      "loss": 1.5628,
+      "step": 4064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871288729747908e-05,
+      "loss": 1.5882,
+      "step": 4065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8712070038406285e-05,
+      "loss": 1.5542,
+      "step": 4066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871125252681434e-05,
+      "loss": 1.5311,
+      "step": 4067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8710434762711956e-05,
+      "loss": 1.535,
+      "step": 4068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.870961674610782e-05,
+      "loss": 1.4681,
+      "step": 4069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8708798477010675e-05,
+      "loss": 1.5182,
+      "step": 4070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8707979955429215e-05,
+      "loss": 1.5098,
+      "step": 4071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.870716118137216e-05,
+      "loss": 1.4816,
+      "step": 4072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.870634215484824e-05,
+      "loss": 1.5314,
+      "step": 4073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.870552287586616e-05,
+      "loss": 1.5695,
+      "step": 4074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8704703344434655e-05,
+      "loss": 1.5264,
+      "step": 4075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8703883560562454e-05,
+      "loss": 1.5843,
+      "step": 4076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.870306352425827e-05,
+      "loss": 1.5859,
+      "step": 4077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.870224323553087e-05,
+      "loss": 1.5319,
+      "step": 4078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8701422694388965e-05,
+      "loss": 1.5508,
+      "step": 4079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.870060190084129e-05,
+      "loss": 1.528,
+      "step": 4080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.86997808548966e-05,
+      "loss": 1.5272,
+      "step": 4081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869895955656363e-05,
+      "loss": 1.5369,
+      "step": 4082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8698138005851125e-05,
+      "loss": 1.5506,
+      "step": 4083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869731620276784e-05,
+      "loss": 1.5041,
+      "step": 4084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869649414732253e-05,
+      "loss": 1.5687,
+      "step": 4085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8695671839523935e-05,
+      "loss": 1.599,
+      "step": 4086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869484927938083e-05,
+      "loss": 1.508,
+      "step": 4087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8694026466901965e-05,
+      "loss": 1.571,
+      "step": 4088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.86932034020961e-05,
+      "loss": 1.5238,
+      "step": 4089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8692380084972e-05,
+      "loss": 1.4635,
+      "step": 4090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869155651553844e-05,
+      "loss": 1.5415,
+      "step": 4091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8690732693804184e-05,
+      "loss": 1.4766,
+      "step": 4092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868990861977801e-05,
+      "loss": 1.5352,
+      "step": 4093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868908429346869e-05,
+      "loss": 1.4985,
+      "step": 4094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868825971488501e-05,
+      "loss": 1.5466,
+      "step": 4095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8687434884035734e-05,
+      "loss": 1.5086,
+      "step": 4096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868660980092966e-05,
+      "loss": 1.5293,
+      "step": 4097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868578446557558e-05,
+      "loss": 1.5353,
+      "step": 4098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8684958877982275e-05,
+      "loss": 1.5031,
+      "step": 4099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868413303815853e-05,
+      "loss": 1.4658,
+      "step": 4100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868330694611315e-05,
+      "loss": 1.513,
+      "step": 4101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8682480601854936e-05,
+      "loss": 1.5062,
+      "step": 4102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868165400539267e-05,
+      "loss": 1.5415,
+      "step": 4103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868082715673518e-05,
+      "loss": 1.5521,
+      "step": 4104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8680000055891254e-05,
+      "loss": 1.5194,
+      "step": 4105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.867917270286969e-05,
+      "loss": 1.523,
+      "step": 4106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8678345097679325e-05,
+      "loss": 1.5508,
+      "step": 4107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8677517240328966e-05,
+      "loss": 1.5677,
+      "step": 4108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8676689130827414e-05,
+      "loss": 1.5238,
+      "step": 4109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.867586076918349e-05,
+      "loss": 1.4714,
+      "step": 4110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8675032155406034e-05,
+      "loss": 1.5708,
+      "step": 4111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8674203289503864e-05,
+      "loss": 1.4715,
+      "step": 4112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.867337417148579e-05,
+      "loss": 1.4544,
+      "step": 4113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.867254480136065e-05,
+      "loss": 1.4855,
+      "step": 4114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8671715179137295e-05,
+      "loss": 1.5086,
+      "step": 4115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.867088530482454e-05,
+      "loss": 1.4915,
+      "step": 4116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.867005517843122e-05,
+      "loss": 1.4658,
+      "step": 4117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866922479996619e-05,
+      "loss": 1.5085,
+      "step": 4118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866839416943828e-05,
+      "loss": 1.4974,
+      "step": 4119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866756328685635e-05,
+      "loss": 1.4792,
+      "step": 4120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8666732152229236e-05,
+      "loss": 1.4898,
+      "step": 4121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8665900765565784e-05,
+      "loss": 1.5037,
+      "step": 4122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8665069126874874e-05,
+      "loss": 1.5181,
+      "step": 4123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866423723616533e-05,
+      "loss": 1.508,
+      "step": 4124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8663405093446036e-05,
+      "loss": 1.5362,
+      "step": 4125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866257269872584e-05,
+      "loss": 1.5721,
+      "step": 4126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866174005201361e-05,
+      "loss": 1.542,
+      "step": 4127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8660907153318214e-05,
+      "loss": 1.4837,
+      "step": 4128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8660074002648524e-05,
+      "loss": 1.5056,
+      "step": 4129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865924060001341e-05,
+      "loss": 1.5636,
+      "step": 4130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8658406945421744e-05,
+      "loss": 1.4995,
+      "step": 4131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865757303888241e-05,
+      "loss": 1.5319,
+      "step": 4132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8656738880404285e-05,
+      "loss": 1.5168,
+      "step": 4133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865590446999626e-05,
+      "loss": 1.486,
+      "step": 4134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865506980766721e-05,
+      "loss": 1.5037,
+      "step": 4135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865423489342602e-05,
+      "loss": 1.5132,
+      "step": 4136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.86533997272816e-05,
+      "loss": 1.4873,
+      "step": 4137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865256430924283e-05,
+      "loss": 1.5408,
+      "step": 4138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8651728639318605e-05,
+      "loss": 1.534,
+      "step": 4139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865089271751783e-05,
+      "loss": 1.6043,
+      "step": 4140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8650056543849406e-05,
+      "loss": 1.5321,
+      "step": 4141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8649220118322246e-05,
+      "loss": 1.5715,
+      "step": 4142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864838344094525e-05,
+      "loss": 1.5047,
+      "step": 4143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8647546511727315e-05,
+      "loss": 1.5011,
+      "step": 4144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8646709330677374e-05,
+      "loss": 1.4707,
+      "step": 4145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864587189780434e-05,
+      "loss": 1.5697,
+      "step": 4146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864503421311713e-05,
+      "loss": 1.5361,
+      "step": 4147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864419627662464e-05,
+      "loss": 1.6011,
+      "step": 4148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864335808833583e-05,
+      "loss": 1.5086,
+      "step": 4149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.86425196482596e-05,
+      "loss": 1.4709,
+      "step": 4150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.86416809564049e-05,
+      "loss": 1.5288,
+      "step": 4151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8640842012780655e-05,
+      "loss": 1.5143,
+      "step": 4152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864000281739578e-05,
+      "loss": 1.5481,
+      "step": 4153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8639163370259235e-05,
+      "loss": 1.4644,
+      "step": 4154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863832367137995e-05,
+      "loss": 1.5303,
+      "step": 4155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8637483720766874e-05,
+      "loss": 1.5567,
+      "step": 4156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863664351842894e-05,
+      "loss": 1.5401,
+      "step": 4157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863580306437511e-05,
+      "loss": 1.4818,
+      "step": 4158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863496235861432e-05,
+      "loss": 1.5391,
+      "step": 4159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8634121401155536e-05,
+      "loss": 1.5537,
+      "step": 4160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8633280192007705e-05,
+      "loss": 1.5365,
+      "step": 4161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863243873117979e-05,
+      "loss": 1.4709,
+      "step": 4162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8631597018680745e-05,
+      "loss": 1.5355,
+      "step": 4163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8630755054519546e-05,
+      "loss": 1.5096,
+      "step": 4164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862991283870515e-05,
+      "loss": 1.5601,
+      "step": 4165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862907037124652e-05,
+      "loss": 1.5762,
+      "step": 4166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8628227652152645e-05,
+      "loss": 1.478,
+      "step": 4167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862738468143249e-05,
+      "loss": 1.4937,
+      "step": 4168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8626541459095034e-05,
+      "loss": 1.5241,
+      "step": 4169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862569798514925e-05,
+      "loss": 1.5845,
+      "step": 4170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862485425960412e-05,
+      "loss": 1.4966,
+      "step": 4171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862401028246865e-05,
+      "loss": 1.4766,
+      "step": 4172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862316605375181e-05,
+      "loss": 1.4979,
+      "step": 4173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862232157346259e-05,
+      "loss": 1.5057,
+      "step": 4174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.862147684160999e-05,
+      "loss": 1.519,
+      "step": 4175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.862063185820299e-05,
+      "loss": 1.4937,
+      "step": 4176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861978662325062e-05,
+      "loss": 1.4691,
+      "step": 4177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861894113676185e-05,
+      "loss": 1.5796,
+      "step": 4178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.86180953987457e-05,
+      "loss": 1.5654,
+      "step": 4179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861724940921117e-05,
+      "loss": 1.5238,
+      "step": 4180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861640316816727e-05,
+      "loss": 1.5316,
+      "step": 4181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861555667562302e-05,
+      "loss": 1.4953,
+      "step": 4182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8614709931587434e-05,
+      "loss": 1.5046,
+      "step": 4183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861386293606951e-05,
+      "loss": 1.5301,
+      "step": 4184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861301568907829e-05,
+      "loss": 1.5203,
+      "step": 4185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861216819062279e-05,
+      "loss": 1.5475,
+      "step": 4186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861132044071203e-05,
+      "loss": 1.5708,
+      "step": 4187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861047243935505e-05,
+      "loss": 1.5033,
+      "step": 4188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8609624186560865e-05,
+      "loss": 1.527,
+      "step": 4189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860877568233852e-05,
+      "loss": 1.5238,
+      "step": 4190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8607926926697043e-05,
+      "loss": 1.5155,
+      "step": 4191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860707791964548e-05,
+      "loss": 1.5228,
+      "step": 4192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860622866119287e-05,
+      "loss": 1.5528,
+      "step": 4193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860537915134825e-05,
+      "loss": 1.4663,
+      "step": 4194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860452939012068e-05,
+      "loss": 1.5003,
+      "step": 4195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.86036793775192e-05,
+      "loss": 1.5461,
+      "step": 4196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8602829113552865e-05,
+      "loss": 1.4741,
+      "step": 4197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8601978598230734e-05,
+      "loss": 1.5143,
+      "step": 4198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8601127831561855e-05,
+      "loss": 1.4998,
+      "step": 4199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8600276813555295e-05,
+      "loss": 1.54,
+      "step": 4200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8599425544220115e-05,
+      "loss": 1.4855,
+      "step": 4201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859857402356538e-05,
+      "loss": 1.5313,
+      "step": 4202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859772225160016e-05,
+      "loss": 1.4884,
+      "step": 4203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8596870228333527e-05,
+      "loss": 1.5422,
+      "step": 4204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859601795377454e-05,
+      "loss": 1.5345,
+      "step": 4205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8595165427932296e-05,
+      "loss": 1.5231,
+      "step": 4206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8594312650815864e-05,
+      "loss": 1.5452,
+      "step": 4207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859345962243434e-05,
+      "loss": 1.5536,
+      "step": 4208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859260634279678e-05,
+      "loss": 1.5439,
+      "step": 4209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8591752811912285e-05,
+      "loss": 1.471,
+      "step": 4210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8590899029789945e-05,
+      "loss": 1.5348,
+      "step": 4211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8590044996438854e-05,
+      "loss": 1.5649,
+      "step": 4212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858919071186811e-05,
+      "loss": 1.519,
+      "step": 4213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.85883361760868e-05,
+      "loss": 1.4787,
+      "step": 4214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858748138910404e-05,
+      "loss": 1.6068,
+      "step": 4215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858662635092892e-05,
+      "loss": 1.4681,
+      "step": 4216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8585771061570546e-05,
+      "loss": 1.4894,
+      "step": 4217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858491552103803e-05,
+      "loss": 1.5189,
+      "step": 4218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858405972934048e-05,
+      "loss": 1.5205,
+      "step": 4219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858320368648702e-05,
+      "loss": 1.6156,
+      "step": 4220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8582347392486756e-05,
+      "loss": 1.5264,
+      "step": 4221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858149084734881e-05,
+      "loss": 1.5868,
+      "step": 4222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8580634051082294e-05,
+      "loss": 1.4784,
+      "step": 4223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857977700369635e-05,
+      "loss": 1.5309,
+      "step": 4224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.85789197052001e-05,
+      "loss": 1.6204,
+      "step": 4225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857806215560266e-05,
+      "loss": 1.5179,
+      "step": 4226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857720435491317e-05,
+      "loss": 1.4694,
+      "step": 4227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857634630314078e-05,
+      "loss": 1.5164,
+      "step": 4228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857548800029461e-05,
+      "loss": 1.5799,
+      "step": 4229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8574629446383815e-05,
+      "loss": 1.5308,
+      "step": 4230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857377064141751e-05,
+      "loss": 1.4837,
+      "step": 4231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857291158540488e-05,
+      "loss": 1.5171,
+      "step": 4232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857205227835504e-05,
+      "loss": 1.4688,
+      "step": 4233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857119272027716e-05,
+      "loss": 1.5094,
+      "step": 4234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8570332911180384e-05,
+      "loss": 1.4676,
+      "step": 4235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.856947285107387e-05,
+      "loss": 1.5285,
+      "step": 4236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856861253996679e-05,
+      "loss": 1.4684,
+      "step": 4237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8567751977868284e-05,
+      "loss": 1.5225,
+      "step": 4238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856689116478753e-05,
+      "loss": 1.5428,
+      "step": 4239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8566030100733696e-05,
+      "loss": 1.5431,
+      "step": 4240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8565168785715954e-05,
+      "loss": 1.4974,
+      "step": 4241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8564307219743465e-05,
+      "loss": 1.5575,
+      "step": 4242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856344540282541e-05,
+      "loss": 1.569,
+      "step": 4243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8562583334970965e-05,
+      "loss": 1.4583,
+      "step": 4244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856172101618932e-05,
+      "loss": 1.5332,
+      "step": 4245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856085844648964e-05,
+      "loss": 1.4775,
+      "step": 4246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8559995625881124e-05,
+      "loss": 1.4881,
+      "step": 4247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855913255437297e-05,
+      "loss": 1.5239,
+      "step": 4248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8558269231974346e-05,
+      "loss": 1.5081,
+      "step": 4249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855740565869446e-05,
+      "loss": 1.5081,
+      "step": 4250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8556541834542505e-05,
+      "loss": 1.5335,
+      "step": 4251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8555677759527685e-05,
+      "loss": 1.5563,
+      "step": 4252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85548134336592e-05,
+      "loss": 1.506,
+      "step": 4253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855394885694624e-05,
+      "loss": 1.5625,
+      "step": 4254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855308402939803e-05,
+      "loss": 1.5361,
+      "step": 4255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855221895102379e-05,
+      "loss": 1.5938,
+      "step": 4256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855135362183269e-05,
+      "loss": 1.5106,
+      "step": 4257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8550488041834e-05,
+      "loss": 1.5273,
+      "step": 4258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854962221103689e-05,
+      "loss": 1.5078,
+      "step": 4259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854875612945061e-05,
+      "loss": 1.5205,
+      "step": 4260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854788979708437e-05,
+      "loss": 1.514,
+      "step": 4261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8547023213947394e-05,
+      "loss": 1.4661,
+      "step": 4262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854615638004893e-05,
+      "loss": 1.5153,
+      "step": 4263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854528929539819e-05,
+      "loss": 1.4546,
+      "step": 4264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854442196000441e-05,
+      "loss": 1.5094,
+      "step": 4265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8543554373876834e-05,
+      "loss": 1.5088,
+      "step": 4266
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8542686537024696e-05,
+      "loss": 1.5379,
+      "step": 4267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854181844945724e-05,
+      "loss": 1.4334,
+      "step": 4268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854095011118371e-05,
+      "loss": 1.5161,
+      "step": 4269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854008152221336e-05,
+      "loss": 1.474,
+      "step": 4270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853921268255542e-05,
+      "loss": 1.5262,
+      "step": 4271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853834359221917e-05,
+      "loss": 1.6439,
+      "step": 4272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853747425121384e-05,
+      "loss": 1.5002,
+      "step": 4273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8536604659548706e-05,
+      "loss": 1.5558,
+      "step": 4274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853573481723302e-05,
+      "loss": 1.5225,
+      "step": 4275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8534864724276043e-05,
+      "loss": 1.5085,
+      "step": 4276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8533994380687056e-05,
+      "loss": 1.5402,
+      "step": 4277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85331237864753e-05,
+      "loss": 1.5863,
+      "step": 4278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853225294165008e-05,
+      "loss": 1.5838,
+      "step": 4279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853138184622064e-05,
+      "loss": 1.5228,
+      "step": 4280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853051050019628e-05,
+      "loss": 1.5008,
+      "step": 4281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852963890358626e-05,
+      "loss": 1.3898,
+      "step": 4282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852876705639987e-05,
+      "loss": 1.5335,
+      "step": 4283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85278949586464e-05,
+      "loss": 1.458,
+      "step": 4284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852702261033514e-05,
+      "loss": 1.5754,
+      "step": 4285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8526150011475365e-05,
+      "loss": 1.4736,
+      "step": 4286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852527716207638e-05,
+      "loss": 1.4883,
+      "step": 4287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852440406214747e-05,
+      "loss": 1.5249,
+      "step": 4288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852353071169794e-05,
+      "loss": 1.4995,
+      "step": 4289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8522657110737094e-05,
+      "loss": 1.5155,
+      "step": 4290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8521783259274224e-05,
+      "loss": 1.527,
+      "step": 4291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852090915731864e-05,
+      "loss": 1.4769,
+      "step": 4292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852003480487966e-05,
+      "loss": 1.4959,
+      "step": 4293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851916020196658e-05,
+      "loss": 1.5215,
+      "step": 4294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851828534858874e-05,
+      "loss": 1.4878,
+      "step": 4295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8517410244755427e-05,
+      "loss": 1.5562,
+      "step": 4296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8516534890475975e-05,
+      "loss": 1.5534,
+      "step": 4297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8515659285759705e-05,
+      "loss": 1.5041,
+      "step": 4298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851478343061594e-05,
+      "loss": 1.5002,
+      "step": 4299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.851390732505401e-05,
+      "loss": 1.5558,
+      "step": 4300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8513030969083236e-05,
+      "loss": 1.528,
+      "step": 4301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.851215436271296e-05,
+      "loss": 1.5445,
+      "step": 4302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.851127750595251e-05,
+      "loss": 1.4697,
+      "step": 4303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.851040039881124e-05,
+      "loss": 1.4419,
+      "step": 4304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850952304129847e-05,
+      "loss": 1.544,
+      "step": 4305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850864543342356e-05,
+      "loss": 1.4828,
+      "step": 4306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850776757519584e-05,
+      "loss": 1.4694,
+      "step": 4307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8506889466624675e-05,
+      "loss": 1.4645,
+      "step": 4308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8506011107719405e-05,
+      "loss": 1.4868,
+      "step": 4309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8505132498489384e-05,
+      "loss": 1.5264,
+      "step": 4310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8504253638943976e-05,
+      "loss": 1.4785,
+      "step": 4311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850337452909254e-05,
+      "loss": 1.5832,
+      "step": 4312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8502495168944426e-05,
+      "loss": 1.4884,
+      "step": 4313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850161555850902e-05,
+      "loss": 1.5156,
+      "step": 4314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8500735697795666e-05,
+      "loss": 1.4979,
+      "step": 4315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849985558681375e-05,
+      "loss": 1.4933,
+      "step": 4316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849897522557263e-05,
+      "loss": 1.4315,
+      "step": 4317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.84980946140817e-05,
+      "loss": 1.5073,
+      "step": 4318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8497213752350326e-05,
+      "loss": 1.5911,
+      "step": 4319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849633264038789e-05,
+      "loss": 1.5693,
+      "step": 4320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849545127820377e-05,
+      "loss": 1.5521,
+      "step": 4321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8494569665807365e-05,
+      "loss": 1.474,
+      "step": 4322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849368780320806e-05,
+      "loss": 1.5218,
+      "step": 4323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8492805690415245e-05,
+      "loss": 1.4604,
+      "step": 4324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8491923327438304e-05,
+      "loss": 1.5031,
+      "step": 4325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849104071428664e-05,
+      "loss": 1.5034,
+      "step": 4326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8490157850969666e-05,
+      "loss": 1.5566,
+      "step": 4327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848927473749676e-05,
+      "loss": 1.4743,
+      "step": 4328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8488391373877345e-05,
+      "loss": 1.5562,
+      "step": 4329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848750776012082e-05,
+      "loss": 1.5376,
+      "step": 4330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.84866238962366e-05,
+      "loss": 1.5446,
+      "step": 4331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848573978223409e-05,
+      "loss": 1.5418,
+      "step": 4332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848485541812271e-05,
+      "loss": 1.5036,
+      "step": 4333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8483970803911886e-05,
+      "loss": 1.4798,
+      "step": 4334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848308593961102e-05,
+      "loss": 1.5145,
+      "step": 4335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848220082522955e-05,
+      "loss": 1.5799,
+      "step": 4336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.84813154607769e-05,
+      "loss": 1.57,
+      "step": 4337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8480429846262496e-05,
+      "loss": 1.5506,
+      "step": 4338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847954398169576e-05,
+      "loss": 1.4782,
+      "step": 4339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847865786708615e-05,
+      "loss": 1.5466,
+      "step": 4340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8477771502443084e-05,
+      "loss": 1.5462,
+      "step": 4341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8476884887776e-05,
+      "loss": 1.5355,
+      "step": 4342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847599802309435e-05,
+      "loss": 1.5565,
+      "step": 4343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8475110908407566e-05,
+      "loss": 1.5347,
+      "step": 4344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8474223543725106e-05,
+      "loss": 1.5433,
+      "step": 4345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847333592905642e-05,
+      "loss": 1.5345,
+      "step": 4346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847244806441096e-05,
+      "loss": 1.5087,
+      "step": 4347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8471559949798176e-05,
+      "loss": 1.4896,
+      "step": 4348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847067158522752e-05,
+      "loss": 1.5855,
+      "step": 4349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8469782970708464e-05,
+      "loss": 1.5597,
+      "step": 4350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846889410625048e-05,
+      "loss": 1.5046,
+      "step": 4351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846800499186301e-05,
+      "loss": 1.4219,
+      "step": 4352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846711562755554e-05,
+      "loss": 1.5503,
+      "step": 4353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8466226013337535e-05,
+      "loss": 1.4572,
+      "step": 4354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846533614921847e-05,
+      "loss": 1.5252,
+      "step": 4355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846444603520782e-05,
+      "loss": 1.5291,
+      "step": 4356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8463555671315066e-05,
+      "loss": 1.568,
+      "step": 4357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846266505754969e-05,
+      "loss": 1.501,
+      "step": 4358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846177419392117e-05,
+      "loss": 1.523,
+      "step": 4359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846088308043901e-05,
+      "loss": 1.5164,
+      "step": 4360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.845999171711268e-05,
+      "loss": 1.5467,
+      "step": 4361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.845910010395168e-05,
+      "loss": 1.5887,
+      "step": 4362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845820824096551e-05,
+      "loss": 1.5181,
+      "step": 4363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8457316128163666e-05,
+      "loss": 1.4601,
+      "step": 4364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845642376555565e-05,
+      "loss": 1.562,
+      "step": 4365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8455531153150945e-05,
+      "loss": 1.4754,
+      "step": 4366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845463829095909e-05,
+      "loss": 1.5218,
+      "step": 4367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845374517898957e-05,
+      "loss": 1.5171,
+      "step": 4368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84528518172519e-05,
+      "loss": 1.5028,
+      "step": 4369
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845195820575561e-05,
+      "loss": 1.5291,
+      "step": 4370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845106434451019e-05,
+      "loss": 1.5571,
+      "step": 4371
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845017023352517e-05,
+      "loss": 1.4512,
+      "step": 4372
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8449275872810086e-05,
+      "loss": 1.4778,
+      "step": 4373
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844838126237444e-05,
+      "loss": 1.4842,
+      "step": 4374
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844748640222778e-05,
+      "loss": 1.495,
+      "step": 4375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844659129237962e-05,
+      "loss": 1.5109,
+      "step": 4376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844569593283949e-05,
+      "loss": 1.5415,
+      "step": 4377
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844480032361694e-05,
+      "loss": 1.4984,
+      "step": 4378
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84439044647215e-05,
+      "loss": 1.5007,
+      "step": 4379
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844300835616271e-05,
+      "loss": 1.4915,
+      "step": 4380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8442111997950114e-05,
+      "loss": 1.5711,
+      "step": 4381
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844121539009325e-05,
+      "loss": 1.5324,
+      "step": 4382
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8440318532601687e-05,
+      "loss": 1.4565,
+      "step": 4383
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843942142548495e-05,
+      "loss": 1.4954,
+      "step": 4384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8438524068752613e-05,
+      "loss": 1.4439,
+      "step": 4385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843762646241422e-05,
+      "loss": 1.5422,
+      "step": 4386
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8436728606479344e-05,
+      "loss": 1.561,
+      "step": 4387
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843583050095752e-05,
+      "loss": 1.5729,
+      "step": 4388
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843493214585835e-05,
+      "loss": 1.5498,
+      "step": 4389
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843403354119137e-05,
+      "loss": 1.5418,
+      "step": 4390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843313468696616e-05,
+      "loss": 1.43,
+      "step": 4391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843223558319229e-05,
+      "loss": 1.5389,
+      "step": 4392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8431336229879345e-05,
+      "loss": 1.492,
+      "step": 4393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8430436627036876e-05,
+      "loss": 1.5519,
+      "step": 4394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84295367746745e-05,
+      "loss": 1.5322,
+      "step": 4395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842863667280178e-05,
+      "loss": 1.4759,
+      "step": 4396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84277363214283e-05,
+      "loss": 1.4852,
+      "step": 4397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842683572056365e-05,
+      "loss": 1.5381,
+      "step": 4398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842593487021742e-05,
+      "loss": 1.4842,
+      "step": 4399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842503377039921e-05,
+      "loss": 1.5269,
+      "step": 4400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8424132421118615e-05,
+      "loss": 1.518,
+      "step": 4401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8423230822385227e-05,
+      "loss": 1.6738,
+      "step": 4402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842232897420865e-05,
+      "loss": 1.5667,
+      "step": 4403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842142687659849e-05,
+      "loss": 1.5275,
+      "step": 4404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842052452956436e-05,
+      "loss": 1.5791,
+      "step": 4405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.841962193311585e-05,
+      "loss": 1.5845,
+      "step": 4406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84187190872626e-05,
+      "loss": 1.6177,
+      "step": 4407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84178159920142e-05,
+      "loss": 1.6394,
+      "step": 4408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8416912647380275e-05,
+      "loss": 1.5094,
+      "step": 4409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8416009053370445e-05,
+      "loss": 1.5812,
+      "step": 4410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.841510520999435e-05,
+      "loss": 1.5791,
+      "step": 4411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8414201117261585e-05,
+      "loss": 1.5853,
+      "step": 4412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84132967751818e-05,
+      "loss": 1.563,
+      "step": 4413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.841239218376462e-05,
+      "loss": 1.5596,
+      "step": 4414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.841148734301967e-05,
+      "loss": 1.5936,
+      "step": 4415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84105822529566e-05,
+      "loss": 1.4618,
+      "step": 4416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8409676913585044e-05,
+      "loss": 1.4958,
+      "step": 4417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.840877132491464e-05,
+      "loss": 1.5373,
+      "step": 4418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8407865486955025e-05,
+      "loss": 1.5719,
+      "step": 4419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.840695939971586e-05,
+      "loss": 1.5589,
+      "step": 4420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.840605306320679e-05,
+      "loss": 1.4588,
+      "step": 4421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.840514647743746e-05,
+      "loss": 1.4704,
+      "step": 4422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.840423964241753e-05,
+      "loss": 1.6159,
+      "step": 4423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.840333255815666e-05,
+      "loss": 1.5842,
+      "step": 4424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84024252246645e-05,
+      "loss": 1.5477,
+      "step": 4425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.840151764195073e-05,
+      "loss": 1.4948,
+      "step": 4426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8400609810025e-05,
+      "loss": 1.5189,
+      "step": 4427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8399701728896984e-05,
+      "loss": 1.5054,
+      "step": 4428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839879339857635e-05,
+      "loss": 1.5004,
+      "step": 4429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839788481907277e-05,
+      "loss": 1.5715,
+      "step": 4430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839697599039592e-05,
+      "loss": 1.5964,
+      "step": 4431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839606691255548e-05,
+      "loss": 1.557,
+      "step": 4432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839515758556114e-05,
+      "loss": 1.516,
+      "step": 4433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8394248009422574e-05,
+      "loss": 1.5445,
+      "step": 4434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839333818414946e-05,
+      "loss": 1.653,
+      "step": 4435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.83924281097515e-05,
+      "loss": 1.5366,
+      "step": 4436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839151778623839e-05,
+      "loss": 1.4832,
+      "step": 4437
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839060721361981e-05,
+      "loss": 1.5031,
+      "step": 4438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8389696391905466e-05,
+      "loss": 1.5724,
+      "step": 4439
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838878532110506e-05,
+      "loss": 1.5654,
+      "step": 4440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838787400122828e-05,
+      "loss": 1.535,
+      "step": 4441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838696243228484e-05,
+      "loss": 1.5293,
+      "step": 4442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838605061428446e-05,
+      "loss": 1.5682,
+      "step": 4443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838513854723683e-05,
+      "loss": 1.529,
+      "step": 4444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838422623115168e-05,
+      "loss": 1.4668,
+      "step": 4445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838331366603871e-05,
+      "loss": 1.4919,
+      "step": 4446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838240085190765e-05,
+      "loss": 1.5216,
+      "step": 4447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838148778876822e-05,
+      "loss": 1.5376,
+      "step": 4448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838057447663013e-05,
+      "loss": 1.5169,
+      "step": 4449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837966091550313e-05,
+      "loss": 1.4946,
+      "step": 4450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8378747105396926e-05,
+      "loss": 1.5267,
+      "step": 4451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8377833046321264e-05,
+      "loss": 1.5391,
+      "step": 4452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837691873828587e-05,
+      "loss": 1.5402,
+      "step": 4453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837600418130048e-05,
+      "loss": 1.5065,
+      "step": 4454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837508937537484e-05,
+      "loss": 1.5137,
+      "step": 4455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8374174320518694e-05,
+      "loss": 1.4821,
+      "step": 4456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837325901674178e-05,
+      "loss": 1.4694,
+      "step": 4457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8372343464053846e-05,
+      "loss": 1.4906,
+      "step": 4458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837142766246465e-05,
+      "loss": 1.4963,
+      "step": 4459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8370511611983924e-05,
+      "loss": 1.5915,
+      "step": 4460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8369595312621455e-05,
+      "loss": 1.485,
+      "step": 4461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836867876438698e-05,
+      "loss": 1.5199,
+      "step": 4462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8367761967290255e-05,
+      "loss": 1.5423,
+      "step": 4463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836684492134106e-05,
+      "loss": 1.5643,
+      "step": 4464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8365927626549144e-05,
+      "loss": 1.5015,
+      "step": 4465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836501008292429e-05,
+      "loss": 1.5361,
+      "step": 4466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8364092290476266e-05,
+      "loss": 1.4954,
+      "step": 4467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836317424921484e-05,
+      "loss": 1.549,
+      "step": 4468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836225595914979e-05,
+      "loss": 1.4873,
+      "step": 4469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8361337420290906e-05,
+      "loss": 1.5431,
+      "step": 4470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836041863264796e-05,
+      "loss": 1.5168,
+      "step": 4471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8359499596230734e-05,
+      "loss": 1.5104,
+      "step": 4472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835858031104902e-05,
+      "loss": 1.5186,
+      "step": 4473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835766077711261e-05,
+      "loss": 1.4958,
+      "step": 4474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8356740994431297e-05,
+      "loss": 1.613,
+      "step": 4475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835582096301486e-05,
+      "loss": 1.4795,
+      "step": 4476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835490068287312e-05,
+      "loss": 1.4596,
+      "step": 4477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835398015401586e-05,
+      "loss": 1.4974,
+      "step": 4478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8353059376452895e-05,
+      "loss": 1.4434,
+      "step": 4479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835213835019402e-05,
+      "loss": 1.5464,
+      "step": 4480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835121707524906e-05,
+      "loss": 1.4945,
+      "step": 4481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.83502955516278e-05,
+      "loss": 1.5384,
+      "step": 4482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.834937377934008e-05,
+      "loss": 1.4897,
+      "step": 4483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.83484517583957e-05,
+      "loss": 1.5977,
+      "step": 4484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.834752948880449e-05,
+      "loss": 1.5239,
+      "step": 4485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.834660697057626e-05,
+      "loss": 1.4759,
+      "step": 4486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.834568420372084e-05,
+      "loss": 1.5213,
+      "step": 4487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8344761188248056e-05,
+      "loss": 1.6038,
+      "step": 4488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8343837924167736e-05,
+      "loss": 1.5535,
+      "step": 4489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.834291441148972e-05,
+      "loss": 1.533,
+      "step": 4490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.834199065022384e-05,
+      "loss": 1.5181,
+      "step": 4491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.834106664037992e-05,
+      "loss": 1.4798,
+      "step": 4492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.834014238196782e-05,
+      "loss": 1.5846,
+      "step": 4493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833921787499737e-05,
+      "loss": 1.5392,
+      "step": 4494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833829311947842e-05,
+      "loss": 1.6034,
+      "step": 4495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833736811542082e-05,
+      "loss": 1.563,
+      "step": 4496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8336442862834416e-05,
+      "loss": 1.5093,
+      "step": 4497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833551736172906e-05,
+      "loss": 1.5243,
+      "step": 4498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833459161211462e-05,
+      "loss": 1.5194,
+      "step": 4499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8333665614000935e-05,
+      "loss": 1.5711,
+      "step": 4500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8332739367397883e-05,
+      "loss": 1.4564,
+      "step": 4501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8331812872315325e-05,
+      "loss": 1.523,
+      "step": 4502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833088612876312e-05,
+      "loss": 1.5137,
+      "step": 4503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8329959136751146e-05,
+      "loss": 1.5077,
+      "step": 4504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8329031896289264e-05,
+      "loss": 1.495,
+      "step": 4505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832810440738737e-05,
+      "loss": 1.5067,
+      "step": 4506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832717667005532e-05,
+      "loss": 1.4738,
+      "step": 4507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832624868430299e-05,
+      "loss": 1.5663,
+      "step": 4508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832532045014029e-05,
+      "loss": 1.5111,
+      "step": 4509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8324391967577075e-05,
+      "loss": 1.5549,
+      "step": 4510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832346323662326e-05,
+      "loss": 1.5736,
+      "step": 4511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832253425728871e-05,
+      "loss": 1.5111,
+      "step": 4512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8321605029583326e-05,
+      "loss": 1.5488,
+      "step": 4513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832067555351702e-05,
+      "loss": 1.5028,
+      "step": 4514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8319745829099675e-05,
+      "loss": 1.5278,
+      "step": 4515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8318815856341185e-05,
+      "loss": 1.5469,
+      "step": 4516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.831788563525147e-05,
+      "loss": 1.5824,
+      "step": 4517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8316955165840434e-05,
+      "loss": 1.5272,
+      "step": 4518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.831602444811797e-05,
+      "loss": 1.5208,
+      "step": 4519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8315093482094015e-05,
+      "loss": 1.578,
+      "step": 4520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.831416226777846e-05,
+      "loss": 1.4445,
+      "step": 4521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.831323080518124e-05,
+      "loss": 1.5122,
+      "step": 4522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.831229909431225e-05,
+      "loss": 1.5285,
+      "step": 4523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.831136713518144e-05,
+      "loss": 1.5195,
+      "step": 4524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.831043492779872e-05,
+      "loss": 1.4948,
+      "step": 4525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8309502472174025e-05,
+      "loss": 1.5031,
+      "step": 4526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8308569768317276e-05,
+      "loss": 1.5476,
+      "step": 4527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830763681623841e-05,
+      "loss": 1.4653,
+      "step": 4528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8306703615947354e-05,
+      "loss": 1.5955,
+      "step": 4529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830577016745407e-05,
+      "loss": 1.52,
+      "step": 4530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8304836470768475e-05,
+      "loss": 1.5186,
+      "step": 4531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830390252590052e-05,
+      "loss": 1.4741,
+      "step": 4532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830296833286016e-05,
+      "loss": 1.4842,
+      "step": 4533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8302033891657316e-05,
+      "loss": 1.4997,
+      "step": 4534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830109920230197e-05,
+      "loss": 1.4891,
+      "step": 4535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830016426480406e-05,
+      "loss": 1.5161,
+      "step": 4536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.829922907917355e-05,
+      "loss": 1.5117,
+      "step": 4537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8298293645420395e-05,
+      "loss": 1.5479,
+      "step": 4538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.829735796355456e-05,
+      "loss": 1.4237,
+      "step": 4539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8296422033586e-05,
+      "loss": 1.513,
+      "step": 4540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8295485855524694e-05,
+      "loss": 1.505,
+      "step": 4541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.829454942938061e-05,
+      "loss": 1.5203,
+      "step": 4542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.829361275516371e-05,
+      "loss": 1.4536,
+      "step": 4543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.829267583288398e-05,
+      "loss": 1.4603,
+      "step": 4544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.829173866255139e-05,
+      "loss": 1.5192,
+      "step": 4545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.829080124417593e-05,
+      "loss": 1.5508,
+      "step": 4546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.828986357776757e-05,
+      "loss": 1.4626,
+      "step": 4547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.82889256633363e-05,
+      "loss": 1.5726,
+      "step": 4548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.828798750089212e-05,
+      "loss": 1.4154,
+      "step": 4549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.828704909044501e-05,
+      "loss": 1.6354,
+      "step": 4550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.828611043200496e-05,
+      "loss": 1.4585,
+      "step": 4551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8285171525581976e-05,
+      "loss": 1.4679,
+      "step": 4552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.828423237118605e-05,
+      "loss": 1.4562,
+      "step": 4553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.828329296882719e-05,
+      "loss": 1.5021,
+      "step": 4554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.828235331851539e-05,
+      "loss": 1.4985,
+      "step": 4555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8281413420260666e-05,
+      "loss": 1.5651,
+      "step": 4556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.828047327407302e-05,
+      "loss": 1.4971,
+      "step": 4557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827953287996246e-05,
+      "loss": 1.5905,
+      "step": 4558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827859223793902e-05,
+      "loss": 1.5389,
+      "step": 4559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827765134801271e-05,
+      "loss": 1.4749,
+      "step": 4560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8276710210193535e-05,
+      "loss": 1.5234,
+      "step": 4561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827576882449153e-05,
+      "loss": 1.4966,
+      "step": 4562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8274827190916715e-05,
+      "loss": 1.5166,
+      "step": 4563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8273885309479125e-05,
+      "loss": 1.5132,
+      "step": 4564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827294318018879e-05,
+      "loss": 1.4461,
+      "step": 4565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8272000803055737e-05,
+      "loss": 1.5156,
+      "step": 4566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827105817809e-05,
+      "loss": 1.4727,
+      "step": 4567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827011530530162e-05,
+      "loss": 1.5594,
+      "step": 4568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826917218470065e-05,
+      "loss": 1.5076,
+      "step": 4569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826822881629711e-05,
+      "loss": 1.5288,
+      "step": 4570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826728520010107e-05,
+      "loss": 1.519,
+      "step": 4571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826634133612257e-05,
+      "loss": 1.4797,
+      "step": 4572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826539722437165e-05,
+      "loss": 1.5244,
+      "step": 4573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826445286485838e-05,
+      "loss": 1.5552,
+      "step": 4574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8263508257592814e-05,
+      "loss": 1.4927,
+      "step": 4575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8262563402585e-05,
+      "loss": 1.5498,
+      "step": 4576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8261618299845016e-05,
+      "loss": 1.4839,
+      "step": 4577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826067294938292e-05,
+      "loss": 1.4559,
+      "step": 4578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8259727351208776e-05,
+      "loss": 1.5653,
+      "step": 4579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825878150533266e-05,
+      "loss": 1.4779,
+      "step": 4580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8257835411764643e-05,
+      "loss": 1.5941,
+      "step": 4581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825688907051479e-05,
+      "loss": 1.5462,
+      "step": 4582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8255942481593194e-05,
+      "loss": 1.5101,
+      "step": 4583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825499564500993e-05,
+      "loss": 1.4829,
+      "step": 4584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825404856077508e-05,
+      "loss": 1.5107,
+      "step": 4585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825310122889873e-05,
+      "loss": 1.5226,
+      "step": 4586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825215364939097e-05,
+      "loss": 1.5277,
+      "step": 4587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825120582226189e-05,
+      "loss": 1.527,
+      "step": 4588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825025774752159e-05,
+      "loss": 1.5295,
+      "step": 4589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824930942518016e-05,
+      "loss": 1.4884,
+      "step": 4590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824836085524769e-05,
+      "loss": 1.5736,
+      "step": 4591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8247412037734296e-05,
+      "loss": 1.4668,
+      "step": 4592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824646297265008e-05,
+      "loss": 1.5088,
+      "step": 4593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824551366000514e-05,
+      "loss": 1.5706,
+      "step": 4594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8244564099809596e-05,
+      "loss": 1.4862,
+      "step": 4595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824361429207356e-05,
+      "loss": 1.509,
+      "step": 4596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8242664236807135e-05,
+      "loss": 1.5422,
+      "step": 4597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824171393402046e-05,
+      "loss": 1.5781,
+      "step": 4598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824076338372363e-05,
+      "loss": 1.5197,
+      "step": 4599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823981258592678e-05,
+      "loss": 1.5073,
+      "step": 4600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823886154064004e-05,
+      "loss": 1.4902,
+      "step": 4601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823791024787353e-05,
+      "loss": 1.4731,
+      "step": 4602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823695870763739e-05,
+      "loss": 1.4894,
+      "step": 4603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823600691994174e-05,
+      "loss": 1.5684,
+      "step": 4604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823505488479673e-05,
+      "loss": 1.5073,
+      "step": 4605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8234102602212485e-05,
+      "loss": 1.4626,
+      "step": 4606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8233150072199165e-05,
+      "loss": 1.5329,
+      "step": 4607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823219729476689e-05,
+      "loss": 1.5156,
+      "step": 4608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8231244269925826e-05,
+      "loss": 1.4491,
+      "step": 4609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823029099768611e-05,
+      "loss": 1.4924,
+      "step": 4610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.82293374780579e-05,
+      "loss": 1.5176,
+      "step": 4611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.822838371105134e-05,
+      "loss": 1.5395,
+      "step": 4612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.822742969667661e-05,
+      "loss": 1.4718,
+      "step": 4613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8226475434943844e-05,
+      "loss": 1.4953,
+      "step": 4614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.822552092586322e-05,
+      "loss": 1.5067,
+      "step": 4615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.82245661694449e-05,
+      "loss": 1.5288,
+      "step": 4616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.822361116569905e-05,
+      "loss": 1.5321,
+      "step": 4617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.822265591463584e-05,
+      "loss": 1.479,
+      "step": 4618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8221700416265444e-05,
+      "loss": 1.4665,
+      "step": 4619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.822074467059804e-05,
+      "loss": 1.7119,
+      "step": 4620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8219788677643796e-05,
+      "loss": 1.5133,
+      "step": 4621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.82188324374129e-05,
+      "loss": 1.4587,
+      "step": 4622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821787594991554e-05,
+      "loss": 1.6393,
+      "step": 4623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8216919215161894e-05,
+      "loss": 1.5221,
+      "step": 4624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821596223316215e-05,
+      "loss": 1.5314,
+      "step": 4625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821500500392651e-05,
+      "loss": 1.5682,
+      "step": 4626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821404752746516e-05,
+      "loss": 1.5615,
+      "step": 4627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8213089803788294e-05,
+      "loss": 1.5622,
+      "step": 4628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8212131832906115e-05,
+      "loss": 1.5648,
+      "step": 4629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821117361482883e-05,
+      "loss": 1.4982,
+      "step": 4630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8210215149566634e-05,
+      "loss": 1.5195,
+      "step": 4631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8209256437129735e-05,
+      "loss": 1.4549,
+      "step": 4632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820829747752835e-05,
+      "loss": 1.5326,
+      "step": 4633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820733827077268e-05,
+      "loss": 1.5325,
+      "step": 4634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8206378816872956e-05,
+      "loss": 1.5461,
+      "step": 4635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820541911583938e-05,
+      "loss": 1.5049,
+      "step": 4636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820445916768219e-05,
+      "loss": 1.5839,
+      "step": 4637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8203498972411585e-05,
+      "loss": 1.5876,
+      "step": 4638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820253853003781e-05,
+      "loss": 1.5612,
+      "step": 4639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8201577840571075e-05,
+      "loss": 1.6125,
+      "step": 4640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820061690402163e-05,
+      "loss": 1.5329,
+      "step": 4641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819965572039971e-05,
+      "loss": 1.4556,
+      "step": 4642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819869428971553e-05,
+      "loss": 1.5304,
+      "step": 4643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819773261197934e-05,
+      "loss": 1.4801,
+      "step": 4644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8196770687201385e-05,
+      "loss": 1.5272,
+      "step": 4645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819580851539191e-05,
+      "loss": 1.5638,
+      "step": 4646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819484609656114e-05,
+      "loss": 1.5103,
+      "step": 4647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819388343071936e-05,
+      "loss": 1.5601,
+      "step": 4648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8192920517876796e-05,
+      "loss": 1.5337,
+      "step": 4649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8191957358043704e-05,
+      "loss": 1.4842,
+      "step": 4650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8190993951230354e-05,
+      "loss": 1.5347,
+      "step": 4651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819003029744698e-05,
+      "loss": 1.4863,
+      "step": 4652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8189066396703886e-05,
+      "loss": 1.4821,
+      "step": 4653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.81881022490113e-05,
+      "loss": 1.513,
+      "step": 4654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.818713785437951e-05,
+      "loss": 1.4771,
+      "step": 4655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.818617321281877e-05,
+      "loss": 1.5286,
+      "step": 4656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8185208324339374e-05,
+      "loss": 1.5031,
+      "step": 4657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8184243188951586e-05,
+      "loss": 1.5633,
+      "step": 4658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8183277806665675e-05,
+      "loss": 1.5706,
+      "step": 4659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8182312177491926e-05,
+      "loss": 1.4969,
+      "step": 4660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.818134630144063e-05,
+      "loss": 1.4196,
+      "step": 4661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8180380178522076e-05,
+      "loss": 1.5599,
+      "step": 4662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817941380874654e-05,
+      "loss": 1.5112,
+      "step": 4663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8178447192124326e-05,
+      "loss": 1.5099,
+      "step": 4664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8177480328665713e-05,
+      "loss": 1.5186,
+      "step": 4665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8176513218381015e-05,
+      "loss": 1.5236,
+      "step": 4666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817554586128051e-05,
+      "loss": 1.4842,
+      "step": 4667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817457825737452e-05,
+      "loss": 1.5098,
+      "step": 4668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817361040667335e-05,
+      "loss": 1.4766,
+      "step": 4669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817264230918729e-05,
+      "loss": 1.4943,
+      "step": 4670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817167396492665e-05,
+      "loss": 1.3714,
+      "step": 4671
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817070537390176e-05,
+      "loss": 1.5306,
+      "step": 4672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.816973653612292e-05,
+      "loss": 1.4579,
+      "step": 4673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8168767451600456e-05,
+      "loss": 1.5773,
+      "step": 4674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8167798120344684e-05,
+      "loss": 1.4963,
+      "step": 4675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.816682854236593e-05,
+      "loss": 1.5386,
+      "step": 4676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.816585871767451e-05,
+      "loss": 1.529,
+      "step": 4677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8164888646280756e-05,
+      "loss": 1.5573,
+      "step": 4678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8163918328195015e-05,
+      "loss": 1.5397,
+      "step": 4679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.81629477634276e-05,
+      "loss": 1.5124,
+      "step": 4680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.816197695198885e-05,
+      "loss": 1.5025,
+      "step": 4681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.816100589388911e-05,
+      "loss": 1.499,
+      "step": 4682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.816003458913872e-05,
+      "loss": 1.4823,
+      "step": 4683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815906303774802e-05,
+      "loss": 1.4722,
+      "step": 4684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8158091239727364e-05,
+      "loss": 1.5848,
+      "step": 4685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8157119195087096e-05,
+      "loss": 1.5309,
+      "step": 4686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815614690383756e-05,
+      "loss": 1.5658,
+      "step": 4687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8155174365989125e-05,
+      "loss": 1.5347,
+      "step": 4688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815420158155214e-05,
+      "loss": 1.5374,
+      "step": 4689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815322855053697e-05,
+      "loss": 1.5132,
+      "step": 4690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815225527295396e-05,
+      "loss": 1.5474,
+      "step": 4691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8151281748813496e-05,
+      "loss": 1.4692,
+      "step": 4692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815030797812593e-05,
+      "loss": 1.5007,
+      "step": 4693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8149333960901647e-05,
+      "loss": 1.5285,
+      "step": 4694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8148359697151005e-05,
+      "loss": 1.495,
+      "step": 4695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8147385186884394e-05,
+      "loss": 1.5646,
+      "step": 4696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8146410430112175e-05,
+      "loss": 1.5228,
+      "step": 4697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.814543542684474e-05,
+      "loss": 1.5327,
+      "step": 4698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8144460177092475e-05,
+      "loss": 1.4665,
+      "step": 4699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.814348468086576e-05,
+      "loss": 1.5205,
+      "step": 4700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.814250893817498e-05,
+      "loss": 1.4992,
+      "step": 4701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.814153294903053e-05,
+      "loss": 1.5313,
+      "step": 4702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8140556713442805e-05,
+      "loss": 1.5306,
+      "step": 4703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8139580231422196e-05,
+      "loss": 1.445,
+      "step": 4704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813860350297911e-05,
+      "loss": 1.5444,
+      "step": 4705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813762652812394e-05,
+      "loss": 1.4997,
+      "step": 4706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.81366493068671e-05,
+      "loss": 1.4922,
+      "step": 4707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813567183921899e-05,
+      "loss": 1.514,
+      "step": 4708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813469412519002e-05,
+      "loss": 1.5225,
+      "step": 4709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8133716164790606e-05,
+      "loss": 1.5197,
+      "step": 4710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813273795803116e-05,
+      "loss": 1.4883,
+      "step": 4711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813175950492209e-05,
+      "loss": 1.4369,
+      "step": 4712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8130780805473827e-05,
+      "loss": 1.5378,
+      "step": 4713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.81298018596968e-05,
+      "loss": 1.5929,
+      "step": 4714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8128822667601424e-05,
+      "loss": 1.585,
+      "step": 4715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8127843229198125e-05,
+      "loss": 1.5667,
+      "step": 4716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8126863544497336e-05,
+      "loss": 1.5041,
+      "step": 4717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8125883613509494e-05,
+      "loss": 1.5002,
+      "step": 4718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8124903436245026e-05,
+      "loss": 1.5843,
+      "step": 4719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812392301271438e-05,
+      "loss": 1.5443,
+      "step": 4720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8122942342928e-05,
+      "loss": 1.5917,
+      "step": 4721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812196142689631e-05,
+      "loss": 1.5169,
+      "step": 4722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812098026462978e-05,
+      "loss": 1.5475,
+      "step": 4723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811999885613884e-05,
+      "loss": 1.5563,
+      "step": 4724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8119017201433946e-05,
+      "loss": 1.5073,
+      "step": 4725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811803530052556e-05,
+      "loss": 1.4772,
+      "step": 4726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8117053153424127e-05,
+      "loss": 1.5591,
+      "step": 4727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811607076014012e-05,
+      "loss": 1.5199,
+      "step": 4728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811508812068399e-05,
+      "loss": 1.5371,
+      "step": 4729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8114105235066206e-05,
+      "loss": 1.4795,
+      "step": 4730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811312210329724e-05,
+      "loss": 1.5171,
+      "step": 4731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8112138725387544e-05,
+      "loss": 1.5044,
+      "step": 4732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8111155101347605e-05,
+      "loss": 1.4401,
+      "step": 4733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.81101712311879e-05,
+      "loss": 1.4889,
+      "step": 4734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.81091871149189e-05,
+      "loss": 1.478,
+      "step": 4735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.810820275255109e-05,
+      "loss": 1.4661,
+      "step": 4736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.810721814409494e-05,
+      "loss": 1.5267,
+      "step": 4737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.810623328956096e-05,
+      "loss": 1.4995,
+      "step": 4738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.810524818895962e-05,
+      "loss": 1.5054,
+      "step": 4739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.81042628423014e-05,
+      "loss": 1.4855,
+      "step": 4740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8103277249596826e-05,
+      "loss": 1.4535,
+      "step": 4741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.810229141085637e-05,
+      "loss": 1.4943,
+      "step": 4742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.810130532609053e-05,
+      "loss": 1.6147,
+      "step": 4743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8100318995309826e-05,
+      "loss": 1.465,
+      "step": 4744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8099332418524735e-05,
+      "loss": 1.5049,
+      "step": 4745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8098345595745787e-05,
+      "loss": 1.5111,
+      "step": 4746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8097358526983476e-05,
+      "loss": 1.5711,
+      "step": 4747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8096371212248327e-05,
+      "loss": 1.5104,
+      "step": 4748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809538365155084e-05,
+      "loss": 1.5135,
+      "step": 4749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809439584490154e-05,
+      "loss": 1.5701,
+      "step": 4750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809340779231094e-05,
+      "loss": 1.4779,
+      "step": 4751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809241949378958e-05,
+      "loss": 1.519,
+      "step": 4752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809143094934796e-05,
+      "loss": 1.4201,
+      "step": 4753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809044215899663e-05,
+      "loss": 1.4707,
+      "step": 4754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.80894531227461e-05,
+      "loss": 1.5018,
+      "step": 4755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808846384060691e-05,
+      "loss": 1.5592,
+      "step": 4756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808747431258961e-05,
+      "loss": 1.5527,
+      "step": 4757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808648453870472e-05,
+      "loss": 1.4746,
+      "step": 4758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8085494518962785e-05,
+      "loss": 1.5838,
+      "step": 4759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808450425337435e-05,
+      "loss": 1.4852,
+      "step": 4760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808351374194996e-05,
+      "loss": 1.4945,
+      "step": 4761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8082522984700154e-05,
+      "loss": 1.5713,
+      "step": 4762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8081531981635504e-05,
+      "loss": 1.5267,
+      "step": 4763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808054073276655e-05,
+      "loss": 1.4849,
+      "step": 4764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807954923810384e-05,
+      "loss": 1.4811,
+      "step": 4765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807855749765795e-05,
+      "loss": 1.5194,
+      "step": 4766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8077565511439435e-05,
+      "loss": 1.5042,
+      "step": 4767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8076573279458855e-05,
+      "loss": 1.4998,
+      "step": 4768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8075580801726784e-05,
+      "loss": 1.5781,
+      "step": 4769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807458807825378e-05,
+      "loss": 1.48,
+      "step": 4770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807359510905043e-05,
+      "loss": 1.472,
+      "step": 4771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8072601894127295e-05,
+      "loss": 1.4046,
+      "step": 4772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807160843349496e-05,
+      "loss": 1.4855,
+      "step": 4773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8070614727164e-05,
+      "loss": 1.4808,
+      "step": 4774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8069620775144996e-05,
+      "loss": 1.4528,
+      "step": 4775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8068626577448544e-05,
+      "loss": 1.4871,
+      "step": 4776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806763213408522e-05,
+      "loss": 1.5623,
+      "step": 4777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806663744506562e-05,
+      "loss": 1.5645,
+      "step": 4778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806564251040034e-05,
+      "loss": 1.5832,
+      "step": 4779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8064647330099964e-05,
+      "loss": 1.4893,
+      "step": 4780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.80636519041751e-05,
+      "loss": 1.5648,
+      "step": 4781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806265623263634e-05,
+      "loss": 1.4844,
+      "step": 4782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.80616603154943e-05,
+      "loss": 1.5503,
+      "step": 4783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806066415275958e-05,
+      "loss": 1.4383,
+      "step": 4784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805966774444278e-05,
+      "loss": 1.534,
+      "step": 4785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805867109055452e-05,
+      "loss": 1.5225,
+      "step": 4786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805767419110542e-05,
+      "loss": 1.5534,
+      "step": 4787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805667704610608e-05,
+      "loss": 1.5313,
+      "step": 4788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8055679655567123e-05,
+      "loss": 1.4334,
+      "step": 4789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805468201949919e-05,
+      "loss": 1.584,
+      "step": 4790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805368413791288e-05,
+      "loss": 1.5164,
+      "step": 4791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805268601081883e-05,
+      "loss": 1.5459,
+      "step": 4792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8051687638227674e-05,
+      "loss": 1.5116,
+      "step": 4793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805068902015003e-05,
+      "loss": 1.4972,
+      "step": 4794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804969015659655e-05,
+      "loss": 1.4536,
+      "step": 4795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804869104757787e-05,
+      "loss": 1.5244,
+      "step": 4796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804769169310462e-05,
+      "loss": 1.4878,
+      "step": 4797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804669209318744e-05,
+      "loss": 1.507,
+      "step": 4798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804569224783698e-05,
+      "loss": 1.5223,
+      "step": 4799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.80446921570639e-05,
+      "loss": 1.5452,
+      "step": 4800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804369182087883e-05,
+      "loss": 1.5036,
+      "step": 4801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8042691239292434e-05,
+      "loss": 1.5799,
+      "step": 4802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.804169041231537e-05,
+      "loss": 1.5153,
+      "step": 4803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.804068933995829e-05,
+      "loss": 1.4985,
+      "step": 4804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803968802223186e-05,
+      "loss": 1.4769,
+      "step": 4805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803868645914673e-05,
+      "loss": 1.5173,
+      "step": 4806
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8037684650713585e-05,
+      "loss": 1.4644,
+      "step": 4807
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8036682596943086e-05,
+      "loss": 1.5181,
+      "step": 4808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803568029784589e-05,
+      "loss": 1.4489,
+      "step": 4809
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.80346777534327e-05,
+      "loss": 1.556,
+      "step": 4810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803367496371417e-05,
+      "loss": 1.4603,
+      "step": 4811
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803267192870098e-05,
+      "loss": 1.488,
+      "step": 4812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8031668648403825e-05,
+      "loss": 1.5501,
+      "step": 4813
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803066512283338e-05,
+      "loss": 1.4956,
+      "step": 4814
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8029661352000335e-05,
+      "loss": 1.4997,
+      "step": 4815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8028657335915376e-05,
+      "loss": 1.5129,
+      "step": 4816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.80276530745892e-05,
+      "loss": 1.5052,
+      "step": 4817
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8026648568032495e-05,
+      "loss": 1.4925,
+      "step": 4818
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802564381625597e-05,
+      "loss": 1.5369,
+      "step": 4819
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802463881927031e-05,
+      "loss": 1.4797,
+      "step": 4820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802363357708623e-05,
+      "loss": 1.4909,
+      "step": 4821
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802262808971443e-05,
+      "loss": 1.4891,
+      "step": 4822
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802162235716562e-05,
+      "loss": 1.5915,
+      "step": 4823
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.80206163794505e-05,
+      "loss": 1.5226,
+      "step": 4824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.80196101565798e-05,
+      "loss": 1.4746,
+      "step": 4825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8018603688564226e-05,
+      "loss": 1.5573,
+      "step": 4826
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8017596975414494e-05,
+      "loss": 1.4569,
+      "step": 4827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801659001714133e-05,
+      "loss": 1.5977,
+      "step": 4828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8015582813755454e-05,
+      "loss": 1.492,
+      "step": 4829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.80145753652676e-05,
+      "loss": 1.3973,
+      "step": 4830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801356767168849e-05,
+      "loss": 1.5646,
+      "step": 4831
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801255973302885e-05,
+      "loss": 1.4606,
+      "step": 4832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801155154929943e-05,
+      "loss": 1.5233,
+      "step": 4833
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801054312051094e-05,
+      "loss": 1.5306,
+      "step": 4834
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800953444667415e-05,
+      "loss": 1.4823,
+      "step": 4835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800852552779978e-05,
+      "loss": 1.5475,
+      "step": 4836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800751636389859e-05,
+      "loss": 1.5156,
+      "step": 4837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800650695498132e-05,
+      "loss": 1.5181,
+      "step": 4838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800549730105871e-05,
+      "loss": 1.495,
+      "step": 4839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8004487402141525e-05,
+      "loss": 1.4658,
+      "step": 4840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800347725824052e-05,
+      "loss": 1.5054,
+      "step": 4841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8002466869366435e-05,
+      "loss": 1.5295,
+      "step": 4842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800145623553005e-05,
+      "loss": 1.4959,
+      "step": 4843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800044535674213e-05,
+      "loss": 1.4639,
+      "step": 4844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.799943423301342e-05,
+      "loss": 1.5142,
+      "step": 4845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7998422864354694e-05,
+      "loss": 1.4855,
+      "step": 4846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7997411250776734e-05,
+      "loss": 1.5596,
+      "step": 4847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.799639939229031e-05,
+      "loss": 1.5166,
+      "step": 4848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.799538728890619e-05,
+      "loss": 1.4906,
+      "step": 4849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7994374940635155e-05,
+      "loss": 1.502,
+      "step": 4850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.799336234748799e-05,
+      "loss": 1.4762,
+      "step": 4851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7992349509475476e-05,
+      "loss": 1.4723,
+      "step": 4852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.799133642660839e-05,
+      "loss": 1.5588,
+      "step": 4853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.799032309889754e-05,
+      "loss": 1.4777,
+      "step": 4854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7989309526353696e-05,
+      "loss": 1.4631,
+      "step": 4855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7988295708987674e-05,
+      "loss": 1.4518,
+      "step": 4856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798728164681026e-05,
+      "loss": 1.498,
+      "step": 4857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798626733983225e-05,
+      "loss": 1.5143,
+      "step": 4858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7985252788064436e-05,
+      "loss": 1.472,
+      "step": 4859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798423799151765e-05,
+      "loss": 1.5318,
+      "step": 4860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798322295020268e-05,
+      "loss": 1.4857,
+      "step": 4861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798220766413033e-05,
+      "loss": 1.5392,
+      "step": 4862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7981192133311424e-05,
+      "loss": 1.5438,
+      "step": 4863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798017635775678e-05,
+      "loss": 1.547,
+      "step": 4864
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7979160337477205e-05,
+      "loss": 1.4821,
+      "step": 4865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.797814407248353e-05,
+      "loss": 1.5208,
+      "step": 4866
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7977127562786565e-05,
+      "loss": 1.5005,
+      "step": 4867
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.797611080839714e-05,
+      "loss": 1.5067,
+      "step": 4868
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.797509380932609e-05,
+      "loss": 1.5404,
+      "step": 4869
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7974076565584234e-05,
+      "loss": 1.5161,
+      "step": 4870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.797305907718241e-05,
+      "loss": 1.4795,
+      "step": 4871
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.797204134413146e-05,
+      "loss": 1.5104,
+      "step": 4872
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7971023366442216e-05,
+      "loss": 1.4668,
+      "step": 4873
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.797000514412552e-05,
+      "loss": 1.4683,
+      "step": 4874
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7968986677192205e-05,
+      "loss": 1.5111,
+      "step": 4875
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796796796565313e-05,
+      "loss": 1.5511,
+      "step": 4876
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796694900951914e-05,
+      "loss": 1.4935,
+      "step": 4877
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.79659298088011e-05,
+      "loss": 1.4456,
+      "step": 4878
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7964910363509833e-05,
+      "loss": 1.5905,
+      "step": 4879
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796389067365622e-05,
+      "loss": 1.513,
+      "step": 4880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7962870739251114e-05,
+      "loss": 1.4108,
+      "step": 4881
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796185056030538e-05,
+      "loss": 1.5177,
+      "step": 4882
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796083013682987e-05,
+      "loss": 1.4622,
+      "step": 4883
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795980946883546e-05,
+      "loss": 1.4455,
+      "step": 4884
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795878855633302e-05,
+      "loss": 1.5841,
+      "step": 4885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795776739933342e-05,
+      "loss": 1.5334,
+      "step": 4886
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795674599784754e-05,
+      "loss": 1.4785,
+      "step": 4887
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795572435188624e-05,
+      "loss": 1.4878,
+      "step": 4888
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795470246146043e-05,
+      "loss": 1.5013,
+      "step": 4889
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7953680326580966e-05,
+      "loss": 1.5077,
+      "step": 4890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795265794725874e-05,
+      "loss": 1.4886,
+      "step": 4891
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795163532350464e-05,
+      "loss": 1.481,
+      "step": 4892
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795061245532956e-05,
+      "loss": 1.4551,
+      "step": 4893
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.79495893427444e-05,
+      "loss": 1.4784,
+      "step": 4894
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.794856598576003e-05,
+      "loss": 1.4696,
+      "step": 4895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7947542384387376e-05,
+      "loss": 1.4797,
+      "step": 4896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7946518538637326e-05,
+      "loss": 1.5384,
+      "step": 4897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7945494448520785e-05,
+      "loss": 1.4626,
+      "step": 4898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7944470114048656e-05,
+      "loss": 1.5483,
+      "step": 4899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7943445535231855e-05,
+      "loss": 1.4896,
+      "step": 4900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7942420712081285e-05,
+      "loss": 1.4938,
+      "step": 4901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.794139564460786e-05,
+      "loss": 1.4886,
+      "step": 4902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7940370332822505e-05,
+      "loss": 1.4831,
+      "step": 4903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793934477673613e-05,
+      "loss": 1.4571,
+      "step": 4904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7938318976359656e-05,
+      "loss": 1.5145,
+      "step": 4905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793729293170402e-05,
+      "loss": 1.4375,
+      "step": 4906
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793626664278014e-05,
+      "loss": 1.5283,
+      "step": 4907
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793524010959893e-05,
+      "loss": 1.5402,
+      "step": 4908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793421333217135e-05,
+      "loss": 1.4465,
+      "step": 4909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7933186310508325e-05,
+      "loss": 1.534,
+      "step": 4910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7932159044620784e-05,
+      "loss": 1.5532,
+      "step": 4911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793113153451967e-05,
+      "loss": 1.5402,
+      "step": 4912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793010378021593e-05,
+      "loss": 1.5413,
+      "step": 4913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.79290757817205e-05,
+      "loss": 1.5062,
+      "step": 4914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.792804753904434e-05,
+      "loss": 1.5436,
+      "step": 4915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7927019052198394e-05,
+      "loss": 1.5042,
+      "step": 4916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7925990321193614e-05,
+      "loss": 1.5518,
+      "step": 4917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.792496134604095e-05,
+      "loss": 1.5643,
+      "step": 4918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.792393212675137e-05,
+      "loss": 1.5138,
+      "step": 4919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7922902663335835e-05,
+      "loss": 1.5068,
+      "step": 4920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.792187295580529e-05,
+      "loss": 1.4666,
+      "step": 4921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7920843004170724e-05,
+      "loss": 1.5072,
+      "step": 4922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7919812808443096e-05,
+      "loss": 1.4811,
+      "step": 4923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.791878236863338e-05,
+      "loss": 1.5697,
+      "step": 4924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.791775168475253e-05,
+      "loss": 1.5332,
+      "step": 4925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.791672075681155e-05,
+      "loss": 1.5516,
+      "step": 4926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.791568958482141e-05,
+      "loss": 1.4998,
+      "step": 4927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.791465816879308e-05,
+      "loss": 1.4992,
+      "step": 4928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.791362650873756e-05,
+      "loss": 1.4525,
+      "step": 4929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.791259460466583e-05,
+      "loss": 1.542,
+      "step": 4930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.791156245658887e-05,
+      "loss": 1.6008,
+      "step": 4931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.791053006451768e-05,
+      "loss": 1.5202,
+      "step": 4932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790949742846326e-05,
+      "loss": 1.5158,
+      "step": 4933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.79084645484366e-05,
+      "loss": 1.5438,
+      "step": 4934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7907431424448704e-05,
+      "loss": 1.5085,
+      "step": 4935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790639805651056e-05,
+      "loss": 1.519,
+      "step": 4936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790536444463319e-05,
+      "loss": 1.4989,
+      "step": 4937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7904330588827595e-05,
+      "loss": 1.5146,
+      "step": 4938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7903296489104784e-05,
+      "loss": 1.5581,
+      "step": 4939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7902262145475774e-05,
+      "loss": 1.5666,
+      "step": 4940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790122755795157e-05,
+      "loss": 1.4735,
+      "step": 4941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.79001927265432e-05,
+      "loss": 1.5288,
+      "step": 4942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.789915765126168e-05,
+      "loss": 1.55,
+      "step": 4943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7898122332118036e-05,
+      "loss": 1.4881,
+      "step": 4944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7897086769123285e-05,
+      "loss": 1.4841,
+      "step": 4945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.789605096228846e-05,
+      "loss": 1.4697,
+      "step": 4946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.78950149116246e-05,
+      "loss": 1.4704,
+      "step": 4947
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.789397861714273e-05,
+      "loss": 1.4933,
+      "step": 4948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.789294207885388e-05,
+      "loss": 1.5174,
+      "step": 4949
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.78919052967691e-05,
+      "loss": 1.4941,
+      "step": 4950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7890868270899424e-05,
+      "loss": 1.5911,
+      "step": 4951
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.78898310012559e-05,
+      "loss": 1.4964,
+      "step": 4952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7888793487849584e-05,
+      "loss": 1.4736,
+      "step": 4953
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.78877557306915e-05,
+      "loss": 1.5252,
+      "step": 4954
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.788671772979272e-05,
+      "loss": 1.4787,
+      "step": 4955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.788567948516428e-05,
+      "loss": 1.4535,
+      "step": 4956
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.788464099681726e-05,
+      "loss": 1.5163,
+      "step": 4957
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.78836022647627e-05,
+      "loss": 1.436,
+      "step": 4958
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7882563289011674e-05,
+      "loss": 1.4372,
+      "step": 4959
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.788152406957524e-05,
+      "loss": 1.4753,
+      "step": 4960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7880484606464456e-05,
+      "loss": 1.5031,
+      "step": 4961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.787944489969042e-05,
+      "loss": 1.4887,
+      "step": 4962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.787840494926417e-05,
+      "loss": 1.5423,
+      "step": 4963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.787736475519681e-05,
+      "loss": 1.5806,
+      "step": 4964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.787632431749939e-05,
+      "loss": 1.5109,
+      "step": 4965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.787528363618301e-05,
+      "loss": 1.48,
+      "step": 4966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7874242711258754e-05,
+      "loss": 1.5444,
+      "step": 4967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.78732015427377e-05,
+      "loss": 1.5112,
+      "step": 4968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7872160130630935e-05,
+      "loss": 1.5617,
+      "step": 4969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.787111847494955e-05,
+      "loss": 1.514,
+      "step": 4970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7870076575704634e-05,
+      "loss": 1.443,
+      "step": 4971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786903443290729e-05,
+      "loss": 1.5111,
+      "step": 4972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786799204656862e-05,
+      "loss": 1.5241,
+      "step": 4973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786694941669971e-05,
+      "loss": 1.4948,
+      "step": 4974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786590654331167e-05,
+      "loss": 1.5308,
+      "step": 4975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786486342641561e-05,
+      "loss": 1.473,
+      "step": 4976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7863820066022636e-05,
+      "loss": 1.4886,
+      "step": 4977
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7862776462143866e-05,
+      "loss": 1.5887,
+      "step": 4978
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.78617326147904e-05,
+      "loss": 1.5226,
+      "step": 4979
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786068852397336e-05,
+      "loss": 1.4932,
+      "step": 4980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785964418970387e-05,
+      "loss": 1.5135,
+      "step": 4981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785859961199305e-05,
+      "loss": 1.4292,
+      "step": 4982
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7857554790852025e-05,
+      "loss": 1.5055,
+      "step": 4983
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785650972629191e-05,
+      "loss": 1.5382,
+      "step": 4984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7855464418323834e-05,
+      "loss": 1.5199,
+      "step": 4985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785441886695895e-05,
+      "loss": 1.5664,
+      "step": 4986
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7853373072208384e-05,
+      "loss": 1.5535,
+      "step": 4987
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785232703408327e-05,
+      "loss": 1.5817,
+      "step": 4988
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785128075259474e-05,
+      "loss": 1.4959,
+      "step": 4989
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7850234227753943e-05,
+      "loss": 1.5065,
+      "step": 4990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784918745957203e-05,
+      "loss": 1.508,
+      "step": 4991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784814044806014e-05,
+      "loss": 1.5311,
+      "step": 4992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784709319322942e-05,
+      "loss": 1.5182,
+      "step": 4993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784604569509104e-05,
+      "loss": 1.5579,
+      "step": 4994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784499795365613e-05,
+      "loss": 1.4562,
+      "step": 4995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784394996893587e-05,
+      "loss": 1.5008,
+      "step": 4996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784290174094141e-05,
+      "loss": 1.4797,
+      "step": 4997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784185326968391e-05,
+      "loss": 1.4551,
+      "step": 4998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7840804555174554e-05,
+      "loss": 1.4889,
+      "step": 4999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783975559742449e-05,
+      "loss": 1.5228,
+      "step": 5000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783870639644489e-05,
+      "loss": 1.5005,
+      "step": 5001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783765695224693e-05,
+      "loss": 1.5949,
+      "step": 5002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783660726484179e-05,
+      "loss": 1.4723,
+      "step": 5003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7835557334240653e-05,
+      "loss": 1.5461,
+      "step": 5004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7834507160454695e-05,
+      "loss": 1.4691,
+      "step": 5005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.78334567434951e-05,
+      "loss": 1.4517,
+      "step": 5006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783240608337304e-05,
+      "loss": 1.4873,
+      "step": 5007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783135518009973e-05,
+      "loss": 1.4956,
+      "step": 5008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7830304033686344e-05,
+      "loss": 1.4994,
+      "step": 5009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782925264414408e-05,
+      "loss": 1.4487,
+      "step": 5010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782820101148414e-05,
+      "loss": 1.4815,
+      "step": 5011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782714913571771e-05,
+      "loss": 1.5013,
+      "step": 5012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782609701685601e-05,
+      "loss": 1.5177,
+      "step": 5013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782504465491023e-05,
+      "loss": 1.5356,
+      "step": 5014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782399204989159e-05,
+      "loss": 1.4816,
+      "step": 5015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782293920181128e-05,
+      "loss": 1.4981,
+      "step": 5016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782188611068053e-05,
+      "loss": 1.5454,
+      "step": 5017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782083277651054e-05,
+      "loss": 1.5026,
+      "step": 5018
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.781977919931254e-05,
+      "loss": 1.4321,
+      "step": 5019
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.781872537909775e-05,
+      "loss": 1.5194,
+      "step": 5020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7817671315877385e-05,
+      "loss": 1.4719,
+      "step": 5021
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.781661700966267e-05,
+      "loss": 1.5181,
+      "step": 5022
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.781556246046485e-05,
+      "loss": 1.4282,
+      "step": 5023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7814507668295125e-05,
+      "loss": 1.4452,
+      "step": 5024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.781345263316476e-05,
+      "loss": 1.4864,
+      "step": 5025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.781239735508496e-05,
+      "loss": 1.4972,
+      "step": 5026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.781134183406698e-05,
+      "loss": 1.4961,
+      "step": 5027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.781028607012207e-05,
+      "loss": 1.478,
+      "step": 5028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7809230063261454e-05,
+      "loss": 1.57,
+      "step": 5029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780817381349638e-05,
+      "loss": 1.4805,
+      "step": 5030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7807117320838115e-05,
+      "loss": 1.5552,
+      "step": 5031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780606058529788e-05,
+      "loss": 1.5353,
+      "step": 5032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780500360688696e-05,
+      "loss": 1.4904,
+      "step": 5033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780394638561659e-05,
+      "loss": 1.4749,
+      "step": 5034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780288892149804e-05,
+      "loss": 1.5488,
+      "step": 5035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780183121454257e-05,
+      "loss": 1.5055,
+      "step": 5036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7800773264761436e-05,
+      "loss": 1.576,
+      "step": 5037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779971507216591e-05,
+      "loss": 1.4333,
+      "step": 5038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779865663676726e-05,
+      "loss": 1.5002,
+      "step": 5039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7797597958576754e-05,
+      "loss": 1.5125,
+      "step": 5040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7796539037605677e-05,
+      "loss": 1.4788,
+      "step": 5041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7795479873865293e-05,
+      "loss": 1.4924,
+      "step": 5042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7794420467366894e-05,
+      "loss": 1.6165,
+      "step": 5043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779336081812176e-05,
+      "loss": 1.4671,
+      "step": 5044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779230092614116e-05,
+      "loss": 1.5002,
+      "step": 5045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7791240791436395e-05,
+      "loss": 1.4715,
+      "step": 5046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779018041401876e-05,
+      "loss": 1.4746,
+      "step": 5047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.778911979389953e-05,
+      "loss": 1.4294,
+      "step": 5048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.778805893109002e-05,
+      "loss": 1.528,
+      "step": 5049
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.778699782560151e-05,
+      "loss": 1.5763,
+      "step": 5050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.778593647744531e-05,
+      "loss": 1.5815,
+      "step": 5051
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.778487488663272e-05,
+      "loss": 1.4504,
+      "step": 5052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7783813053175045e-05,
+      "loss": 1.4819,
+      "step": 5053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.778275097708359e-05,
+      "loss": 1.4409,
+      "step": 5054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.778168865836967e-05,
+      "loss": 1.4377,
+      "step": 5055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7780626097044604e-05,
+      "loss": 1.4626,
+      "step": 5056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.77795632931197e-05,
+      "loss": 1.5223,
+      "step": 5057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7778500246606273e-05,
+      "loss": 1.4994,
+      "step": 5058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777743695751564e-05,
+      "loss": 1.5099,
+      "step": 5059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777637342585914e-05,
+      "loss": 1.5213,
+      "step": 5060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7775309651648096e-05,
+      "loss": 1.4751,
+      "step": 5061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777424563489382e-05,
+      "loss": 1.5226,
+      "step": 5062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777318137560767e-05,
+      "loss": 1.5503,
+      "step": 5063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777211687380095e-05,
+      "loss": 1.5392,
+      "step": 5064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7771052129485015e-05,
+      "loss": 1.4769,
+      "step": 5065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7769987142671203e-05,
+      "loss": 1.5295,
+      "step": 5066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.776892191337085e-05,
+      "loss": 1.5311,
+      "step": 5067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7767856441595294e-05,
+      "loss": 1.508,
+      "step": 5068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.77667907273559e-05,
+      "loss": 1.52,
+      "step": 5069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7765724770664e-05,
+      "loss": 1.5967,
+      "step": 5070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.776465857153095e-05,
+      "loss": 1.5579,
+      "step": 5071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7763592129968115e-05,
+      "loss": 1.5514,
+      "step": 5072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7762525445986845e-05,
+      "loss": 1.529,
+      "step": 5073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.776145851959849e-05,
+      "loss": 1.4764,
+      "step": 5074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.776039135081443e-05,
+      "loss": 1.5225,
+      "step": 5075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775932393964602e-05,
+      "loss": 1.4596,
+      "step": 5076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775825628610462e-05,
+      "loss": 1.4342,
+      "step": 5077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7757188390201615e-05,
+      "loss": 1.536,
+      "step": 5078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7756120251948354e-05,
+      "loss": 1.4461,
+      "step": 5079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775505187135624e-05,
+      "loss": 1.5285,
+      "step": 5080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775398324843664e-05,
+      "loss": 1.4946,
+      "step": 5081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775291438320093e-05,
+      "loss": 1.5461,
+      "step": 5082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.77518452756605e-05,
+      "loss": 1.5506,
+      "step": 5083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775077592582673e-05,
+      "loss": 1.5653,
+      "step": 5084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7749706333710996e-05,
+      "loss": 1.592,
+      "step": 5085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774863649932472e-05,
+      "loss": 1.5532,
+      "step": 5086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774756642267926e-05,
+      "loss": 1.5023,
+      "step": 5087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7746496103786036e-05,
+      "loss": 1.5636,
+      "step": 5088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774542554265644e-05,
+      "loss": 1.4985,
+      "step": 5089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7744354739301866e-05,
+      "loss": 1.5277,
+      "step": 5090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774328369373373e-05,
+      "loss": 1.5124,
+      "step": 5091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774221240596342e-05,
+      "loss": 1.4878,
+      "step": 5092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774114087600237e-05,
+      "loss": 1.4942,
+      "step": 5093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774006910386196e-05,
+      "loss": 1.5143,
+      "step": 5094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7738997089553626e-05,
+      "loss": 1.467,
+      "step": 5095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7737924833088774e-05,
+      "loss": 1.5138,
+      "step": 5096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.773685233447883e-05,
+      "loss": 1.5127,
+      "step": 5097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.773577959373522e-05,
+      "loss": 1.4801,
+      "step": 5098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.773470661086935e-05,
+      "loss": 1.5234,
+      "step": 5099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.773363338589266e-05,
+      "loss": 1.4927,
+      "step": 5100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.773255991881658e-05,
+      "loss": 1.4774,
+      "step": 5101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7731486209652544e-05,
+      "loss": 1.4733,
+      "step": 5102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.773041225841197e-05,
+      "loss": 1.4696,
+      "step": 5103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772933806510631e-05,
+      "loss": 1.5729,
+      "step": 5104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7728263629746994e-05,
+      "loss": 1.501,
+      "step": 5105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772718895234547e-05,
+      "loss": 1.5363,
+      "step": 5106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772611403291318e-05,
+      "loss": 1.5382,
+      "step": 5107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772503887146158e-05,
+      "loss": 1.4406,
+      "step": 5108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7723963468002107e-05,
+      "loss": 1.5708,
+      "step": 5109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772288782254622e-05,
+      "loss": 1.5158,
+      "step": 5110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7721811935105384e-05,
+      "loss": 1.4658,
+      "step": 5111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772073580569103e-05,
+      "loss": 1.5252,
+      "step": 5112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7719659434314644e-05,
+      "loss": 1.471,
+      "step": 5113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.771858282098768e-05,
+      "loss": 1.548,
+      "step": 5114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.771750596572159e-05,
+      "loss": 1.528,
+      "step": 5115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7716428868527865e-05,
+      "loss": 1.5184,
+      "step": 5116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7715351529417954e-05,
+      "loss": 1.4972,
+      "step": 5117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.771427394840334e-05,
+      "loss": 1.4893,
+      "step": 5118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7713196125495505e-05,
+      "loss": 1.4727,
+      "step": 5119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7712118060705926e-05,
+      "loss": 1.487,
+      "step": 5120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7711039754046075e-05,
+      "loss": 1.4707,
+      "step": 5121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.770996120552743e-05,
+      "loss": 1.502,
+      "step": 5122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7708882415161494e-05,
+      "loss": 1.5479,
+      "step": 5123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.770780338295974e-05,
+      "loss": 1.5423,
+      "step": 5124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7706724108933676e-05,
+      "loss": 1.4526,
+      "step": 5125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.770564459309478e-05,
+      "loss": 1.4907,
+      "step": 5126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7704564835454555e-05,
+      "loss": 1.4824,
+      "step": 5127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.770348483602449e-05,
+      "loss": 1.4852,
+      "step": 5128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.770240459481611e-05,
+      "loss": 1.4832,
+      "step": 5129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.77013241118409e-05,
+      "loss": 1.4246,
+      "step": 5130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7700243387110366e-05,
+      "loss": 1.4844,
+      "step": 5131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769916242063602e-05,
+      "loss": 1.4779,
+      "step": 5132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769808121242938e-05,
+      "loss": 1.4601,
+      "step": 5133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769699976250195e-05,
+      "loss": 1.4821,
+      "step": 5134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7695918070865264e-05,
+      "loss": 1.5329,
+      "step": 5135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7694836137530816e-05,
+      "loss": 1.4997,
+      "step": 5136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769375396251015e-05,
+      "loss": 1.4634,
+      "step": 5137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769267154581478e-05,
+      "loss": 1.4929,
+      "step": 5138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7691588887456236e-05,
+      "loss": 1.4797,
+      "step": 5139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769050598744605e-05,
+      "loss": 1.4831,
+      "step": 5140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768942284579574e-05,
+      "loss": 1.5954,
+      "step": 5141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7688339462516854e-05,
+      "loss": 1.4741,
+      "step": 5142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768725583762094e-05,
+      "loss": 1.4565,
+      "step": 5143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768617197111951e-05,
+      "loss": 1.5407,
+      "step": 5144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768508786302413e-05,
+      "loss": 1.5622,
+      "step": 5145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768400351334633e-05,
+      "loss": 1.5104,
+      "step": 5146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768291892209768e-05,
+      "loss": 1.563,
+      "step": 5147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.76818340892897e-05,
+      "loss": 1.4795,
+      "step": 5148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7680749014933956e-05,
+      "loss": 1.5933,
+      "step": 5149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767966369904201e-05,
+      "loss": 1.5352,
+      "step": 5150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767857814162542e-05,
+      "loss": 1.5324,
+      "step": 5151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767749234269574e-05,
+      "loss": 1.5098,
+      "step": 5152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767640630226452e-05,
+      "loss": 1.5231,
+      "step": 5153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7675320020343355e-05,
+      "loss": 1.4774,
+      "step": 5154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767423349694379e-05,
+      "loss": 1.4792,
+      "step": 5155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767314673207741e-05,
+      "loss": 1.4763,
+      "step": 5156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767205972575578e-05,
+      "loss": 1.4554,
+      "step": 5157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7670972477990485e-05,
+      "loss": 1.4982,
+      "step": 5158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766988498879309e-05,
+      "loss": 1.535,
+      "step": 5159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766879725817518e-05,
+      "loss": 1.541,
+      "step": 5160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7667709286148345e-05,
+      "loss": 1.4969,
+      "step": 5161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766662107272417e-05,
+      "loss": 1.4727,
+      "step": 5162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7665532617914235e-05,
+      "loss": 1.4404,
+      "step": 5163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766444392173015e-05,
+      "loss": 1.5303,
+      "step": 5164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766335498418349e-05,
+      "loss": 1.4696,
+      "step": 5165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7662265805285855e-05,
+      "loss": 1.5391,
+      "step": 5166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7661176385048857e-05,
+      "loss": 1.4808,
+      "step": 5167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766008672348408e-05,
+      "loss": 1.4277,
+      "step": 5168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.765899682060314e-05,
+      "loss": 1.4878,
+      "step": 5169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.765790667641764e-05,
+      "loss": 1.5145,
+      "step": 5170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.765681629093919e-05,
+      "loss": 1.5551,
+      "step": 5171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7655725664179405e-05,
+      "loss": 1.4198,
+      "step": 5172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.765463479614989e-05,
+      "loss": 1.4427,
+      "step": 5173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.765354368686228e-05,
+      "loss": 1.4313,
+      "step": 5174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7652452336328174e-05,
+      "loss": 1.5278,
+      "step": 5175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7651360744559206e-05,
+      "loss": 1.5399,
+      "step": 5176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7650268911567e-05,
+      "loss": 1.4439,
+      "step": 5177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.764917683736318e-05,
+      "loss": 1.4479,
+      "step": 5178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7648084521959366e-05,
+      "loss": 1.5736,
+      "step": 5179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.764699196536721e-05,
+      "loss": 1.5233,
+      "step": 5180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.764589916759834e-05,
+      "loss": 1.4616,
+      "step": 5181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.764480612866439e-05,
+      "loss": 1.4963,
+      "step": 5182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.764371284857701e-05,
+      "loss": 1.4994,
+      "step": 5183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.764261932734783e-05,
+      "loss": 1.4785,
+      "step": 5184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.76415255649885e-05,
+      "loss": 1.5555,
+      "step": 5185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7640431561510674e-05,
+      "loss": 1.5211,
+      "step": 5186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7639337316925994e-05,
+      "loss": 1.4995,
+      "step": 5187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763824283124612e-05,
+      "loss": 1.548,
+      "step": 5188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7637148104482696e-05,
+      "loss": 1.5999,
+      "step": 5189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763605313664739e-05,
+      "loss": 1.5107,
+      "step": 5190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763495792775186e-05,
+      "loss": 1.472,
+      "step": 5191
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763386247780777e-05,
+      "loss": 1.4907,
+      "step": 5192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763276678682679e-05,
+      "loss": 1.5804,
+      "step": 5193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7631670854820586e-05,
+      "loss": 1.5173,
+      "step": 5194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763057468180082e-05,
+      "loss": 1.4915,
+      "step": 5195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762947826777918e-05,
+      "loss": 1.4855,
+      "step": 5196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762838161276734e-05,
+      "loss": 1.5026,
+      "step": 5197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762728471677696e-05,
+      "loss": 1.4476,
+      "step": 5198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7626187579819734e-05,
+      "loss": 1.5091,
+      "step": 5199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7625090201907364e-05,
+      "loss": 1.4447,
+      "step": 5200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7623992583051505e-05,
+      "loss": 1.472,
+      "step": 5201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762289472326387e-05,
+      "loss": 1.5181,
+      "step": 5202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762179662255614e-05,
+      "loss": 1.5369,
+      "step": 5203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762069828094e-05,
+      "loss": 1.4639,
+      "step": 5204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.761959969842717e-05,
+      "loss": 1.4336,
+      "step": 5205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.761850087502932e-05,
+      "loss": 1.4409,
+      "step": 5206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.761740181075818e-05,
+      "loss": 1.5527,
+      "step": 5207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7616302505625434e-05,
+      "loss": 1.5628,
+      "step": 5208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.76152029596428e-05,
+      "loss": 1.5405,
+      "step": 5209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7614103172821986e-05,
+      "loss": 1.4653,
+      "step": 5210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7613003145174706e-05,
+      "loss": 1.4396,
+      "step": 5211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7611902876712666e-05,
+      "loss": 1.4626,
+      "step": 5212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7610802367447596e-05,
+      "loss": 1.4779,
+      "step": 5213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.76097016173912e-05,
+      "loss": 1.4857,
+      "step": 5214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760860062655521e-05,
+      "loss": 1.514,
+      "step": 5215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760749939495135e-05,
+      "loss": 1.5205,
+      "step": 5216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760639792259134e-05,
+      "loss": 1.4971,
+      "step": 5217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760529620948693e-05,
+      "loss": 1.4147,
+      "step": 5218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760419425564983e-05,
+      "loss": 1.4771,
+      "step": 5219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760309206109179e-05,
+      "loss": 1.5142,
+      "step": 5220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760198962582454e-05,
+      "loss": 1.5038,
+      "step": 5221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760088694985982e-05,
+      "loss": 1.4596,
+      "step": 5222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759978403320937e-05,
+      "loss": 1.506,
+      "step": 5223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759868087588496e-05,
+      "loss": 1.5073,
+      "step": 5224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.75975774778983e-05,
+      "loss": 1.4172,
+      "step": 5225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7596473839261166e-05,
+      "loss": 1.4816,
+      "step": 5226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.75953699599853e-05,
+      "loss": 1.498,
+      "step": 5227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759426584008247e-05,
+      "loss": 1.527,
+      "step": 5228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759316147956442e-05,
+      "loss": 1.5477,
+      "step": 5229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759205687844293e-05,
+      "loss": 1.4606,
+      "step": 5230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759095203672973e-05,
+      "loss": 1.5254,
+      "step": 5231
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758984695443662e-05,
+      "loss": 1.5124,
+      "step": 5232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7588741631575344e-05,
+      "loss": 1.4823,
+      "step": 5233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7587636068157695e-05,
+      "loss": 1.4741,
+      "step": 5234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758653026419543e-05,
+      "loss": 1.485,
+      "step": 5235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758542421970033e-05,
+      "loss": 1.5576,
+      "step": 5236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758431793468417e-05,
+      "loss": 1.4681,
+      "step": 5237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758321140915874e-05,
+      "loss": 1.4168,
+      "step": 5238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758210464313582e-05,
+      "loss": 1.5187,
+      "step": 5239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758099763662719e-05,
+      "loss": 1.5111,
+      "step": 5240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.757989038964466e-05,
+      "loss": 1.5524,
+      "step": 5241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757878290219999e-05,
+      "loss": 1.4499,
+      "step": 5242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757767517430499e-05,
+      "loss": 1.4621,
+      "step": 5243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757656720597146e-05,
+      "loss": 1.5077,
+      "step": 5244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7575458997211195e-05,
+      "loss": 1.5398,
+      "step": 5245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7574350548036e-05,
+      "loss": 1.5522,
+      "step": 5246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757324185845768e-05,
+      "loss": 1.4466,
+      "step": 5247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757213292848803e-05,
+      "loss": 1.5138,
+      "step": 5248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757102375813888e-05,
+      "loss": 1.5433,
+      "step": 5249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756991434742202e-05,
+      "loss": 1.5448,
+      "step": 5250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756880469634928e-05,
+      "loss": 1.5148,
+      "step": 5251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756769480493247e-05,
+      "loss": 1.4878,
+      "step": 5252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7566584673183415e-05,
+      "loss": 1.5109,
+      "step": 5253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7565474301113924e-05,
+      "loss": 1.5466,
+      "step": 5254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756436368873584e-05,
+      "loss": 1.5801,
+      "step": 5255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756325283606098e-05,
+      "loss": 1.4432,
+      "step": 5256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756214174310117e-05,
+      "loss": 1.497,
+      "step": 5257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7561030409868255e-05,
+      "loss": 1.4631,
+      "step": 5258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.755991883637406e-05,
+      "loss": 1.5114,
+      "step": 5259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.755880702263043e-05,
+      "loss": 1.4867,
+      "step": 5260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.75576949686492e-05,
+      "loss": 1.5064,
+      "step": 5261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7556582674442214e-05,
+      "loss": 1.4652,
+      "step": 5262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7555470140021316e-05,
+      "loss": 1.5778,
+      "step": 5263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.755435736539835e-05,
+      "loss": 1.5508,
+      "step": 5264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.755324435058518e-05,
+      "loss": 1.5155,
+      "step": 5265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7552131095593645e-05,
+      "loss": 1.5083,
+      "step": 5266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.75510176004356e-05,
+      "loss": 1.4601,
+      "step": 5267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.754990386512291e-05,
+      "loss": 1.5335,
+      "step": 5268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.754878988966745e-05,
+      "loss": 1.4243,
+      "step": 5269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.754767567408105e-05,
+      "loss": 1.5158,
+      "step": 5270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.75465612183756e-05,
+      "loss": 1.4814,
+      "step": 5271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.754544652256296e-05,
+      "loss": 1.4596,
+      "step": 5272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7544331586655e-05,
+      "loss": 1.5464,
+      "step": 5273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.75432164106636e-05,
+      "loss": 1.5163,
+      "step": 5274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7542100994600635e-05,
+      "loss": 1.4688,
+      "step": 5275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7540985338477974e-05,
+      "loss": 1.5361,
+      "step": 5276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7539869442307504e-05,
+      "loss": 1.554,
+      "step": 5277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.753875330610111e-05,
+      "loss": 1.5915,
+      "step": 5278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.753763692987068e-05,
+      "loss": 1.5218,
+      "step": 5279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7536520313628094e-05,
+      "loss": 1.5428,
+      "step": 5280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.753540345738525e-05,
+      "loss": 1.5413,
+      "step": 5281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.753428636115404e-05,
+      "loss": 1.486,
+      "step": 5282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.753316902494637e-05,
+      "loss": 1.4676,
+      "step": 5283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.753205144877412e-05,
+      "loss": 1.4639,
+      "step": 5284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.75309336326492e-05,
+      "loss": 1.4531,
+      "step": 5285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752981557658353e-05,
+      "loss": 1.4875,
+      "step": 5286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7528697280588984e-05,
+      "loss": 1.6038,
+      "step": 5287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7527578744677495e-05,
+      "loss": 1.4707,
+      "step": 5288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752645996886098e-05,
+      "loss": 1.5656,
+      "step": 5289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752534095315133e-05,
+      "loss": 1.4635,
+      "step": 5290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752422169756048e-05,
+      "loss": 1.512,
+      "step": 5291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7523102202100345e-05,
+      "loss": 1.5162,
+      "step": 5292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752198246678284e-05,
+      "loss": 1.4663,
+      "step": 5293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.75208624916199e-05,
+      "loss": 1.487,
+      "step": 5294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7519742276623444e-05,
+      "loss": 1.5016,
+      "step": 5295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7518621821805406e-05,
+      "loss": 1.4997,
+      "step": 5296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7517501127177724e-05,
+      "loss": 1.5627,
+      "step": 5297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.751638019275232e-05,
+      "loss": 1.4172,
+      "step": 5298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7515259018541135e-05,
+      "loss": 1.5168,
+      "step": 5299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7514137604556106e-05,
+      "loss": 1.4924,
+      "step": 5300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.751301595080919e-05,
+      "loss": 1.4758,
+      "step": 5301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.751189405731232e-05,
+      "loss": 1.534,
+      "step": 5302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7510771924077444e-05,
+      "loss": 1.4282,
+      "step": 5303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7509649551116516e-05,
+      "loss": 1.4911,
+      "step": 5304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7508526938441475e-05,
+      "loss": 1.4183,
+      "step": 5305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7507404086064297e-05,
+      "loss": 1.4894,
+      "step": 5306
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.750628099399693e-05,
+      "loss": 1.4505,
+      "step": 5307
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7505157662251344e-05,
+      "loss": 1.5146,
+      "step": 5308
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.750403409083948e-05,
+      "loss": 1.5837,
+      "step": 5309
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.750291027977332e-05,
+      "loss": 1.5829,
+      "step": 5310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7501786229064816e-05,
+      "loss": 1.5073,
+      "step": 5311
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.750066193872596e-05,
+      "loss": 1.4985,
+      "step": 5312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749953740876871e-05,
+      "loss": 1.4896,
+      "step": 5313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749841263920505e-05,
+      "loss": 1.45,
+      "step": 5314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749728763004696e-05,
+      "loss": 1.5034,
+      "step": 5315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7496162381306407e-05,
+      "loss": 1.599,
+      "step": 5316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7495036892995385e-05,
+      "loss": 1.4976,
+      "step": 5317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7493911165125874e-05,
+      "loss": 1.4443,
+      "step": 5318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749278519770987e-05,
+      "loss": 1.4989,
+      "step": 5319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749165899075936e-05,
+      "loss": 1.458,
+      "step": 5320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749053254428634e-05,
+      "loss": 1.5238,
+      "step": 5321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7489405858302795e-05,
+      "loss": 1.5249,
+      "step": 5322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7488278932820726e-05,
+      "loss": 1.4684,
+      "step": 5323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.748715176785215e-05,
+      "loss": 1.5501,
+      "step": 5324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7486024363409055e-05,
+      "loss": 1.4712,
+      "step": 5325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7484896719503456e-05,
+      "loss": 1.5176,
+      "step": 5326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7483768836147354e-05,
+      "loss": 1.568,
+      "step": 5327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7482640713352765e-05,
+      "loss": 1.4707,
+      "step": 5328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.748151235113171e-05,
+      "loss": 1.5329,
+      "step": 5329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.748038374949618e-05,
+      "loss": 1.4774,
+      "step": 5330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7479254908458224e-05,
+      "loss": 1.5008,
+      "step": 5331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.747812582802985e-05,
+      "loss": 1.5392,
+      "step": 5332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7476996508223084e-05,
+      "loss": 1.472,
+      "step": 5333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7475866949049945e-05,
+      "loss": 1.5679,
+      "step": 5334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7474737150522474e-05,
+      "loss": 1.4891,
+      "step": 5335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.74736071126527e-05,
+      "loss": 1.5177,
+      "step": 5336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.747247683545264e-05,
+      "loss": 1.5252,
+      "step": 5337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.747134631893436e-05,
+      "loss": 1.4977,
+      "step": 5338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7470215563109874e-05,
+      "loss": 1.4819,
+      "step": 5339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7469084567991237e-05,
+      "loss": 1.4964,
+      "step": 5340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.74679533335905e-05,
+      "loss": 1.459,
+      "step": 5341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746682185991969e-05,
+      "loss": 1.4889,
+      "step": 5342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7465690146990873e-05,
+      "loss": 1.4653,
+      "step": 5343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746455819481609e-05,
+      "loss": 1.4424,
+      "step": 5344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7463426003407405e-05,
+      "loss": 1.4964,
+      "step": 5345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746229357277686e-05,
+      "loss": 1.5513,
+      "step": 5346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746116090293654e-05,
+      "loss": 1.5182,
+      "step": 5347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746002799389849e-05,
+      "loss": 1.5046,
+      "step": 5348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.745889484567477e-05,
+      "loss": 1.5033,
+      "step": 5349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7457761458277464e-05,
+      "loss": 1.4979,
+      "step": 5350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.745662783171863e-05,
+      "loss": 1.5171,
+      "step": 5351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.745549396601034e-05,
+      "loss": 1.4933,
+      "step": 5352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7454359861164674e-05,
+      "loss": 1.5815,
+      "step": 5353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.745322551719371e-05,
+      "loss": 1.4702,
+      "step": 5354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7452090934109525e-05,
+      "loss": 1.5436,
+      "step": 5355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7450956111924206e-05,
+      "loss": 1.5658,
+      "step": 5356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7449821050649824e-05,
+      "loss": 1.5272,
+      "step": 5357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.744868575029848e-05,
+      "loss": 1.5226,
+      "step": 5358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7447550210882275e-05,
+      "loss": 1.4652,
+      "step": 5359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.744641443241328e-05,
+      "loss": 1.5107,
+      "step": 5360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7445278414903596e-05,
+      "loss": 1.4613,
+      "step": 5361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.744414215836532e-05,
+      "loss": 1.4671,
+      "step": 5362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7443005662810566e-05,
+      "loss": 1.4935,
+      "step": 5363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.744186892825142e-05,
+      "loss": 1.4902,
+      "step": 5364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.74407319547e-05,
+      "loss": 1.4881,
+      "step": 5365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7439594742168414e-05,
+      "loss": 1.4961,
+      "step": 5366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743845729066876e-05,
+      "loss": 1.5257,
+      "step": 5367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7437319600213156e-05,
+      "loss": 1.5122,
+      "step": 5368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7436181670813726e-05,
+      "loss": 1.4823,
+      "step": 5369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743504350248258e-05,
+      "loss": 1.4684,
+      "step": 5370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743390509523183e-05,
+      "loss": 1.5444,
+      "step": 5371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743276644907363e-05,
+      "loss": 1.5,
+      "step": 5372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743162756402008e-05,
+      "loss": 1.5008,
+      "step": 5373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7430488440083306e-05,
+      "loss": 1.6105,
+      "step": 5374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742934907727546e-05,
+      "loss": 1.4028,
+      "step": 5375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742820947560866e-05,
+      "loss": 1.5469,
+      "step": 5376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7427069635095045e-05,
+      "loss": 1.4408,
+      "step": 5377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742592955574675e-05,
+      "loss": 1.5062,
+      "step": 5378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742478923757593e-05,
+      "loss": 1.5877,
+      "step": 5379
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742364868059471e-05,
+      "loss": 1.5422,
+      "step": 5380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7422507884815246e-05,
+      "loss": 1.5524,
+      "step": 5381
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7421366850249694e-05,
+      "loss": 1.5431,
+      "step": 5382
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742022557691019e-05,
+      "loss": 1.4577,
+      "step": 5383
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7419084064808894e-05,
+      "loss": 1.4806,
+      "step": 5384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7417942313957966e-05,
+      "loss": 1.5623,
+      "step": 5385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.741680032436956e-05,
+      "loss": 1.5392,
+      "step": 5386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7415658096055846e-05,
+      "loss": 1.4899,
+      "step": 5387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.741451562902898e-05,
+      "loss": 1.4598,
+      "step": 5388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.741337292330113e-05,
+      "loss": 1.5557,
+      "step": 5389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7412229978884473e-05,
+      "loss": 1.5327,
+      "step": 5390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.741108679579116e-05,
+      "loss": 1.5739,
+      "step": 5391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7409943374033384e-05,
+      "loss": 1.5736,
+      "step": 5392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7408799713623316e-05,
+      "loss": 1.5238,
+      "step": 5393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.740765581457314e-05,
+      "loss": 1.4644,
+      "step": 5394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7406511676895034e-05,
+      "loss": 1.4967,
+      "step": 5395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.740536730060118e-05,
+      "loss": 1.4595,
+      "step": 5396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.740422268570376e-05,
+      "loss": 1.4373,
+      "step": 5397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7403077832214984e-05,
+      "loss": 1.536,
+      "step": 5398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7401932740147024e-05,
+      "loss": 1.4774,
+      "step": 5399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.740078740951207e-05,
+      "loss": 1.4681,
+      "step": 5400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7399641840322344e-05,
+      "loss": 1.4948,
+      "step": 5401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7398496032590026e-05,
+      "loss": 1.5607,
+      "step": 5402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.739734998632732e-05,
+      "loss": 1.5129,
+      "step": 5403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7396203701546446e-05,
+      "loss": 1.5534,
+      "step": 5404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7395057178259593e-05,
+      "loss": 1.4785,
+      "step": 5405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.739391041647898e-05,
+      "loss": 1.4987,
+      "step": 5406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7392763416216815e-05,
+      "loss": 1.5052,
+      "step": 5407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.739161617748531e-05,
+      "loss": 1.4979,
+      "step": 5408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7390468700296685e-05,
+      "loss": 1.4072,
+      "step": 5409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738932098466317e-05,
+      "loss": 1.498,
+      "step": 5410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738817303059698e-05,
+      "loss": 1.481,
+      "step": 5411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738702483811034e-05,
+      "loss": 1.4951,
+      "step": 5412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738587640721547e-05,
+      "loss": 1.5011,
+      "step": 5413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738472773792461e-05,
+      "loss": 1.5412,
+      "step": 5414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7383578830249983e-05,
+      "loss": 1.502,
+      "step": 5415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738242968420384e-05,
+      "loss": 1.5111,
+      "step": 5416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738128029979841e-05,
+      "loss": 1.4668,
+      "step": 5417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738013067704593e-05,
+      "loss": 1.5226,
+      "step": 5418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737898081595864e-05,
+      "loss": 1.4858,
+      "step": 5419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7377830716548803e-05,
+      "loss": 1.5013,
+      "step": 5420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737668037882865e-05,
+      "loss": 1.4819,
+      "step": 5421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7375529802810434e-05,
+      "loss": 1.5423,
+      "step": 5422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737437898850641e-05,
+      "loss": 1.5174,
+      "step": 5423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7373227935928834e-05,
+      "loss": 1.5487,
+      "step": 5424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737207664508997e-05,
+      "loss": 1.5145,
+      "step": 5425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737092511600206e-05,
+      "loss": 1.5021,
+      "step": 5426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.736977334867739e-05,
+      "loss": 1.5555,
+      "step": 5427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7368621343128205e-05,
+      "loss": 1.5148,
+      "step": 5428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.736746909936679e-05,
+      "loss": 1.4787,
+      "step": 5429
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.73663166174054e-05,
+      "loss": 1.5246,
+      "step": 5430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7365163897256326e-05,
+      "loss": 1.4919,
+      "step": 5431
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.736401093893184e-05,
+      "loss": 1.5132,
+      "step": 5432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.736285774244421e-05,
+      "loss": 1.5838,
+      "step": 5433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.736170430780572e-05,
+      "loss": 1.5187,
+      "step": 5434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.736055063502865e-05,
+      "loss": 1.4911,
+      "step": 5435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.735939672412529e-05,
+      "loss": 1.4276,
+      "step": 5436
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.735824257510794e-05,
+      "loss": 1.4948,
+      "step": 5437
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7357088187988874e-05,
+      "loss": 1.512,
+      "step": 5438
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7355933562780395e-05,
+      "loss": 1.5011,
+      "step": 5439
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.735477869949479e-05,
+      "loss": 1.543,
+      "step": 5440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7353623598144374e-05,
+      "loss": 1.4691,
+      "step": 5441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.735246825874144e-05,
+      "loss": 1.4747,
+      "step": 5442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7351312681298274e-05,
+      "loss": 1.5589,
+      "step": 5443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.735015686582721e-05,
+      "loss": 1.4772,
+      "step": 5444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.734900081234054e-05,
+      "loss": 1.5575,
+      "step": 5445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.734784452085058e-05,
+      "loss": 1.4792,
+      "step": 5446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.734668799136965e-05,
+      "loss": 1.4839,
+      "step": 5447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.734553122391005e-05,
+      "loss": 1.4744,
+      "step": 5448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.734437421848411e-05,
+      "loss": 1.5158,
+      "step": 5449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7343216975104154e-05,
+      "loss": 1.4769,
+      "step": 5450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.73420594937825e-05,
+      "loss": 1.4883,
+      "step": 5451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.734090177453148e-05,
+      "loss": 1.5692,
+      "step": 5452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.733974381736342e-05,
+      "loss": 1.5514,
+      "step": 5453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7338585622290644e-05,
+      "loss": 1.5233,
+      "step": 5454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.73374271893255e-05,
+      "loss": 1.5098,
+      "step": 5455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7336268518480314e-05,
+      "loss": 1.4561,
+      "step": 5456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7335109609767436e-05,
+      "loss": 1.4801,
+      "step": 5457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7333950463199196e-05,
+      "loss": 1.4025,
+      "step": 5458
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.733279107878794e-05,
+      "loss": 1.4299,
+      "step": 5459
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.733163145654602e-05,
+      "loss": 1.4891,
+      "step": 5460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.733047159648578e-05,
+      "loss": 1.5155,
+      "step": 5461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7329311498619576e-05,
+      "loss": 1.5461,
+      "step": 5462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7328151162959765e-05,
+      "loss": 1.4434,
+      "step": 5463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.732699058951869e-05,
+      "loss": 1.5514,
+      "step": 5464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7325829778308725e-05,
+      "loss": 1.4696,
+      "step": 5465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7324668729342236e-05,
+      "loss": 1.5068,
+      "step": 5466
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.732350744263157e-05,
+      "loss": 1.485,
+      "step": 5467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7322345918189104e-05,
+      "loss": 1.4501,
+      "step": 5468
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.732118415602721e-05,
+      "loss": 1.449,
+      "step": 5469
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.732002215615825e-05,
+      "loss": 1.5438,
+      "step": 5470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.731885991859461e-05,
+      "loss": 1.5046,
+      "step": 5471
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.731769744334866e-05,
+      "loss": 1.4644,
+      "step": 5472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7316534730432783e-05,
+      "loss": 1.5135,
+      "step": 5473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.731537177985936e-05,
+      "loss": 1.4715,
+      "step": 5474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7314208591640774e-05,
+      "loss": 1.4657,
+      "step": 5475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.731304516578941e-05,
+      "loss": 1.4777,
+      "step": 5476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.731188150231767e-05,
+      "loss": 1.5013,
+      "step": 5477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.731071760123793e-05,
+      "loss": 1.4992,
+      "step": 5478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7309553462562595e-05,
+      "loss": 1.502,
+      "step": 5479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7308389086304064e-05,
+      "loss": 1.4816,
+      "step": 5480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7307224472474726e-05,
+      "loss": 1.4844,
+      "step": 5481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7306059621086995e-05,
+      "loss": 1.4733,
+      "step": 5482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730489453215327e-05,
+      "loss": 1.5103,
+      "step": 5483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7303729205685964e-05,
+      "loss": 1.4645,
+      "step": 5484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730256364169747e-05,
+      "loss": 1.5029,
+      "step": 5485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730139784020023e-05,
+      "loss": 1.5313,
+      "step": 5486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730023180120663e-05,
+      "loss": 1.5485,
+      "step": 5487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.729906552472911e-05,
+      "loss": 1.5189,
+      "step": 5488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.729789901078008e-05,
+      "loss": 1.4585,
+      "step": 5489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7296732259371956e-05,
+      "loss": 1.4355,
+      "step": 5490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7295565270517176e-05,
+      "loss": 1.4766,
+      "step": 5491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.729439804422815e-05,
+      "loss": 1.5365,
+      "step": 5492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7293230580517336e-05,
+      "loss": 1.5581,
+      "step": 5493
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.729206287939714e-05,
+      "loss": 1.4704,
+      "step": 5494
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.729089494088001e-05,
+      "loss": 1.6011,
+      "step": 5495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.728972676497838e-05,
+      "loss": 1.4785,
+      "step": 5496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.728855835170471e-05,
+      "loss": 1.429,
+      "step": 5497
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7287389701071404e-05,
+      "loss": 1.4736,
+      "step": 5498
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.728622081309093e-05,
+      "loss": 1.48,
+      "step": 5499
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.728505168777574e-05,
+      "loss": 1.4665,
+      "step": 5500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7283882325138286e-05,
+      "loss": 1.5241,
+      "step": 5501
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.728271272519101e-05,
+      "loss": 1.4808,
+      "step": 5502
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7281542887946366e-05,
+      "loss": 1.4928,
+      "step": 5503
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7280372813416824e-05,
+      "loss": 1.4691,
+      "step": 5504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7279202501614834e-05,
+      "loss": 1.5493,
+      "step": 5505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7278031952552865e-05,
+      "loss": 1.4653,
+      "step": 5506
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7276861166243383e-05,
+      "loss": 1.4533,
+      "step": 5507
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7275690142698846e-05,
+      "loss": 1.5094,
+      "step": 5508
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.727451888193174e-05,
+      "loss": 1.5392,
+      "step": 5509
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7273347383954535e-05,
+      "loss": 1.5457,
+      "step": 5510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.727217564877969e-05,
+      "loss": 1.5324,
+      "step": 5511
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.727100367641971e-05,
+      "loss": 1.5207,
+      "step": 5512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.726983146688705e-05,
+      "loss": 1.508,
+      "step": 5513
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.726865902019422e-05,
+      "loss": 1.4964,
+      "step": 5514
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7267486336353676e-05,
+      "loss": 1.472,
+      "step": 5515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7266313415377924e-05,
+      "loss": 1.4953,
+      "step": 5516
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.726514025727946e-05,
+      "loss": 1.5379,
+      "step": 5517
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.726396686207077e-05,
+      "loss": 1.4756,
+      "step": 5518
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7262793229764346e-05,
+      "loss": 1.5299,
+      "step": 5519
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.726161936037269e-05,
+      "loss": 1.4652,
+      "step": 5520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7260445253908306e-05,
+      "loss": 1.5098,
+      "step": 5521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725927091038369e-05,
+      "loss": 1.4352,
+      "step": 5522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725809632981135e-05,
+      "loss": 1.5435,
+      "step": 5523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.72569215122038e-05,
+      "loss": 1.5521,
+      "step": 5524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725574645757355e-05,
+      "loss": 1.5252,
+      "step": 5525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725457116593312e-05,
+      "loss": 1.543,
+      "step": 5526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7253395637295e-05,
+      "loss": 1.4893,
+      "step": 5527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7252219871671747e-05,
+      "loss": 1.521,
+      "step": 5528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7251043869075844e-05,
+      "loss": 1.4629,
+      "step": 5529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.724986762951984e-05,
+      "loss": 1.5436,
+      "step": 5530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7248691153016254e-05,
+      "loss": 1.5091,
+      "step": 5531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.724751443957762e-05,
+      "loss": 1.5549,
+      "step": 5532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.724633748921646e-05,
+      "loss": 1.4544,
+      "step": 5533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7245160301945314e-05,
+      "loss": 1.48,
+      "step": 5534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.724398287777671e-05,
+      "loss": 1.4194,
+      "step": 5535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.724280521672319e-05,
+      "loss": 1.5268,
+      "step": 5536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.724162731879731e-05,
+      "loss": 1.4512,
+      "step": 5537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7240449184011596e-05,
+      "loss": 1.5024,
+      "step": 5538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.72392708123786e-05,
+      "loss": 1.5208,
+      "step": 5539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.723809220391088e-05,
+      "loss": 1.5067,
+      "step": 5540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7236913358620974e-05,
+      "loss": 1.5227,
+      "step": 5541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.723573427652143e-05,
+      "loss": 1.5602,
+      "step": 5542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.723455495762483e-05,
+      "loss": 1.6009,
+      "step": 5543
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7233375401943716e-05,
+      "loss": 1.4339,
+      "step": 5544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7232195609490646e-05,
+      "loss": 1.5226,
+      "step": 5545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.723101558027819e-05,
+      "loss": 1.5392,
+      "step": 5546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7229835314318926e-05,
+      "loss": 1.4753,
+      "step": 5547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.72286548116254e-05,
+      "loss": 1.5098,
+      "step": 5548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7227474072210206e-05,
+      "loss": 1.4824,
+      "step": 5549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.72262930960859e-05,
+      "loss": 1.5402,
+      "step": 5550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.722511188326507e-05,
+      "loss": 1.5352,
+      "step": 5551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.722393043376029e-05,
+      "loss": 1.4958,
+      "step": 5552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7222748747584146e-05,
+      "loss": 1.4683,
+      "step": 5553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7221566824749214e-05,
+      "loss": 1.4326,
+      "step": 5554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.722038466526809e-05,
+      "loss": 1.5047,
+      "step": 5555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721920226915336e-05,
+      "loss": 1.4222,
+      "step": 5556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7218019636417615e-05,
+      "loss": 1.5443,
+      "step": 5557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721683676707345e-05,
+      "loss": 1.5519,
+      "step": 5558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721565366113346e-05,
+      "loss": 1.4717,
+      "step": 5559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721447031861025e-05,
+      "loss": 1.4227,
+      "step": 5560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721328673951642e-05,
+      "loss": 1.4948,
+      "step": 5561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721210292386456e-05,
+      "loss": 1.5124,
+      "step": 5562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7210918871667296e-05,
+      "loss": 1.4611,
+      "step": 5563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720973458293723e-05,
+      "loss": 1.4855,
+      "step": 5564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720855005768697e-05,
+      "loss": 1.507,
+      "step": 5565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720736529592914e-05,
+      "loss": 1.4979,
+      "step": 5566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720618029767636e-05,
+      "loss": 1.5155,
+      "step": 5567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7204995062941235e-05,
+      "loss": 1.5247,
+      "step": 5568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720380959173639e-05,
+      "loss": 1.432,
+      "step": 5569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720262388407446e-05,
+      "loss": 1.4192,
+      "step": 5570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720143793996806e-05,
+      "loss": 1.4831,
+      "step": 5571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720025175942983e-05,
+      "loss": 1.4911,
+      "step": 5572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7199065342472386e-05,
+      "loss": 1.4556,
+      "step": 5573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7197878689108375e-05,
+      "loss": 1.5163,
+      "step": 5574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7196691799350444e-05,
+      "loss": 1.5501,
+      "step": 5575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.719550467321121e-05,
+      "loss": 1.4552,
+      "step": 5576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7194317310703336e-05,
+      "loss": 1.5251,
+      "step": 5577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.719312971183945e-05,
+      "loss": 1.481,
+      "step": 5578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.719194187663221e-05,
+      "loss": 1.4349,
+      "step": 5579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.719075380509426e-05,
+      "loss": 1.5163,
+      "step": 5580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7189565497238254e-05,
+      "loss": 1.4468,
+      "step": 5581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.718837695307685e-05,
+      "loss": 1.478,
+      "step": 5582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.71871881726227e-05,
+      "loss": 1.4294,
+      "step": 5583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7185999155888475e-05,
+      "loss": 1.4715,
+      "step": 5584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7184809902886815e-05,
+      "loss": 1.438,
+      "step": 5585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.71836204136304e-05,
+      "loss": 1.4974,
+      "step": 5586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7182430688131904e-05,
+      "loss": 1.4297,
+      "step": 5587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.718124072640398e-05,
+      "loss": 1.4671,
+      "step": 5588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.718005052845931e-05,
+      "loss": 1.5449,
+      "step": 5589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.717886009431057e-05,
+      "loss": 1.563,
+      "step": 5590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.717766942397044e-05,
+      "loss": 1.5179,
+      "step": 5591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.717647851745159e-05,
+      "loss": 1.4847,
+      "step": 5592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7175287374766705e-05,
+      "loss": 1.4595,
+      "step": 5593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.717409599592848e-05,
+      "loss": 1.408,
+      "step": 5594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7172904380949586e-05,
+      "loss": 1.4079,
+      "step": 5595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7171712529842724e-05,
+      "loss": 1.4627,
+      "step": 5596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.717052044262059e-05,
+      "loss": 1.4348,
+      "step": 5597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.716932811929587e-05,
+      "loss": 1.5435,
+      "step": 5598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7168135559881266e-05,
+      "loss": 1.4307,
+      "step": 5599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.716694276438947e-05,
+      "loss": 1.5334,
+      "step": 5600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7165749732833196e-05,
+      "loss": 1.5036,
+      "step": 5601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.716455646522515e-05,
+      "loss": 1.509,
+      "step": 5602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7163362961578025e-05,
+      "loss": 1.471,
+      "step": 5603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.716216922190454e-05,
+      "loss": 1.5011,
+      "step": 5604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.716097524621742e-05,
+      "loss": 1.4154,
+      "step": 5605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715978103452936e-05,
+      "loss": 1.4504,
+      "step": 5606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7158586586853084e-05,
+      "loss": 1.5098,
+      "step": 5607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7157391903201314e-05,
+      "loss": 1.4248,
+      "step": 5608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715619698358678e-05,
+      "loss": 1.4159,
+      "step": 5609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715500182802218e-05,
+      "loss": 1.4904,
+      "step": 5610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715380643652028e-05,
+      "loss": 1.4455,
+      "step": 5611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7152610809093786e-05,
+      "loss": 1.5025,
+      "step": 5612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715141494575543e-05,
+      "loss": 1.4136,
+      "step": 5613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715021884651797e-05,
+      "loss": 1.4673,
+      "step": 5614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7149022511394115e-05,
+      "loss": 1.5072,
+      "step": 5615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7147825940396616e-05,
+      "loss": 1.4406,
+      "step": 5616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.714662913353822e-05,
+      "loss": 1.4046,
+      "step": 5617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7145432090831665e-05,
+      "loss": 1.5228,
+      "step": 5618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7144234812289704e-05,
+      "loss": 1.4618,
+      "step": 5619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.714303729792509e-05,
+      "loss": 1.3828,
+      "step": 5620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.714183954775057e-05,
+      "loss": 1.4271,
+      "step": 5621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.71406415617789e-05,
+      "loss": 1.3853,
+      "step": 5622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7139443340022846e-05,
+      "loss": 1.4645,
+      "step": 5623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7138244882495156e-05,
+      "loss": 1.4136,
+      "step": 5624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.71370461892086e-05,
+      "loss": 1.4818,
+      "step": 5625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713584726017594e-05,
+      "loss": 1.4393,
+      "step": 5626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713464809540995e-05,
+      "loss": 1.4337,
+      "step": 5627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713344869492339e-05,
+      "loss": 1.4898,
+      "step": 5628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7132249058729046e-05,
+      "loss": 1.5083,
+      "step": 5629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713104918683968e-05,
+      "loss": 1.5169,
+      "step": 5630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712984907926808e-05,
+      "loss": 1.4435,
+      "step": 5631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712864873602702e-05,
+      "loss": 1.5203,
+      "step": 5632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7127448157129284e-05,
+      "loss": 1.4834,
+      "step": 5633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7126247342587664e-05,
+      "loss": 1.4489,
+      "step": 5634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712504629241494e-05,
+      "loss": 1.4688,
+      "step": 5635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7123845006623914e-05,
+      "loss": 1.4426,
+      "step": 5636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7122643485227357e-05,
+      "loss": 1.4331,
+      "step": 5637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712144172823808e-05,
+      "loss": 1.4728,
+      "step": 5638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7120239735668894e-05,
+      "loss": 1.3875,
+      "step": 5639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.711903750753257e-05,
+      "loss": 1.4829,
+      "step": 5640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7117835043841924e-05,
+      "loss": 1.4919,
+      "step": 5641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.711663234460977e-05,
+      "loss": 1.5717,
+      "step": 5642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.71154294098489e-05,
+      "loss": 1.4492,
+      "step": 5643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7114226239572144e-05,
+      "loss": 1.4893,
+      "step": 5644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.71130228337923e-05,
+      "loss": 1.4243,
+      "step": 5645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.711181919252219e-05,
+      "loss": 1.4481,
+      "step": 5646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7110615315774625e-05,
+      "loss": 1.402,
+      "step": 5647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.710941120356243e-05,
+      "loss": 1.4943,
+      "step": 5648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7108206855898435e-05,
+      "loss": 1.387,
+      "step": 5649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.710700227279545e-05,
+      "loss": 1.4006,
+      "step": 5650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.710579745426632e-05,
+      "loss": 1.446,
+      "step": 5651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7104592400323866e-05,
+      "loss": 1.4779,
+      "step": 5652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7103387110980925e-05,
+      "loss": 1.563,
+      "step": 5653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.710218158625033e-05,
+      "loss": 1.4289,
+      "step": 5654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7100975826144925e-05,
+      "loss": 1.4408,
+      "step": 5655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.709976983067753e-05,
+      "loss": 1.4482,
+      "step": 5656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.709856359986101e-05,
+      "loss": 1.4491,
+      "step": 5657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7097357133708206e-05,
+      "loss": 1.4277,
+      "step": 5658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.709615043223197e-05,
+      "loss": 1.4499,
+      "step": 5659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.709494349544514e-05,
+      "loss": 1.5632,
+      "step": 5660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.709373632336058e-05,
+      "loss": 1.4448,
+      "step": 5661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.709252891599113e-05,
+      "loss": 1.3996,
+      "step": 5662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7091321273349673e-05,
+      "loss": 1.4315,
+      "step": 5663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7090113395449055e-05,
+      "loss": 1.4793,
+      "step": 5664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708890528230214e-05,
+      "loss": 1.48,
+      "step": 5665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7087696933921786e-05,
+      "loss": 1.4528,
+      "step": 5666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708648835032088e-05,
+      "loss": 1.4019,
+      "step": 5667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708527953151227e-05,
+      "loss": 1.417,
+      "step": 5668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708407047750885e-05,
+      "loss": 1.4468,
+      "step": 5669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708286118832349e-05,
+      "loss": 1.396,
+      "step": 5670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708165166396906e-05,
+      "loss": 1.4504,
+      "step": 5671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708044190445844e-05,
+      "loss": 1.4564,
+      "step": 5672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707923190980453e-05,
+      "loss": 1.4139,
+      "step": 5673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.70780216800202e-05,
+      "loss": 1.4792,
+      "step": 5674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707681121511834e-05,
+      "loss": 1.4909,
+      "step": 5675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707560051511185e-05,
+      "loss": 1.4186,
+      "step": 5676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707438958001361e-05,
+      "loss": 1.4326,
+      "step": 5677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707317840983653e-05,
+      "loss": 1.4662,
+      "step": 5678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7071967004593496e-05,
+      "loss": 1.403,
+      "step": 5679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707075536429741e-05,
+      "loss": 1.4051,
+      "step": 5680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7069543488961185e-05,
+      "loss": 1.3688,
+      "step": 5681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.706833137859773e-05,
+      "loss": 1.4419,
+      "step": 5682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.706711903321993e-05,
+      "loss": 1.3457,
+      "step": 5683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7065906452840716e-05,
+      "loss": 1.3964,
+      "step": 5684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.706469363747299e-05,
+      "loss": 1.389,
+      "step": 5685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.706348058712967e-05,
+      "loss": 1.4262,
+      "step": 5686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.706226730182368e-05,
+      "loss": 1.4253,
+      "step": 5687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7061053781567946e-05,
+      "loss": 1.4212,
+      "step": 5688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7059840026375375e-05,
+      "loss": 1.4758,
+      "step": 5689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.70586260362589e-05,
+      "loss": 1.3609,
+      "step": 5690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705741181123145e-05,
+      "loss": 1.4823,
+      "step": 5691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705619735130595e-05,
+      "loss": 1.4461,
+      "step": 5692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705498265649535e-05,
+      "loss": 1.4302,
+      "step": 5693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705376772681257e-05,
+      "loss": 1.4225,
+      "step": 5694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705255256227056e-05,
+      "loss": 1.3999,
+      "step": 5695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7051337162882236e-05,
+      "loss": 1.3289,
+      "step": 5696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7050121528660574e-05,
+      "loss": 1.4202,
+      "step": 5697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.70489056596185e-05,
+      "loss": 1.4137,
+      "step": 5698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.704768955576897e-05,
+      "loss": 1.3739,
+      "step": 5699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7046473217124926e-05,
+      "loss": 1.4006,
+      "step": 5700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.704525664369933e-05,
+      "loss": 1.3293,
+      "step": 5701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.704403983550513e-05,
+      "loss": 1.4373,
+      "step": 5702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7042822792555295e-05,
+      "loss": 1.4002,
+      "step": 5703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.704160551486278e-05,
+      "loss": 1.4396,
+      "step": 5704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.704038800244055e-05,
+      "loss": 1.3298,
+      "step": 5705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7039170255301564e-05,
+      "loss": 1.5239,
+      "step": 5706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7037952273458795e-05,
+      "loss": 1.3968,
+      "step": 5707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703673405692521e-05,
+      "loss": 1.3229,
+      "step": 5708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.70355156057138e-05,
+      "loss": 1.3734,
+      "step": 5709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7034296919837516e-05,
+      "loss": 1.4261,
+      "step": 5710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703307799930935e-05,
+      "loss": 1.3875,
+      "step": 5711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703185884414227e-05,
+      "loss": 1.3465,
+      "step": 5712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7030639454349276e-05,
+      "loss": 1.3658,
+      "step": 5713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702941982994335e-05,
+      "loss": 1.3937,
+      "step": 5714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702819997093748e-05,
+      "loss": 1.3682,
+      "step": 5715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7026979877344656e-05,
+      "loss": 1.3714,
+      "step": 5716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7025759549177864e-05,
+      "loss": 1.3392,
+      "step": 5717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.70245389864501e-05,
+      "loss": 1.4125,
+      "step": 5718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702331818917437e-05,
+      "loss": 1.3488,
+      "step": 5719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7022097157363675e-05,
+      "loss": 1.4461,
+      "step": 5720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702087589103101e-05,
+      "loss": 1.5228,
+      "step": 5721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7019654390189394e-05,
+      "loss": 1.3936,
+      "step": 5722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701843265485182e-05,
+      "loss": 1.436,
+      "step": 5723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701721068503132e-05,
+      "loss": 1.4202,
+      "step": 5724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701598848074088e-05,
+      "loss": 1.464,
+      "step": 5725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701476604199353e-05,
+      "loss": 1.3809,
+      "step": 5726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701354336880228e-05,
+      "loss": 1.4302,
+      "step": 5727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701232046118017e-05,
+      "loss": 1.3916,
+      "step": 5728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701109731914021e-05,
+      "loss": 1.4079,
+      "step": 5729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.700987394269542e-05,
+      "loss": 1.3545,
+      "step": 5730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.700865033185884e-05,
+      "loss": 1.3969,
+      "step": 5731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.70074264866435e-05,
+      "loss": 1.4346,
+      "step": 5732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7006202407062414e-05,
+      "loss": 1.3149,
+      "step": 5733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.700497809312865e-05,
+      "loss": 1.3984,
+      "step": 5734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.700375354485522e-05,
+      "loss": 1.3551,
+      "step": 5735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7002528762255174e-05,
+      "loss": 1.3877,
+      "step": 5736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7001303745341555e-05,
+      "loss": 1.4051,
+      "step": 5737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.700007849412741e-05,
+      "loss": 1.3219,
+      "step": 5738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.699885300862579e-05,
+      "loss": 1.4549,
+      "step": 5739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6997627288849734e-05,
+      "loss": 1.4333,
+      "step": 5740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.69964013348123e-05,
+      "loss": 1.3828,
+      "step": 5741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.699517514652655e-05,
+      "loss": 1.3921,
+      "step": 5742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.699394872400554e-05,
+      "loss": 1.4209,
+      "step": 5743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.699272206726233e-05,
+      "loss": 1.4027,
+      "step": 5744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.699149517630998e-05,
+      "loss": 1.3687,
+      "step": 5745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.699026805116156e-05,
+      "loss": 1.4009,
+      "step": 5746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6989040691830134e-05,
+      "loss": 1.4356,
+      "step": 5747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.698781309832877e-05,
+      "loss": 1.3651,
+      "step": 5748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6986585270670544e-05,
+      "loss": 1.2827,
+      "step": 5749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.698535720886854e-05,
+      "loss": 1.3696,
+      "step": 5750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.698412891293583e-05,
+      "loss": 1.3714,
+      "step": 5751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6982900382885496e-05,
+      "loss": 1.3376,
+      "step": 5752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.698167161873062e-05,
+      "loss": 1.4528,
+      "step": 5753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.698044262048428e-05,
+      "loss": 1.3789,
+      "step": 5754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.697921338815957e-05,
+      "loss": 1.3148,
+      "step": 5755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.697798392176958e-05,
+      "loss": 1.3425,
+      "step": 5756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6976754221327413e-05,
+      "loss": 1.3711,
+      "step": 5757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.697552428684615e-05,
+      "loss": 1.3812,
+      "step": 5758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.697429411833889e-05,
+      "loss": 1.3175,
+      "step": 5759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6973063715818746e-05,
+      "loss": 1.3884,
+      "step": 5760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.697183307929882e-05,
+      "loss": 1.3886,
+      "step": 5761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6970602208792205e-05,
+      "loss": 1.3191,
+      "step": 5762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696937110431201e-05,
+      "loss": 1.3317,
+      "step": 5763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696813976587136e-05,
+      "loss": 1.3682,
+      "step": 5764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6966908193483346e-05,
+      "loss": 1.3874,
+      "step": 5765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6965676387161106e-05,
+      "loss": 1.4191,
+      "step": 5766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696444434691774e-05,
+      "loss": 1.3595,
+      "step": 5767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696321207276638e-05,
+      "loss": 1.3262,
+      "step": 5768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696197956472015e-05,
+      "loss": 1.3815,
+      "step": 5769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696074682279217e-05,
+      "loss": 1.3389,
+      "step": 5770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695951384699557e-05,
+      "loss": 1.339,
+      "step": 5771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6958280637343476e-05,
+      "loss": 1.4476,
+      "step": 5772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6957047193849026e-05,
+      "loss": 1.3841,
+      "step": 5773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695581351652535e-05,
+      "loss": 1.297,
+      "step": 5774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6954579605385596e-05,
+      "loss": 1.4061,
+      "step": 5775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6953345460442895e-05,
+      "loss": 1.3503,
+      "step": 5776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695211108171039e-05,
+      "loss": 1.3384,
+      "step": 5777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6950876469201235e-05,
+      "loss": 1.3289,
+      "step": 5778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.694964162292857e-05,
+      "loss": 1.348,
+      "step": 5779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6948406542905546e-05,
+      "loss": 1.3187,
+      "step": 5780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.694717122914532e-05,
+      "loss": 1.3114,
+      "step": 5781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.694593568166104e-05,
+      "loss": 1.3105,
+      "step": 5782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6944699900465875e-05,
+      "loss": 1.4136,
+      "step": 5783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.694346388557298e-05,
+      "loss": 1.3071,
+      "step": 5784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6942227636995515e-05,
+      "loss": 1.3792,
+      "step": 5785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.694099115474665e-05,
+      "loss": 1.305,
+      "step": 5786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.693975443883954e-05,
+      "loss": 1.3211,
+      "step": 5787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6938517489287374e-05,
+      "loss": 1.3862,
+      "step": 5788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.693728030610331e-05,
+      "loss": 1.3512,
+      "step": 5789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6936042889300536e-05,
+      "loss": 1.3464,
+      "step": 5790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6934805238892216e-05,
+      "loss": 1.3628,
+      "step": 5791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.693356735489154e-05,
+      "loss": 1.3361,
+      "step": 5792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6932329237311687e-05,
+      "loss": 1.3703,
+      "step": 5793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6931090886165844e-05,
+      "loss": 1.4186,
+      "step": 5794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.69298523014672e-05,
+      "loss": 1.4116,
+      "step": 5795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692861348322894e-05,
+      "loss": 1.4067,
+      "step": 5796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6927374431464254e-05,
+      "loss": 1.3192,
+      "step": 5797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6926135146186354e-05,
+      "loss": 1.3654,
+      "step": 5798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692489562740842e-05,
+      "loss": 1.3257,
+      "step": 5799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692365587514366e-05,
+      "loss": 1.4163,
+      "step": 5800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692241588940527e-05,
+      "loss": 1.3641,
+      "step": 5801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692117567020648e-05,
+      "loss": 1.3717,
+      "step": 5802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6919935217560453e-05,
+      "loss": 1.2826,
+      "step": 5803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.691869453148044e-05,
+      "loss": 1.3765,
+      "step": 5804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.691745361197963e-05,
+      "loss": 1.3394,
+      "step": 5805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.691621245907124e-05,
+      "loss": 1.267,
+      "step": 5806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.69149710727685e-05,
+      "loss": 1.3953,
+      "step": 5807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.691372945308463e-05,
+      "loss": 1.2802,
+      "step": 5808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6912487600032834e-05,
+      "loss": 1.3962,
+      "step": 5809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6911245513626355e-05,
+      "loss": 1.3591,
+      "step": 5810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6910003193878406e-05,
+      "loss": 1.3571,
+      "step": 5811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.690876064080223e-05,
+      "loss": 1.325,
+      "step": 5812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6907517854411056e-05,
+      "loss": 1.369,
+      "step": 5813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6906274834718116e-05,
+      "loss": 1.3228,
+      "step": 5814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.690503158173665e-05,
+      "loss": 1.4548,
+      "step": 5815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6903788095479895e-05,
+      "loss": 1.3296,
+      "step": 5816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.69025443759611e-05,
+      "loss": 1.3265,
+      "step": 5817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6901300423193494e-05,
+      "loss": 1.3398,
+      "step": 5818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6900056237190335e-05,
+      "loss": 1.3162,
+      "step": 5819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689881181796488e-05,
+      "loss": 1.4002,
+      "step": 5820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6897567165530376e-05,
+      "loss": 1.3894,
+      "step": 5821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689632227990007e-05,
+      "loss": 1.2912,
+      "step": 5822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689507716108723e-05,
+      "loss": 1.2762,
+      "step": 5823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6893831809105115e-05,
+      "loss": 1.3597,
+      "step": 5824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689258622396697e-05,
+      "loss": 1.2793,
+      "step": 5825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689134040568608e-05,
+      "loss": 1.2917,
+      "step": 5826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689009435427571e-05,
+      "loss": 1.3216,
+      "step": 5827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.688884806974913e-05,
+      "loss": 1.3763,
+      "step": 5828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6887601552119586e-05,
+      "loss": 1.3063,
+      "step": 5829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.688635480140039e-05,
+      "loss": 1.2384,
+      "step": 5830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.68851078176048e-05,
+      "loss": 1.2993,
+      "step": 5831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6883860600746094e-05,
+      "loss": 1.2866,
+      "step": 5832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.688261315083756e-05,
+      "loss": 1.3066,
+      "step": 5833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.688136546789249e-05,
+      "loss": 1.3154,
+      "step": 5834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.688011755192415e-05,
+      "loss": 1.3167,
+      "step": 5835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.687886940294585e-05,
+      "loss": 1.3508,
+      "step": 5836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6877621020970866e-05,
+      "loss": 1.3267,
+      "step": 5837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6876372406012505e-05,
+      "loss": 1.3272,
+      "step": 5838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.687512355808406e-05,
+      "loss": 1.2977,
+      "step": 5839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6873874477198834e-05,
+      "loss": 1.2181,
+      "step": 5840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.687262516337012e-05,
+      "loss": 1.2914,
+      "step": 5841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6871375616611225e-05,
+      "loss": 1.2634,
+      "step": 5842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.687012583693546e-05,
+      "loss": 1.2759,
+      "step": 5843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6868875824356126e-05,
+      "loss": 1.3285,
+      "step": 5844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6867625578886555e-05,
+      "loss": 1.3641,
+      "step": 5845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.686637510054004e-05,
+      "loss": 1.3021,
+      "step": 5846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6865124389329904e-05,
+      "loss": 1.2529,
+      "step": 5847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6863873445269476e-05,
+      "loss": 1.2969,
+      "step": 5848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6862622268372057e-05,
+      "loss": 1.403,
+      "step": 5849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6861370858650996e-05,
+      "loss": 1.3763,
+      "step": 5850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6860119216119603e-05,
+      "loss": 1.2716,
+      "step": 5851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6858867340791214e-05,
+      "loss": 1.3455,
+      "step": 5852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6857615232679156e-05,
+      "loss": 1.3329,
+      "step": 5853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6856362891796765e-05,
+      "loss": 1.418,
+      "step": 5854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.685511031815738e-05,
+      "loss": 1.2892,
+      "step": 5855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6853857511774337e-05,
+      "loss": 1.3418,
+      "step": 5856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.685260447266098e-05,
+      "loss": 1.3698,
+      "step": 5857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6851351200830654e-05,
+      "loss": 1.3407,
+      "step": 5858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6850097696296704e-05,
+      "loss": 1.3052,
+      "step": 5859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684884395907248e-05,
+      "loss": 1.3591,
+      "step": 5860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684758998917133e-05,
+      "loss": 1.3682,
+      "step": 5861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684633578660661e-05,
+      "loss": 1.4043,
+      "step": 5862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684508135139168e-05,
+      "loss": 1.3927,
+      "step": 5863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684382668353989e-05,
+      "loss": 1.3289,
+      "step": 5864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684257178306462e-05,
+      "loss": 1.2951,
+      "step": 5865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684131664997921e-05,
+      "loss": 1.383,
+      "step": 5866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6840061284297046e-05,
+      "loss": 1.3116,
+      "step": 5867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6838805686031476e-05,
+      "loss": 1.2904,
+      "step": 5868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.683754985519589e-05,
+      "loss": 1.2882,
+      "step": 5869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6836293791803667e-05,
+      "loss": 1.3065,
+      "step": 5870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.683503749586816e-05,
+      "loss": 1.2881,
+      "step": 5871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6833780967402764e-05,
+      "loss": 1.3488,
+      "step": 5872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.683252420642086e-05,
+      "loss": 1.2354,
+      "step": 5873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.683126721293582e-05,
+      "loss": 1.2932,
+      "step": 5874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6830009986961044e-05,
+      "loss": 1.3247,
+      "step": 5875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.682875252850991e-05,
+      "loss": 1.3203,
+      "step": 5876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.682749483759582e-05,
+      "loss": 1.3178,
+      "step": 5877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.682623691423216e-05,
+      "loss": 1.4146,
+      "step": 5878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6824978758432325e-05,
+      "loss": 1.2929,
+      "step": 5879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6823720370209714e-05,
+      "loss": 1.2211,
+      "step": 5880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.682246174957774e-05,
+      "loss": 1.2743,
+      "step": 5881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6821202896549786e-05,
+      "loss": 1.3677,
+      "step": 5882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6819943811139275e-05,
+      "loss": 1.2726,
+      "step": 5883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681868449335961e-05,
+      "loss": 1.356,
+      "step": 5884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681742494322421e-05,
+      "loss": 1.2702,
+      "step": 5885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681616516074647e-05,
+      "loss": 1.3024,
+      "step": 5886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681490514593981e-05,
+      "loss": 1.4111,
+      "step": 5887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6813644898817664e-05,
+      "loss": 1.2993,
+      "step": 5888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681238441939344e-05,
+      "loss": 1.3566,
+      "step": 5889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681112370768057e-05,
+      "loss": 1.3436,
+      "step": 5890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.680986276369247e-05,
+      "loss": 1.2598,
+      "step": 5891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.680860158744257e-05,
+      "loss": 1.341,
+      "step": 5892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.680734017894431e-05,
+      "loss": 1.3096,
+      "step": 5893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.680607853821111e-05,
+      "loss": 1.3298,
+      "step": 5894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.680481666525641e-05,
+      "loss": 1.355,
+      "step": 5895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.680355456009366e-05,
+      "loss": 1.3301,
+      "step": 5896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.680229222273629e-05,
+      "loss": 1.2705,
+      "step": 5897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.680102965319774e-05,
+      "loss": 1.3089,
+      "step": 5898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.679976685149146e-05,
+      "loss": 1.3283,
+      "step": 5899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.67985038176309e-05,
+      "loss": 1.2666,
+      "step": 5900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6797240551629505e-05,
+      "loss": 1.3503,
+      "step": 5901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6795977053500736e-05,
+      "loss": 1.3197,
+      "step": 5902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6794713323258046e-05,
+      "loss": 1.2736,
+      "step": 5903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.679344936091488e-05,
+      "loss": 1.3835,
+      "step": 5904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.679218516648471e-05,
+      "loss": 1.2787,
+      "step": 5905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6790920739981014e-05,
+      "loss": 1.3447,
+      "step": 5906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678965608141723e-05,
+      "loss": 1.2588,
+      "step": 5907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678839119080684e-05,
+      "loss": 1.1911,
+      "step": 5908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678712606816331e-05,
+      "loss": 1.2661,
+      "step": 5909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678586071350012e-05,
+      "loss": 1.2059,
+      "step": 5910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6784595126830735e-05,
+      "loss": 1.265,
+      "step": 5911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678332930816865e-05,
+      "loss": 1.2767,
+      "step": 5912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6782063257527316e-05,
+      "loss": 1.3171,
+      "step": 5913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678079697492024e-05,
+      "loss": 1.2494,
+      "step": 5914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677953046036089e-05,
+      "loss": 1.3089,
+      "step": 5915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677826371386278e-05,
+      "loss": 1.2713,
+      "step": 5916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677699673543937e-05,
+      "loss": 1.2297,
+      "step": 5917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677572952510417e-05,
+      "loss": 1.3338,
+      "step": 5918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6774462082870675e-05,
+      "loss": 1.3324,
+      "step": 5919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6773194408752376e-05,
+      "loss": 1.3969,
+      "step": 5920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677192650276277e-05,
+      "loss": 1.2731,
+      "step": 5921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677065836491537e-05,
+      "loss": 1.2788,
+      "step": 5922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6769389995223675e-05,
+      "loss": 1.2307,
+      "step": 5923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.676812139370119e-05,
+      "loss": 1.363,
+      "step": 5924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.676685256036143e-05,
+      "loss": 1.3372,
+      "step": 5925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.676558349521791e-05,
+      "loss": 1.2842,
+      "step": 5926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.676431419828414e-05,
+      "loss": 1.1863,
+      "step": 5927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.676304466957363e-05,
+      "loss": 1.2769,
+      "step": 5928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.676177490909991e-05,
+      "loss": 1.2733,
+      "step": 5929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6760504916876504e-05,
+      "loss": 1.2852,
+      "step": 5930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.675923469291692e-05,
+      "loss": 1.2397,
+      "step": 5931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.675796423723471e-05,
+      "loss": 1.3263,
+      "step": 5932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.675669354984338e-05,
+      "loss": 1.3281,
+      "step": 5933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.675542263075648e-05,
+      "loss": 1.2218,
+      "step": 5934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.675415147998753e-05,
+      "loss": 1.2308,
+      "step": 5935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.675288009755008e-05,
+      "loss": 1.3031,
+      "step": 5936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6751608483457665e-05,
+      "loss": 1.2868,
+      "step": 5937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.675033663772382e-05,
+      "loss": 1.3309,
+      "step": 5938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.67490645603621e-05,
+      "loss": 1.2471,
+      "step": 5939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6747792251386034e-05,
+      "loss": 1.2917,
+      "step": 5940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.67465197108092e-05,
+      "loss": 1.2312,
+      "step": 5941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6745246938645124e-05,
+      "loss": 1.2194,
+      "step": 5942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.674397393490737e-05,
+      "loss": 1.1947,
+      "step": 5943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.67427006996095e-05,
+      "loss": 1.2949,
+      "step": 5944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.674142723276507e-05,
+      "loss": 1.3626,
+      "step": 5945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.674015353438763e-05,
+      "loss": 1.3465,
+      "step": 5946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.673887960449076e-05,
+      "loss": 1.3533,
+      "step": 5947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6737605443088015e-05,
+      "loss": 1.2321,
+      "step": 5948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.673633105019298e-05,
+      "loss": 1.348,
+      "step": 5949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.67350564258192e-05,
+      "loss": 1.3136,
+      "step": 5950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.673378156998027e-05,
+      "loss": 1.215,
+      "step": 5951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.673250648268978e-05,
+      "loss": 1.2319,
+      "step": 5952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.673123116396127e-05,
+      "loss": 1.2183,
+      "step": 5953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672995561380834e-05,
+      "loss": 1.2752,
+      "step": 5954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672867983224458e-05,
+      "loss": 1.2453,
+      "step": 5955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6727403819283576e-05,
+      "loss": 1.276,
+      "step": 5956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672612757493891e-05,
+      "loss": 1.381,
+      "step": 5957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6724851099224173e-05,
+      "loss": 1.3885,
+      "step": 5958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672357439215297e-05,
+      "loss": 1.27,
+      "step": 5959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672229745373888e-05,
+      "loss": 1.3138,
+      "step": 5960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672102028399551e-05,
+      "loss": 1.2688,
+      "step": 5961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6719742882936466e-05,
+      "loss": 1.3132,
+      "step": 5962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.671846525057534e-05,
+      "loss": 1.3231,
+      "step": 5963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6717187386925746e-05,
+      "loss": 1.2113,
+      "step": 5964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6715909292001296e-05,
+      "loss": 1.3901,
+      "step": 5965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.671463096581559e-05,
+      "loss": 1.2606,
+      "step": 5966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.671335240838225e-05,
+      "loss": 1.2992,
+      "step": 5967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6712073619714894e-05,
+      "loss": 1.2882,
+      "step": 5968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.671079459982714e-05,
+      "loss": 1.2511,
+      "step": 5969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.670951534873259e-05,
+      "loss": 1.3428,
+      "step": 5970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6708235866444886e-05,
+      "loss": 1.2028,
+      "step": 5971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.670695615297765e-05,
+      "loss": 1.2596,
+      "step": 5972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6705676208344515e-05,
+      "loss": 1.1964,
+      "step": 5973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.67043960325591e-05,
+      "loss": 1.2661,
+      "step": 5974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6703115625635046e-05,
+      "loss": 1.3242,
+      "step": 5975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.670183498758599e-05,
+      "loss": 1.2622,
+      "step": 5976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6700554118425566e-05,
+      "loss": 1.2326,
+      "step": 5977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.669927301816741e-05,
+      "loss": 1.2883,
+      "step": 5978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.669799168682518e-05,
+      "loss": 1.265,
+      "step": 5979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.66967101244125e-05,
+      "loss": 1.2754,
+      "step": 5980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.669542833094303e-05,
+      "loss": 1.2507,
+      "step": 5981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.669414630643043e-05,
+      "loss": 1.2983,
+      "step": 5982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.669286405088834e-05,
+      "loss": 1.3037,
+      "step": 5983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.669158156433041e-05,
+      "loss": 1.3299,
+      "step": 5984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6690298846770317e-05,
+      "loss": 1.2075,
+      "step": 5985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.66890158982217e-05,
+      "loss": 1.2612,
+      "step": 5986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.668773271869824e-05,
+      "loss": 1.2344,
+      "step": 5987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.668644930821359e-05,
+      "loss": 1.2044,
+      "step": 5988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.668516566678142e-05,
+      "loss": 1.2435,
+      "step": 5989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.668388179441539e-05,
+      "loss": 1.2306,
+      "step": 5990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6682597691129206e-05,
+      "loss": 1.2362,
+      "step": 5991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6681313356936505e-05,
+      "loss": 1.3164,
+      "step": 5992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6680028791850984e-05,
+      "loss": 1.262,
+      "step": 5993
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.667874399588632e-05,
+      "loss": 1.2264,
+      "step": 5994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6677458969056195e-05,
+      "loss": 1.2712,
+      "step": 5995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.667617371137429e-05,
+      "loss": 1.3271,
+      "step": 5996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.66748882228543e-05,
+      "loss": 1.2396,
+      "step": 5997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.66736025035099e-05,
+      "loss": 1.2474,
+      "step": 5998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6672316553354794e-05,
+      "loss": 1.271,
+      "step": 5999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.667103037240268e-05,
+      "loss": 1.2158,
+      "step": 6000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6669743960667244e-05,
+      "loss": 1.2266,
+      "step": 6001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6668457318162195e-05,
+      "loss": 1.2363,
+      "step": 6002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666717044490123e-05,
+      "loss": 1.2118,
+      "step": 6003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666588334089805e-05,
+      "loss": 1.2811,
+      "step": 6004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666459600616637e-05,
+      "loss": 1.2433,
+      "step": 6005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666330844071989e-05,
+      "loss": 1.1961,
+      "step": 6006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6662020644572336e-05,
+      "loss": 1.2285,
+      "step": 6007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666073261773741e-05,
+      "loss": 1.2749,
+      "step": 6008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6659444360228824e-05,
+      "loss": 1.2778,
+      "step": 6009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6658155872060315e-05,
+      "loss": 1.2393,
+      "step": 6010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6656867153245594e-05,
+      "loss": 1.2814,
+      "step": 6011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6655578203798384e-05,
+      "loss": 1.2319,
+      "step": 6012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6654289023732415e-05,
+      "loss": 1.2083,
+      "step": 6013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.665299961306141e-05,
+      "loss": 1.3473,
+      "step": 6014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.665170997179911e-05,
+      "loss": 1.194,
+      "step": 6015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6650420099959244e-05,
+      "loss": 1.3436,
+      "step": 6016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.664912999755555e-05,
+      "loss": 1.1849,
+      "step": 6017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.664783966460175e-05,
+      "loss": 1.3162,
+      "step": 6018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.664654910111162e-05,
+      "loss": 1.2749,
+      "step": 6019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6645258307098873e-05,
+      "loss": 1.2749,
+      "step": 6020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.664396728257727e-05,
+      "loss": 1.25,
+      "step": 6021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6642676027560556e-05,
+      "loss": 1.1877,
+      "step": 6022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.664138454206248e-05,
+      "loss": 1.2594,
+      "step": 6023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.66400928260968e-05,
+      "loss": 1.2795,
+      "step": 6024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.663880087967727e-05,
+      "loss": 1.2876,
+      "step": 6025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.663750870281764e-05,
+      "loss": 1.1813,
+      "step": 6026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.663621629553169e-05,
+      "loss": 1.2329,
+      "step": 6027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.663492365783316e-05,
+      "loss": 1.1997,
+      "step": 6028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.663363078973584e-05,
+      "loss": 1.3302,
+      "step": 6029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6632337691253483e-05,
+      "loss": 1.2386,
+      "step": 6030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6631044362399857e-05,
+      "loss": 1.2507,
+      "step": 6031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.662975080318874e-05,
+      "loss": 1.1849,
+      "step": 6032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.662845701363392e-05,
+      "loss": 1.2357,
+      "step": 6033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.662716299374915e-05,
+      "loss": 1.172,
+      "step": 6034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.662586874354823e-05,
+      "loss": 1.3276,
+      "step": 6035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.662457426304494e-05,
+      "loss": 1.1639,
+      "step": 6036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.662327955225306e-05,
+      "loss": 1.1524,
+      "step": 6037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6621984611186386e-05,
+      "loss": 1.2746,
+      "step": 6038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.66206894398587e-05,
+      "loss": 1.2954,
+      "step": 6039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6619394038283795e-05,
+      "loss": 1.2954,
+      "step": 6040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6618098406475466e-05,
+      "loss": 1.2795,
+      "step": 6041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6616802544447526e-05,
+      "loss": 1.2035,
+      "step": 6042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.661550645221375e-05,
+      "loss": 1.1828,
+      "step": 6043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6614210129787964e-05,
+      "loss": 1.2127,
+      "step": 6044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.661291357718395e-05,
+      "loss": 1.2338,
+      "step": 6045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6611616794415535e-05,
+      "loss": 1.2472,
+      "step": 6046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6610319781496526e-05,
+      "loss": 1.2412,
+      "step": 6047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6609022538440725e-05,
+      "loss": 1.2043,
+      "step": 6048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.660772506526196e-05,
+      "loss": 1.3566,
+      "step": 6049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.660642736197402e-05,
+      "loss": 1.3283,
+      "step": 6050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.660512942859077e-05,
+      "loss": 1.1706,
+      "step": 6051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6603831265126e-05,
+      "loss": 1.2131,
+      "step": 6052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.660253287159355e-05,
+      "loss": 1.0871,
+      "step": 6053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.660123424800723e-05,
+      "loss": 1.1784,
+      "step": 6054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.659993539438088e-05,
+      "loss": 1.1395,
+      "step": 6055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.659863631072834e-05,
+      "loss": 1.2254,
+      "step": 6056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.659733699706343e-05,
+      "loss": 1.2412,
+      "step": 6057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.65960374534e-05,
+      "loss": 1.1807,
+      "step": 6058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.659473767975187e-05,
+      "loss": 1.2038,
+      "step": 6059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.65934376761329e-05,
+      "loss": 1.2311,
+      "step": 6060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.659213744255693e-05,
+      "loss": 1.342,
+      "step": 6061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.659083697903781e-05,
+      "loss": 1.1755,
+      "step": 6062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658953628558937e-05,
+      "loss": 1.1914,
+      "step": 6063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658823536222549e-05,
+      "loss": 1.2512,
+      "step": 6064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6586934208960005e-05,
+      "loss": 1.2821,
+      "step": 6065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658563282580678e-05,
+      "loss": 1.1899,
+      "step": 6066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658433121277966e-05,
+      "loss": 1.1642,
+      "step": 6067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658302936989253e-05,
+      "loss": 1.2176,
+      "step": 6068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6581727297159225e-05,
+      "loss": 1.1169,
+      "step": 6069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658042499459364e-05,
+      "loss": 1.1761,
+      "step": 6070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6579122462209626e-05,
+      "loss": 1.2026,
+      "step": 6071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.657781970002106e-05,
+      "loss": 1.2375,
+      "step": 6072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6576516708041814e-05,
+      "loss": 1.242,
+      "step": 6073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6575213486285765e-05,
+      "loss": 1.173,
+      "step": 6074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6573910034766785e-05,
+      "loss": 1.2852,
+      "step": 6075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.657260635349877e-05,
+      "loss": 1.2072,
+      "step": 6076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.657130244249559e-05,
+      "loss": 1.2573,
+      "step": 6077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656999830177113e-05,
+      "loss": 1.1598,
+      "step": 6078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656869393133929e-05,
+      "loss": 1.221,
+      "step": 6079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656738933121396e-05,
+      "loss": 1.2619,
+      "step": 6080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656608450140902e-05,
+      "loss": 1.3192,
+      "step": 6081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656477944193837e-05,
+      "loss": 1.215,
+      "step": 6082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656347415281592e-05,
+      "loss": 1.2026,
+      "step": 6083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656216863405556e-05,
+      "loss": 1.2132,
+      "step": 6084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656086288567119e-05,
+      "loss": 1.2367,
+      "step": 6085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655955690767672e-05,
+      "loss": 1.2352,
+      "step": 6086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655825070008606e-05,
+      "loss": 1.1722,
+      "step": 6087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655694426291311e-05,
+      "loss": 1.2184,
+      "step": 6088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655563759617181e-05,
+      "loss": 1.2126,
+      "step": 6089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655433069987605e-05,
+      "loss": 1.245,
+      "step": 6090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655302357403975e-05,
+      "loss": 1.2311,
+      "step": 6091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655171621867683e-05,
+      "loss": 1.2861,
+      "step": 6092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655040863380122e-05,
+      "loss": 1.2682,
+      "step": 6093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.654910081942684e-05,
+      "loss": 1.2813,
+      "step": 6094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6547792775567625e-05,
+      "loss": 1.271,
+      "step": 6095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6546484502237495e-05,
+      "loss": 1.2985,
+      "step": 6096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.654517599945039e-05,
+      "loss": 1.1542,
+      "step": 6097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.654386726722023e-05,
+      "loss": 1.2538,
+      "step": 6098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.654255830556097e-05,
+      "loss": 1.3236,
+      "step": 6099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.654124911448655e-05,
+      "loss": 1.3154,
+      "step": 6100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6539939694010894e-05,
+      "loss": 1.2868,
+      "step": 6101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6538630044147964e-05,
+      "loss": 1.1755,
+      "step": 6102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.65373201649117e-05,
+      "loss": 1.2427,
+      "step": 6103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6536010056316046e-05,
+      "loss": 1.3027,
+      "step": 6104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.653469971837496e-05,
+      "loss": 1.223,
+      "step": 6105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6533389151102405e-05,
+      "loss": 1.2458,
+      "step": 6106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.653207835451232e-05,
+      "loss": 1.2262,
+      "step": 6107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.653076732861868e-05,
+      "loss": 1.2134,
+      "step": 6108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652945607343544e-05,
+      "loss": 1.2065,
+      "step": 6109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652814458897656e-05,
+      "loss": 1.214,
+      "step": 6110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652683287525601e-05,
+      "loss": 1.1734,
+      "step": 6111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652552093228776e-05,
+      "loss": 1.1932,
+      "step": 6112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6524208760085774e-05,
+      "loss": 1.1777,
+      "step": 6113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6522896358664044e-05,
+      "loss": 1.304,
+      "step": 6114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6521583728036526e-05,
+      "loss": 1.2159,
+      "step": 6115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652027086821721e-05,
+      "loss": 1.2344,
+      "step": 6116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6518957779220074e-05,
+      "loss": 1.2835,
+      "step": 6117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6517644461059105e-05,
+      "loss": 1.228,
+      "step": 6118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.651633091374829e-05,
+      "loss": 1.2217,
+      "step": 6119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.65150171373016e-05,
+      "loss": 1.1467,
+      "step": 6120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.651370313173305e-05,
+      "loss": 1.271,
+      "step": 6121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.651238889705662e-05,
+      "loss": 1.2427,
+      "step": 6122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6511074433286306e-05,
+      "loss": 1.1735,
+      "step": 6123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6509759740436115e-05,
+      "loss": 1.2036,
+      "step": 6124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.650844481852003e-05,
+      "loss": 1.1504,
+      "step": 6125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.650712966755209e-05,
+      "loss": 1.1597,
+      "step": 6126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.650581428754625e-05,
+      "loss": 1.2616,
+      "step": 6127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.650449867851656e-05,
+      "loss": 1.1754,
+      "step": 6128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.650318284047701e-05,
+      "loss": 1.1976,
+      "step": 6129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.650186677344162e-05,
+      "loss": 1.1491,
+      "step": 6130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.65005504774244e-05,
+      "loss": 1.2271,
+      "step": 6131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.649923395243937e-05,
+      "loss": 1.2661,
+      "step": 6132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.649791719850055e-05,
+      "loss": 1.1701,
+      "step": 6133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6496600215621965e-05,
+      "loss": 1.2145,
+      "step": 6134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6495283003817634e-05,
+      "loss": 1.2306,
+      "step": 6135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.64939655631016e-05,
+      "loss": 1.2283,
+      "step": 6136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.649264789348787e-05,
+      "loss": 1.2013,
+      "step": 6137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.64913299949905e-05,
+      "loss": 1.2196,
+      "step": 6138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6490011867623506e-05,
+      "loss": 1.2184,
+      "step": 6139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.648869351140094e-05,
+      "loss": 1.1372,
+      "step": 6140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.648737492633683e-05,
+      "loss": 1.2087,
+      "step": 6141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.648605611244522e-05,
+      "loss": 1.1958,
+      "step": 6142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.648473706974016e-05,
+      "loss": 1.2393,
+      "step": 6143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.648341779823569e-05,
+      "loss": 1.2315,
+      "step": 6144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6482098297945867e-05,
+      "loss": 1.1345,
+      "step": 6145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.648077856888473e-05,
+      "loss": 1.1392,
+      "step": 6146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.647945861106635e-05,
+      "loss": 1.155,
+      "step": 6147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.647813842450478e-05,
+      "loss": 1.17,
+      "step": 6148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6476818009214076e-05,
+      "loss": 1.2142,
+      "step": 6149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.647549736520829e-05,
+      "loss": 1.1628,
+      "step": 6150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.64741764925015e-05,
+      "loss": 1.2489,
+      "step": 6151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.647285539110777e-05,
+      "loss": 1.1862,
+      "step": 6152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.647153406104117e-05,
+      "loss": 1.2469,
+      "step": 6153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.647021250231576e-05,
+      "loss": 1.2801,
+      "step": 6154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6468890714945624e-05,
+      "loss": 1.2033,
+      "step": 6155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.646756869894483e-05,
+      "loss": 1.207,
+      "step": 6156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6466246454327465e-05,
+      "loss": 1.187,
+      "step": 6157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6464923981107616e-05,
+      "loss": 1.2415,
+      "step": 6158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.646360127929935e-05,
+      "loss": 1.214,
+      "step": 6159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.646227834891677e-05,
+      "loss": 1.2062,
+      "step": 6160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.646095518997396e-05,
+      "loss": 1.1977,
+      "step": 6161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.645963180248498e-05,
+      "loss": 1.1761,
+      "step": 6162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6458308186463974e-05,
+      "loss": 1.2296,
+      "step": 6163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6456984341925004e-05,
+      "loss": 1.229,
+      "step": 6164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6455660268882187e-05,
+      "loss": 1.1779,
+      "step": 6165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.64543359673496e-05,
+      "loss": 1.1942,
+      "step": 6166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.645301143734137e-05,
+      "loss": 1.1981,
+      "step": 6167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.645168667887159e-05,
+      "loss": 1.2212,
+      "step": 6168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.645036169195437e-05,
+      "loss": 1.1194,
+      "step": 6169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644903647660382e-05,
+      "loss": 1.1791,
+      "step": 6170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644771103283405e-05,
+      "loss": 1.1743,
+      "step": 6171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6446385360659184e-05,
+      "loss": 1.2157,
+      "step": 6172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6445059460093336e-05,
+      "loss": 1.1216,
+      "step": 6173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644373333115062e-05,
+      "loss": 1.2469,
+      "step": 6174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644240697384517e-05,
+      "loss": 1.2384,
+      "step": 6175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.64410803881911e-05,
+      "loss": 1.182,
+      "step": 6176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.643975357420254e-05,
+      "loss": 1.2208,
+      "step": 6177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.643842653189362e-05,
+      "loss": 1.1195,
+      "step": 6178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.643709926127847e-05,
+      "loss": 1.1139,
+      "step": 6179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.643577176237123e-05,
+      "loss": 1.163,
+      "step": 6180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6434444035186044e-05,
+      "loss": 1.1724,
+      "step": 6181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6433116079737023e-05,
+      "loss": 1.0762,
+      "step": 6182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.643178789603834e-05,
+      "loss": 1.1766,
+      "step": 6183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6430459484104134e-05,
+      "loss": 1.2337,
+      "step": 6184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642913084394853e-05,
+      "loss": 1.2982,
+      "step": 6185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642780197558571e-05,
+      "loss": 1.2845,
+      "step": 6186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642647287902979e-05,
+      "loss": 1.1263,
+      "step": 6187
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6425143554294954e-05,
+      "loss": 1.2589,
+      "step": 6188
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642381400139535e-05,
+      "loss": 1.3027,
+      "step": 6189
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642248422034512e-05,
+      "loss": 1.1719,
+      "step": 6190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6421154211158446e-05,
+      "loss": 1.2171,
+      "step": 6191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.641982397384949e-05,
+      "loss": 1.1878,
+      "step": 6192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6418493508432404e-05,
+      "loss": 1.2384,
+      "step": 6193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.641716281492137e-05,
+      "loss": 1.1231,
+      "step": 6194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.641583189333055e-05,
+      "loss": 1.1863,
+      "step": 6195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6414500743674125e-05,
+      "loss": 1.2764,
+      "step": 6196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.641316936596627e-05,
+      "loss": 1.2422,
+      "step": 6197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.641183776022117e-05,
+      "loss": 1.2083,
+      "step": 6198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6410505926452985e-05,
+      "loss": 1.1849,
+      "step": 6199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6409173864675914e-05,
+      "loss": 1.1436,
+      "step": 6200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.640784157490413e-05,
+      "loss": 1.1353,
+      "step": 6201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.640650905715184e-05,
+      "loss": 1.1813,
+      "step": 6202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.640517631143323e-05,
+      "loss": 1.1541,
+      "step": 6203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6403843337762474e-05,
+      "loss": 1.1338,
+      "step": 6204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.640251013615379e-05,
+      "loss": 1.2017,
+      "step": 6205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6401176706621366e-05,
+      "loss": 1.269,
+      "step": 6206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6399843049179404e-05,
+      "loss": 1.2202,
+      "step": 6207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.639850916384211e-05,
+      "loss": 1.2489,
+      "step": 6208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.639717505062368e-05,
+      "loss": 1.2583,
+      "step": 6209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.639584070953832e-05,
+      "loss": 1.1375,
+      "step": 6210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6394506140600247e-05,
+      "loss": 1.1885,
+      "step": 6211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.639317134382368e-05,
+      "loss": 1.1827,
+      "step": 6212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.639183631922283e-05,
+      "loss": 1.2072,
+      "step": 6213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.63905010668119e-05,
+      "loss": 1.2205,
+      "step": 6214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638916558660512e-05,
+      "loss": 1.0991,
+      "step": 6215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6387829878616715e-05,
+      "loss": 1.1841,
+      "step": 6216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638649394286091e-05,
+      "loss": 1.1686,
+      "step": 6217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6385157779351925e-05,
+      "loss": 1.1387,
+      "step": 6218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638382138810399e-05,
+      "loss": 1.1986,
+      "step": 6219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638248476913135e-05,
+      "loss": 1.1897,
+      "step": 6220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638114792244822e-05,
+      "loss": 1.2031,
+      "step": 6221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.637981084806885e-05,
+      "loss": 1.1374,
+      "step": 6222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.637847354600747e-05,
+      "loss": 1.2224,
+      "step": 6223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6377136016278325e-05,
+      "loss": 1.1182,
+      "step": 6224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6375798258895666e-05,
+      "loss": 1.209,
+      "step": 6225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.637446027387373e-05,
+      "loss": 1.2065,
+      "step": 6226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.637312206122677e-05,
+      "loss": 1.2306,
+      "step": 6227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.637178362096903e-05,
+      "loss": 1.1229,
+      "step": 6228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.637044495311478e-05,
+      "loss": 1.2044,
+      "step": 6229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.636910605767826e-05,
+      "loss": 1.1816,
+      "step": 6230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6367766934673726e-05,
+      "loss": 1.119,
+      "step": 6231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6366427584115456e-05,
+      "loss": 1.1746,
+      "step": 6232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.636508800601771e-05,
+      "loss": 1.1748,
+      "step": 6233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6363748200394744e-05,
+      "loss": 1.1873,
+      "step": 6234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6362408167260815e-05,
+      "loss": 1.1271,
+      "step": 6235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.636106790663023e-05,
+      "loss": 1.1094,
+      "step": 6236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635972741851723e-05,
+      "loss": 1.2,
+      "step": 6237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.63583867029361e-05,
+      "loss": 1.1515,
+      "step": 6238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635704575990112e-05,
+      "loss": 1.1658,
+      "step": 6239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635570458942657e-05,
+      "loss": 1.2153,
+      "step": 6240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635436319152673e-05,
+      "loss": 1.1226,
+      "step": 6241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635302156621588e-05,
+      "loss": 1.2008,
+      "step": 6242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635167971350832e-05,
+      "loss": 1.111,
+      "step": 6243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6350337633418336e-05,
+      "loss": 1.1612,
+      "step": 6244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.634899532596022e-05,
+      "loss": 1.251,
+      "step": 6245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.634765279114827e-05,
+      "loss": 1.2358,
+      "step": 6246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6346310028996765e-05,
+      "loss": 1.1506,
+      "step": 6247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6344967039520024e-05,
+      "loss": 1.1771,
+      "step": 6248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.634362382273234e-05,
+      "loss": 1.2118,
+      "step": 6249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6342280378648024e-05,
+      "loss": 1.2598,
+      "step": 6250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.634093670728138e-05,
+      "loss": 1.1768,
+      "step": 6251
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6339592808646706e-05,
+      "loss": 1.3053,
+      "step": 6252
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6338248682758335e-05,
+      "loss": 1.2277,
+      "step": 6253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6336904329630574e-05,
+      "loss": 1.1865,
+      "step": 6254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633555974927773e-05,
+      "loss": 1.2203,
+      "step": 6255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633421494171413e-05,
+      "loss": 1.2598,
+      "step": 6256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633286990695408e-05,
+      "loss": 1.1648,
+      "step": 6257
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633152464501193e-05,
+      "loss": 1.1565,
+      "step": 6258
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633017915590199e-05,
+      "loss": 1.1944,
+      "step": 6259
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.63288334396386e-05,
+      "loss": 1.2444,
+      "step": 6260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6327487496236066e-05,
+      "loss": 1.1421,
+      "step": 6261
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.632614132570875e-05,
+      "loss": 1.1722,
+      "step": 6262
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.632479492807097e-05,
+      "loss": 1.1364,
+      "step": 6263
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.632344830333708e-05,
+      "loss": 1.2007,
+      "step": 6264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6322101451521407e-05,
+      "loss": 1.1842,
+      "step": 6265
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.63207543726383e-05,
+      "loss": 1.1987,
+      "step": 6266
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.63194070667021e-05,
+      "loss": 1.2505,
+      "step": 6267
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.631805953372716e-05,
+      "loss": 1.204,
+      "step": 6268
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.631671177372782e-05,
+      "loss": 1.1621,
+      "step": 6269
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6315363786718454e-05,
+      "loss": 1.2342,
+      "step": 6270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.63140155727134e-05,
+      "loss": 1.2057,
+      "step": 6271
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.631266713172702e-05,
+      "loss": 1.1341,
+      "step": 6272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.631131846377367e-05,
+      "loss": 1.2516,
+      "step": 6273
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630996956886772e-05,
+      "loss": 1.181,
+      "step": 6274
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630862044702353e-05,
+      "loss": 1.1846,
+      "step": 6275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630727109825548e-05,
+      "loss": 1.2079,
+      "step": 6276
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630592152257791e-05,
+      "loss": 1.1916,
+      "step": 6277
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630457172000522e-05,
+      "loss": 1.1146,
+      "step": 6278
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630322169055178e-05,
+      "loss": 1.1294,
+      "step": 6279
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630187143423196e-05,
+      "loss": 1.1717,
+      "step": 6280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6300520951060136e-05,
+      "loss": 1.1213,
+      "step": 6281
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6299170241050706e-05,
+      "loss": 1.092,
+      "step": 6282
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.629781930421804e-05,
+      "loss": 1.106,
+      "step": 6283
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.629646814057652e-05,
+      "loss": 1.1348,
+      "step": 6284
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.629511675014056e-05,
+      "loss": 1.061,
+      "step": 6285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.629376513292452e-05,
+      "loss": 1.1469,
+      "step": 6286
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.629241328894281e-05,
+      "loss": 1.1733,
+      "step": 6287
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.629106121820984e-05,
+      "loss": 1.1987,
+      "step": 6288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6289708920739986e-05,
+      "loss": 1.145,
+      "step": 6289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.628835639654765e-05,
+      "loss": 1.1115,
+      "step": 6290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6287003645647246e-05,
+      "loss": 1.1647,
+      "step": 6291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6285650668053184e-05,
+      "loss": 1.2146,
+      "step": 6292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.628429746377986e-05,
+      "loss": 1.1629,
+      "step": 6293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6282944032841694e-05,
+      "loss": 1.1793,
+      "step": 6294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6281590375253085e-05,
+      "loss": 1.1833,
+      "step": 6295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6280236491028465e-05,
+      "loss": 1.2292,
+      "step": 6296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.627888238018225e-05,
+      "loss": 1.099,
+      "step": 6297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.627752804272885e-05,
+      "loss": 1.1115,
+      "step": 6298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.62761734786827e-05,
+      "loss": 1.1647,
+      "step": 6299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.627481868805822e-05,
+      "loss": 1.2163,
+      "step": 6300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.627346367086983e-05,
+      "loss": 1.0788,
+      "step": 6301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.627210842713197e-05,
+      "loss": 1.1846,
+      "step": 6302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6270752956859076e-05,
+      "loss": 1.1587,
+      "step": 6303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.626939726006557e-05,
+      "loss": 1.1896,
+      "step": 6304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.62680413367659e-05,
+      "loss": 1.1769,
+      "step": 6305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.62666851869745e-05,
+      "loss": 1.1051,
+      "step": 6306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.626532881070581e-05,
+      "loss": 1.2173,
+      "step": 6307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.626397220797428e-05,
+      "loss": 1.0825,
+      "step": 6308
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.626261537879436e-05,
+      "loss": 1.2119,
+      "step": 6309
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6261258323180504e-05,
+      "loss": 1.1582,
+      "step": 6310
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6259901041147136e-05,
+      "loss": 1.1818,
+      "step": 6311
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.625854353270874e-05,
+      "loss": 1.2525,
+      "step": 6312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.625718579787976e-05,
+      "loss": 1.1803,
+      "step": 6313
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.625582783667466e-05,
+      "loss": 1.1574,
+      "step": 6314
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.625446964910789e-05,
+      "loss": 1.1953,
+      "step": 6315
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.625311123519393e-05,
+      "loss": 1.1953,
+      "step": 6316
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6251752594947226e-05,
+      "loss": 1.1352,
+      "step": 6317
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6250393728382274e-05,
+      "loss": 1.0667,
+      "step": 6318
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.624903463551352e-05,
+      "loss": 1.2266,
+      "step": 6319
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6247675316355455e-05,
+      "loss": 1.2461,
+      "step": 6320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.624631577092254e-05,
+      "loss": 1.1701,
+      "step": 6321
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.624495599922927e-05,
+      "loss": 1.1296,
+      "step": 6322
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6243596001290114e-05,
+      "loss": 1.22,
+      "step": 6323
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.624223577711956e-05,
+      "loss": 1.1025,
+      "step": 6324
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.624087532673208e-05,
+      "loss": 1.151,
+      "step": 6325
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.623951465014219e-05,
+      "loss": 1.1046,
+      "step": 6326
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.623815374736435e-05,
+      "loss": 1.211,
+      "step": 6327
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.623679261841307e-05,
+      "loss": 1.3438,
+      "step": 6328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.623543126330284e-05,
+      "loss": 1.2149,
+      "step": 6329
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.623406968204816e-05,
+      "loss": 1.2276,
+      "step": 6330
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6232707874663536e-05,
+      "loss": 1.1623,
+      "step": 6331
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6231345841163465e-05,
+      "loss": 1.075,
+      "step": 6332
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6229983581562443e-05,
+      "loss": 1.2073,
+      "step": 6333
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.622862109587499e-05,
+      "loss": 1.1732,
+      "step": 6334
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.622725838411561e-05,
+      "loss": 1.222,
+      "step": 6335
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6225895446298814e-05,
+      "loss": 1.2415,
+      "step": 6336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6224532282439116e-05,
+      "loss": 1.1794,
+      "step": 6337
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.622316889255104e-05,
+      "loss": 1.1309,
+      "step": 6338
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.622180527664909e-05,
+      "loss": 1.2078,
+      "step": 6339
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6220441434747806e-05,
+      "loss": 1.1005,
+      "step": 6340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.621907736686171e-05,
+      "loss": 1.1201,
+      "step": 6341
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.621771307300531e-05,
+      "loss": 1.1642,
+      "step": 6342
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.621634855319315e-05,
+      "loss": 1.2231,
+      "step": 6343
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.621498380743976e-05,
+      "loss": 1.2407,
+      "step": 6344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6213618835759674e-05,
+      "loss": 1.1857,
+      "step": 6345
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6212253638167426e-05,
+      "loss": 1.2101,
+      "step": 6346
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.621088821467756e-05,
+      "loss": 1.1331,
+      "step": 6347
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6209522565304604e-05,
+      "loss": 1.1141,
+      "step": 6348
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.620815669006311e-05,
+      "loss": 1.1042,
+      "step": 6349
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.620679058896763e-05,
+      "loss": 1.0973,
+      "step": 6350
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.62054242620327e-05,
+      "loss": 1.2165,
+      "step": 6351
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6204057709272873e-05,
+      "loss": 1.1047,
+      "step": 6352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.620269093070271e-05,
+      "loss": 1.2261,
+      "step": 6353
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6201323926336756e-05,
+      "loss": 1.1074,
+      "step": 6354
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619995669618958e-05,
+      "loss": 1.1658,
+      "step": 6355
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619858924027573e-05,
+      "loss": 1.1865,
+      "step": 6356
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619722155860978e-05,
+      "loss": 1.1587,
+      "step": 6357
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619585365120629e-05,
+      "loss": 1.1751,
+      "step": 6358
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619448551807982e-05,
+      "loss": 1.136,
+      "step": 6359
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619311715924495e-05,
+      "loss": 1.1597,
+      "step": 6360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619174857471625e-05,
+      "loss": 1.17,
+      "step": 6361
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6190379764508296e-05,
+      "loss": 1.1693,
+      "step": 6362
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.618901072863566e-05,
+      "loss": 1.1812,
+      "step": 6363
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.618764146711292e-05,
+      "loss": 1.1341,
+      "step": 6364
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6186271979954666e-05,
+      "loss": 1.1504,
+      "step": 6365
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.618490226717548e-05,
+      "loss": 1.1438,
+      "step": 6366
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6183532328789937e-05,
+      "loss": 1.1235,
+      "step": 6367
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.618216216481265e-05,
+      "loss": 1.2138,
+      "step": 6368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.618079177525818e-05,
+      "loss": 1.1727,
+      "step": 6369
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6179421160141145e-05,
+      "loss": 1.1517,
+      "step": 6370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6178050319476136e-05,
+      "loss": 1.16,
+      "step": 6371
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.617667925327774e-05,
+      "loss": 1.1048,
+      "step": 6372
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6175307961560574e-05,
+      "loss": 1.2108,
+      "step": 6373
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6173936444339234e-05,
+      "loss": 1.1001,
+      "step": 6374
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6172564701628324e-05,
+      "loss": 1.208,
+      "step": 6375
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6171192733442455e-05,
+      "loss": 1.2,
+      "step": 6376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6169820539796236e-05,
+      "loss": 1.1322,
+      "step": 6377
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6168448120704277e-05,
+      "loss": 1.1411,
+      "step": 6378
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.61670754761812e-05,
+      "loss": 1.1584,
+      "step": 6379
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.616570260624162e-05,
+      "loss": 1.0811,
+      "step": 6380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.616432951090016e-05,
+      "loss": 1.2131,
+      "step": 6381
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.616295619017144e-05,
+      "loss": 1.1745,
+      "step": 6382
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.616158264407008e-05,
+      "loss": 1.1143,
+      "step": 6383
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6160208872610714e-05,
+      "loss": 1.2281,
+      "step": 6384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.615883487580798e-05,
+      "loss": 1.197,
+      "step": 6385
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6157460653676485e-05,
+      "loss": 1.2484,
+      "step": 6386
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6156086206230885e-05,
+      "loss": 1.1056,
+      "step": 6387
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6154711533485816e-05,
+      "loss": 1.1644,
+      "step": 6388
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6153336635455904e-05,
+      "loss": 1.147,
+      "step": 6389
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.61519615121558e-05,
+      "loss": 1.272,
+      "step": 6390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.615058616360016e-05,
+      "loss": 1.1889,
+      "step": 6391
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6149210589803606e-05,
+      "loss": 1.1467,
+      "step": 6392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.61478347907808e-05,
+      "loss": 1.1296,
+      "step": 6393
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6146458766546394e-05,
+      "loss": 1.1849,
+      "step": 6394
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.614508251711505e-05,
+      "loss": 1.2209,
+      "step": 6395
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.61437060425014e-05,
+      "loss": 1.1017,
+      "step": 6396
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.614232934272013e-05,
+      "loss": 1.2139,
+      "step": 6397
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6140952417785875e-05,
+      "loss": 1.1642,
+      "step": 6398
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.613957526771332e-05,
+      "loss": 1.1919,
+      "step": 6399
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.613819789251712e-05,
+      "loss": 1.2054,
+      "step": 6400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.613682029221195e-05,
+      "loss": 1.0975,
+      "step": 6401
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6135442466812475e-05,
+      "loss": 1.1534,
+      "step": 6402
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.613406441633337e-05,
+      "loss": 1.1353,
+      "step": 6403
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6132686140789305e-05,
+      "loss": 1.1136,
+      "step": 6404
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6131307640194965e-05,
+      "loss": 1.0786,
+      "step": 6405
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.612992891456503e-05,
+      "loss": 1.153,
+      "step": 6406
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6128549963914175e-05,
+      "loss": 1.1252,
+      "step": 6407
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.612717078825709e-05,
+      "loss": 1.1401,
+      "step": 6408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.612579138760848e-05,
+      "loss": 1.2767,
+      "step": 6409
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.612441176198299e-05,
+      "loss": 1.2022,
+      "step": 6410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.612303191139536e-05,
+      "loss": 1.1338,
+      "step": 6411
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.612165183586026e-05,
+      "loss": 1.2342,
+      "step": 6412
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.612027153539238e-05,
+      "loss": 1.2028,
+      "step": 6413
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6118891010006446e-05,
+      "loss": 1.1077,
+      "step": 6414
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6117510259717135e-05,
+      "loss": 1.151,
+      "step": 6415
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6116129284539167e-05,
+      "loss": 1.1206,
+      "step": 6416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.611474808448724e-05,
+      "loss": 1.1429,
+      "step": 6417
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6113366659576064e-05,
+      "loss": 1.0888,
+      "step": 6418
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6111985009820345e-05,
+      "loss": 1.1063,
+      "step": 6419
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.611060313523481e-05,
+      "loss": 1.2179,
+      "step": 6420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.610922103583417e-05,
+      "loss": 1.1489,
+      "step": 6421
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6107838711633135e-05,
+      "loss": 1.096,
+      "step": 6422
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6106456162646436e-05,
+      "loss": 1.1128,
+      "step": 6423
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.610507338888879e-05,
+      "loss": 1.179,
+      "step": 6424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.610369039037493e-05,
+      "loss": 1.1211,
+      "step": 6425
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.610230716711959e-05,
+      "loss": 1.123,
+      "step": 6426
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6100923719137465e-05,
+      "loss": 1.1291,
+      "step": 6427
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.609954004644333e-05,
+      "loss": 1.1234,
+      "step": 6428
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6098156149051905e-05,
+      "loss": 1.2354,
+      "step": 6429
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.609677202697792e-05,
+      "loss": 1.1248,
+      "step": 6430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.609538768023612e-05,
+      "loss": 1.1611,
+      "step": 6431
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.609400310884125e-05,
+      "loss": 1.1914,
+      "step": 6432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.609261831280806e-05,
+      "loss": 1.0996,
+      "step": 6433
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.609123329215128e-05,
+      "loss": 1.1147,
+      "step": 6434
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.608984804688568e-05,
+      "loss": 1.2095,
+      "step": 6435
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.608846257702599e-05,
+      "loss": 1.1715,
+      "step": 6436
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.608707688258699e-05,
+      "loss": 1.0675,
+      "step": 6437
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.608569096358341e-05,
+      "loss": 1.0695,
+      "step": 6438
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.608430482003003e-05,
+      "loss": 1.1034,
+      "step": 6439
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6082918451941607e-05,
+      "loss": 1.1365,
+      "step": 6440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6081531859332896e-05,
+      "loss": 1.1182,
+      "step": 6441
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.608014504221867e-05,
+      "loss": 1.0879,
+      "step": 6442
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607875800061371e-05,
+      "loss": 1.164,
+      "step": 6443
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607737073453276e-05,
+      "loss": 1.1268,
+      "step": 6444
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6075983243990616e-05,
+      "loss": 1.2206,
+      "step": 6445
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607459552900204e-05,
+      "loss": 1.1869,
+      "step": 6446
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607320758958182e-05,
+      "loss": 1.1221,
+      "step": 6447
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607181942574473e-05,
+      "loss": 1.1354,
+      "step": 6448
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607043103750556e-05,
+      "loss": 1.1517,
+      "step": 6449
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.606904242487909e-05,
+      "loss": 1.1044,
+      "step": 6450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.606765358788011e-05,
+      "loss": 1.1763,
+      "step": 6451
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.60662645265234e-05,
+      "loss": 1.2041,
+      "step": 6452
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.606487524082378e-05,
+      "loss": 1.158,
+      "step": 6453
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.606348573079602e-05,
+      "loss": 1.1071,
+      "step": 6454
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6062095996454926e-05,
+      "loss": 1.1704,
+      "step": 6455
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.606070603781529e-05,
+      "loss": 1.0963,
+      "step": 6456
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6059315854891925e-05,
+      "loss": 1.1193,
+      "step": 6457
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605792544769964e-05,
+      "loss": 1.105,
+      "step": 6458
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605653481625322e-05,
+      "loss": 1.1414,
+      "step": 6459
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605514396056749e-05,
+      "loss": 1.0104,
+      "step": 6460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605375288065727e-05,
+      "loss": 1.1191,
+      "step": 6461
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605236157653736e-05,
+      "loss": 1.1032,
+      "step": 6462
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6050970048222576e-05,
+      "loss": 1.0978,
+      "step": 6463
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6049578295727744e-05,
+      "loss": 1.121,
+      "step": 6464
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6048186319067674e-05,
+      "loss": 1.1078,
+      "step": 6465
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6046794118257215e-05,
+      "loss": 1.0623,
+      "step": 6466
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6045401693311166e-05,
+      "loss": 1.1308,
+      "step": 6467
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6044009044244366e-05,
+      "loss": 1.2447,
+      "step": 6468
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.604261617107165e-05,
+      "loss": 1.2501,
+      "step": 6469
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.604122307380784e-05,
+      "loss": 1.2866,
+      "step": 6470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603982975246779e-05,
+      "loss": 1.102,
+      "step": 6471
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603843620706631e-05,
+      "loss": 1.255,
+      "step": 6472
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603704243761827e-05,
+      "loss": 1.181,
+      "step": 6473
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.60356484441385e-05,
+      "loss": 1.1147,
+      "step": 6474
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603425422664184e-05,
+      "loss": 1.3288,
+      "step": 6475
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603285978514314e-05,
+      "loss": 1.1409,
+      "step": 6476
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603146511965727e-05,
+      "loss": 1.1697,
+      "step": 6477
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603007023019905e-05,
+      "loss": 1.114,
+      "step": 6478
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6028675116783346e-05,
+      "loss": 1.1711,
+      "step": 6479
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.602727977942503e-05,
+      "loss": 1.1165,
+      "step": 6480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.602588421813894e-05,
+      "loss": 1.1757,
+      "step": 6481
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.602448843293996e-05,
+      "loss": 1.1875,
+      "step": 6482
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6023092423842935e-05,
+      "loss": 1.1309,
+      "step": 6483
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.602169619086274e-05,
+      "loss": 1.1462,
+      "step": 6484
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.602029973401424e-05,
+      "loss": 1.1073,
+      "step": 6485
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.601890305331231e-05,
+      "loss": 1.1401,
+      "step": 6486
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6017506148771824e-05,
+      "loss": 1.1144,
+      "step": 6487
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.601610902040767e-05,
+      "loss": 1.1126,
+      "step": 6488
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.60147116682347e-05,
+      "loss": 1.0761,
+      "step": 6489
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6013314092267814e-05,
+      "loss": 1.065,
+      "step": 6490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.601191629252189e-05,
+      "loss": 1.1333,
+      "step": 6491
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6010518269011815e-05,
+      "loss": 1.1389,
+      "step": 6492
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.600912002175247e-05,
+      "loss": 1.0881,
+      "step": 6493
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.600772155075876e-05,
+      "loss": 1.1191,
+      "step": 6494
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.600632285604556e-05,
+      "loss": 1.1952,
+      "step": 6495
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.600492393762778e-05,
+      "loss": 1.2059,
+      "step": 6496
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.600352479552032e-05,
+      "loss": 1.2006,
+      "step": 6497
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6002125429738066e-05,
+      "loss": 1.1676,
+      "step": 6498
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.600072584029592e-05,
+      "loss": 1.2075,
+      "step": 6499
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.599932602720881e-05,
+      "loss": 1.1729,
+      "step": 6500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5997925990491615e-05,
+      "loss": 1.1365,
+      "step": 6501
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5996525730159254e-05,
+      "loss": 1.1161,
+      "step": 6502
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.599512524622665e-05,
+      "loss": 1.0994,
+      "step": 6503
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5993724538708704e-05,
+      "loss": 1.1532,
+      "step": 6504
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.599232360762034e-05,
+      "loss": 1.1548,
+      "step": 6505
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.599092245297647e-05,
+      "loss": 1.1603,
+      "step": 6506
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598952107479202e-05,
+      "loss": 1.0815,
+      "step": 6507
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5988119473081914e-05,
+      "loss": 1.1138,
+      "step": 6508
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5986717647861076e-05,
+      "loss": 1.1022,
+      "step": 6509
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598531559914444e-05,
+      "loss": 1.0879,
+      "step": 6510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598391332694694e-05,
+      "loss": 1.1461,
+      "step": 6511
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598251083128349e-05,
+      "loss": 1.1286,
+      "step": 6512
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5981108112169044e-05,
+      "loss": 1.2135,
+      "step": 6513
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5979705169618534e-05,
+      "loss": 1.1355,
+      "step": 6514
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.59783020036469e-05,
+      "loss": 1.2052,
+      "step": 6515
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.597689861426909e-05,
+      "loss": 1.1667,
+      "step": 6516
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.597549500150005e-05,
+      "loss": 1.1912,
+      "step": 6517
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.597409116535471e-05,
+      "loss": 1.1882,
+      "step": 6518
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.597268710584804e-05,
+      "loss": 1.0461,
+      "step": 6519
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.597128282299498e-05,
+      "loss": 1.147,
+      "step": 6520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.59698783168105e-05,
+      "loss": 1.1453,
+      "step": 6521
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.596847358730954e-05,
+      "loss": 1.0948,
+      "step": 6522
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.596706863450706e-05,
+      "loss": 1.0628,
+      "step": 6523
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.596566345841804e-05,
+      "loss": 1.066,
+      "step": 6524
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.596425805905742e-05,
+      "loss": 1.1867,
+      "step": 6525
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.596285243644019e-05,
+      "loss": 1.174,
+      "step": 6526
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.596144659058131e-05,
+      "loss": 1.1427,
+      "step": 6527
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5960040521495746e-05,
+      "loss": 1.102,
+      "step": 6528
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.595863422919847e-05,
+      "loss": 1.1073,
+      "step": 6529
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5957227713704474e-05,
+      "loss": 1.1219,
+      "step": 6530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.595582097502872e-05,
+      "loss": 1.2093,
+      "step": 6531
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5954414013186196e-05,
+      "loss": 1.2235,
+      "step": 6532
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.595300682819189e-05,
+      "loss": 1.1068,
+      "step": 6533
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5951599420060775e-05,
+      "loss": 1.146,
+      "step": 6534
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.595019178880784e-05,
+      "loss": 1.1558,
+      "step": 6535
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594878393444809e-05,
+      "loss": 1.083,
+      "step": 6536
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594737585699652e-05,
+      "loss": 1.1657,
+      "step": 6537
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.59459675564681e-05,
+      "loss": 1.1535,
+      "step": 6538
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594455903287784e-05,
+      "loss": 1.1472,
+      "step": 6539
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594315028624075e-05,
+      "loss": 1.1746,
+      "step": 6540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594174131657183e-05,
+      "loss": 1.1334,
+      "step": 6541
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594033212388606e-05,
+      "loss": 1.0865,
+      "step": 6542
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.593892270819848e-05,
+      "loss": 1.1344,
+      "step": 6543
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5937513069524086e-05,
+      "loss": 1.1173,
+      "step": 6544
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.593610320787789e-05,
+      "loss": 1.115,
+      "step": 6545
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.59346931232749e-05,
+      "loss": 1.1315,
+      "step": 6546
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.593328281573014e-05,
+      "loss": 1.099,
+      "step": 6547
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.593187228525862e-05,
+      "loss": 1.1825,
+      "step": 6548
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.593046153187538e-05,
+      "loss": 1.0951,
+      "step": 6549
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.592905055559543e-05,
+      "loss": 1.1681,
+      "step": 6550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.592763935643379e-05,
+      "loss": 1.0511,
+      "step": 6551
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.592622793440551e-05,
+      "loss": 1.195,
+      "step": 6552
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5924816289525594e-05,
+      "loss": 1.1576,
+      "step": 6553
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.592340442180909e-05,
+      "loss": 1.0983,
+      "step": 6554
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.592199233127104e-05,
+      "loss": 1.1895,
+      "step": 6555
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5920580017926464e-05,
+      "loss": 1.1258,
+      "step": 6556
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.591916748179042e-05,
+      "loss": 1.1585,
+      "step": 6557
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.591775472287794e-05,
+      "loss": 1.1549,
+      "step": 6558
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.591634174120406e-05,
+      "loss": 1.1823,
+      "step": 6559
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5914928536783854e-05,
+      "loss": 1.068,
+      "step": 6560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5913515109632356e-05,
+      "loss": 1.1266,
+      "step": 6561
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.591210145976461e-05,
+      "loss": 1.1919,
+      "step": 6562
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.591068758719569e-05,
+      "loss": 1.0623,
+      "step": 6563
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.590927349194063e-05,
+      "loss": 1.1912,
+      "step": 6564
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.590785917401451e-05,
+      "loss": 1.1208,
+      "step": 6565
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5906444633432385e-05,
+      "loss": 1.0731,
+      "step": 6566
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5905029870209315e-05,
+      "loss": 1.1343,
+      "step": 6567
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.590361488436037e-05,
+      "loss": 1.1091,
+      "step": 6568
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5902199675900606e-05,
+      "loss": 1.1009,
+      "step": 6569
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.590078424484512e-05,
+      "loss": 1.1759,
+      "step": 6570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589936859120896e-05,
+      "loss": 1.1117,
+      "step": 6571
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589795271500722e-05,
+      "loss": 1.1344,
+      "step": 6572
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589653661625497e-05,
+      "loss": 1.189,
+      "step": 6573
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.58951202949673e-05,
+      "loss": 1.2453,
+      "step": 6574
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5893703751159266e-05,
+      "loss": 1.1683,
+      "step": 6575
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589228698484599e-05,
+      "loss": 1.1336,
+      "step": 6576
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589086999604253e-05,
+      "loss": 1.1571,
+      "step": 6577
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.588945278476399e-05,
+      "loss": 1.1921,
+      "step": 6578
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.588803535102546e-05,
+      "loss": 1.1012,
+      "step": 6579
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.588661769484204e-05,
+      "loss": 1.0545,
+      "step": 6580
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.588519981622882e-05,
+      "loss": 1.162,
+      "step": 6581
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.58837817152009e-05,
+      "loss": 1.1442,
+      "step": 6582
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5882363391773384e-05,
+      "loss": 1.0712,
+      "step": 6583
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.588094484596137e-05,
+      "loss": 1.0937,
+      "step": 6584
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.587952607777998e-05,
+      "loss": 1.1701,
+      "step": 6585
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.58781070872443e-05,
+      "loss": 1.1179,
+      "step": 6586
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.587668787436947e-05,
+      "loss": 1.1028,
+      "step": 6587
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.587526843917058e-05,
+      "loss": 1.1664,
+      "step": 6588
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5873848781662765e-05,
+      "loss": 1.0864,
+      "step": 6589
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.587242890186112e-05,
+      "loss": 1.1577,
+      "step": 6590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.587100879978079e-05,
+      "loss": 1.1213,
+      "step": 6591
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586958847543688e-05,
+      "loss": 1.139,
+      "step": 6592
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586816792884453e-05,
+      "loss": 1.0731,
+      "step": 6593
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586674716001885e-05,
+      "loss": 1.0466,
+      "step": 6594
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586532616897499e-05,
+      "loss": 1.1162,
+      "step": 6595
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5863904955728075e-05,
+      "loss": 1.0718,
+      "step": 6596
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586248352029323e-05,
+      "loss": 1.1117,
+      "step": 6597
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586106186268561e-05,
+      "loss": 1.0098,
+      "step": 6598
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5859639982920346e-05,
+      "loss": 1.1049,
+      "step": 6599
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5858217881012576e-05,
+      "loss": 1.1702,
+      "step": 6600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.585679555697745e-05,
+      "loss": 1.0909,
+      "step": 6601
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.585537301083012e-05,
+      "loss": 1.2122,
+      "step": 6602
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.585395024258572e-05,
+      "loss": 1.0958,
+      "step": 6603
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.585252725225941e-05,
+      "loss": 1.0596,
+      "step": 6604
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5851104039866354e-05,
+      "loss": 1.2583,
+      "step": 6605
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584968060542169e-05,
+      "loss": 1.1377,
+      "step": 6606
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584825694894059e-05,
+      "loss": 1.0913,
+      "step": 6607
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584683307043821e-05,
+      "loss": 1.058,
+      "step": 6608
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584540896992972e-05,
+      "loss": 1.1374,
+      "step": 6609
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584398464743027e-05,
+      "loss": 1.0229,
+      "step": 6610
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584256010295504e-05,
+      "loss": 1.0894,
+      "step": 6611
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.58411353365192e-05,
+      "loss": 1.0427,
+      "step": 6612
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.583971034813793e-05,
+      "loss": 1.0139,
+      "step": 6613
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5838285137826386e-05,
+      "loss": 1.1265,
+      "step": 6614
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5836859705599766e-05,
+      "loss": 1.106,
+      "step": 6615
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.583543405147323e-05,
+      "loss": 1.1479,
+      "step": 6616
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.583400817546197e-05,
+      "loss": 1.0339,
+      "step": 6617
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.583258207758119e-05,
+      "loss": 0.998,
+      "step": 6618
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.583115575784604e-05,
+      "loss": 1.1121,
+      "step": 6619
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5829729216271735e-05,
+      "loss": 1.147,
+      "step": 6620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.582830245287346e-05,
+      "loss": 1.1759,
+      "step": 6621
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.582687546766641e-05,
+      "loss": 0.98,
+      "step": 6622
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.582544826066578e-05,
+      "loss": 1.0601,
+      "step": 6623
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.582402083188676e-05,
+      "loss": 1.0968,
+      "step": 6624
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5822593181344575e-05,
+      "loss": 1.0686,
+      "step": 6625
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.58211653090544e-05,
+      "loss": 1.1077,
+      "step": 6626
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.581973721503147e-05,
+      "loss": 1.1398,
+      "step": 6627
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5818308899290974e-05,
+      "loss": 1.1934,
+      "step": 6628
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.581688036184812e-05,
+      "loss": 1.1044,
+      "step": 6629
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.581545160271814e-05,
+      "loss": 1.1935,
+      "step": 6630
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.581402262191623e-05,
+      "loss": 1.1544,
+      "step": 6631
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.581259341945762e-05,
+      "loss": 1.0132,
+      "step": 6632
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.581116399535752e-05,
+      "loss": 1.0858,
+      "step": 6633
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.580973434963116e-05,
+      "loss": 1.1458,
+      "step": 6634
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.580830448229376e-05,
+      "loss": 1.0623,
+      "step": 6635
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5806874393360555e-05,
+      "loss": 1.1649,
+      "step": 6636
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.580544408284677e-05,
+      "loss": 1.2349,
+      "step": 6637
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.580401355076764e-05,
+      "loss": 1.1365,
+      "step": 6638
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5802582797138395e-05,
+      "loss": 1.0702,
+      "step": 6639
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.580115182197427e-05,
+      "loss": 1.1053,
+      "step": 6640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5799720625290505e-05,
+      "loss": 1.1834,
+      "step": 6641
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.579828920710234e-05,
+      "loss": 1.1393,
+      "step": 6642
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.579685756742503e-05,
+      "loss": 1.1026,
+      "step": 6643
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.579542570627381e-05,
+      "loss": 1.1709,
+      "step": 6644
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.579399362366393e-05,
+      "loss": 1.179,
+      "step": 6645
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.579256131961064e-05,
+      "loss": 1.1975,
+      "step": 6646
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.57911287941292e-05,
+      "loss": 1.065,
+      "step": 6647
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.578969604723486e-05,
+      "loss": 1.1574,
+      "step": 6648
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.578826307894288e-05,
+      "loss": 1.06,
+      "step": 6649
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5786829889268523e-05,
+      "loss": 1.1091,
+      "step": 6650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5785396478227036e-05,
+      "loss": 1.104,
+      "step": 6651
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.578396284583371e-05,
+      "loss": 1.1554,
+      "step": 6652
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5782528992103786e-05,
+      "loss": 1.099,
+      "step": 6653
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5781094917052544e-05,
+      "loss": 1.2183,
+      "step": 6654
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.577966062069527e-05,
+      "loss": 1.0902,
+      "step": 6655
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5778226103047216e-05,
+      "loss": 1.0763,
+      "step": 6656
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.577679136412367e-05,
+      "loss": 1.0574,
+      "step": 6657
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.577535640393992e-05,
+      "loss": 1.0845,
+      "step": 6658
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.577392122251122e-05,
+      "loss": 1.2161,
+      "step": 6659
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.577248581985288e-05,
+      "loss": 1.0702,
+      "step": 6660
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.577105019598018e-05,
+      "loss": 1.1717,
+      "step": 6661
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5769614350908404e-05,
+      "loss": 1.0825,
+      "step": 6662
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.576817828465285e-05,
+      "loss": 1.0968,
+      "step": 6663
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5766741997228805e-05,
+      "loss": 1.1091,
+      "step": 6664
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.576530548865155e-05,
+      "loss": 1.0406,
+      "step": 6665
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5763868758936415e-05,
+      "loss": 1.1086,
+      "step": 6666
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.576243180809867e-05,
+      "loss": 1.057,
+      "step": 6667
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.576099463615364e-05,
+      "loss": 1.1321,
+      "step": 6668
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.575955724311663e-05,
+      "loss": 1.0732,
+      "step": 6669
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.575811962900293e-05,
+      "loss": 1.1224,
+      "step": 6670
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.575668179382785e-05,
+      "loss": 1.0581,
+      "step": 6671
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.575524373760672e-05,
+      "loss": 1.1146,
+      "step": 6672
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.575380546035485e-05,
+      "loss": 1.0705,
+      "step": 6673
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.575236696208754e-05,
+      "loss": 1.0793,
+      "step": 6674
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.575092824282013e-05,
+      "loss": 1.0352,
+      "step": 6675
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.574948930256793e-05,
+      "loss": 1.1006,
+      "step": 6676
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.574805014134626e-05,
+      "loss": 1.1122,
+      "step": 6677
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.574661075917046e-05,
+      "loss": 1.1236,
+      "step": 6678
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5745171156055844e-05,
+      "loss": 1.1619,
+      "step": 6679
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.574373133201776e-05,
+      "loss": 1.0216,
+      "step": 6680
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.574229128707152e-05,
+      "loss": 1.1541,
+      "step": 6681
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.574085102123248e-05,
+      "loss": 1.0233,
+      "step": 6682
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5739410534515966e-05,
+      "loss": 1.0944,
+      "step": 6683
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5737969826937324e-05,
+      "loss": 1.1032,
+      "step": 6684
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5736528898511884e-05,
+      "loss": 1.1053,
+      "step": 6685
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5735087749255e-05,
+      "loss": 1.0209,
+      "step": 6686
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.573364637918203e-05,
+      "loss": 1.1524,
+      "step": 6687
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.573220478830831e-05,
+      "loss": 1.1027,
+      "step": 6688
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.57307629766492e-05,
+      "loss": 1.094,
+      "step": 6689
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.572932094422004e-05,
+      "loss": 1.0859,
+      "step": 6690
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.572787869103621e-05,
+      "loss": 1.0777,
+      "step": 6691
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5726436217113045e-05,
+      "loss": 1.1641,
+      "step": 6692
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.572499352246592e-05,
+      "loss": 1.1806,
+      "step": 6693
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.57235506071102e-05,
+      "loss": 1.0214,
+      "step": 6694
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5722107471061246e-05,
+      "loss": 1.1597,
+      "step": 6695
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5720664114334424e-05,
+      "loss": 1.1737,
+      "step": 6696
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.571922053694512e-05,
+      "loss": 1.1128,
+      "step": 6697
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.571777673890869e-05,
+      "loss": 1.1895,
+      "step": 6698
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.571633272024051e-05,
+      "loss": 1.0747,
+      "step": 6699
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.571488848095597e-05,
+      "loss": 1.0742,
+      "step": 6700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.571344402107045e-05,
+      "loss": 1.1388,
+      "step": 6701
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.571199934059931e-05,
+      "loss": 1.1016,
+      "step": 6702
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.571055443955796e-05,
+      "loss": 1.0879,
+      "step": 6703
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5709109317961786e-05,
+      "loss": 1.1144,
+      "step": 6704
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5707663975826165e-05,
+      "loss": 1.061,
+      "step": 6705
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.570621841316649e-05,
+      "loss": 1.1142,
+      "step": 6706
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.570477262999816e-05,
+      "loss": 1.0972,
+      "step": 6707
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5703326626336586e-05,
+      "loss": 1.1532,
+      "step": 6708
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.570188040219714e-05,
+      "loss": 1.2085,
+      "step": 6709
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5700433957595234e-05,
+      "loss": 1.0962,
+      "step": 6710
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.569898729254628e-05,
+      "loss": 1.1296,
+      "step": 6711
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5697540407065674e-05,
+      "loss": 1.1743,
+      "step": 6712
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.569609330116884e-05,
+      "loss": 1.1026,
+      "step": 6713
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.569464597487116e-05,
+      "loss": 1.1192,
+      "step": 6714
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5693198428188077e-05,
+      "loss": 1.0553,
+      "step": 6715
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.569175066113499e-05,
+      "loss": 1.1649,
+      "step": 6716
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5690302673727313e-05,
+      "loss": 1.159,
+      "step": 6717
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.568885446598049e-05,
+      "loss": 1.0996,
+      "step": 6718
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.568740603790992e-05,
+      "loss": 1.0905,
+      "step": 6719
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.568595738953103e-05,
+      "loss": 1.1593,
+      "step": 6720
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5684508520859256e-05,
+      "loss": 1.0569,
+      "step": 6721
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5683059431910026e-05,
+      "loss": 1.098,
+      "step": 6722
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.568161012269877e-05,
+      "loss": 1.1535,
+      "step": 6723
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.568016059324092e-05,
+      "loss": 1.1059,
+      "step": 6724
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5678710843551914e-05,
+      "loss": 1.1196,
+      "step": 6725
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.567726087364719e-05,
+      "loss": 1.1111,
+      "step": 6726
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.56758106835422e-05,
+      "loss": 1.0807,
+      "step": 6727
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.567436027325237e-05,
+      "loss": 1.1361,
+      "step": 6728
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.567290964279316e-05,
+      "loss": 1.1884,
+      "step": 6729
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.567145879218001e-05,
+      "loss": 1.1128,
+      "step": 6730
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.567000772142838e-05,
+      "loss": 1.171,
+      "step": 6731
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.566855643055371e-05,
+      "loss": 1.1382,
+      "step": 6732
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5667104919571466e-05,
+      "loss": 1.0147,
+      "step": 6733
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5665653188497096e-05,
+      "loss": 1.081,
+      "step": 6734
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.566420123734607e-05,
+      "loss": 1.0664,
+      "step": 6735
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5662749066133845e-05,
+      "loss": 1.0882,
+      "step": 6736
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.566129667487589e-05,
+      "loss": 1.0823,
+      "step": 6737
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.565984406358767e-05,
+      "loss": 1.0294,
+      "step": 6738
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.565839123228465e-05,
+      "loss": 1.0594,
+      "step": 6739
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.565693818098231e-05,
+      "loss": 1.0475,
+      "step": 6740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.565548490969611e-05,
+      "loss": 1.1784,
+      "step": 6741
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5654031418441545e-05,
+      "loss": 1.1423,
+      "step": 6742
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5652577707234084e-05,
+      "loss": 1.105,
+      "step": 6743
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.565112377608921e-05,
+      "loss": 1.1034,
+      "step": 6744
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.56496696250224e-05,
+      "loss": 1.0971,
+      "step": 6745
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.564821525404915e-05,
+      "loss": 1.1483,
+      "step": 6746
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.564676066318494e-05,
+      "loss": 1.0925,
+      "step": 6747
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5645305852445266e-05,
+      "loss": 1.1117,
+      "step": 6748
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.564385082184561e-05,
+      "loss": 1.0649,
+      "step": 6749
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.564239557140149e-05,
+      "loss": 1.0614,
+      "step": 6750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.564094010112838e-05,
+      "loss": 1.0866,
+      "step": 6751
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563948441104179e-05,
+      "loss": 1.086,
+      "step": 6752
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5638028501157224e-05,
+      "loss": 1.1819,
+      "step": 6753
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563657237149019e-05,
+      "loss": 1.0518,
+      "step": 6754
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563511602205618e-05,
+      "loss": 1.0395,
+      "step": 6755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5633659452870717e-05,
+      "loss": 1.2299,
+      "step": 6756
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563220266394931e-05,
+      "loss": 1.2229,
+      "step": 6757
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563074565530746e-05,
+      "loss": 1.0503,
+      "step": 6758
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.56292884269607e-05,
+      "loss": 1.2074,
+      "step": 6759
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.562783097892455e-05,
+      "loss": 1.1008,
+      "step": 6760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5626373311214515e-05,
+      "loss": 1.1765,
+      "step": 6761
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5624915423846124e-05,
+      "loss": 1.0459,
+      "step": 6762
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5623457316834914e-05,
+      "loss": 1.0938,
+      "step": 6763
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.56219989901964e-05,
+      "loss": 1.1751,
+      "step": 6764
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.562054044394612e-05,
+      "loss": 1.0734,
+      "step": 6765
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.56190816780996e-05,
+      "loss": 1.1475,
+      "step": 6766
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561762269267238e-05,
+      "loss": 1.0257,
+      "step": 6767
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561616348767999e-05,
+      "loss": 1.118,
+      "step": 6768
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561470406313798e-05,
+      "loss": 1.1418,
+      "step": 6769
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561324441906189e-05,
+      "loss": 1.1024,
+      "step": 6770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561178455546726e-05,
+      "loss": 1.1309,
+      "step": 6771
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561032447236964e-05,
+      "loss": 1.0575,
+      "step": 6772
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5608864169784574e-05,
+      "loss": 1.0799,
+      "step": 6773
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.560740364772761e-05,
+      "loss": 1.1204,
+      "step": 6774
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.560594290621432e-05,
+      "loss": 1.1523,
+      "step": 6775
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5604481945260236e-05,
+      "loss": 1.0871,
+      "step": 6776
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.560302076488094e-05,
+      "loss": 1.0871,
+      "step": 6777
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.560155936509197e-05,
+      "loss": 1.1302,
+      "step": 6778
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.560009774590891e-05,
+      "loss": 1.0155,
+      "step": 6779
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5598635907347314e-05,
+      "loss": 1.0656,
+      "step": 6780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.559717384942275e-05,
+      "loss": 1.1349,
+      "step": 6781
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.559571157215079e-05,
+      "loss": 1.0815,
+      "step": 6782
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5594249075547e-05,
+      "loss": 1.0977,
+      "step": 6783
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.559278635962697e-05,
+      "loss": 0.988,
+      "step": 6784
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.559132342440626e-05,
+      "loss": 1.1525,
+      "step": 6785
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.558986026990046e-05,
+      "loss": 1.0483,
+      "step": 6786
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.558839689612515e-05,
+      "loss": 1.1423,
+      "step": 6787
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.558693330309591e-05,
+      "loss": 0.9985,
+      "step": 6788
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.558546949082833e-05,
+      "loss": 1.1728,
+      "step": 6789
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5584005459337995e-05,
+      "loss": 1.1227,
+      "step": 6790
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.55825412086405e-05,
+      "loss": 1.0981,
+      "step": 6791
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5581076738751444e-05,
+      "loss": 1.0737,
+      "step": 6792
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.557961204968641e-05,
+      "loss": 1.0835,
+      "step": 6793
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5578147141461006e-05,
+      "loss": 1.083,
+      "step": 6794
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5576682014090824e-05,
+      "loss": 1.03,
+      "step": 6795
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.557521666759147e-05,
+      "loss": 1.0824,
+      "step": 6796
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5573751101978545e-05,
+      "loss": 1.1033,
+      "step": 6797
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.557228531726767e-05,
+      "loss": 1.023,
+      "step": 6798
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5570819313474445e-05,
+      "loss": 1.0732,
+      "step": 6799
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.556935309061448e-05,
+      "loss": 1.0929,
+      "step": 6800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.55678866487034e-05,
+      "loss": 1.1804,
+      "step": 6801
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.55664199877568e-05,
+      "loss": 1.0457,
+      "step": 6802
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.556495310779032e-05,
+      "loss": 1.1765,
+      "step": 6803
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.556348600881958e-05,
+      "loss": 1.023,
+      "step": 6804
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5562018690860195e-05,
+      "loss": 1.0896,
+      "step": 6805
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.55605511539278e-05,
+      "loss": 1.019,
+      "step": 6806
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5559083398038e-05,
+      "loss": 1.1094,
+      "step": 6807
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.555761542320646e-05,
+      "loss": 1.0963,
+      "step": 6808
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5556147229448795e-05,
+      "loss": 1.0458,
+      "step": 6809
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5554678816780636e-05,
+      "loss": 1.0125,
+      "step": 6810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.555321018521763e-05,
+      "loss": 1.093,
+      "step": 6811
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.555174133477542e-05,
+      "loss": 1.1909,
+      "step": 6812
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5550272265469626e-05,
+      "loss": 1.1089,
+      "step": 6813
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.554880297731593e-05,
+      "loss": 1.1151,
+      "step": 6814
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5547333470329945e-05,
+      "loss": 0.9906,
+      "step": 6815
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.554586374452733e-05,
+      "loss": 1.0876,
+      "step": 6816
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5544393799923744e-05,
+      "loss": 1.1986,
+      "step": 6817
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.554292363653484e-05,
+      "loss": 1.0299,
+      "step": 6818
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.554145325437627e-05,
+      "loss": 1.0369,
+      "step": 6819
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5539982653463684e-05,
+      "loss": 1.0861,
+      "step": 6820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553851183381277e-05,
+      "loss": 1.0413,
+      "step": 6821
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553704079543916e-05,
+      "loss": 1.0595,
+      "step": 6822
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5535569538358536e-05,
+      "loss": 1.1673,
+      "step": 6823
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553409806258657e-05,
+      "loss": 1.066,
+      "step": 6824
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553262636813892e-05,
+      "loss": 1.0655,
+      "step": 6825
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553115445503127e-05,
+      "loss": 1.0837,
+      "step": 6826
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.552968232327929e-05,
+      "loss": 1.0698,
+      "step": 6827
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.552820997289865e-05,
+      "loss": 1.0638,
+      "step": 6828
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.552673740390504e-05,
+      "loss": 1.0529,
+      "step": 6829
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5525264616314146e-05,
+      "loss": 1.1742,
+      "step": 6830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.552379161014164e-05,
+      "loss": 1.0702,
+      "step": 6831
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5522318385403215e-05,
+      "loss": 1.1351,
+      "step": 6832
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.552084494211455e-05,
+      "loss": 1.0236,
+      "step": 6833
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551937128029136e-05,
+      "loss": 1.1283,
+      "step": 6834
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551789739994931e-05,
+      "loss": 1.1707,
+      "step": 6835
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551642330110412e-05,
+      "loss": 1.0755,
+      "step": 6836
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5514948983771466e-05,
+      "loss": 1.1138,
+      "step": 6837
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551347444796706e-05,
+      "loss": 1.0718,
+      "step": 6838
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551199969370662e-05,
+      "loss": 1.0558,
+      "step": 6839
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551052472100582e-05,
+      "loss": 1.1325,
+      "step": 6840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.55090495298804e-05,
+      "loss": 1.1166,
+      "step": 6841
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550757412034604e-05,
+      "loss": 1.1174,
+      "step": 6842
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550609849241847e-05,
+      "loss": 1.1243,
+      "step": 6843
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.55046226461134e-05,
+      "loss": 1.1589,
+      "step": 6844
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550314658144655e-05,
+      "loss": 1.2494,
+      "step": 6845
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550167029843364e-05,
+      "loss": 1.0377,
+      "step": 6846
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550019379709038e-05,
+      "loss": 1.0556,
+      "step": 6847
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.549871707743251e-05,
+      "loss": 1.1125,
+      "step": 6848
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.549724013947575e-05,
+      "loss": 1.0394,
+      "step": 6849
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5495762983235827e-05,
+      "loss": 1.1534,
+      "step": 6850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.549428560872846e-05,
+      "loss": 1.0655,
+      "step": 6851
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5492808015969405e-05,
+      "loss": 1.0503,
+      "step": 6852
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.549133020497437e-05,
+      "loss": 1.0624,
+      "step": 6853
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.548985217575912e-05,
+      "loss": 1.1185,
+      "step": 6854
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.548837392833939e-05,
+      "loss": 1.075,
+      "step": 6855
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.548689546273091e-05,
+      "loss": 1.0985,
+      "step": 6856
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.548541677894943e-05,
+      "loss": 1.0642,
+      "step": 6857
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.54839378770107e-05,
+      "loss": 1.0713,
+      "step": 6858
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.548245875693047e-05,
+      "loss": 1.0819,
+      "step": 6859
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5480979418724486e-05,
+      "loss": 1.1446,
+      "step": 6860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.54794998624085e-05,
+      "loss": 1.0101,
+      "step": 6861
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.547802008799828e-05,
+      "loss": 1.0387,
+      "step": 6862
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5476540095509585e-05,
+      "loss": 1.1335,
+      "step": 6863
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.547505988495816e-05,
+      "loss": 0.9934,
+      "step": 6864
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5473579456359783e-05,
+      "loss": 1.0927,
+      "step": 6865
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5472098809730215e-05,
+      "loss": 0.9716,
+      "step": 6866
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.547061794508522e-05,
+      "loss": 1.0912,
+      "step": 6867
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.546913686244058e-05,
+      "loss": 1.1646,
+      "step": 6868
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5467655561812046e-05,
+      "loss": 1.0243,
+      "step": 6869
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5466174043215415e-05,
+      "loss": 1.1138,
+      "step": 6870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.546469230666645e-05,
+      "loss": 1.0313,
+      "step": 6871
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.546321035218094e-05,
+      "loss": 1.047,
+      "step": 6872
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.546172817977467e-05,
+      "loss": 1.0817,
+      "step": 6873
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.546024578946341e-05,
+      "loss": 1.077,
+      "step": 6874
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5458763181262955e-05,
+      "loss": 1.05,
+      "step": 6875
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.545728035518909e-05,
+      "loss": 1.0841,
+      "step": 6876
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5455797311257616e-05,
+      "loss": 1.0372,
+      "step": 6877
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5454314049484316e-05,
+      "loss": 1.1287,
+      "step": 6878
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.545283056988499e-05,
+      "loss": 1.0306,
+      "step": 6879
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5451346872475434e-05,
+      "loss": 1.1196,
+      "step": 6880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544986295727145e-05,
+      "loss": 1.0552,
+      "step": 6881
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544837882428884e-05,
+      "loss": 1.0813,
+      "step": 6882
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544689447354341e-05,
+      "loss": 1.1256,
+      "step": 6883
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544540990505097e-05,
+      "loss": 1.1689,
+      "step": 6884
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5443925118827334e-05,
+      "loss": 1.1214,
+      "step": 6885
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5442440114888296e-05,
+      "loss": 1.1331,
+      "step": 6886
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544095489324969e-05,
+      "loss": 1.0941,
+      "step": 6887
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.543946945392732e-05,
+      "loss": 1.0711,
+      "step": 6888
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5437983796937014e-05,
+      "loss": 1.1472,
+      "step": 6889
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.543649792229458e-05,
+      "loss": 1.0676,
+      "step": 6890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.543501183001585e-05,
+      "loss": 1.1444,
+      "step": 6891
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.543352552011666e-05,
+      "loss": 1.0724,
+      "step": 6892
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5432038992612824e-05,
+      "loss": 1.0718,
+      "step": 6893
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.543055224752018e-05,
+      "loss": 1.1056,
+      "step": 6894
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542906528485456e-05,
+      "loss": 0.992,
+      "step": 6895
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542757810463178e-05,
+      "loss": 1.0372,
+      "step": 6896
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542609070686771e-05,
+      "loss": 1.0767,
+      "step": 6897
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542460309157818e-05,
+      "loss": 1.0566,
+      "step": 6898
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542311525877902e-05,
+      "loss": 1.0111,
+      "step": 6899
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542162720848608e-05,
+      "loss": 1.0395,
+      "step": 6900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5420138940715204e-05,
+      "loss": 0.9849,
+      "step": 6901
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.541865045548226e-05,
+      "loss": 1.13,
+      "step": 6902
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.541716175280307e-05,
+      "loss": 1.0783,
+      "step": 6903
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.541567283269351e-05,
+      "loss": 1.1191,
+      "step": 6904
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.541418369516943e-05,
+      "loss": 1.1003,
+      "step": 6905
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5412694340246694e-05,
+      "loss": 1.087,
+      "step": 6906
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5411204767941145e-05,
+      "loss": 1.1143,
+      "step": 6907
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540971497826866e-05,
+      "loss": 1.0855,
+      "step": 6908
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540822497124511e-05,
+      "loss": 1.0882,
+      "step": 6909
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540673474688634e-05,
+      "loss": 1.0778,
+      "step": 6910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540524430520824e-05,
+      "loss": 1.0745,
+      "step": 6911
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5403753646226676e-05,
+      "loss": 1.1367,
+      "step": 6912
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540226276995753e-05,
+      "loss": 1.0374,
+      "step": 6913
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540077167641667e-05,
+      "loss": 1.1299,
+      "step": 6914
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5399280365619965e-05,
+      "loss": 1.0236,
+      "step": 6915
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.539778883758332e-05,
+      "loss": 1.1743,
+      "step": 6916
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.53962970923226e-05,
+      "loss": 1.0357,
+      "step": 6917
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.539480512985371e-05,
+      "loss": 1.1455,
+      "step": 6918
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.539331295019252e-05,
+      "loss": 1.0506,
+      "step": 6919
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5391820553354925e-05,
+      "loss": 1.0255,
+      "step": 6920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5390327939356827e-05,
+      "loss": 1.1348,
+      "step": 6921
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5388835108214114e-05,
+      "loss": 1.1243,
+      "step": 6922
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5387342059942685e-05,
+      "loss": 1.1603,
+      "step": 6923
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5385848794558436e-05,
+      "loss": 1.0441,
+      "step": 6924
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5384355312077284e-05,
+      "loss": 1.0506,
+      "step": 6925
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.538286161251513e-05,
+      "loss": 1.0341,
+      "step": 6926
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5381367695887855e-05,
+      "loss": 1.0243,
+      "step": 6927
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537987356221139e-05,
+      "loss": 1.0627,
+      "step": 6928
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5378379211501655e-05,
+      "loss": 1.0076,
+      "step": 6929
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5376884643774556e-05,
+      "loss": 1.0992,
+      "step": 6930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537538985904599e-05,
+      "loss": 1.0702,
+      "step": 6931
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537389485733191e-05,
+      "loss": 1.0209,
+      "step": 6932
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537239963864821e-05,
+      "loss": 1.0067,
+      "step": 6933
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537090420301082e-05,
+      "loss": 1.1221,
+      "step": 6934
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.536940855043567e-05,
+      "loss": 1.091,
+      "step": 6935
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.536791268093869e-05,
+      "loss": 1.0678,
+      "step": 6936
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.53664165945358e-05,
+      "loss": 1.1232,
+      "step": 6937
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.536492029124294e-05,
+      "loss": 0.9357,
+      "step": 6938
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.536342377107604e-05,
+      "loss": 1.0851,
+      "step": 6939
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.536192703405105e-05,
+      "loss": 1.0156,
+      "step": 6940
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.536043008018389e-05,
+      "loss": 1.0885,
+      "step": 6941
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.535893290949051e-05,
+      "loss": 1.1479,
+      "step": 6942
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5357435521986855e-05,
+      "loss": 1.1214,
+      "step": 6943
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5355937917688877e-05,
+      "loss": 1.0964,
+      "step": 6944
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5354440096612514e-05,
+      "loss": 1.1117,
+      "step": 6945
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5352942058773725e-05,
+      "loss": 1.106,
+      "step": 6946
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5351443804188454e-05,
+      "loss": 1.0152,
+      "step": 6947
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5349945332872667e-05,
+      "loss": 1.0835,
+      "step": 6948
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5348446644842314e-05,
+      "loss": 1.0158,
+      "step": 6949
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5346947740113354e-05,
+      "loss": 1.0216,
+      "step": 6950
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5345448618701765e-05,
+      "loss": 1.0586,
+      "step": 6951
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.534394928062349e-05,
+      "loss": 1.0878,
+      "step": 6952
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.534244972589451e-05,
+      "loss": 1.0829,
+      "step": 6953
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.534094995453079e-05,
+      "loss": 1.0911,
+      "step": 6954
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.533944996654831e-05,
+      "loss": 1.0908,
+      "step": 6955
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.533794976196303e-05,
+      "loss": 1.0293,
+      "step": 6956
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.533644934079093e-05,
+      "loss": 1.0187,
+      "step": 6957
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.533494870304799e-05,
+      "loss": 1.0246,
+      "step": 6958
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.533344784875019e-05,
+      "loss": 1.0117,
+      "step": 6959
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5331946777913525e-05,
+      "loss": 1.0479,
+      "step": 6960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5330445490553967e-05,
+      "loss": 0.9695,
+      "step": 6961
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.532894398668751e-05,
+      "loss": 1.012,
+      "step": 6962
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5327442266330134e-05,
+      "loss": 1.085,
+      "step": 6963
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.532594032949784e-05,
+      "loss": 1.1744,
+      "step": 6964
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5324438176206624e-05,
+      "loss": 1.1852,
+      "step": 6965
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.532293580647248e-05,
+      "loss": 1.1115,
+      "step": 6966
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.532143322031141e-05,
+      "loss": 1.0413,
+      "step": 6967
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.531993041773941e-05,
+      "loss": 1.0749,
+      "step": 6968
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.531842739877249e-05,
+      "loss": 1.1073,
+      "step": 6969
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.531692416342666e-05,
+      "loss": 1.1038,
+      "step": 6970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5315420711717914e-05,
+      "loss": 1.0914,
+      "step": 6971
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.531391704366227e-05,
+      "loss": 1.1176,
+      "step": 6972
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.531241315927575e-05,
+      "loss": 1.0513,
+      "step": 6973
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.531090905857436e-05,
+      "loss": 1.0538,
+      "step": 6974
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5309404741574114e-05,
+      "loss": 1.1516,
+      "step": 6975
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.530790020829104e-05,
+      "loss": 1.0935,
+      "step": 6976
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5306395458741155e-05,
+      "loss": 1.1694,
+      "step": 6977
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.530489049294049e-05,
+      "loss": 1.0793,
+      "step": 6978
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.530338531090507e-05,
+      "loss": 1.1183,
+      "step": 6979
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5301879912650926e-05,
+      "loss": 1.0695,
+      "step": 6980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.530037429819408e-05,
+      "loss": 1.0807,
+      "step": 6981
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.529886846755058e-05,
+      "loss": 1.0731,
+      "step": 6982
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5297362420736444e-05,
+      "loss": 1.1438,
+      "step": 6983
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.529585615776773e-05,
+      "loss": 1.0426,
+      "step": 6984
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.529434967866046e-05,
+      "loss": 1.0802,
+      "step": 6985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.52928429834307e-05,
+      "loss": 0.974,
+      "step": 6986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.529133607209447e-05,
+      "loss": 1.082,
+      "step": 6987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.528982894466784e-05,
+      "loss": 1.0518,
+      "step": 6988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5288321601166844e-05,
+      "loss": 1.103,
+      "step": 6989
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.528681404160754e-05,
+      "loss": 1.1676,
+      "step": 6990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.528530626600599e-05,
+      "loss": 1.0697,
+      "step": 6991
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5283798274378234e-05,
+      "loss": 1.1316,
+      "step": 6992
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.528229006674035e-05,
+      "loss": 1.0344,
+      "step": 6993
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5280781643108386e-05,
+      "loss": 1.1147,
+      "step": 6994
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5279273003498414e-05,
+      "loss": 1.0676,
+      "step": 6995
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.527776414792649e-05,
+      "loss": 1.0106,
+      "step": 6996
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.527625507640869e-05,
+      "loss": 1.1328,
+      "step": 6997
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.527474578896108e-05,
+      "loss": 1.1195,
+      "step": 6998
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.527323628559975e-05,
+      "loss": 1.1611,
+      "step": 6999
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.527172656634075e-05,
+      "loss": 1.0661,
+      "step": 7000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5270216631200166e-05,
+      "loss": 1.082,
+      "step": 7001
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.526870648019409e-05,
+      "loss": 1.1837,
+      "step": 7002
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5267196113338594e-05,
+      "loss": 1.1375,
+      "step": 7003
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5265685530649765e-05,
+      "loss": 1.0362,
+      "step": 7004
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5264174732143684e-05,
+      "loss": 1.1167,
+      "step": 7005
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.526266371783645e-05,
+      "loss": 1.075,
+      "step": 7006
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5261152487744144e-05,
+      "loss": 1.0437,
+      "step": 7007
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5259641041882874e-05,
+      "loss": 1.0361,
+      "step": 7008
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.525812938026872e-05,
+      "loss": 1.0169,
+      "step": 7009
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5256617502917785e-05,
+      "loss": 1.0894,
+      "step": 7010
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5255105409846175e-05,
+      "loss": 1.0183,
+      "step": 7011
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.525359310106999e-05,
+      "loss": 1.0864,
+      "step": 7012
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5252080576605324e-05,
+      "loss": 1.0687,
+      "step": 7013
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.525056783646831e-05,
+      "loss": 1.0287,
+      "step": 7014
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.524905488067504e-05,
+      "loss": 0.9806,
+      "step": 7015
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.524754170924163e-05,
+      "loss": 1.0658,
+      "step": 7016
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5246028322184184e-05,
+      "loss": 1.15,
+      "step": 7017
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.524451471951884e-05,
+      "loss": 0.9984,
+      "step": 7018
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.524300090126169e-05,
+      "loss": 1.0763,
+      "step": 7019
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.524148686742888e-05,
+      "loss": 0.9967,
+      "step": 7020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.523997261803652e-05,
+      "loss": 1.1485,
+      "step": 7021
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.523845815310075e-05,
+      "loss": 1.0257,
+      "step": 7022
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5236943472637674e-05,
+      "loss": 1.1313,
+      "step": 7023
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5235428576663446e-05,
+      "loss": 1.1577,
+      "step": 7024
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.523391346519419e-05,
+      "loss": 1.0921,
+      "step": 7025
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.523239813824603e-05,
+      "loss": 1.0438,
+      "step": 7026
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.523088259583512e-05,
+      "loss": 1.0264,
+      "step": 7027
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5229366837977594e-05,
+      "loss": 0.9759,
+      "step": 7028
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5227850864689584e-05,
+      "loss": 1.1213,
+      "step": 7029
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5226334675987245e-05,
+      "loss": 1.0901,
+      "step": 7030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.522481827188673e-05,
+      "loss": 1.0706,
+      "step": 7031
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.522330165240418e-05,
+      "loss": 0.956,
+      "step": 7032
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.522178481755574e-05,
+      "loss": 1.1418,
+      "step": 7033
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.522026776735756e-05,
+      "loss": 1.0996,
+      "step": 7034
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.521875050182581e-05,
+      "loss": 1.0413,
+      "step": 7035
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.521723302097665e-05,
+      "loss": 1.017,
+      "step": 7036
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.521571532482623e-05,
+      "loss": 1.0451,
+      "step": 7037
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.521419741339071e-05,
+      "loss": 1.1082,
+      "step": 7038
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.521267928668627e-05,
+      "loss": 1.0588,
+      "step": 7039
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.521116094472905e-05,
+      "loss": 1.1209,
+      "step": 7040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.520964238753524e-05,
+      "loss": 1.1055,
+      "step": 7041
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.520812361512102e-05,
+      "loss": 1.0715,
+      "step": 7042
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5206604627502546e-05,
+      "loss": 1.1222,
+      "step": 7043
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5205085424696e-05,
+      "loss": 1.022,
+      "step": 7044
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.520356600671757e-05,
+      "loss": 0.9652,
+      "step": 7045
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.520204637358342e-05,
+      "loss": 1.0944,
+      "step": 7046
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.520052652530974e-05,
+      "loss": 1.0959,
+      "step": 7047
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5199006461912716e-05,
+      "loss": 1.062,
+      "step": 7048
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5197486183408536e-05,
+      "loss": 1.1106,
+      "step": 7049
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.519596568981339e-05,
+      "loss": 1.1357,
+      "step": 7050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5194444981143466e-05,
+      "loss": 1.0495,
+      "step": 7051
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5192924057414974e-05,
+      "loss": 1.1122,
+      "step": 7052
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.51914029186441e-05,
+      "loss": 1.1532,
+      "step": 7053
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518988156484704e-05,
+      "loss": 1.0698,
+      "step": 7054
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518835999604e-05,
+      "loss": 1.0863,
+      "step": 7055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518683821223918e-05,
+      "loss": 1.0819,
+      "step": 7056
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5185316213460786e-05,
+      "loss": 1.069,
+      "step": 7057
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518379399972104e-05,
+      "loss": 1.1063,
+      "step": 7058
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518227157103613e-05,
+      "loss": 1.1339,
+      "step": 7059
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518074892742229e-05,
+      "loss": 1.1009,
+      "step": 7060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5179226068895716e-05,
+      "loss": 1.1071,
+      "step": 7061
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.517770299547264e-05,
+      "loss": 1.1063,
+      "step": 7062
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.517617970716927e-05,
+      "loss": 1.1071,
+      "step": 7063
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5174656204001854e-05,
+      "loss": 1.1057,
+      "step": 7064
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.517313248598658e-05,
+      "loss": 1.0553,
+      "step": 7065
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.51716085531397e-05,
+      "loss": 1.1047,
+      "step": 7066
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.517008440547744e-05,
+      "loss": 1.116,
+      "step": 7067
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5168560043016016e-05,
+      "loss": 1.0374,
+      "step": 7068
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.516703546577168e-05,
+      "loss": 1.0827,
+      "step": 7069
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5165510673760656e-05,
+      "loss": 1.0773,
+      "step": 7070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.516398566699919e-05,
+      "loss": 1.0778,
+      "step": 7071
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5162460445503515e-05,
+      "loss": 1.0526,
+      "step": 7072
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5160935009289884e-05,
+      "loss": 1.0418,
+      "step": 7073
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.515940935837453e-05,
+      "loss": 0.9492,
+      "step": 7074
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.51578834927737e-05,
+      "loss": 0.9565,
+      "step": 7075
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.515635741250366e-05,
+      "loss": 1.0781,
+      "step": 7076
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.515483111758064e-05,
+      "loss": 1.0225,
+      "step": 7077
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.515330460802091e-05,
+      "loss": 1.0681,
+      "step": 7078
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5151777883840724e-05,
+      "loss": 1.185,
+      "step": 7079
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.515025094505633e-05,
+      "loss": 1.0736,
+      "step": 7080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5148723791684006e-05,
+      "loss": 1.0531,
+      "step": 7081
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.514719642374e-05,
+      "loss": 1.0719,
+      "step": 7082
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5145668841240587e-05,
+      "loss": 1.0762,
+      "step": 7083
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5144141044202034e-05,
+      "loss": 1.0378,
+      "step": 7084
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5142613032640604e-05,
+      "loss": 1.1046,
+      "step": 7085
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.514108480657258e-05,
+      "loss": 1.0302,
+      "step": 7086
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.513955636601422e-05,
+      "loss": 1.0044,
+      "step": 7087
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.513802771098182e-05,
+      "loss": 1.044,
+      "step": 7088
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.513649884149165e-05,
+      "loss": 0.996,
+      "step": 7089
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5134969757559985e-05,
+      "loss": 1.1592,
+      "step": 7090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5133440459203124e-05,
+      "loss": 1.0582,
+      "step": 7091
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.513191094643735e-05,
+      "loss": 1.04,
+      "step": 7092
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.513038121927894e-05,
+      "loss": 1.0596,
+      "step": 7093
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.512885127774419e-05,
+      "loss": 0.9698,
+      "step": 7094
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.51273211218494e-05,
+      "loss": 1.0324,
+      "step": 7095
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.512579075161085e-05,
+      "loss": 1.0606,
+      "step": 7096
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.512426016704485e-05,
+      "loss": 1.085,
+      "step": 7097
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.51227293681677e-05,
+      "loss": 1.1076,
+      "step": 7098
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.512119835499571e-05,
+      "loss": 1.0184,
+      "step": 7099
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5119667127545155e-05,
+      "loss": 0.9925,
+      "step": 7100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.511813568583237e-05,
+      "loss": 1.0441,
+      "step": 7101
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.511660402987366e-05,
+      "loss": 1.1041,
+      "step": 7102
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5115072159685315e-05,
+      "loss": 1.0713,
+      "step": 7103
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.511354007528367e-05,
+      "loss": 1.0533,
+      "step": 7104
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.511200777668504e-05,
+      "loss": 1.129,
+      "step": 7105
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5110475263905725e-05,
+      "loss": 1.0061,
+      "step": 7106
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.510894253696207e-05,
+      "loss": 1.0759,
+      "step": 7107
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.510740959587038e-05,
+      "loss": 1.0487,
+      "step": 7108
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.510587644064699e-05,
+      "loss": 0.9531,
+      "step": 7109
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.510434307130821e-05,
+      "loss": 1.0601,
+      "step": 7110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.510280948787039e-05,
+      "loss": 0.9943,
+      "step": 7111
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5101275690349856e-05,
+      "loss": 1.0951,
+      "step": 7112
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5099741678762944e-05,
+      "loss": 1.047,
+      "step": 7113
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5098207453125974e-05,
+      "loss": 0.9847,
+      "step": 7114
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5096673013455306e-05,
+      "loss": 1.0214,
+      "step": 7115
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.509513835976726e-05,
+      "loss": 1.0361,
+      "step": 7116
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.509360349207821e-05,
+      "loss": 1.0531,
+      "step": 7117
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5092068410404465e-05,
+      "loss": 1.0347,
+      "step": 7118
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5090533114762395e-05,
+      "loss": 1.0444,
+      "step": 7119
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.508899760516834e-05,
+      "loss": 1.1092,
+      "step": 7120
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5087461881638666e-05,
+      "loss": 1.0259,
+      "step": 7121
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.508592594418971e-05,
+      "loss": 1.0761,
+      "step": 7122
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.508438979283784e-05,
+      "loss": 1.0945,
+      "step": 7123
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5082853427599416e-05,
+      "loss": 1.1079,
+      "step": 7124
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5081316848490795e-05,
+      "loss": 1.0619,
+      "step": 7125
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5079780055528336e-05,
+      "loss": 1.092,
+      "step": 7126
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.507824304872841e-05,
+      "loss": 1.044,
+      "step": 7127
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5076705828107394e-05,
+      "loss": 1.1378,
+      "step": 7128
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5075168393681646e-05,
+      "loss": 1.0615,
+      "step": 7129
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.507363074546754e-05,
+      "loss": 1.0178,
+      "step": 7130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5072092883481456e-05,
+      "loss": 0.983,
+      "step": 7131
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.507055480773976e-05,
+      "loss": 1.0718,
+      "step": 7132
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.506901651825885e-05,
+      "loss": 1.0134,
+      "step": 7133
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.50674780150551e-05,
+      "loss": 1.1199,
+      "step": 7134
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.506593929814489e-05,
+      "loss": 1.1637,
+      "step": 7135
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.50644003675446e-05,
+      "loss": 1.1012,
+      "step": 7136
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.506286122327063e-05,
+      "loss": 1.1021,
+      "step": 7137
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.506132186533938e-05,
+      "loss": 1.1624,
+      "step": 7138
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5059782293767224e-05,
+      "loss": 1.0983,
+      "step": 7139
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5058242508570565e-05,
+      "loss": 1.0885,
+      "step": 7140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5056702509765804e-05,
+      "loss": 0.9801,
+      "step": 7141
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.505516229736934e-05,
+      "loss": 1.0796,
+      "step": 7142
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.505362187139756e-05,
+      "loss": 0.9667,
+      "step": 7143
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.50520812318669e-05,
+      "loss": 0.9674,
+      "step": 7144
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.505054037879373e-05,
+      "loss": 1.0671,
+      "step": 7145
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.504899931219449e-05,
+      "loss": 1.1996,
+      "step": 7146
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.504745803208558e-05,
+      "loss": 1.0718,
+      "step": 7147
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.504591653848341e-05,
+      "loss": 1.0308,
+      "step": 7148
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.50443748314044e-05,
+      "loss": 0.9739,
+      "step": 7149
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.504283291086496e-05,
+      "loss": 1.0153,
+      "step": 7150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.504129077688153e-05,
+      "loss": 1.1005,
+      "step": 7151
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5039748429470505e-05,
+      "loss": 1.1258,
+      "step": 7152
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.503820586864834e-05,
+      "loss": 1.0374,
+      "step": 7153
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.503666309443145e-05,
+      "loss": 1.0181,
+      "step": 7154
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5035120106836245e-05,
+      "loss": 0.9963,
+      "step": 7155
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5033576905879195e-05,
+      "loss": 1.0654,
+      "step": 7156
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.50320334915767e-05,
+      "loss": 1.0604,
+      "step": 7157
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.503048986394521e-05,
+      "loss": 0.9733,
+      "step": 7158
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.502894602300117e-05,
+      "loss": 1.0494,
+      "step": 7159
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.502740196876102e-05,
+      "loss": 1.0181,
+      "step": 7160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5025857701241183e-05,
+      "loss": 1.0426,
+      "step": 7161
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.502431322045812e-05,
+      "loss": 1.0488,
+      "step": 7162
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.502276852642829e-05,
+      "loss": 1.0248,
+      "step": 7163
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5021223619168126e-05,
+      "loss": 1.0558,
+      "step": 7164
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5019678498694095e-05,
+      "loss": 0.9361,
+      "step": 7165
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.501813316502263e-05,
+      "loss": 1.0089,
+      "step": 7166
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5016587618170205e-05,
+      "loss": 1.054,
+      "step": 7167
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.501504185815327e-05,
+      "loss": 1.0184,
+      "step": 7168
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.50134958849883e-05,
+      "loss": 1.019,
+      "step": 7169
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.501194969869174e-05,
+      "loss": 1.113,
+      "step": 7170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.501040329928007e-05,
+      "loss": 1.014,
+      "step": 7171
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.500885668676975e-05,
+      "loss": 1.0155,
+      "step": 7172
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.500730986117725e-05,
+      "loss": 1.0733,
+      "step": 7173
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.500576282251906e-05,
+      "loss": 0.9713,
+      "step": 7174
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.500421557081163e-05,
+      "loss": 1.0129,
+      "step": 7175
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.500266810607146e-05,
+      "loss": 1.1197,
+      "step": 7176
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5001120428315e-05,
+      "loss": 1.0252,
+      "step": 7177
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4999572537558766e-05,
+      "loss": 1.0766,
+      "step": 7178
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.499802443381922e-05,
+      "loss": 1.1893,
+      "step": 7179
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4996476117112865e-05,
+      "loss": 1.068,
+      "step": 7180
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.499492758745616e-05,
+      "loss": 1.0619,
+      "step": 7181
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.499337884486563e-05,
+      "loss": 1.1108,
+      "step": 7182
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.499182988935775e-05,
+      "loss": 0.9903,
+      "step": 7183
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4990280720949024e-05,
+      "loss": 1.2133,
+      "step": 7184
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4988731339655946e-05,
+      "loss": 1.0964,
+      "step": 7185
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4987181745495e-05,
+      "loss": 1.1293,
+      "step": 7186
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.498563193848272e-05,
+      "loss": 1.0725,
+      "step": 7187
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.498408191863558e-05,
+      "loss": 1.1119,
+      "step": 7188
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.498253168597011e-05,
+      "loss": 0.9985,
+      "step": 7189
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4980981240502805e-05,
+      "loss": 1.0389,
+      "step": 7190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4979430582250184e-05,
+      "loss": 1.1086,
+      "step": 7191
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.497787971122874e-05,
+      "loss": 1.0432,
+      "step": 7192
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4976328627455026e-05,
+      "loss": 1.0535,
+      "step": 7193
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4974777330945536e-05,
+      "loss": 1.0583,
+      "step": 7194
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.497322582171679e-05,
+      "loss": 1.0152,
+      "step": 7195
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.497167409978532e-05,
+      "loss": 1.0426,
+      "step": 7196
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4970122165167635e-05,
+      "loss": 1.0736,
+      "step": 7197
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.496857001788028e-05,
+      "loss": 1.0424,
+      "step": 7198
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4967017657939786e-05,
+      "loss": 1.0805,
+      "step": 7199
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4965465085362665e-05,
+      "loss": 1.0342,
+      "step": 7200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.496391230016547e-05,
+      "loss": 0.9769,
+      "step": 7201
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.496235930236472e-05,
+      "loss": 1.1004,
+      "step": 7202
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.496080609197697e-05,
+      "loss": 1.0806,
+      "step": 7203
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.495925266901876e-05,
+      "loss": 0.9984,
+      "step": 7204
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.495769903350662e-05,
+      "loss": 1.0,
+      "step": 7205
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4956145185457096e-05,
+      "loss": 1.0858,
+      "step": 7206
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.495459112488675e-05,
+      "loss": 1.0375,
+      "step": 7207
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.495303685181211e-05,
+      "loss": 1.0366,
+      "step": 7208
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.495148236624975e-05,
+      "loss": 1.1218,
+      "step": 7209
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.494992766821622e-05,
+      "loss": 1.0179,
+      "step": 7210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.494837275772806e-05,
+      "loss": 1.0685,
+      "step": 7211
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.494681763480185e-05,
+      "loss": 1.2041,
+      "step": 7212
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.494526229945415e-05,
+      "loss": 1.0278,
+      "step": 7213
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.494370675170151e-05,
+      "loss": 1.0632,
+      "step": 7214
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.494215099156049e-05,
+      "loss": 0.9911,
+      "step": 7215
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4940595019047685e-05,
+      "loss": 0.9845,
+      "step": 7216
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.493903883417964e-05,
+      "loss": 0.988,
+      "step": 7217
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.493748243697294e-05,
+      "loss": 1.0578,
+      "step": 7218
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4935925827444144e-05,
+      "loss": 0.9752,
+      "step": 7219
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.493436900560986e-05,
+      "loss": 1.0833,
+      "step": 7220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4932811971486646e-05,
+      "loss": 1.1007,
+      "step": 7221
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.493125472509108e-05,
+      "loss": 1.0649,
+      "step": 7222
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.492969726643975e-05,
+      "loss": 1.0198,
+      "step": 7223
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4928139595549246e-05,
+      "loss": 0.9957,
+      "step": 7224
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.492658171243616e-05,
+      "loss": 1.1302,
+      "step": 7225
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4925023617117066e-05,
+      "loss": 1.0334,
+      "step": 7226
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.492346530960857e-05,
+      "loss": 1.0622,
+      "step": 7227
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4921906789927264e-05,
+      "loss": 0.9725,
+      "step": 7228
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4920348058089745e-05,
+      "loss": 0.9958,
+      "step": 7229
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.491878911411262e-05,
+      "loss": 1.0267,
+      "step": 7230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.491722995801248e-05,
+      "loss": 1.0387,
+      "step": 7231
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.491567058980593e-05,
+      "loss": 1.1043,
+      "step": 7232
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4914111009509586e-05,
+      "loss": 1.0688,
+      "step": 7233
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.491255121714004e-05,
+      "loss": 1.0446,
+      "step": 7234
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.491099121271393e-05,
+      "loss": 0.9841,
+      "step": 7235
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.490943099624784e-05,
+      "loss": 1.029,
+      "step": 7236
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4907870567758394e-05,
+      "loss": 1.072,
+      "step": 7237
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.490630992726221e-05,
+      "loss": 1.0263,
+      "step": 7238
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.490474907477592e-05,
+      "loss": 1.0853,
+      "step": 7239
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.490318801031613e-05,
+      "loss": 1.0079,
+      "step": 7240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.490162673389948e-05,
+      "loss": 1.035,
+      "step": 7241
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4900065245542585e-05,
+      "loss": 1.054,
+      "step": 7242
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.489850354526207e-05,
+      "loss": 1.0203,
+      "step": 7243
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4896941633074586e-05,
+      "loss": 0.9881,
+      "step": 7244
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4895379508996734e-05,
+      "loss": 1.0724,
+      "step": 7245
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4893817173045186e-05,
+      "loss": 0.9758,
+      "step": 7246
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4892254625236555e-05,
+      "loss": 1.077,
+      "step": 7247
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.489069186558749e-05,
+      "loss": 0.9466,
+      "step": 7248
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4889128894114644e-05,
+      "loss": 1.1043,
+      "step": 7249
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4887565710834635e-05,
+      "loss": 1.0918,
+      "step": 7250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4886002315764134e-05,
+      "loss": 0.9827,
+      "step": 7251
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.488443870891977e-05,
+      "loss": 1.0845,
+      "step": 7252
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4882874890318214e-05,
+      "loss": 1.1258,
+      "step": 7253
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.488131085997611e-05,
+      "loss": 1.0589,
+      "step": 7254
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4879746617910124e-05,
+      "loss": 1.0194,
+      "step": 7255
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.48781821641369e-05,
+      "loss": 0.9857,
+      "step": 7256
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.48766174986731e-05,
+      "loss": 1.0758,
+      "step": 7257
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.48750526215354e-05,
+      "loss": 1.0897,
+      "step": 7258
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4873487532740445e-05,
+      "loss": 1.0303,
+      "step": 7259
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.487192223230491e-05,
+      "loss": 1.0133,
+      "step": 7260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4870356720245474e-05,
+      "loss": 1.0391,
+      "step": 7261
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4868790996578815e-05,
+      "loss": 1.0293,
+      "step": 7262
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4867225061321585e-05,
+      "loss": 1.1346,
+      "step": 7263
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.486565891449046e-05,
+      "loss": 0.9919,
+      "step": 7264
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.486409255610214e-05,
+      "loss": 1.1473,
+      "step": 7265
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4862525986173296e-05,
+      "loss": 1.0286,
+      "step": 7266
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4860959204720597e-05,
+      "loss": 1.0638,
+      "step": 7267
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4859392211760755e-05,
+      "loss": 1.0336,
+      "step": 7268
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4857825007310434e-05,
+      "loss": 1.0777,
+      "step": 7269
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.485625759138633e-05,
+      "loss": 0.9272,
+      "step": 7270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.485468996400514e-05,
+      "loss": 1.1052,
+      "step": 7271
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.485312212518356e-05,
+      "loss": 1.1121,
+      "step": 7272
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.485155407493828e-05,
+      "loss": 0.9402,
+      "step": 7273
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4849985813286e-05,
+      "loss": 0.9552,
+      "step": 7274
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.484841734024343e-05,
+      "loss": 1.0195,
+      "step": 7275
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.484684865582726e-05,
+      "loss": 1.0508,
+      "step": 7276
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4845279760054195e-05,
+      "loss": 1.035,
+      "step": 7277
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.484371065294096e-05,
+      "loss": 1.0777,
+      "step": 7278
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.484214133450424e-05,
+      "loss": 1.0413,
+      "step": 7279
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.484057180476077e-05,
+      "loss": 1.0505,
+      "step": 7280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4839002063727257e-05,
+      "loss": 1.0158,
+      "step": 7281
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.483743211142042e-05,
+      "loss": 1.1022,
+      "step": 7282
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.483586194785697e-05,
+      "loss": 1.0515,
+      "step": 7283
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4834291573053636e-05,
+      "loss": 1.0584,
+      "step": 7284
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4832720987027134e-05,
+      "loss": 1.0892,
+      "step": 7285
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.483115018979419e-05,
+      "loss": 1.0827,
+      "step": 7286
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.482957918137154e-05,
+      "loss": 1.0518,
+      "step": 7287
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.482800796177591e-05,
+      "loss": 1.0518,
+      "step": 7288
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4826436531024043e-05,
+      "loss": 1.0961,
+      "step": 7289
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.482486488913265e-05,
+      "loss": 1.0595,
+      "step": 7290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4823293036118483e-05,
+      "loss": 0.9928,
+      "step": 7291
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.482172097199829e-05,
+      "loss": 1.1017,
+      "step": 7292
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.48201486967888e-05,
+      "loss": 1.027,
+      "step": 7293
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4818576210506755e-05,
+      "loss": 1.054,
+      "step": 7294
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.481700351316891e-05,
+      "loss": 0.9947,
+      "step": 7295
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.481543060479201e-05,
+      "loss": 1.0364,
+      "step": 7296
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.48138574853928e-05,
+      "loss": 1.067,
+      "step": 7297
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.481228415498803e-05,
+      "loss": 0.9529,
+      "step": 7298
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.481071061359448e-05,
+      "loss": 1.0475,
+      "step": 7299
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480913686122887e-05,
+      "loss": 1.0492,
+      "step": 7300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480756289790799e-05,
+      "loss": 1.0389,
+      "step": 7301
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480598872364858e-05,
+      "loss": 1.0241,
+      "step": 7302
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480441433846743e-05,
+      "loss": 1.0264,
+      "step": 7303
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4802839742381285e-05,
+      "loss": 1.0772,
+      "step": 7304
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480126493540692e-05,
+      "loss": 1.0006,
+      "step": 7305
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4799689917561106e-05,
+      "loss": 1.0201,
+      "step": 7306
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.479811468886061e-05,
+      "loss": 1.0656,
+      "step": 7307
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4796539249322214e-05,
+      "loss": 1.1144,
+      "step": 7308
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.47949635989627e-05,
+      "loss": 1.0649,
+      "step": 7309
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.479338773779883e-05,
+      "loss": 1.0083,
+      "step": 7310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4791811665847406e-05,
+      "loss": 1.0187,
+      "step": 7311
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.47902353831252e-05,
+      "loss": 1.0741,
+      "step": 7312
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4788658889649e-05,
+      "loss": 1.1071,
+      "step": 7313
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4787082185435605e-05,
+      "loss": 0.9842,
+      "step": 7314
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4785505270501795e-05,
+      "loss": 1.0084,
+      "step": 7315
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4783928144864365e-05,
+      "loss": 1.0416,
+      "step": 7316
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.478235080854011e-05,
+      "loss": 1.0835,
+      "step": 7317
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4780773261545827e-05,
+      "loss": 0.9828,
+      "step": 7318
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.477919550389832e-05,
+      "loss": 0.9272,
+      "step": 7319
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.477761753561439e-05,
+      "loss": 0.9448,
+      "step": 7320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.477603935671084e-05,
+      "loss": 1.05,
+      "step": 7321
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.477446096720447e-05,
+      "loss": 1.007,
+      "step": 7322
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4772882367112106e-05,
+      "loss": 1.0614,
+      "step": 7323
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.477130355645055e-05,
+      "loss": 1.0859,
+      "step": 7324
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4769724535236614e-05,
+      "loss": 1.0389,
+      "step": 7325
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4768145303487106e-05,
+      "loss": 1.0245,
+      "step": 7326
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4766565861218855e-05,
+      "loss": 0.9943,
+      "step": 7327
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.476498620844869e-05,
+      "loss": 1.026,
+      "step": 7328
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4763406345193406e-05,
+      "loss": 0.9595,
+      "step": 7329
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4761826271469845e-05,
+      "loss": 1.1523,
+      "step": 7330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4760245987294836e-05,
+      "loss": 0.961,
+      "step": 7331
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4758665492685204e-05,
+      "loss": 1.0767,
+      "step": 7332
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4757084787657776e-05,
+      "loss": 1.0698,
+      "step": 7333
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4755503872229396e-05,
+      "loss": 1.1033,
+      "step": 7334
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.47539227464169e-05,
+      "loss": 1.0107,
+      "step": 7335
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.475234141023711e-05,
+      "loss": 1.0657,
+      "step": 7336
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.475075986370687e-05,
+      "loss": 1.0687,
+      "step": 7337
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.474917810684304e-05,
+      "loss": 0.9849,
+      "step": 7338
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.474759613966245e-05,
+      "loss": 1.0485,
+      "step": 7339
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4746013962181944e-05,
+      "loss": 1.0563,
+      "step": 7340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.474443157441839e-05,
+      "loss": 1.0441,
+      "step": 7341
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.474284897638861e-05,
+      "loss": 1.0182,
+      "step": 7342
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.474126616810949e-05,
+      "loss": 1.0197,
+      "step": 7343
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.473968314959786e-05,
+      "loss": 0.9626,
+      "step": 7344
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.473809992087059e-05,
+      "loss": 0.9985,
+      "step": 7345
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.473651648194455e-05,
+      "loss": 1.001,
+      "step": 7346
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4734932832836585e-05,
+      "loss": 1.0544,
+      "step": 7347
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.473334897356356e-05,
+      "loss": 0.975,
+      "step": 7348
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4731764904142354e-05,
+      "loss": 1.0083,
+      "step": 7349
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4730180624589836e-05,
+      "loss": 1.0465,
+      "step": 7350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.472859613492286e-05,
+      "loss": 0.9972,
+      "step": 7351
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.472701143515833e-05,
+      "loss": 1.0873,
+      "step": 7352
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.47254265253131e-05,
+      "loss": 1.0175,
+      "step": 7353
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.472384140540404e-05,
+      "loss": 0.976,
+      "step": 7354
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4722256075448054e-05,
+      "loss": 0.9876,
+      "step": 7355
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.472067053546202e-05,
+      "loss": 1.0006,
+      "step": 7356
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4719084785462806e-05,
+      "loss": 1.0613,
+      "step": 7357
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4717498825467313e-05,
+      "loss": 1.0392,
+      "step": 7358
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4715912655492434e-05,
+      "loss": 1.062,
+      "step": 7359
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4714326275555055e-05,
+      "loss": 0.9984,
+      "step": 7360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.471273968567206e-05,
+      "loss": 0.9345,
+      "step": 7361
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4711152885860363e-05,
+      "loss": 0.9727,
+      "step": 7362
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.470956587613686e-05,
+      "loss": 1.0213,
+      "step": 7363
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.470797865651843e-05,
+      "loss": 1.0684,
+      "step": 7364
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4706391227022006e-05,
+      "loss": 1.022,
+      "step": 7365
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.470480358766448e-05,
+      "loss": 0.9303,
+      "step": 7366
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.470321573846275e-05,
+      "loss": 1.0122,
+      "step": 7367
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4701627679433736e-05,
+      "loss": 1.0139,
+      "step": 7368
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.470003941059435e-05,
+      "loss": 1.0042,
+      "step": 7369
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.469845093196151e-05,
+      "loss": 1.0615,
+      "step": 7370
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.469686224355213e-05,
+      "loss": 1.0014,
+      "step": 7371
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.469527334538312e-05,
+      "loss": 1.0597,
+      "step": 7372
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.46936842374714e-05,
+      "loss": 0.9936,
+      "step": 7373
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.46920949198339e-05,
+      "loss": 1.0137,
+      "step": 7374
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4690505392487545e-05,
+      "loss": 0.9955,
+      "step": 7375
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.468891565544927e-05,
+      "loss": 1.1194,
+      "step": 7376
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4687325708736e-05,
+      "loss": 1.0754,
+      "step": 7377
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4685735552364646e-05,
+      "loss": 1.1458,
+      "step": 7378
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4684145186352166e-05,
+      "loss": 0.9954,
+      "step": 7379
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46825546107155e-05,
+      "loss": 0.9914,
+      "step": 7380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.468096382547157e-05,
+      "loss": 0.8726,
+      "step": 7381
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4679372830637325e-05,
+      "loss": 1.0894,
+      "step": 7382
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46777816262297e-05,
+      "loss": 1.018,
+      "step": 7383
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4676190212265654e-05,
+      "loss": 0.9699,
+      "step": 7384
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.467459858876213e-05,
+      "loss": 1.0263,
+      "step": 7385
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.467300675573608e-05,
+      "loss": 0.9359,
+      "step": 7386
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.467141471320444e-05,
+      "loss": 1.1257,
+      "step": 7387
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.466982246118418e-05,
+      "loss": 0.994,
+      "step": 7388
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.466822999969226e-05,
+      "loss": 1.0124,
+      "step": 7389
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.466663732874562e-05,
+      "loss": 1.1152,
+      "step": 7390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.466504444836124e-05,
+      "loss": 0.9858,
+      "step": 7391
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4663451358556076e-05,
+      "loss": 0.9739,
+      "step": 7392
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4661858059347084e-05,
+      "loss": 0.9367,
+      "step": 7393
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.466026455075125e-05,
+      "loss": 1.0029,
+      "step": 7394
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.465867083278553e-05,
+      "loss": 0.963,
+      "step": 7395
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46570769054669e-05,
+      "loss": 0.973,
+      "step": 7396
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.465548276881234e-05,
+      "loss": 1.0395,
+      "step": 7397
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.465388842283882e-05,
+      "loss": 0.9176,
+      "step": 7398
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.465229386756332e-05,
+      "loss": 1.0268,
+      "step": 7399
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.465069910300282e-05,
+      "loss": 0.9921,
+      "step": 7400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46491041291743e-05,
+      "loss": 0.972,
+      "step": 7401
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.464750894609476e-05,
+      "loss": 1.0651,
+      "step": 7402
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.464591355378117e-05,
+      "loss": 1.0331,
+      "step": 7403
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4644317952250535e-05,
+      "loss": 0.9979,
+      "step": 7404
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.464272214151984e-05,
+      "loss": 1.0468,
+      "step": 7405
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4641126121606074e-05,
+      "loss": 0.9727,
+      "step": 7406
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4639529892526244e-05,
+      "loss": 1.0122,
+      "step": 7407
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4637933454297345e-05,
+      "loss": 1.0921,
+      "step": 7408
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4636336806936376e-05,
+      "loss": 1.0008,
+      "step": 7409
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4634739950460334e-05,
+      "loss": 1.0552,
+      "step": 7410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.463314288488624e-05,
+      "loss": 1.0584,
+      "step": 7411
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.463154561023109e-05,
+      "loss": 1.0379,
+      "step": 7412
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46299481265119e-05,
+      "loss": 0.9142,
+      "step": 7413
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.462835043374568e-05,
+      "loss": 0.9477,
+      "step": 7414
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4626752531949444e-05,
+      "loss": 0.9914,
+      "step": 7415
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.462515442114021e-05,
+      "loss": 0.9967,
+      "step": 7416
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4623556101334994e-05,
+      "loss": 1.0472,
+      "step": 7417
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.462195757255082e-05,
+      "loss": 0.9579,
+      "step": 7418
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.462035883480471e-05,
+      "loss": 1.0051,
+      "step": 7419
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4618759888113694e-05,
+      "loss": 1.039,
+      "step": 7420
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.461716073249479e-05,
+      "loss": 1.0127,
+      "step": 7421
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4615561367965044e-05,
+      "loss": 1.0986,
+      "step": 7422
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.461396179454147e-05,
+      "loss": 0.9917,
+      "step": 7423
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.461236201224111e-05,
+      "loss": 0.9518,
+      "step": 7424
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.461076202108101e-05,
+      "loss": 1.0686,
+      "step": 7425
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.460916182107819e-05,
+      "loss": 1.0758,
+      "step": 7426
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.460756141224971e-05,
+      "loss": 0.9758,
+      "step": 7427
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46059607946126e-05,
+      "loss": 1.0224,
+      "step": 7428
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.460435996818391e-05,
+      "loss": 1.0327,
+      "step": 7429
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4602758932980694e-05,
+      "loss": 1.0079,
+      "step": 7430
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.460115768901999e-05,
+      "loss": 0.9867,
+      "step": 7431
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4599556236318864e-05,
+      "loss": 1.0792,
+      "step": 7432
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.459795457489435e-05,
+      "loss": 0.9876,
+      "step": 7433
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.459635270476353e-05,
+      "loss": 0.9883,
+      "step": 7434
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.459475062594345e-05,
+      "loss": 1.0682,
+      "step": 7435
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4593148338451174e-05,
+      "loss": 0.979,
+      "step": 7436
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.459154584230376e-05,
+      "loss": 1.0295,
+      "step": 7437
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4589943137518275e-05,
+      "loss": 1.0249,
+      "step": 7438
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.458834022411179e-05,
+      "loss": 1.0081,
+      "step": 7439
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.458673710210137e-05,
+      "loss": 1.0511,
+      "step": 7440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4585133771504095e-05,
+      "loss": 1.0638,
+      "step": 7441
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.458353023233704e-05,
+      "loss": 0.9528,
+      "step": 7442
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4581926484617275e-05,
+      "loss": 1.045,
+      "step": 7443
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.458032252836188e-05,
+      "loss": 1.0709,
+      "step": 7444
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4578718363587934e-05,
+      "loss": 0.9736,
+      "step": 7445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.457711399031253e-05,
+      "loss": 0.9794,
+      "step": 7446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.457550940855275e-05,
+      "loss": 0.9851,
+      "step": 7447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.457390461832567e-05,
+      "loss": 1.0104,
+      "step": 7448
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.457229961964839e-05,
+      "loss": 0.9788,
+      "step": 7449
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.457069441253801e-05,
+      "loss": 0.9382,
+      "step": 7450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.456908899701161e-05,
+      "loss": 1.1463,
+      "step": 7451
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.45674833730863e-05,
+      "loss": 0.9967,
+      "step": 7452
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.456587754077917e-05,
+      "loss": 0.9736,
+      "step": 7453
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.456427150010731e-05,
+      "loss": 1.0566,
+      "step": 7454
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.456266525108786e-05,
+      "loss": 0.9757,
+      "step": 7455
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.456105879373787e-05,
+      "loss": 1.0107,
+      "step": 7456
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.455945212807451e-05,
+      "loss": 0.9768,
+      "step": 7457
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.455784525411484e-05,
+      "loss": 1.0127,
+      "step": 7458
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4556238171876005e-05,
+      "loss": 0.9591,
+      "step": 7459
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4554630881375096e-05,
+      "loss": 1.0085,
+      "step": 7460
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4553023382629246e-05,
+      "loss": 0.9989,
+      "step": 7461
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4551415675655566e-05,
+      "loss": 1.0501,
+      "step": 7462
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.454980776047118e-05,
+      "loss": 1.0772,
+      "step": 7463
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4548199637093214e-05,
+      "loss": 0.9958,
+      "step": 7464
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4546591305538784e-05,
+      "loss": 1.0355,
+      "step": 7465
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4544982765825024e-05,
+      "loss": 1.0492,
+      "step": 7466
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4543374017969065e-05,
+      "loss": 0.9448,
+      "step": 7467
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.454176506198804e-05,
+      "loss": 0.9837,
+      "step": 7468
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.454015589789907e-05,
+      "loss": 1.0592,
+      "step": 7469
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.453854652571931e-05,
+      "loss": 1.0124,
+      "step": 7470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.45369369454659e-05,
+      "loss": 1.0964,
+      "step": 7471
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4535327157155966e-05,
+      "loss": 1.0129,
+      "step": 7472
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4533717160806654e-05,
+      "loss": 1.0426,
+      "step": 7473
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.453210695643512e-05,
+      "loss": 0.9865,
+      "step": 7474
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4530496544058504e-05,
+      "loss": 1.0107,
+      "step": 7475
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4528885923693954e-05,
+      "loss": 1.019,
+      "step": 7476
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4527275095358625e-05,
+      "loss": 1.0199,
+      "step": 7477
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.452566405906967e-05,
+      "loss": 0.9788,
+      "step": 7478
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.452405281484425e-05,
+      "loss": 1.033,
+      "step": 7479
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4522441362699526e-05,
+      "loss": 1.0105,
+      "step": 7480
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.452082970265265e-05,
+      "loss": 0.961,
+      "step": 7481
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.451921783472079e-05,
+      "loss": 1.0299,
+      "step": 7482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4517605758921106e-05,
+      "loss": 0.9888,
+      "step": 7483
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4515993475270776e-05,
+      "loss": 0.9983,
+      "step": 7484
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.451438098378696e-05,
+      "loss": 1.0221,
+      "step": 7485
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.451276828448683e-05,
+      "loss": 1.0449,
+      "step": 7486
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.451115537738757e-05,
+      "loss": 0.9823,
+      "step": 7487
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.450954226250635e-05,
+      "loss": 1.0188,
+      "step": 7488
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.450792893986035e-05,
+      "loss": 1.0719,
+      "step": 7489
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.450631540946675e-05,
+      "loss": 1.0688,
+      "step": 7490
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.450470167134273e-05,
+      "loss": 1.0453,
+      "step": 7491
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.450308772550548e-05,
+      "loss": 1.036,
+      "step": 7492
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.450147357197219e-05,
+      "loss": 1.018,
+      "step": 7493
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4499859210760044e-05,
+      "loss": 0.9793,
+      "step": 7494
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.449824464188623e-05,
+      "loss": 1.0857,
+      "step": 7495
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.449662986536796e-05,
+      "loss": 0.9854,
+      "step": 7496
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4495014881222414e-05,
+      "loss": 1.0304,
+      "step": 7497
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4493399689466795e-05,
+      "loss": 1.0182,
+      "step": 7498
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.44917842901183e-05,
+      "loss": 0.93,
+      "step": 7499
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4490168683194145e-05,
+      "loss": 0.9384,
+      "step": 7500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.448855286871152e-05,
+      "loss": 1.0645,
+      "step": 7501
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.448693684668763e-05,
+      "loss": 1.02,
+      "step": 7502
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.448532061713971e-05,
+      "loss": 0.9673,
+      "step": 7503
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4483704180084944e-05,
+      "loss": 0.9734,
+      "step": 7504
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.448208753554056e-05,
+      "loss": 0.9891,
+      "step": 7505
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.448047068352378e-05,
+      "loss": 1.0526,
+      "step": 7506
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4478853624051805e-05,
+      "loss": 0.9753,
+      "step": 7507
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.447723635714187e-05,
+      "loss": 1.0055,
+      "step": 7508
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4475618882811186e-05,
+      "loss": 0.9648,
+      "step": 7509
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4474001201076986e-05,
+      "loss": 0.9476,
+      "step": 7510
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.44723833119565e-05,
+      "loss": 1.0099,
+      "step": 7511
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4470765215466946e-05,
+      "loss": 0.9621,
+      "step": 7512
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.446914691162557e-05,
+      "loss": 0.9924,
+      "step": 7513
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.44675284004496e-05,
+      "loss": 0.9668,
+      "step": 7514
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.446590968195628e-05,
+      "loss": 1.1059,
+      "step": 7515
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4464290756162826e-05,
+      "loss": 1.1115,
+      "step": 7516
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4462671623086494e-05,
+      "loss": 0.9802,
+      "step": 7517
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.446105228274453e-05,
+      "loss": 0.9891,
+      "step": 7518
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.445943273515417e-05,
+      "loss": 1.0706,
+      "step": 7519
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.445781298033267e-05,
+      "loss": 0.9744,
+      "step": 7520
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.445619301829728e-05,
+      "loss": 0.9797,
+      "step": 7521
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.445457284906523e-05,
+      "loss": 1.0083,
+      "step": 7522
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.44529524726538e-05,
+      "loss": 1.0221,
+      "step": 7523
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4451331889080236e-05,
+      "loss": 1.0599,
+      "step": 7524
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.44497110983618e-05,
+      "loss": 0.9176,
+      "step": 7525
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4448090100515745e-05,
+      "loss": 0.9707,
+      "step": 7526
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.444646889555934e-05,
+      "loss": 1.0102,
+      "step": 7527
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4444847483509846e-05,
+      "loss": 0.9928,
+      "step": 7528
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.444322586438453e-05,
+      "loss": 1.052,
+      "step": 7529
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.444160403820066e-05,
+      "loss": 1.0541,
+      "step": 7530
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4439982004975525e-05,
+      "loss": 1.0508,
+      "step": 7531
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.443835976472637e-05,
+      "loss": 1.0061,
+      "step": 7532
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.443673731747049e-05,
+      "loss": 0.9941,
+      "step": 7533
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4435114663225154e-05,
+      "loss": 0.9862,
+      "step": 7534
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.443349180200764e-05,
+      "loss": 1.018,
+      "step": 7535
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4431868733835244e-05,
+      "loss": 0.9395,
+      "step": 7536
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.443024545872525e-05,
+      "loss": 1.0714,
+      "step": 7537
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.442862197669493e-05,
+      "loss": 1.095,
+      "step": 7538
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.442699828776158e-05,
+      "loss": 1.0435,
+      "step": 7539
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4425374391942506e-05,
+      "loss": 0.9404,
+      "step": 7540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.442375028925497e-05,
+      "loss": 1.0466,
+      "step": 7541
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.44221259797163e-05,
+      "loss": 0.939,
+      "step": 7542
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4420501463343775e-05,
+      "loss": 0.9943,
+      "step": 7543
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.44188767401547e-05,
+      "loss": 0.9859,
+      "step": 7544
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.441725181016637e-05,
+      "loss": 1.0892,
+      "step": 7545
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4415626673396106e-05,
+      "loss": 0.996,
+      "step": 7546
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.441400132986121e-05,
+      "loss": 0.9329,
+      "step": 7547
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.441237577957898e-05,
+      "loss": 1.0256,
+      "step": 7548
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.441075002256673e-05,
+      "loss": 1.0722,
+      "step": 7549
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.440912405884178e-05,
+      "loss": 1.0269,
+      "step": 7550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.440749788842143e-05,
+      "loss": 0.9738,
+      "step": 7551
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.440587151132303e-05,
+      "loss": 1.0435,
+      "step": 7552
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.440424492756387e-05,
+      "loss": 1.0392,
+      "step": 7553
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.440261813716128e-05,
+      "loss": 0.974,
+      "step": 7554
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4400991140132595e-05,
+      "loss": 0.9731,
+      "step": 7555
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.439936393649513e-05,
+      "loss": 1.0109,
+      "step": 7556
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4397736526266214e-05,
+      "loss": 1.0225,
+      "step": 7557
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.439610890946318e-05,
+      "loss": 0.9765,
+      "step": 7558
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.439448108610337e-05,
+      "loss": 1.0199,
+      "step": 7559
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4392853056204097e-05,
+      "loss": 0.998,
+      "step": 7560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.439122481978272e-05,
+      "loss": 0.9006,
+      "step": 7561
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4389596376856575e-05,
+      "loss": 0.9845,
+      "step": 7562
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4387967727442994e-05,
+      "loss": 0.9312,
+      "step": 7563
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.438633887155933e-05,
+      "loss": 0.9462,
+      "step": 7564
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.438470980922292e-05,
+      "loss": 1.0376,
+      "step": 7565
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.438308054045113e-05,
+      "loss": 1.003,
+      "step": 7566
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.43814510652613e-05,
+      "loss": 1.0013,
+      "step": 7567
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4379821383670786e-05,
+      "loss": 0.89,
+      "step": 7568
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.437819149569693e-05,
+      "loss": 0.9818,
+      "step": 7569
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.43765614013571e-05,
+      "loss": 0.9395,
+      "step": 7570
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.437493110066866e-05,
+      "loss": 0.9696,
+      "step": 7571
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.437330059364896e-05,
+      "loss": 1.0316,
+      "step": 7572
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4371669880315375e-05,
+      "loss": 0.9887,
+      "step": 7573
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.437003896068527e-05,
+      "loss": 0.899,
+      "step": 7574
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.436840783477601e-05,
+      "loss": 1.0098,
+      "step": 7575
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4366776502604954e-05,
+      "loss": 1.0643,
+      "step": 7576
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4365144964189496e-05,
+      "loss": 1.0112,
+      "step": 7577
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4363513219546995e-05,
+      "loss": 1.0487,
+      "step": 7578
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.436188126869484e-05,
+      "loss": 1.0094,
+      "step": 7579
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.43602491116504e-05,
+      "loss": 0.9948,
+      "step": 7580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.435861674843106e-05,
+      "loss": 0.9713,
+      "step": 7581
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.435698417905421e-05,
+      "loss": 1.0517,
+      "step": 7582
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.435535140353723e-05,
+      "loss": 1.039,
+      "step": 7583
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.435371842189751e-05,
+      "loss": 1.029,
+      "step": 7584
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.435208523415243e-05,
+      "loss": 0.9747,
+      "step": 7585
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.43504518403194e-05,
+      "loss": 0.9766,
+      "step": 7586
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.43488182404158e-05,
+      "loss": 0.95,
+      "step": 7587
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.434718443445903e-05,
+      "loss": 1.0856,
+      "step": 7588
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.43455504224665e-05,
+      "loss": 0.9842,
+      "step": 7589
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.43439162044556e-05,
+      "loss": 0.9292,
+      "step": 7590
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.434228178044373e-05,
+      "loss": 0.9969,
+      "step": 7591
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4340647150448314e-05,
+      "loss": 1.0205,
+      "step": 7592
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.433901231448674e-05,
+      "loss": 1.0133,
+      "step": 7593
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.433737727257643e-05,
+      "loss": 0.9725,
+      "step": 7594
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.433574202473478e-05,
+      "loss": 0.9395,
+      "step": 7595
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.433410657097923e-05,
+      "loss": 0.9827,
+      "step": 7596
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.433247091132719e-05,
+      "loss": 0.9585,
+      "step": 7597
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.433083504579605e-05,
+      "loss": 0.9839,
+      "step": 7598
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.432919897440326e-05,
+      "loss": 1.0327,
+      "step": 7599
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4327562697166244e-05,
+      "loss": 1.0291,
+      "step": 7600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.432592621410242e-05,
+      "loss": 1.1207,
+      "step": 7601
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.432428952522921e-05,
+      "loss": 1.0091,
+      "step": 7602
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4322652630564046e-05,
+      "loss": 0.9742,
+      "step": 7603
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4321015530124365e-05,
+      "loss": 0.9612,
+      "step": 7604
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.43193782239276e-05,
+      "loss": 0.9934,
+      "step": 7605
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.431774071199118e-05,
+      "loss": 1.0094,
+      "step": 7606
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.431610299433255e-05,
+      "loss": 1.0229,
+      "step": 7607
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.431446507096916e-05,
+      "loss": 1.0251,
+      "step": 7608
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.431282694191844e-05,
+      "loss": 1.0142,
+      "step": 7609
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4311188607197836e-05,
+      "loss": 1.068,
+      "step": 7610
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.43095500668248e-05,
+      "loss": 1.0098,
+      "step": 7611
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.430791132081678e-05,
+      "loss": 1.0338,
+      "step": 7612
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.430627236919121e-05,
+      "loss": 0.9907,
+      "step": 7613
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4304633211965576e-05,
+      "loss": 0.9967,
+      "step": 7614
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.430299384915731e-05,
+      "loss": 0.9478,
+      "step": 7615
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.430135428078389e-05,
+      "loss": 0.9553,
+      "step": 7616
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.429971450686276e-05,
+      "loss": 0.9849,
+      "step": 7617
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4298074527411384e-05,
+      "loss": 0.9817,
+      "step": 7618
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.429643434244723e-05,
+      "loss": 0.9363,
+      "step": 7619
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.429479395198777e-05,
+      "loss": 0.9403,
+      "step": 7620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.429315335605047e-05,
+      "loss": 0.9701,
+      "step": 7621
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.42915125546528e-05,
+      "loss": 0.9859,
+      "step": 7622
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.428987154781223e-05,
+      "loss": 1.0054,
+      "step": 7623
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.428823033554623e-05,
+      "loss": 0.9683,
+      "step": 7624
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.428658891787231e-05,
+      "loss": 0.9119,
+      "step": 7625
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.428494729480791e-05,
+      "loss": 0.971,
+      "step": 7626
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.428330546637053e-05,
+      "loss": 0.9656,
+      "step": 7627
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.428166343257766e-05,
+      "loss": 0.9456,
+      "step": 7628
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4280021193446774e-05,
+      "loss": 1.0749,
+      "step": 7629
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4278378748995375e-05,
+      "loss": 1.0123,
+      "step": 7630
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.427673609924094e-05,
+      "loss": 1.0151,
+      "step": 7631
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4275093244200974e-05,
+      "loss": 0.954,
+      "step": 7632
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.427345018389296e-05,
+      "loss": 0.9543,
+      "step": 7633
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4271806918334414e-05,
+      "loss": 1.0641,
+      "step": 7634
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4270163447542814e-05,
+      "loss": 0.8915,
+      "step": 7635
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.426851977153568e-05,
+      "loss": 1.0212,
+      "step": 7636
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4266875890330506e-05,
+      "loss": 1.1001,
+      "step": 7637
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.42652318039448e-05,
+      "loss": 1.001,
+      "step": 7638
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.426358751239607e-05,
+      "loss": 0.8862,
+      "step": 7639
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.426194301570183e-05,
+      "loss": 0.924,
+      "step": 7640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.426029831387959e-05,
+      "loss": 1.0257,
+      "step": 7641
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.425865340694687e-05,
+      "loss": 0.9058,
+      "step": 7642
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4257008294921173e-05,
+      "loss": 0.9676,
+      "step": 7643
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4255362977820044e-05,
+      "loss": 1.0357,
+      "step": 7644
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.425371745566098e-05,
+      "loss": 1.0462,
+      "step": 7645
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.425207172846151e-05,
+      "loss": 1.0098,
+      "step": 7646
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4250425796239174e-05,
+      "loss": 1.0732,
+      "step": 7647
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.424877965901149e-05,
+      "loss": 1.03,
+      "step": 7648
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.424713331679598e-05,
+      "loss": 1.0346,
+      "step": 7649
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4245486769610196e-05,
+      "loss": 1.0078,
+      "step": 7650
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.424384001747166e-05,
+      "loss": 1.0091,
+      "step": 7651
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.424219306039791e-05,
+      "loss": 0.9299,
+      "step": 7652
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4240545898406486e-05,
+      "loss": 1.0086,
+      "step": 7653
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4238898531514925e-05,
+      "loss": 1.008,
+      "step": 7654
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.423725095974077e-05,
+      "loss": 1.0544,
+      "step": 7655
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.423560318310158e-05,
+      "loss": 1.0887,
+      "step": 7656
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4233955201614896e-05,
+      "loss": 0.9285,
+      "step": 7657
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.423230701529827e-05,
+      "loss": 1.0149,
+      "step": 7658
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.423065862416924e-05,
+      "loss": 0.9153,
+      "step": 7659
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.422901002824538e-05,
+      "loss": 1.0133,
+      "step": 7660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.422736122754423e-05,
+      "loss": 0.9694,
+      "step": 7661
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.422571222208336e-05,
+      "loss": 1.0174,
+      "step": 7662
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.422406301188032e-05,
+      "loss": 0.9405,
+      "step": 7663
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.422241359695269e-05,
+      "loss": 1.0633,
+      "step": 7664
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4220763977318026e-05,
+      "loss": 0.9471,
+      "step": 7665
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4219114152993886e-05,
+      "loss": 0.9601,
+      "step": 7666
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4217464123997854e-05,
+      "loss": 0.9889,
+      "step": 7667
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4215813890347495e-05,
+      "loss": 0.926,
+      "step": 7668
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4214163452060386e-05,
+      "loss": 1.0076,
+      "step": 7669
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.421251280915409e-05,
+      "loss": 1.0003,
+      "step": 7670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.421086196164621e-05,
+      "loss": 0.8953,
+      "step": 7671
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.420921090955431e-05,
+      "loss": 1.0133,
+      "step": 7672
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.420755965289598e-05,
+      "loss": 0.9593,
+      "step": 7673
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.420590819168879e-05,
+      "loss": 0.9033,
+      "step": 7674
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.420425652595035e-05,
+      "loss": 0.9806,
+      "step": 7675
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4202604655698234e-05,
+      "loss": 1.0207,
+      "step": 7676
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.420095258095004e-05,
+      "loss": 0.8814,
+      "step": 7677
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.419930030172334e-05,
+      "loss": 0.9959,
+      "step": 7678
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.419764781803577e-05,
+      "loss": 0.9596,
+      "step": 7679
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.419599512990489e-05,
+      "loss": 0.9951,
+      "step": 7680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.419434223734833e-05,
+      "loss": 1.0194,
+      "step": 7681
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.419268914038367e-05,
+      "loss": 0.9587,
+      "step": 7682
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.419103583902852e-05,
+      "loss": 1.0178,
+      "step": 7683
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.41893823333005e-05,
+      "loss": 0.9264,
+      "step": 7684
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.41877286232172e-05,
+      "loss": 0.939,
+      "step": 7685
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.418607470879623e-05,
+      "loss": 0.9857,
+      "step": 7686
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.418442059005523e-05,
+      "loss": 0.9791,
+      "step": 7687
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.418276626701179e-05,
+      "loss": 0.884,
+      "step": 7688
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4181111739683535e-05,
+      "loss": 0.9405,
+      "step": 7689
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.417945700808809e-05,
+      "loss": 0.9675,
+      "step": 7690
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.417780207224307e-05,
+      "loss": 0.9233,
+      "step": 7691
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.41761469321661e-05,
+      "loss": 0.9346,
+      "step": 7692
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4174491587874805e-05,
+      "loss": 0.9691,
+      "step": 7693
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4172836039386825e-05,
+      "loss": 0.9164,
+      "step": 7694
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.417118028671977e-05,
+      "loss": 1.0636,
+      "step": 7695
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.416952432989129e-05,
+      "loss": 0.998,
+      "step": 7696
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.416786816891901e-05,
+      "loss": 1.0449,
+      "step": 7697
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.416621180382057e-05,
+      "loss": 1.0012,
+      "step": 7698
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4164555234613616e-05,
+      "loss": 1.019,
+      "step": 7699
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4162898461315785e-05,
+      "loss": 1.0141,
+      "step": 7700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.416124148394471e-05,
+      "loss": 0.9823,
+      "step": 7701
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4159584302518054e-05,
+      "loss": 1.0021,
+      "step": 7702
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.415792691705345e-05,
+      "loss": 0.9498,
+      "step": 7703
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4156269327568564e-05,
+      "loss": 1.0033,
+      "step": 7704
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4154611534081035e-05,
+      "loss": 1.0199,
+      "step": 7705
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4152953536608525e-05,
+      "loss": 1.0633,
+      "step": 7706
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4151295335168686e-05,
+      "loss": 0.9969,
+      "step": 7707
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.414963692977918e-05,
+      "loss": 1.0042,
+      "step": 7708
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.414797832045767e-05,
+      "loss": 0.9336,
+      "step": 7709
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4146319507221814e-05,
+      "loss": 0.9417,
+      "step": 7710
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4144660490089274e-05,
+      "loss": 1.0005,
+      "step": 7711
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.414300126907772e-05,
+      "loss": 0.9399,
+      "step": 7712
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4141341844204836e-05,
+      "loss": 1.1086,
+      "step": 7713
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4139682215488276e-05,
+      "loss": 0.9835,
+      "step": 7714
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.413802238294572e-05,
+      "loss": 1.0832,
+      "step": 7715
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.413636234659485e-05,
+      "loss": 1.028,
+      "step": 7716
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.413470210645333e-05,
+      "loss": 0.9815,
+      "step": 7717
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4133041662538856e-05,
+      "loss": 1.0212,
+      "step": 7718
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.41313810148691e-05,
+      "loss": 1.0024,
+      "step": 7719
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.412972016346176e-05,
+      "loss": 0.9295,
+      "step": 7720
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.412805910833451e-05,
+      "loss": 1.0128,
+      "step": 7721
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.412639784950503e-05,
+      "loss": 0.9451,
+      "step": 7722
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4124736386991044e-05,
+      "loss": 1.035,
+      "step": 7723
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.412307472081022e-05,
+      "loss": 0.9076,
+      "step": 7724
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4121412850980257e-05,
+      "loss": 1.0169,
+      "step": 7725
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.411975077751885e-05,
+      "loss": 0.9532,
+      "step": 7726
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.411808850044372e-05,
+      "loss": 0.9467,
+      "step": 7727
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.411642601977255e-05,
+      "loss": 0.9938,
+      "step": 7728
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4114763335523044e-05,
+      "loss": 0.9589,
+      "step": 7729
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4113100447712915e-05,
+      "loss": 1.0469,
+      "step": 7730
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.411143735635987e-05,
+      "loss": 0.9518,
+      "step": 7731
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.410977406148162e-05,
+      "loss": 1.0231,
+      "step": 7732
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4108110563095875e-05,
+      "loss": 0.9781,
+      "step": 7733
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4106446861220364e-05,
+      "loss": 0.9486,
+      "step": 7734
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.410478295587278e-05,
+      "loss": 1.0605,
+      "step": 7735
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.410311884707087e-05,
+      "loss": 1.1125,
+      "step": 7736
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.410145453483233e-05,
+      "loss": 0.9582,
+      "step": 7737
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.40997900191749e-05,
+      "loss": 1.0974,
+      "step": 7738
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.40981253001163e-05,
+      "loss": 0.9849,
+      "step": 7739
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.409646037767426e-05,
+      "loss": 0.9873,
+      "step": 7740
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.409479525186652e-05,
+      "loss": 0.9647,
+      "step": 7741
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.409312992271079e-05,
+      "loss": 0.9454,
+      "step": 7742
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.409146439022484e-05,
+      "loss": 0.9688,
+      "step": 7743
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.408979865442636e-05,
+      "loss": 0.9849,
+      "step": 7744
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4088132715333124e-05,
+      "loss": 0.9525,
+      "step": 7745
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.408646657296287e-05,
+      "loss": 0.9373,
+      "step": 7746
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4084800227333325e-05,
+      "loss": 1.023,
+      "step": 7747
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4083133678462246e-05,
+      "loss": 0.9544,
+      "step": 7748
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.408146692636738e-05,
+      "loss": 0.9553,
+      "step": 7749
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.407979997106648e-05,
+      "loss": 0.9269,
+      "step": 7750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.407813281257729e-05,
+      "loss": 0.9474,
+      "step": 7751
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4076465450917564e-05,
+      "loss": 1.0186,
+      "step": 7752
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.407479788610507e-05,
+      "loss": 1.0065,
+      "step": 7753
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.407313011815756e-05,
+      "loss": 0.9281,
+      "step": 7754
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.407146214709278e-05,
+      "loss": 0.9987,
+      "step": 7755
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.406979397292852e-05,
+      "loss": 0.9746,
+      "step": 7756
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.406812559568253e-05,
+      "loss": 0.9134,
+      "step": 7757
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4066457015372575e-05,
+      "loss": 1.0095,
+      "step": 7758
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.406478823201643e-05,
+      "loss": 0.9227,
+      "step": 7759
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.406311924563187e-05,
+      "loss": 0.9931,
+      "step": 7760
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.406145005623665e-05,
+      "loss": 1.0029,
+      "step": 7761
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4059780663848574e-05,
+      "loss": 1.0393,
+      "step": 7762
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.40581110684854e-05,
+      "loss": 0.9505,
+      "step": 7763
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.405644127016492e-05,
+      "loss": 0.967,
+      "step": 7764
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.405477126890489e-05,
+      "loss": 1.0394,
+      "step": 7765
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.405310106472313e-05,
+      "loss": 0.9146,
+      "step": 7766
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.405143065763741e-05,
+      "loss": 0.9333,
+      "step": 7767
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.404976004766552e-05,
+      "loss": 1.0365,
+      "step": 7768
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4048089234825254e-05,
+      "loss": 0.9667,
+      "step": 7769
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.404641821913439e-05,
+      "loss": 1.0075,
+      "step": 7770
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.404474700061074e-05,
+      "loss": 0.9224,
+      "step": 7771
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.404307557927211e-05,
+      "loss": 0.9713,
+      "step": 7772
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4041403955136276e-05,
+      "loss": 1.0044,
+      "step": 7773
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.403973212822105e-05,
+      "loss": 0.9291,
+      "step": 7774
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.403806009854423e-05,
+      "loss": 1.0054,
+      "step": 7775
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.403638786612364e-05,
+      "loss": 0.9974,
+      "step": 7776
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.403471543097707e-05,
+      "loss": 0.9535,
+      "step": 7777
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4033042793122334e-05,
+      "loss": 1.0037,
+      "step": 7778
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.403136995257725e-05,
+      "loss": 0.9465,
+      "step": 7779
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4029696909359636e-05,
+      "loss": 0.9614,
+      "step": 7780
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.40280236634873e-05,
+      "loss": 0.9587,
+      "step": 7781
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4026350214978066e-05,
+      "loss": 0.9882,
+      "step": 7782
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.402467656384974e-05,
+      "loss": 0.9559,
+      "step": 7783
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4023002710120175e-05,
+      "loss": 0.9938,
+      "step": 7784
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4021328653807176e-05,
+      "loss": 0.983,
+      "step": 7785
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.401965439492857e-05,
+      "loss": 0.9489,
+      "step": 7786
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.40179799335022e-05,
+      "loss": 1.0029,
+      "step": 7787
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.401630526954588e-05,
+      "loss": 0.9957,
+      "step": 7788
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.401463040307746e-05,
+      "loss": 0.9283,
+      "step": 7789
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.401295533411477e-05,
+      "loss": 1.0246,
+      "step": 7790
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4011280062675656e-05,
+      "loss": 0.9571,
+      "step": 7791
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.400960458877794e-05,
+      "loss": 1.0037,
+      "step": 7792
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4007928912439475e-05,
+      "loss": 0.9529,
+      "step": 7793
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4006253033678115e-05,
+      "loss": 0.9181,
+      "step": 7794
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4004576952511687e-05,
+      "loss": 0.9442,
+      "step": 7795
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.400290066895807e-05,
+      "loss": 0.9322,
+      "step": 7796
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4001224183035086e-05,
+      "loss": 0.9925,
+      "step": 7797
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3999547494760604e-05,
+      "loss": 0.9163,
+      "step": 7798
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3997870604152476e-05,
+      "loss": 1.068,
+      "step": 7799
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3996193511228556e-05,
+      "loss": 0.9313,
+      "step": 7800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.399451621600671e-05,
+      "loss": 0.9701,
+      "step": 7801
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3992838718504805e-05,
+      "loss": 0.9831,
+      "step": 7802
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.399116101874069e-05,
+      "loss": 0.9705,
+      "step": 7803
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.398948311673224e-05,
+      "loss": 0.9392,
+      "step": 7804
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.398780501249733e-05,
+      "loss": 0.9829,
+      "step": 7805
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.398612670605382e-05,
+      "loss": 0.9523,
+      "step": 7806
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.398444819741959e-05,
+      "loss": 0.9429,
+      "step": 7807
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.398276948661251e-05,
+      "loss": 1.0219,
+      "step": 7808
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3981090573650466e-05,
+      "loss": 0.9721,
+      "step": 7809
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.397941145855132e-05,
+      "loss": 1.0001,
+      "step": 7810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3977732141332975e-05,
+      "loss": 0.9189,
+      "step": 7811
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3976052622013294e-05,
+      "loss": 1.0675,
+      "step": 7812
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.397437290061018e-05,
+      "loss": 0.937,
+      "step": 7813
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3972692977141516e-05,
+      "loss": 1.001,
+      "step": 7814
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.397101285162518e-05,
+      "loss": 1.0008,
+      "step": 7815
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3969332524079085e-05,
+      "loss": 0.9767,
+      "step": 7816
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.396765199452111e-05,
+      "loss": 1.0033,
+      "step": 7817
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3965971262969164e-05,
+      "loss": 0.9623,
+      "step": 7818
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.396429032944113e-05,
+      "loss": 0.9991,
+      "step": 7819
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3962609193954915e-05,
+      "loss": 0.918,
+      "step": 7820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.396092785652844e-05,
+      "loss": 0.9229,
+      "step": 7821
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.395924631717958e-05,
+      "loss": 0.9804,
+      "step": 7822
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3957564575926265e-05,
+      "loss": 0.9342,
+      "step": 7823
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.395588263278639e-05,
+      "loss": 0.9658,
+      "step": 7824
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.395420048777789e-05,
+      "loss": 0.9574,
+      "step": 7825
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.395251814091864e-05,
+      "loss": 0.9194,
+      "step": 7826
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3950835592226594e-05,
+      "loss": 0.9035,
+      "step": 7827
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.394915284171964e-05,
+      "loss": 0.9671,
+      "step": 7828
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3947469889415726e-05,
+      "loss": 0.9929,
+      "step": 7829
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.394578673533276e-05,
+      "loss": 1.1329,
+      "step": 7830
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3944103379488656e-05,
+      "loss": 0.9608,
+      "step": 7831
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.394241982190136e-05,
+      "loss": 0.9186,
+      "step": 7832
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.394073606258879e-05,
+      "loss": 0.921,
+      "step": 7833
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.393905210156889e-05,
+      "loss": 0.9766,
+      "step": 7834
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.393736793885957e-05,
+      "loss": 1.0627,
+      "step": 7835
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3935683574478785e-05,
+      "loss": 0.9669,
+      "step": 7836
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.393399900844447e-05,
+      "loss": 0.9597,
+      "step": 7837
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3932314240774554e-05,
+      "loss": 1.022,
+      "step": 7838
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.393062927148699e-05,
+      "loss": 0.9668,
+      "step": 7839
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.392894410059971e-05,
+      "loss": 0.9063,
+      "step": 7840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.392725872813068e-05,
+      "loss": 0.9487,
+      "step": 7841
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.392557315409782e-05,
+      "loss": 1.0209,
+      "step": 7842
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3923887378519115e-05,
+      "loss": 0.9513,
+      "step": 7843
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3922201401412485e-05,
+      "loss": 0.9695,
+      "step": 7844
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.39205152227959e-05,
+      "loss": 1.0311,
+      "step": 7845
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.391882884268731e-05,
+      "loss": 0.9929,
+      "step": 7846
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3917142261104684e-05,
+      "loss": 0.9463,
+      "step": 7847
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.391545547806598e-05,
+      "loss": 0.9683,
+      "step": 7848
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.391376849358916e-05,
+      "loss": 0.9601,
+      "step": 7849
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.391208130769218e-05,
+      "loss": 1.0612,
+      "step": 7850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.391039392039302e-05,
+      "loss": 1.0409,
+      "step": 7851
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.390870633170965e-05,
+      "loss": 0.9625,
+      "step": 7852
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3907018541660034e-05,
+      "loss": 1.0767,
+      "step": 7853
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.390533055026215e-05,
+      "loss": 0.9789,
+      "step": 7854
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.390364235753397e-05,
+      "loss": 0.9509,
+      "step": 7855
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.390195396349348e-05,
+      "loss": 1.0966,
+      "step": 7856
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.390026536815866e-05,
+      "loss": 1.0051,
+      "step": 7857
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.389857657154747e-05,
+      "loss": 0.944,
+      "step": 7858
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.389688757367793e-05,
+      "loss": 0.9895,
+      "step": 7859
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3895198374568006e-05,
+      "loss": 0.9333,
+      "step": 7860
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3893508974235684e-05,
+      "loss": 0.961,
+      "step": 7861
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.389181937269897e-05,
+      "loss": 0.9816,
+      "step": 7862
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.389012956997585e-05,
+      "loss": 1.0831,
+      "step": 7863
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3888439566084314e-05,
+      "loss": 0.9517,
+      "step": 7864
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.388674936104237e-05,
+      "loss": 0.9956,
+      "step": 7865
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.388505895486801e-05,
+      "loss": 1.0011,
+      "step": 7866
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.388336834757924e-05,
+      "loss": 0.9536,
+      "step": 7867
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.388167753919405e-05,
+      "loss": 0.9757,
+      "step": 7868
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387998652973046e-05,
+      "loss": 0.9535,
+      "step": 7869
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387829531920648e-05,
+      "loss": 0.9192,
+      "step": 7870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387660390764012e-05,
+      "loss": 1.0247,
+      "step": 7871
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387491229504939e-05,
+      "loss": 0.9775,
+      "step": 7872
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3873220481452294e-05,
+      "loss": 0.9738,
+      "step": 7873
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387152846686686e-05,
+      "loss": 0.939,
+      "step": 7874
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.386983625131111e-05,
+      "loss": 0.9858,
+      "step": 7875
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.386814383480306e-05,
+      "loss": 1.0276,
+      "step": 7876
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3866451217360726e-05,
+      "loss": 0.9842,
+      "step": 7877
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3864758399002146e-05,
+      "loss": 0.9652,
+      "step": 7878
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3863065379745336e-05,
+      "loss": 0.972,
+      "step": 7879
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3861372159608346e-05,
+      "loss": 0.9806,
+      "step": 7880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.385967873860918e-05,
+      "loss": 0.9045,
+      "step": 7881
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.385798511676589e-05,
+      "loss": 0.9732,
+      "step": 7882
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3856291294096505e-05,
+      "loss": 0.9784,
+      "step": 7883
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.385459727061907e-05,
+      "loss": 0.898,
+      "step": 7884
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3852903046351625e-05,
+      "loss": 0.9438,
+      "step": 7885
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.38512086213122e-05,
+      "loss": 0.9953,
+      "step": 7886
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3849513995518856e-05,
+      "loss": 1.0687,
+      "step": 7887
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.384781916898962e-05,
+      "loss": 0.9476,
+      "step": 7888
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.384612414174256e-05,
+      "loss": 0.8811,
+      "step": 7889
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.384442891379572e-05,
+      "loss": 0.9624,
+      "step": 7890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.384273348516715e-05,
+      "loss": 1.0217,
+      "step": 7891
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.384103785587491e-05,
+      "loss": 0.9064,
+      "step": 7892
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.383934202593706e-05,
+      "loss": 1.0131,
+      "step": 7893
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3837645995371644e-05,
+      "loss": 0.8924,
+      "step": 7894
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.383594976419674e-05,
+      "loss": 1.0315,
+      "step": 7895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3834253332430406e-05,
+      "loss": 0.9032,
+      "step": 7896
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.383255670009071e-05,
+      "loss": 0.9788,
+      "step": 7897
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.383085986719572e-05,
+      "loss": 1.0322,
+      "step": 7898
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3829162833763505e-05,
+      "loss": 0.9286,
+      "step": 7899
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382746559981214e-05,
+      "loss": 0.9836,
+      "step": 7900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382576816535969e-05,
+      "loss": 0.971,
+      "step": 7901
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3824070530424246e-05,
+      "loss": 0.9463,
+      "step": 7902
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382237269502387e-05,
+      "loss": 0.9338,
+      "step": 7903
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382067465917665e-05,
+      "loss": 0.8869,
+      "step": 7904
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3818976422900683e-05,
+      "loss": 0.9181,
+      "step": 7905
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.381727798621404e-05,
+      "loss": 1.0411,
+      "step": 7906
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.381557934913481e-05,
+      "loss": 0.9478,
+      "step": 7907
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.381388051168108e-05,
+      "loss": 1.0055,
+      "step": 7908
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.381218147387094e-05,
+      "loss": 0.9163,
+      "step": 7909
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3810482235722494e-05,
+      "loss": 0.9843,
+      "step": 7910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3808782797253835e-05,
+      "loss": 0.9762,
+      "step": 7911
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380708315848305e-05,
+      "loss": 1.0173,
+      "step": 7912
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380538331942826e-05,
+      "loss": 0.9043,
+      "step": 7913
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3803683280107543e-05,
+      "loss": 0.9943,
+      "step": 7914
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380198304053902e-05,
+      "loss": 0.9733,
+      "step": 7915
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3800282600740797e-05,
+      "loss": 0.9243,
+      "step": 7916
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379858196073097e-05,
+      "loss": 0.9231,
+      "step": 7917
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379688112052767e-05,
+      "loss": 1.0068,
+      "step": 7918
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3795180080148984e-05,
+      "loss": 1.012,
+      "step": 7919
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379347883961305e-05,
+      "loss": 1.0387,
+      "step": 7920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379177739893797e-05,
+      "loss": 0.9967,
+      "step": 7921
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379007575814187e-05,
+      "loss": 0.9409,
+      "step": 7922
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378837391724287e-05,
+      "loss": 1.0055,
+      "step": 7923
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37866718762591e-05,
+      "loss": 1.0138,
+      "step": 7924
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3784969635208686e-05,
+      "loss": 0.9004,
+      "step": 7925
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378326719410973e-05,
+      "loss": 0.9533,
+      "step": 7926
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37815645529804e-05,
+      "loss": 0.8719,
+      "step": 7927
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37798617118388e-05,
+      "loss": 0.9868,
+      "step": 7928
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.377815867070308e-05,
+      "loss": 0.9784,
+      "step": 7929
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3776455429591374e-05,
+      "loss": 0.9757,
+      "step": 7930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.377475198852181e-05,
+      "loss": 0.9744,
+      "step": 7931
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.377304834751253e-05,
+      "loss": 0.916,
+      "step": 7932
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37713445065817e-05,
+      "loss": 0.9415,
+      "step": 7933
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3769640465747434e-05,
+      "loss": 0.9783,
+      "step": 7934
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.37679362250279e-05,
+      "loss": 0.9429,
+      "step": 7935
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376623178444124e-05,
+      "loss": 0.9127,
+      "step": 7936
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3764527144005594e-05,
+      "loss": 0.9622,
+      "step": 7937
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376282230373913e-05,
+      "loss": 0.9374,
+      "step": 7938
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376111726366e-05,
+      "loss": 0.927,
+      "step": 7939
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3759412023786365e-05,
+      "loss": 0.8859,
+      "step": 7940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375770658413638e-05,
+      "loss": 0.9954,
+      "step": 7941
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375600094472821e-05,
+      "loss": 0.9317,
+      "step": 7942
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375429510558001e-05,
+      "loss": 0.9732,
+      "step": 7943
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375258906670996e-05,
+      "loss": 0.9761,
+      "step": 7944
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375088282813622e-05,
+      "loss": 0.9634,
+      "step": 7945
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374917638987696e-05,
+      "loss": 0.8433,
+      "step": 7946
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374746975195035e-05,
+      "loss": 1.0291,
+      "step": 7947
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374576291437458e-05,
+      "loss": 1.0103,
+      "step": 7948
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374405587716781e-05,
+      "loss": 0.9836,
+      "step": 7949
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374234864034822e-05,
+      "loss": 0.9691,
+      "step": 7950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374064120393401e-05,
+      "loss": 0.9866,
+      "step": 7951
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373893356794333e-05,
+      "loss": 0.974,
+      "step": 7952
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373722573239439e-05,
+      "loss": 1.0153,
+      "step": 7953
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373551769730538e-05,
+      "loss": 0.9735,
+      "step": 7954
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3733809462694474e-05,
+      "loss": 0.9327,
+      "step": 7955
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373210102857988e-05,
+      "loss": 0.9543,
+      "step": 7956
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3730392394979766e-05,
+      "loss": 0.8862,
+      "step": 7957
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.372868356191236e-05,
+      "loss": 0.9677,
+      "step": 7958
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3726974529395835e-05,
+      "loss": 0.9912,
+      "step": 7959
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.372526529744841e-05,
+      "loss": 0.9605,
+      "step": 7960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3723555866088274e-05,
+      "loss": 0.9939,
+      "step": 7961
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.372184623533363e-05,
+      "loss": 0.935,
+      "step": 7962
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.372013640520269e-05,
+      "loss": 0.8727,
+      "step": 7963
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371842637571367e-05,
+      "loss": 0.9457,
+      "step": 7964
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371671614688477e-05,
+      "loss": 0.9483,
+      "step": 7965
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3715005718734206e-05,
+      "loss": 0.9272,
+      "step": 7966
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3713295091280194e-05,
+      "loss": 0.9264,
+      "step": 7967
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3711584264540945e-05,
+      "loss": 0.9795,
+      "step": 7968
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370987323853469e-05,
+      "loss": 0.9498,
+      "step": 7969
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370816201327964e-05,
+      "loss": 0.9626,
+      "step": 7970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3706450588794024e-05,
+      "loss": 1.0161,
+      "step": 7971
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370473896509607e-05,
+      "loss": 0.8694,
+      "step": 7972
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3703027142203996e-05,
+      "loss": 0.9781,
+      "step": 7973
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370131512013604e-05,
+      "loss": 0.9016,
+      "step": 7974
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369960289891043e-05,
+      "loss": 0.9904,
+      "step": 7975
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3697890478545405e-05,
+      "loss": 0.9086,
+      "step": 7976
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3696177859059196e-05,
+      "loss": 0.8932,
+      "step": 7977
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3694465040470046e-05,
+      "loss": 0.905,
+      "step": 7978
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369275202279619e-05,
+      "loss": 0.9191,
+      "step": 7979
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369103880605588e-05,
+      "loss": 0.9452,
+      "step": 7980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368932539026735e-05,
+      "loss": 0.9778,
+      "step": 7981
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368761177544885e-05,
+      "loss": 1.0086,
+      "step": 7982
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368589796161863e-05,
+      "loss": 0.9945,
+      "step": 7983
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368418394879494e-05,
+      "loss": 0.9894,
+      "step": 7984
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368246973699604e-05,
+      "loss": 0.9691,
+      "step": 7985
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3680755326240175e-05,
+      "loss": 1.0115,
+      "step": 7986
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367904071654561e-05,
+      "loss": 0.9386,
+      "step": 7987
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.36773259079306e-05,
+      "loss": 0.9096,
+      "step": 7988
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3675610900413404e-05,
+      "loss": 0.9211,
+      "step": 7989
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.36738956940123e-05,
+      "loss": 0.8557,
+      "step": 7990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367218028874553e-05,
+      "loss": 0.9679,
+      "step": 7991
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367046468463139e-05,
+      "loss": 0.9086,
+      "step": 7992
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366874888168813e-05,
+      "loss": 0.9573,
+      "step": 7993
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366703287993403e-05,
+      "loss": 0.9064,
+      "step": 7994
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366531667938736e-05,
+      "loss": 0.9017,
+      "step": 7995
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366360028006641e-05,
+      "loss": 0.8993,
+      "step": 7996
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3661883681989434e-05,
+      "loss": 0.9487,
+      "step": 7997
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366016688517474e-05,
+      "loss": 0.9886,
+      "step": 7998
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3658449889640584e-05,
+      "loss": 0.9961,
+      "step": 7999
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.365673269540527e-05,
+      "loss": 0.9519,
+      "step": 8000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3655015302487076e-05,
+      "loss": 0.9756,
+      "step": 8001
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3653297710904306e-05,
+      "loss": 0.9522,
+      "step": 8002
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.365157992067523e-05,
+      "loss": 0.9035,
+      "step": 8003
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364986193181815e-05,
+      "loss": 0.9128,
+      "step": 8004
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364814374435137e-05,
+      "loss": 0.9149,
+      "step": 8005
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364642535829318e-05,
+      "loss": 1.0081,
+      "step": 8006
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364470677366188e-05,
+      "loss": 0.9566,
+      "step": 8007
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364298799047577e-05,
+      "loss": 0.9158,
+      "step": 8008
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364126900875316e-05,
+      "loss": 0.8327,
+      "step": 8009
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3639549828512343e-05,
+      "loss": 0.9701,
+      "step": 8010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363783044977164e-05,
+      "loss": 0.9102,
+      "step": 8011
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363611087254936e-05,
+      "loss": 0.9726,
+      "step": 8012
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363439109686383e-05,
+      "loss": 1.0319,
+      "step": 8013
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3632671122733326e-05,
+      "loss": 0.966,
+      "step": 8014
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363095095017619e-05,
+      "loss": 0.8945,
+      "step": 8015
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3629230579210746e-05,
+      "loss": 1.005,
+      "step": 8016
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36275100098553e-05,
+      "loss": 0.9246,
+      "step": 8017
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362578924212818e-05,
+      "loss": 1.021,
+      "step": 8018
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362406827604771e-05,
+      "loss": 0.9082,
+      "step": 8019
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362234711163223e-05,
+      "loss": 0.9455,
+      "step": 8020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3620625748900056e-05,
+      "loss": 0.9258,
+      "step": 8021
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361890418786952e-05,
+      "loss": 0.907,
+      "step": 8022
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361718242855896e-05,
+      "loss": 1.0048,
+      "step": 8023
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361546047098671e-05,
+      "loss": 0.969,
+      "step": 8024
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36137383151711e-05,
+      "loss": 0.892,
+      "step": 8025
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361201596113049e-05,
+      "loss": 0.9811,
+      "step": 8026
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3610293408883206e-05,
+      "loss": 0.9801,
+      "step": 8027
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3608570658447586e-05,
+      "loss": 0.9616,
+      "step": 8028
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3606847709842e-05,
+      "loss": 0.9736,
+      "step": 8029
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3605124563084775e-05,
+      "loss": 0.9014,
+      "step": 8030
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.360340121819427e-05,
+      "loss": 0.9556,
+      "step": 8031
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.360167767518884e-05,
+      "loss": 0.9093,
+      "step": 8032
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3599953934086834e-05,
+      "loss": 0.8938,
+      "step": 8033
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.35982299949066e-05,
+      "loss": 0.9501,
+      "step": 8034
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3596505857666526e-05,
+      "loss": 0.9113,
+      "step": 8035
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3594781522384944e-05,
+      "loss": 0.9405,
+      "step": 8036
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.359305698908023e-05,
+      "loss": 0.9626,
+      "step": 8037
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3591332257770745e-05,
+      "loss": 0.9284,
+      "step": 8038
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358960732847486e-05,
+      "loss": 1.0175,
+      "step": 8039
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358788220121095e-05,
+      "loss": 0.9441,
+      "step": 8040
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358615687599737e-05,
+      "loss": 0.9425,
+      "step": 8041
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3584431352852507e-05,
+      "loss": 0.9872,
+      "step": 8042
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358270563179473e-05,
+      "loss": 1.0086,
+      "step": 8043
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358097971284242e-05,
+      "loss": 0.9059,
+      "step": 8044
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.357925359601396e-05,
+      "loss": 0.8715,
+      "step": 8045
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.357752728132774e-05,
+      "loss": 0.9319,
+      "step": 8046
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.357580076880211e-05,
+      "loss": 1.0006,
+      "step": 8047
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3574074058455494e-05,
+      "loss": 0.9591,
+      "step": 8048
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3572347150306266e-05,
+      "loss": 0.9836,
+      "step": 8049
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.357062004437281e-05,
+      "loss": 0.9128,
+      "step": 8050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.356889274067353e-05,
+      "loss": 0.9595,
+      "step": 8051
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3567165239226813e-05,
+      "loss": 0.9392,
+      "step": 8052
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.356543754005106e-05,
+      "loss": 0.9457,
+      "step": 8053
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.356370964316466e-05,
+      "loss": 0.9552,
+      "step": 8054
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3561981548586034e-05,
+      "loss": 0.855,
+      "step": 8055
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.356025325633357e-05,
+      "loss": 0.953,
+      "step": 8056
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.355852476642568e-05,
+      "loss": 1.0409,
+      "step": 8057
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.355679607888077e-05,
+      "loss": 0.9349,
+      "step": 8058
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.355506719371724e-05,
+      "loss": 0.9057,
+      "step": 8059
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.355333811095351e-05,
+      "loss": 1.0042,
+      "step": 8060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3551608830607996e-05,
+      "loss": 0.9768,
+      "step": 8061
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3549879352699116e-05,
+      "loss": 0.9453,
+      "step": 8062
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.354814967724528e-05,
+      "loss": 0.8971,
+      "step": 8063
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3546419804264904e-05,
+      "loss": 0.9385,
+      "step": 8064
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.354468973377642e-05,
+      "loss": 0.961,
+      "step": 8065
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.354295946579825e-05,
+      "loss": 0.9891,
+      "step": 8066
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3541229000348824e-05,
+      "loss": 1.004,
+      "step": 8067
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.353949833744656e-05,
+      "loss": 1.0298,
+      "step": 8068
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.35377674771099e-05,
+      "loss": 0.9902,
+      "step": 8069
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.353603641935726e-05,
+      "loss": 0.9735,
+      "step": 8070
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3534305164207096e-05,
+      "loss": 0.8975,
+      "step": 8071
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.353257371167783e-05,
+      "loss": 0.9226,
+      "step": 8072
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.353084206178791e-05,
+      "loss": 1.0033,
+      "step": 8073
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.352911021455577e-05,
+      "loss": 0.8856,
+      "step": 8074
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.352737816999986e-05,
+      "loss": 0.8476,
+      "step": 8075
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.352564592813861e-05,
+      "loss": 1.0293,
+      "step": 8076
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.352391348899048e-05,
+      "loss": 0.9005,
+      "step": 8077
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3522180852573925e-05,
+      "loss": 0.9432,
+      "step": 8078
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.352044801890738e-05,
+      "loss": 1.0044,
+      "step": 8079
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3518714988009315e-05,
+      "loss": 0.9789,
+      "step": 8080
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.351698175989817e-05,
+      "loss": 0.9083,
+      "step": 8081
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.351524833459242e-05,
+      "loss": 1.0052,
+      "step": 8082
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3513514712110513e-05,
+      "loss": 1.0058,
+      "step": 8083
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.351178089247091e-05,
+      "loss": 0.9399,
+      "step": 8084
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.351004687569209e-05,
+      "loss": 0.8907,
+      "step": 8085
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.35083126617925e-05,
+      "loss": 0.9584,
+      "step": 8086
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.350657825079062e-05,
+      "loss": 1.0338,
+      "step": 8087
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.350484364270492e-05,
+      "loss": 0.9736,
+      "step": 8088
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3503108837553865e-05,
+      "loss": 0.9995,
+      "step": 8089
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.350137383535593e-05,
+      "loss": 1.058,
+      "step": 8090
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.34996386361296e-05,
+      "loss": 0.9708,
+      "step": 8091
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3497903239893355e-05,
+      "loss": 0.9574,
+      "step": 8092
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3496167646665666e-05,
+      "loss": 1.0255,
+      "step": 8093
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3494431856465025e-05,
+      "loss": 0.949,
+      "step": 8094
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3492695869309904e-05,
+      "loss": 0.9405,
+      "step": 8095
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3490959685218804e-05,
+      "loss": 0.9124,
+      "step": 8096
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3489223304210216e-05,
+      "loss": 0.9562,
+      "step": 8097
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.348748672630262e-05,
+      "loss": 0.8862,
+      "step": 8098
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.348574995151451e-05,
+      "loss": 0.9269,
+      "step": 8099
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.348401297986439e-05,
+      "loss": 0.9188,
+      "step": 8100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.348227581137075e-05,
+      "loss": 0.9843,
+      "step": 8101
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.34805384460521e-05,
+      "loss": 0.9531,
+      "step": 8102
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.347880088392693e-05,
+      "loss": 0.9553,
+      "step": 8103
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.347706312501375e-05,
+      "loss": 0.9229,
+      "step": 8104
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.347532516933107e-05,
+      "loss": 0.9051,
+      "step": 8105
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.347358701689739e-05,
+      "loss": 0.9775,
+      "step": 8106
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.347184866773123e-05,
+      "loss": 0.9766,
+      "step": 8107
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3470110121851087e-05,
+      "loss": 0.9289,
+      "step": 8108
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.346837137927549e-05,
+      "loss": 0.9694,
+      "step": 8109
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3466632440022946e-05,
+      "loss": 0.8281,
+      "step": 8110
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.346489330411198e-05,
+      "loss": 0.8929,
+      "step": 8111
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.346315397156111e-05,
+      "loss": 0.9596,
+      "step": 8112
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3461414442388864e-05,
+      "loss": 0.9842,
+      "step": 8113
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3459674716613755e-05,
+      "loss": 0.9694,
+      "step": 8114
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.345793479425433e-05,
+      "loss": 0.9657,
+      "step": 8115
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3456194675329085e-05,
+      "loss": 0.9019,
+      "step": 8116
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.345445435985659e-05,
+      "loss": 0.893,
+      "step": 8117
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3452713847855356e-05,
+      "loss": 0.9637,
+      "step": 8118
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.345097313934392e-05,
+      "loss": 1.0271,
+      "step": 8119
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.344923223434082e-05,
+      "loss": 0.8337,
+      "step": 8120
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.34474911328646e-05,
+      "loss": 0.8689,
+      "step": 8121
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3445749834933804e-05,
+      "loss": 0.9428,
+      "step": 8122
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.344400834056696e-05,
+      "loss": 0.9578,
+      "step": 8123
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.344226664978263e-05,
+      "loss": 0.9494,
+      "step": 8124
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.344052476259935e-05,
+      "loss": 0.9654,
+      "step": 8125
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.34387826790357e-05,
+      "loss": 0.9642,
+      "step": 8126
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.343704039911018e-05,
+      "loss": 0.9355,
+      "step": 8127
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.343529792284139e-05,
+      "loss": 0.8617,
+      "step": 8128
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.343355525024787e-05,
+      "loss": 1.0171,
+      "step": 8129
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.343181238134817e-05,
+      "loss": 0.9536,
+      "step": 8130
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.343006931616086e-05,
+      "loss": 0.9136,
+      "step": 8131
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.34283260547045e-05,
+      "loss": 0.9076,
+      "step": 8132
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3426582596997664e-05,
+      "loss": 1.0147,
+      "step": 8133
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3424838943058906e-05,
+      "loss": 0.9355,
+      "step": 8134
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3423095092906796e-05,
+      "loss": 0.9906,
+      "step": 8135
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3421351046559914e-05,
+      "loss": 0.9323,
+      "step": 8136
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.341960680403682e-05,
+      "loss": 1.0297,
+      "step": 8137
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.34178623653561e-05,
+      "loss": 0.9037,
+      "step": 8138
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3416117730536324e-05,
+      "loss": 0.9416,
+      "step": 8139
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.341437289959608e-05,
+      "loss": 0.9086,
+      "step": 8140
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.341262787255395e-05,
+      "loss": 0.9276,
+      "step": 8141
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3410882649428494e-05,
+      "loss": 0.9613,
+      "step": 8142
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3409137230238326e-05,
+      "loss": 0.9508,
+      "step": 8143
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.340739161500202e-05,
+      "loss": 0.9661,
+      "step": 8144
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.340564580373818e-05,
+      "loss": 0.9694,
+      "step": 8145
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3403899796465375e-05,
+      "loss": 0.9667,
+      "step": 8146
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.340215359320221e-05,
+      "loss": 0.9631,
+      "step": 8147
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.340040719396728e-05,
+      "loss": 0.957,
+      "step": 8148
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.33986605987792e-05,
+      "loss": 0.9328,
+      "step": 8149
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.339691380765654e-05,
+      "loss": 0.9574,
+      "step": 8150
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.339516682061792e-05,
+      "loss": 0.983,
+      "step": 8151
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.339341963768194e-05,
+      "loss": 1.02,
+      "step": 8152
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.339167225886721e-05,
+      "loss": 0.976,
+      "step": 8153
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.338992468419234e-05,
+      "loss": 1.0601,
+      "step": 8154
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3388176913675934e-05,
+      "loss": 0.9274,
+      "step": 8155
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.33864289473366e-05,
+      "loss": 0.9896,
+      "step": 8156
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.338468078519298e-05,
+      "loss": 0.9146,
+      "step": 8157
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.338293242726365e-05,
+      "loss": 1.0124,
+      "step": 8158
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3381183873567255e-05,
+      "loss": 0.9242,
+      "step": 8159
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.337943512412242e-05,
+      "loss": 0.9724,
+      "step": 8160
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.337768617894775e-05,
+      "loss": 0.9561,
+      "step": 8161
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.337593703806189e-05,
+      "loss": 0.9902,
+      "step": 8162
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.337418770148345e-05,
+      "loss": 0.9471,
+      "step": 8163
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3372438169231066e-05,
+      "loss": 0.9171,
+      "step": 8164
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.337068844132337e-05,
+      "loss": 0.9087,
+      "step": 8165
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.336893851777899e-05,
+      "loss": 0.9093,
+      "step": 8166
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.336718839861658e-05,
+      "loss": 0.9187,
+      "step": 8167
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.336543808385475e-05,
+      "loss": 0.9299,
+      "step": 8168
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.336368757351217e-05,
+      "loss": 0.9681,
+      "step": 8169
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.336193686760745e-05,
+      "loss": 0.9106,
+      "step": 8170
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.336018596615926e-05,
+      "loss": 0.9278,
+      "step": 8171
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3358434869186235e-05,
+      "loss": 0.9443,
+      "step": 8172
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.335668357670702e-05,
+      "loss": 0.8945,
+      "step": 8173
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3354932088740276e-05,
+      "loss": 0.9273,
+      "step": 8174
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3353180405304644e-05,
+      "loss": 0.9522,
+      "step": 8175
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.335142852641879e-05,
+      "loss": 0.8862,
+      "step": 8176
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.334967645210136e-05,
+      "loss": 0.9408,
+      "step": 8177
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3347924182371016e-05,
+      "loss": 0.8958,
+      "step": 8178
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.334617171724641e-05,
+      "loss": 0.9055,
+      "step": 8179
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3344419056746223e-05,
+      "loss": 0.8957,
+      "step": 8180
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3342666200889105e-05,
+      "loss": 0.8698,
+      "step": 8181
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3340913149693726e-05,
+      "loss": 0.8955,
+      "step": 8182
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.333915990317877e-05,
+      "loss": 0.9618,
+      "step": 8183
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.333740646136288e-05,
+      "loss": 0.9483,
+      "step": 8184
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.333565282426476e-05,
+      "loss": 0.9542,
+      "step": 8185
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3333898991903055e-05,
+      "loss": 0.9547,
+      "step": 8186
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3332144964296465e-05,
+      "loss": 0.9746,
+      "step": 8187
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3330390741463654e-05,
+      "loss": 0.984,
+      "step": 8188
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.332863632342332e-05,
+      "loss": 0.9561,
+      "step": 8189
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3326881710194126e-05,
+      "loss": 1.0054,
+      "step": 8190
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.332512690179478e-05,
+      "loss": 0.9105,
+      "step": 8191
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.332337189824395e-05,
+      "loss": 0.9362,
+      "step": 8192
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.332161669956033e-05,
+      "loss": 0.8981,
+      "step": 8193
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.331986130576262e-05,
+      "loss": 0.9317,
+      "step": 8194
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3318105716869504e-05,
+      "loss": 0.8585,
+      "step": 8195
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.331634993289969e-05,
+      "loss": 0.9226,
+      "step": 8196
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.331459395387187e-05,
+      "loss": 0.9422,
+      "step": 8197
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3312837779804746e-05,
+      "loss": 0.9627,
+      "step": 8198
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3311081410717014e-05,
+      "loss": 0.9175,
+      "step": 8199
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3309324846627376e-05,
+      "loss": 0.8912,
+      "step": 8200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.330756808755455e-05,
+      "loss": 0.9729,
+      "step": 8201
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.330581113351724e-05,
+      "loss": 0.9312,
+      "step": 8202
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.330405398453415e-05,
+      "loss": 0.9611,
+      "step": 8203
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3302296640624004e-05,
+      "loss": 0.8766,
+      "step": 8204
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.33005391018055e-05,
+      "loss": 0.9617,
+      "step": 8205
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.329878136809738e-05,
+      "loss": 0.9019,
+      "step": 8206
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3297023439518334e-05,
+      "loss": 1.0267,
+      "step": 8207
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3295265316087105e-05,
+      "loss": 0.9231,
+      "step": 8208
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.329350699782241e-05,
+      "loss": 0.9029,
+      "step": 8209
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3291748484742965e-05,
+      "loss": 0.9362,
+      "step": 8210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.328998977686751e-05,
+      "loss": 1.0409,
+      "step": 8211
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.328823087421476e-05,
+      "loss": 0.9229,
+      "step": 8212
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.328647177680346e-05,
+      "loss": 1.0246,
+      "step": 8213
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.328471248465233e-05,
+      "loss": 0.8857,
+      "step": 8214
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.328295299778013e-05,
+      "loss": 0.9465,
+      "step": 8215
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3281193316205574e-05,
+      "loss": 0.9278,
+      "step": 8216
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.327943343994739e-05,
+      "loss": 0.8972,
+      "step": 8217
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.327767336902436e-05,
+      "loss": 0.917,
+      "step": 8218
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3275913103455194e-05,
+      "loss": 0.9549,
+      "step": 8219
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.327415264325865e-05,
+      "loss": 0.8752,
+      "step": 8220
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.327239198845348e-05,
+      "loss": 1.0422,
+      "step": 8221
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.327063113905842e-05,
+      "loss": 0.9299,
+      "step": 8222
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.326887009509224e-05,
+      "loss": 0.9709,
+      "step": 8223
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.326710885657368e-05,
+      "loss": 1.0094,
+      "step": 8224
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.32653474235215e-05,
+      "loss": 0.9573,
+      "step": 8225
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.326358579595446e-05,
+      "loss": 1.0038,
+      "step": 8226
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.326182397389132e-05,
+      "loss": 0.9332,
+      "step": 8227
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.326006195735084e-05,
+      "loss": 0.8968,
+      "step": 8228
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3258299746351795e-05,
+      "loss": 0.9942,
+      "step": 8229
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.325653734091293e-05,
+      "loss": 0.8206,
+      "step": 8230
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3254774741053025e-05,
+      "loss": 0.8643,
+      "step": 8231
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.325301194679087e-05,
+      "loss": 1.0005,
+      "step": 8232
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.32512489581452e-05,
+      "loss": 0.9115,
+      "step": 8233
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.324948577513482e-05,
+      "loss": 0.9515,
+      "step": 8234
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.324772239777849e-05,
+      "loss": 0.8836,
+      "step": 8235
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3245958826095e-05,
+      "loss": 0.9877,
+      "step": 8236
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.324419506010312e-05,
+      "loss": 0.9574,
+      "step": 8237
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.324243109982164e-05,
+      "loss": 0.9037,
+      "step": 8238
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.324066694526935e-05,
+      "loss": 0.9111,
+      "step": 8239
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3238902596465025e-05,
+      "loss": 0.9431,
+      "step": 8240
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.323713805342747e-05,
+      "loss": 0.9485,
+      "step": 8241
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.323537331617545e-05,
+      "loss": 0.9121,
+      "step": 8242
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.323360838472779e-05,
+      "loss": 0.9234,
+      "step": 8243
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3231843259103265e-05,
+      "loss": 0.9598,
+      "step": 8244
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.323007793932068e-05,
+      "loss": 0.9199,
+      "step": 8245
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.322831242539883e-05,
+      "loss": 0.9513,
+      "step": 8246
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.322654671735652e-05,
+      "loss": 0.9511,
+      "step": 8247
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.322478081521256e-05,
+      "loss": 0.8563,
+      "step": 8248
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3223014718985744e-05,
+      "loss": 0.9555,
+      "step": 8249
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3221248428694884e-05,
+      "loss": 0.9443,
+      "step": 8250
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3219481944358784e-05,
+      "loss": 0.8917,
+      "step": 8251
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.321771526599627e-05,
+      "loss": 0.8949,
+      "step": 8252
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.321594839362615e-05,
+      "loss": 0.8947,
+      "step": 8253
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.321418132726723e-05,
+      "loss": 0.9255,
+      "step": 8254
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.321241406693834e-05,
+      "loss": 0.9837,
+      "step": 8255
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.321064661265831e-05,
+      "loss": 0.9221,
+      "step": 8256
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3208878964445936e-05,
+      "loss": 0.9173,
+      "step": 8257
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.320711112232007e-05,
+      "loss": 0.9905,
+      "step": 8258
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.320534308629951e-05,
+      "loss": 0.8543,
+      "step": 8259
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.320357485640309e-05,
+      "loss": 0.9514,
+      "step": 8260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.320180643264967e-05,
+      "loss": 0.9945,
+      "step": 8261
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.320003781505806e-05,
+      "loss": 0.8411,
+      "step": 8262
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3198269003647085e-05,
+      "loss": 0.9963,
+      "step": 8263
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3196499998435604e-05,
+      "loss": 1.0034,
+      "step": 8264
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.319473079944244e-05,
+      "loss": 0.9242,
+      "step": 8265
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.319296140668644e-05,
+      "loss": 1.0025,
+      "step": 8266
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.319119182018645e-05,
+      "loss": 1.0594,
+      "step": 8267
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3189422039961304e-05,
+      "loss": 0.9137,
+      "step": 8268
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.318765206602986e-05,
+      "loss": 0.9432,
+      "step": 8269
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.318588189841096e-05,
+      "loss": 1.0043,
+      "step": 8270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.318411153712347e-05,
+      "loss": 0.9742,
+      "step": 8271
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.318234098218622e-05,
+      "loss": 0.9256,
+      "step": 8272
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3180570233618086e-05,
+      "loss": 0.8392,
+      "step": 8273
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.317879929143791e-05,
+      "loss": 0.8299,
+      "step": 8274
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3177028155664564e-05,
+      "loss": 0.964,
+      "step": 8275
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.31752568263169e-05,
+      "loss": 0.9389,
+      "step": 8276
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.317348530341378e-05,
+      "loss": 1.0553,
+      "step": 8277
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.317171358697408e-05,
+      "loss": 1.0019,
+      "step": 8278
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.316994167701667e-05,
+      "loss": 0.8952,
+      "step": 8279
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3168169573560405e-05,
+      "loss": 0.9366,
+      "step": 8280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.316639727662416e-05,
+      "loss": 0.9833,
+      "step": 8281
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.316462478622682e-05,
+      "loss": 0.9281,
+      "step": 8282
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.316285210238725e-05,
+      "loss": 0.9419,
+      "step": 8283
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3161079225124334e-05,
+      "loss": 0.9106,
+      "step": 8284
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3159306154456954e-05,
+      "loss": 0.9129,
+      "step": 8285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3157532890403984e-05,
+      "loss": 0.9533,
+      "step": 8286
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.315575943298431e-05,
+      "loss": 0.8684,
+      "step": 8287
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.315398578221682e-05,
+      "loss": 0.9111,
+      "step": 8288
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.315221193812041e-05,
+      "loss": 0.868,
+      "step": 8289
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.315043790071396e-05,
+      "loss": 0.91,
+      "step": 8290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3148663670016364e-05,
+      "loss": 0.914,
+      "step": 8291
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.314688924604652e-05,
+      "loss": 0.8366,
+      "step": 8292
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.314511462882333e-05,
+      "loss": 0.9441,
+      "step": 8293
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.314333981836567e-05,
+      "loss": 0.9111,
+      "step": 8294
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3141564814692466e-05,
+      "loss": 0.8462,
+      "step": 8295
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3139789617822606e-05,
+      "loss": 0.8513,
+      "step": 8296
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3138014227775004e-05,
+      "loss": 0.9413,
+      "step": 8297
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.313623864456856e-05,
+      "loss": 1.0055,
+      "step": 8298
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.313446286822219e-05,
+      "loss": 0.9054,
+      "step": 8299
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3132686898754795e-05,
+      "loss": 0.9191,
+      "step": 8300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3130910736185294e-05,
+      "loss": 0.8474,
+      "step": 8301
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.31291343805326e-05,
+      "loss": 0.8681,
+      "step": 8302
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3127357831815626e-05,
+      "loss": 0.8979,
+      "step": 8303
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3125581090053305e-05,
+      "loss": 0.9049,
+      "step": 8304
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.312380415526455e-05,
+      "loss": 0.9767,
+      "step": 8305
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.312202702746828e-05,
+      "loss": 0.917,
+      "step": 8306
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3120249706683425e-05,
+      "loss": 0.9181,
+      "step": 8307
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.311847219292892e-05,
+      "loss": 0.9272,
+      "step": 8308
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3116694486223675e-05,
+      "loss": 0.9572,
+      "step": 8309
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.311491658658663e-05,
+      "loss": 0.8752,
+      "step": 8310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.311313849403673e-05,
+      "loss": 0.9628,
+      "step": 8311
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.31113602085929e-05,
+      "loss": 0.8747,
+      "step": 8312
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3109581730274076e-05,
+      "loss": 0.8192,
+      "step": 8313
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3107803059099206e-05,
+      "loss": 0.9024,
+      "step": 8314
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.310602419508722e-05,
+      "loss": 0.8569,
+      "step": 8315
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3104245138257084e-05,
+      "loss": 0.9915,
+      "step": 8316
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.310246588862772e-05,
+      "loss": 1.0549,
+      "step": 8317
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3100686446218075e-05,
+      "loss": 0.9105,
+      "step": 8318
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3098906811047116e-05,
+      "loss": 0.937,
+      "step": 8319
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.309712698313379e-05,
+      "loss": 0.924,
+      "step": 8320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.309534696249705e-05,
+      "loss": 0.9244,
+      "step": 8321
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.309356674915585e-05,
+      "loss": 1.0246,
+      "step": 8322
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3091786343129145e-05,
+      "loss": 0.8946,
+      "step": 8323
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.30900057444359e-05,
+      "loss": 0.9613,
+      "step": 8324
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3088224953095076e-05,
+      "loss": 1.0093,
+      "step": 8325
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.308644396912563e-05,
+      "loss": 0.9408,
+      "step": 8326
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.308466279254655e-05,
+      "loss": 0.972,
+      "step": 8327
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.308288142337679e-05,
+      "loss": 0.9633,
+      "step": 8328
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3081099861635315e-05,
+      "loss": 0.9022,
+      "step": 8329
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3079318107341106e-05,
+      "loss": 0.9574,
+      "step": 8330
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.307753616051313e-05,
+      "loss": 0.9425,
+      "step": 8331
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3075754021170375e-05,
+      "loss": 0.9476,
+      "step": 8332
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.307397168933181e-05,
+      "loss": 0.9144,
+      "step": 8333
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.307218916501643e-05,
+      "loss": 0.9143,
+      "step": 8334
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.30704064482432e-05,
+      "loss": 0.9762,
+      "step": 8335
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.306862353903111e-05,
+      "loss": 0.892,
+      "step": 8336
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3066840437399156e-05,
+      "loss": 0.8926,
+      "step": 8337
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3065057143366314e-05,
+      "loss": 1.0117,
+      "step": 8338
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.306327365695158e-05,
+      "loss": 0.9606,
+      "step": 8339
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3061489978173953e-05,
+      "loss": 0.9896,
+      "step": 8340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.305970610705242e-05,
+      "loss": 0.9461,
+      "step": 8341
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.305792204360598e-05,
+      "loss": 0.966,
+      "step": 8342
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.305613778785363e-05,
+      "loss": 0.9481,
+      "step": 8343
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.305435333981438e-05,
+      "loss": 1.0136,
+      "step": 8344
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.305256869950723e-05,
+      "loss": 0.9205,
+      "step": 8345
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.305078386695118e-05,
+      "loss": 0.8635,
+      "step": 8346
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.304899884216525e-05,
+      "loss": 0.8994,
+      "step": 8347
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.304721362516843e-05,
+      "loss": 0.9108,
+      "step": 8348
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3045428215979743e-05,
+      "loss": 0.8927,
+      "step": 8349
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.30436426146182e-05,
+      "loss": 0.9142,
+      "step": 8350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.304185682110282e-05,
+      "loss": 0.9039,
+      "step": 8351
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.304007083545262e-05,
+      "loss": 0.9346,
+      "step": 8352
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.303828465768661e-05,
+      "loss": 0.8486,
+      "step": 8353
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.303649828782383e-05,
+      "loss": 0.9176,
+      "step": 8354
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.303471172588329e-05,
+      "loss": 1.0529,
+      "step": 8355
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3032924971884016e-05,
+      "loss": 0.8666,
+      "step": 8356
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.303113802584504e-05,
+      "loss": 0.9023,
+      "step": 8357
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.302935088778539e-05,
+      "loss": 0.9593,
+      "step": 8358
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.302756355772411e-05,
+      "loss": 0.8999,
+      "step": 8359
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.302577603568021e-05,
+      "loss": 0.9023,
+      "step": 8360
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3023988321672736e-05,
+      "loss": 0.9156,
+      "step": 8361
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.302220041572074e-05,
+      "loss": 0.8856,
+      "step": 8362
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.302041231784325e-05,
+      "loss": 1.0034,
+      "step": 8363
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.301862402805931e-05,
+      "loss": 0.9416,
+      "step": 8364
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3016835546387954e-05,
+      "loss": 0.8901,
+      "step": 8365
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.301504687284824e-05,
+      "loss": 0.8492,
+      "step": 8366
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3013258007459225e-05,
+      "loss": 0.9208,
+      "step": 8367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.301146895023993e-05,
+      "loss": 0.9414,
+      "step": 8368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.300967970120944e-05,
+      "loss": 0.8831,
+      "step": 8369
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3007890260386796e-05,
+      "loss": 0.9076,
+      "step": 8370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.300610062779105e-05,
+      "loss": 0.8705,
+      "step": 8371
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.300431080344126e-05,
+      "loss": 0.8997,
+      "step": 8372
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.300252078735649e-05,
+      "loss": 1.0745,
+      "step": 8373
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.300073057955582e-05,
+      "loss": 0.9672,
+      "step": 8374
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.2998940180058275e-05,
+      "loss": 0.9268,
+      "step": 8375
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.2997149588882956e-05,
+      "loss": 0.9631,
+      "step": 8376
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.2995358806048916e-05,
+      "loss": 0.9221,
+      "step": 8377
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.299356783157523e-05,
+      "loss": 0.9821,
+      "step": 8378
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.2991776665480974e-05,
+      "loss": 0.9028,
+      "step": 8379
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.298998530778523e-05,
+      "loss": 0.9072,
+      "step": 8380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2988193758507044e-05,
+      "loss": 0.934,
+      "step": 8381
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.298640201766553e-05,
+      "loss": 0.9166,
+      "step": 8382
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.298461008527975e-05,
+      "loss": 0.861,
+      "step": 8383
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.298281796136879e-05,
+      "loss": 0.8905,
+      "step": 8384
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.298102564595173e-05,
+      "loss": 0.8931,
+      "step": 8385
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.297923313904768e-05,
+      "loss": 1.0154,
+      "step": 8386
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2977440440675697e-05,
+      "loss": 0.8946,
+      "step": 8387
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.297564755085489e-05,
+      "loss": 0.932,
+      "step": 8388
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2973854469604356e-05,
+      "loss": 0.8878,
+      "step": 8389
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2972061196943185e-05,
+      "loss": 0.8851,
+      "step": 8390
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.297026773289047e-05,
+      "loss": 0.8407,
+      "step": 8391
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.296847407746531e-05,
+      "loss": 0.9628,
+      "step": 8392
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2966680230686816e-05,
+      "loss": 0.9346,
+      "step": 8393
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.296488619257408e-05,
+      "loss": 0.9391,
+      "step": 8394
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.296309196314622e-05,
+      "loss": 1.0248,
+      "step": 8395
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.296129754242233e-05,
+      "loss": 0.9707,
+      "step": 8396
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2959502930421525e-05,
+      "loss": 0.9584,
+      "step": 8397
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.295770812716292e-05,
+      "loss": 0.9477,
+      "step": 8398
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.295591313266563e-05,
+      "loss": 0.9137,
+      "step": 8399
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2954117946948766e-05,
+      "loss": 0.9714,
+      "step": 8400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2952322570031434e-05,
+      "loss": 0.929,
+      "step": 8401
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2950527001932775e-05,
+      "loss": 0.9624,
+      "step": 8402
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.29487312426719e-05,
+      "loss": 0.968,
+      "step": 8403
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.294693529226794e-05,
+      "loss": 0.984,
+      "step": 8404
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.294513915074001e-05,
+      "loss": 0.8555,
+      "step": 8405
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.294334281810724e-05,
+      "loss": 0.9341,
+      "step": 8406
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.294154629438877e-05,
+      "loss": 0.894,
+      "step": 8407
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.293974957960372e-05,
+      "loss": 0.9486,
+      "step": 8408
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.293795267377122e-05,
+      "loss": 0.9254,
+      "step": 8409
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.293615557691043e-05,
+      "loss": 0.9113,
+      "step": 8410
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.293435828904046e-05,
+      "loss": 0.8652,
+      "step": 8411
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.293256081018047e-05,
+      "loss": 0.9564,
+      "step": 8412
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.293076314034959e-05,
+      "loss": 0.9467,
+      "step": 8413
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.292896527956697e-05,
+      "loss": 0.9157,
+      "step": 8414
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2927167227851755e-05,
+      "loss": 0.9445,
+      "step": 8415
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.292536898522309e-05,
+      "loss": 0.9139,
+      "step": 8416
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.292357055170013e-05,
+      "loss": 1.0317,
+      "step": 8417
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.292177192730203e-05,
+      "loss": 0.8706,
+      "step": 8418
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.291997311204794e-05,
+      "loss": 0.9059,
+      "step": 8419
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.291817410595701e-05,
+      "loss": 0.9196,
+      "step": 8420
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.291637490904841e-05,
+      "loss": 0.9392,
+      "step": 8421
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.291457552134128e-05,
+      "loss": 0.9486,
+      "step": 8422
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.291277594285481e-05,
+      "loss": 0.9485,
+      "step": 8423
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2910976173608155e-05,
+      "loss": 0.9805,
+      "step": 8424
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.290917621362047e-05,
+      "loss": 0.9134,
+      "step": 8425
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.290737606291093e-05,
+      "loss": 1.0128,
+      "step": 8426
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.290557572149871e-05,
+      "loss": 1.0002,
+      "step": 8427
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2903775189402976e-05,
+      "loss": 0.8947,
+      "step": 8428
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.290197446664291e-05,
+      "loss": 0.9679,
+      "step": 8429
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2900173553237685e-05,
+      "loss": 0.9139,
+      "step": 8430
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2898372449206474e-05,
+      "loss": 0.9421,
+      "step": 8431
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.289657115456847e-05,
+      "loss": 0.9102,
+      "step": 8432
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2894769669342846e-05,
+      "loss": 0.9508,
+      "step": 8433
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2892967993548784e-05,
+      "loss": 0.9246,
+      "step": 8434
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2891166127205484e-05,
+      "loss": 0.9761,
+      "step": 8435
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.288936407033212e-05,
+      "loss": 0.9395,
+      "step": 8436
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.288756182294789e-05,
+      "loss": 0.8896,
+      "step": 8437
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.288575938507199e-05,
+      "loss": 0.9263,
+      "step": 8438
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.288395675672361e-05,
+      "loss": 0.9221,
+      "step": 8439
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.288215393792194e-05,
+      "loss": 0.9452,
+      "step": 8440
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2880350928686205e-05,
+      "loss": 0.8749,
+      "step": 8441
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.287854772903558e-05,
+      "loss": 0.8789,
+      "step": 8442
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.287674433898927e-05,
+      "loss": 1.0549,
+      "step": 8443
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.287494075856648e-05,
+      "loss": 0.9578,
+      "step": 8444
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2873136987786445e-05,
+      "loss": 1.0129,
+      "step": 8445
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.287133302666834e-05,
+      "loss": 0.9701,
+      "step": 8446
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.286952887523138e-05,
+      "loss": 0.899,
+      "step": 8447
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.28677245334948e-05,
+      "loss": 0.8735,
+      "step": 8448
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2865920001477795e-05,
+      "loss": 0.859,
+      "step": 8449
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.286411527919959e-05,
+      "loss": 0.8564,
+      "step": 8450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.28623103666794e-05,
+      "loss": 0.9324,
+      "step": 8451
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.286050526393646e-05,
+      "loss": 0.9752,
+      "step": 8452
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2858699970989965e-05,
+      "loss": 1.0694,
+      "step": 8453
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2856894487859166e-05,
+      "loss": 0.9139,
+      "step": 8454
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.285508881456329e-05,
+      "loss": 0.9223,
+      "step": 8455
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.285328295112155e-05,
+      "loss": 0.8234,
+      "step": 8456
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2851476897553186e-05,
+      "loss": 0.9139,
+      "step": 8457
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2849670653877426e-05,
+      "loss": 0.9746,
+      "step": 8458
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.284786422011352e-05,
+      "loss": 0.9875,
+      "step": 8459
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.284605759628069e-05,
+      "loss": 0.8535,
+      "step": 8460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.284425078239818e-05,
+      "loss": 0.8805,
+      "step": 8461
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.284244377848523e-05,
+      "loss": 0.8811,
+      "step": 8462
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.28406365845611e-05,
+      "loss": 0.9623,
+      "step": 8463
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2838829200645015e-05,
+      "loss": 0.9,
+      "step": 8464
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.283702162675622e-05,
+      "loss": 0.9361,
+      "step": 8465
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.283521386291398e-05,
+      "loss": 1.0155,
+      "step": 8466
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.283340590913755e-05,
+      "loss": 0.8875,
+      "step": 8467
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.283159776544616e-05,
+      "loss": 0.9597,
+      "step": 8468
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.282978943185909e-05,
+      "loss": 0.9216,
+      "step": 8469
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.282798090839558e-05,
+      "loss": 0.9493,
+      "step": 8470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2826172195074896e-05,
+      "loss": 1.0808,
+      "step": 8471
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.28243632919163e-05,
+      "loss": 0.8396,
+      "step": 8472
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.282255419893907e-05,
+      "loss": 0.9852,
+      "step": 8473
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2820744916162445e-05,
+      "loss": 0.9188,
+      "step": 8474
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.281893544360571e-05,
+      "loss": 0.9857,
+      "step": 8475
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.281712578128812e-05,
+      "loss": 0.9157,
+      "step": 8476
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2815315929228975e-05,
+      "loss": 0.8549,
+      "step": 8477
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2813505887447525e-05,
+      "loss": 0.8912,
+      "step": 8478
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2811695655963045e-05,
+      "loss": 0.8663,
+      "step": 8479
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.280988523479483e-05,
+      "loss": 0.9249,
+      "step": 8480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.280807462396215e-05,
+      "loss": 0.9255,
+      "step": 8481
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.280626382348428e-05,
+      "loss": 0.8511,
+      "step": 8482
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.280445283338051e-05,
+      "loss": 0.9322,
+      "step": 8483
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.280264165367013e-05,
+      "loss": 0.9959,
+      "step": 8484
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.280083028437243e-05,
+      "loss": 0.9002,
+      "step": 8485
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.279901872550669e-05,
+      "loss": 0.9384,
+      "step": 8486
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.27972069770922e-05,
+      "loss": 0.8356,
+      "step": 8487
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.279539503914827e-05,
+      "loss": 0.8628,
+      "step": 8488
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.279358291169418e-05,
+      "loss": 0.8863,
+      "step": 8489
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2791770594749236e-05,
+      "loss": 0.9066,
+      "step": 8490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.278995808833274e-05,
+      "loss": 0.9325,
+      "step": 8491
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.278814539246399e-05,
+      "loss": 0.9681,
+      "step": 8492
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.278633250716229e-05,
+      "loss": 0.9139,
+      "step": 8493
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2784519432446954e-05,
+      "loss": 0.9375,
+      "step": 8494
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.278270616833727e-05,
+      "loss": 0.9505,
+      "step": 8495
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.278089271485257e-05,
+      "loss": 0.885,
+      "step": 8496
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.277907907201216e-05,
+      "loss": 0.9404,
+      "step": 8497
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2777265239835345e-05,
+      "loss": 1.0203,
+      "step": 8498
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.277545121834145e-05,
+      "loss": 0.9691,
+      "step": 8499
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2773637007549796e-05,
+      "loss": 0.9535,
+      "step": 8500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.277182260747969e-05,
+      "loss": 0.8952,
+      "step": 8501
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2770008018150473e-05,
+      "loss": 0.9417,
+      "step": 8502
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2768193239581456e-05,
+      "loss": 0.9556,
+      "step": 8503
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2766378271791965e-05,
+      "loss": 0.8938,
+      "step": 8504
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.276456311480133e-05,
+      "loss": 0.894,
+      "step": 8505
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2762747768628886e-05,
+      "loss": 1.0112,
+      "step": 8506
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.276093223329396e-05,
+      "loss": 0.8972,
+      "step": 8507
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.275911650881589e-05,
+      "loss": 0.9517,
+      "step": 8508
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2757300595214014e-05,
+      "loss": 0.9437,
+      "step": 8509
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.275548449250766e-05,
+      "loss": 1.0062,
+      "step": 8510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.275366820071618e-05,
+      "loss": 0.8583,
+      "step": 8511
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.275185171985892e-05,
+      "loss": 0.9669,
+      "step": 8512
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.27500350499552e-05,
+      "loss": 0.8737,
+      "step": 8513
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2748218191024396e-05,
+      "loss": 0.9036,
+      "step": 8514
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.274640114308584e-05,
+      "loss": 0.8789,
+      "step": 8515
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.274458390615889e-05,
+      "loss": 0.8807,
+      "step": 8516
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2742766480262884e-05,
+      "loss": 0.9357,
+      "step": 8517
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2740948865417195e-05,
+      "loss": 0.8659,
+      "step": 8518
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2739131061641166e-05,
+      "loss": 0.9471,
+      "step": 8519
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.273731306895417e-05,
+      "loss": 0.8972,
+      "step": 8520
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.273549488737555e-05,
+      "loss": 0.9568,
+      "step": 8521
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.273367651692468e-05,
+      "loss": 0.9359,
+      "step": 8522
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2731857957620924e-05,
+      "loss": 0.8931,
+      "step": 8523
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.273003920948364e-05,
+      "loss": 0.9535,
+      "step": 8524
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2728220272532204e-05,
+      "loss": 0.9378,
+      "step": 8525
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2726401146785986e-05,
+      "loss": 0.9031,
+      "step": 8526
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.272458183226437e-05,
+      "loss": 0.8882,
+      "step": 8527
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2722762328986695e-05,
+      "loss": 0.9475,
+      "step": 8528
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2720942636972376e-05,
+      "loss": 0.9124,
+      "step": 8529
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.271912275624077e-05,
+      "loss": 0.914,
+      "step": 8530
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2717302686811275e-05,
+      "loss": 0.9354,
+      "step": 8531
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2715482428703255e-05,
+      "loss": 1.022,
+      "step": 8532
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.27136619819361e-05,
+      "loss": 0.9362,
+      "step": 8533
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.27118413465292e-05,
+      "loss": 0.922,
+      "step": 8534
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.271002052250195e-05,
+      "loss": 0.8943,
+      "step": 8535
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.270819950987373e-05,
+      "loss": 0.8881,
+      "step": 8536
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.270637830866393e-05,
+      "loss": 0.9101,
+      "step": 8537
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.270455691889196e-05,
+      "loss": 0.9063,
+      "step": 8538
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.27027353405772e-05,
+      "loss": 0.9691,
+      "step": 8539
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.270091357373905e-05,
+      "loss": 0.9277,
+      "step": 8540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2699091618396936e-05,
+      "loss": 0.9112,
+      "step": 8541
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.269726947457023e-05,
+      "loss": 0.9078,
+      "step": 8542
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2695447142278347e-05,
+      "loss": 0.9044,
+      "step": 8543
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2693624621540694e-05,
+      "loss": 0.8783,
+      "step": 8544
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.269180191237669e-05,
+      "loss": 0.8923,
+      "step": 8545
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.268997901480573e-05,
+      "loss": 1.0837,
+      "step": 8546
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.268815592884724e-05,
+      "loss": 0.9731,
+      "step": 8547
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.268633265452062e-05,
+      "loss": 0.9709,
+      "step": 8548
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.26845091918453e-05,
+      "loss": 0.9257,
+      "step": 8549
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2682685540840684e-05,
+      "loss": 0.9142,
+      "step": 8550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.268086170152621e-05,
+      "loss": 0.9053,
+      "step": 8551
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.26790376739213e-05,
+      "loss": 0.8635,
+      "step": 8552
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.267721345804536e-05,
+      "loss": 0.9886,
+      "step": 8553
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.267538905391783e-05,
+      "loss": 0.9793,
+      "step": 8554
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2673564461558145e-05,
+      "loss": 0.9491,
+      "step": 8555
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2671739680985724e-05,
+      "loss": 0.9153,
+      "step": 8556
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2669914712220004e-05,
+      "loss": 0.8885,
+      "step": 8557
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.266808955528042e-05,
+      "loss": 0.9611,
+      "step": 8558
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.266626421018641e-05,
+      "loss": 0.9005,
+      "step": 8559
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.266443867695742e-05,
+      "loss": 0.9635,
+      "step": 8560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.266261295561287e-05,
+      "loss": 0.9039,
+      "step": 8561
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.266078704617222e-05,
+      "loss": 0.8128,
+      "step": 8562
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.26589609486549e-05,
+      "loss": 0.913,
+      "step": 8563
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.265713466308038e-05,
+      "loss": 0.9417,
+      "step": 8564
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.265530818946809e-05,
+      "loss": 0.8804,
+      "step": 8565
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2653481527837494e-05,
+      "loss": 0.9513,
+      "step": 8566
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.265165467820803e-05,
+      "loss": 0.9342,
+      "step": 8567
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.264982764059916e-05,
+      "loss": 0.8511,
+      "step": 8568
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2648000415030344e-05,
+      "loss": 1.0083,
+      "step": 8569
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.264617300152104e-05,
+      "loss": 0.9383,
+      "step": 8570
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2644345400090704e-05,
+      "loss": 0.9238,
+      "step": 8571
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2642517610758796e-05,
+      "loss": 0.8712,
+      "step": 8572
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2640689633544794e-05,
+      "loss": 0.9693,
+      "step": 8573
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.263886146846815e-05,
+      "loss": 0.9061,
+      "step": 8574
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2637033115548354e-05,
+      "loss": 0.8529,
+      "step": 8575
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.263520457480485e-05,
+      "loss": 0.9171,
+      "step": 8576
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.263337584625713e-05,
+      "loss": 0.9235,
+      "step": 8577
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.263154692992466e-05,
+      "loss": 0.8544,
+      "step": 8578
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.262971782582692e-05,
+      "loss": 0.8286,
+      "step": 8579
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.262788853398339e-05,
+      "loss": 0.949,
+      "step": 8580
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.262605905441355e-05,
+      "loss": 0.9231,
+      "step": 8581
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.262422938713688e-05,
+      "loss": 0.8741,
+      "step": 8582
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.262239953217286e-05,
+      "loss": 0.8853,
+      "step": 8583
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2620569489540996e-05,
+      "loss": 0.9869,
+      "step": 8584
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.261873925926076e-05,
+      "loss": 0.9745,
+      "step": 8585
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2616908841351644e-05,
+      "loss": 0.9443,
+      "step": 8586
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.261507823583315e-05,
+      "loss": 0.8829,
+      "step": 8587
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2613247442724764e-05,
+      "loss": 0.9023,
+      "step": 8588
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.261141646204599e-05,
+      "loss": 0.9841,
+      "step": 8589
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.260958529381632e-05,
+      "loss": 0.942,
+      "step": 8590
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.260775393805525e-05,
+      "loss": 0.9705,
+      "step": 8591
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.26059223947823e-05,
+      "loss": 0.8822,
+      "step": 8592
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.260409066401696e-05,
+      "loss": 0.9714,
+      "step": 8593
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2602258745778746e-05,
+      "loss": 0.9192,
+      "step": 8594
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.260042664008716e-05,
+      "loss": 0.84,
+      "step": 8595
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2598594346961716e-05,
+      "loss": 0.8997,
+      "step": 8596
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.259676186642193e-05,
+      "loss": 0.9679,
+      "step": 8597
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.259492919848732e-05,
+      "loss": 0.9677,
+      "step": 8598
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2593096343177384e-05,
+      "loss": 0.8332,
+      "step": 8599
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2591263300511656e-05,
+      "loss": 0.9146,
+      "step": 8600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2589430070509656e-05,
+      "loss": 0.8276,
+      "step": 8601
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.25875966531909e-05,
+      "loss": 0.8503,
+      "step": 8602
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.258576304857492e-05,
+      "loss": 0.8695,
+      "step": 8603
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.258392925668124e-05,
+      "loss": 0.9093,
+      "step": 8604
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2582095277529396e-05,
+      "loss": 0.9351,
+      "step": 8605
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2580261111138905e-05,
+      "loss": 0.9041,
+      "step": 8606
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.25784267575293e-05,
+      "loss": 0.9188,
+      "step": 8607
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2576592216720135e-05,
+      "loss": 0.9501,
+      "step": 8608
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.257475748873093e-05,
+      "loss": 0.8954,
+      "step": 8609
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.257292257358122e-05,
+      "loss": 1.0803,
+      "step": 8610
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.257108747129056e-05,
+      "loss": 0.9226,
+      "step": 8611
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.256925218187848e-05,
+      "loss": 0.8498,
+      "step": 8612
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.256741670536454e-05,
+      "loss": 0.9412,
+      "step": 8613
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.256558104176828e-05,
+      "loss": 0.8853,
+      "step": 8614
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.256374519110923e-05,
+      "loss": 0.856,
+      "step": 8615
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2561909153406965e-05,
+      "loss": 0.8514,
+      "step": 8616
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.256007292868103e-05,
+      "loss": 0.8757,
+      "step": 8617
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2558236516950984e-05,
+      "loss": 0.961,
+      "step": 8618
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2556399918236375e-05,
+      "loss": 0.8439,
+      "step": 8619
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2554563132556766e-05,
+      "loss": 0.8667,
+      "step": 8620
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.255272615993172e-05,
+      "loss": 0.8762,
+      "step": 8621
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.255088900038079e-05,
+      "loss": 0.9113,
+      "step": 8622
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.254905165392355e-05,
+      "loss": 0.95,
+      "step": 8623
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.254721412057957e-05,
+      "loss": 0.904,
+      "step": 8624
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.254537640036841e-05,
+      "loss": 0.8871,
+      "step": 8625
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.254353849330964e-05,
+      "loss": 0.9165,
+      "step": 8626
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.254170039942284e-05,
+      "loss": 0.9382,
+      "step": 8627
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.253986211872757e-05,
+      "loss": 0.9548,
+      "step": 8628
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.253802365124343e-05,
+      "loss": 0.9058,
+      "step": 8629
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2536184996989984e-05,
+      "loss": 0.8914,
+      "step": 8630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.253434615598681e-05,
+      "loss": 0.8895,
+      "step": 8631
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.253250712825349e-05,
+      "loss": 0.9214,
+      "step": 8632
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2530667913809615e-05,
+      "loss": 0.931,
+      "step": 8633
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2528828512674784e-05,
+      "loss": 0.9084,
+      "step": 8634
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.252698892486856e-05,
+      "loss": 0.9117,
+      "step": 8635
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2525149150410545e-05,
+      "loss": 0.9745,
+      "step": 8636
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.252330918932034e-05,
+      "loss": 1.0186,
+      "step": 8637
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.252146904161752e-05,
+      "loss": 0.9129,
+      "step": 8638
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2519628707321695e-05,
+      "loss": 1.0288,
+      "step": 8639
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.251778818645246e-05,
+      "loss": 0.8617,
+      "step": 8640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.251594747902942e-05,
+      "loss": 0.9037,
+      "step": 8641
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2514106585072175e-05,
+      "loss": 0.9164,
+      "step": 8642
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2512265504600316e-05,
+      "loss": 0.8881,
+      "step": 8643
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.251042423763347e-05,
+      "loss": 0.8813,
+      "step": 8644
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.250858278419124e-05,
+      "loss": 0.9469,
+      "step": 8645
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2506741144293224e-05,
+      "loss": 0.8839,
+      "step": 8646
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.250489931795905e-05,
+      "loss": 0.9442,
+      "step": 8647
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2503057305208325e-05,
+      "loss": 0.9738,
+      "step": 8648
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.250121510606067e-05,
+      "loss": 0.9545,
+      "step": 8649
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.249937272053569e-05,
+      "loss": 0.9248,
+      "step": 8650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.249753014865303e-05,
+      "loss": 0.9764,
+      "step": 8651
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.249568739043228e-05,
+      "loss": 1.0229,
+      "step": 8652
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.249384444589308e-05,
+      "loss": 0.994,
+      "step": 8653
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.249200131505507e-05,
+      "loss": 0.8526,
+      "step": 8654
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.249015799793786e-05,
+      "loss": 0.8888,
+      "step": 8655
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.248831449456109e-05,
+      "loss": 0.8824,
+      "step": 8656
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.248647080494439e-05,
+      "loss": 0.9893,
+      "step": 8657
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.248462692910738e-05,
+      "loss": 0.8828,
+      "step": 8658
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.248278286706971e-05,
+      "loss": 0.9256,
+      "step": 8659
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.248093861885103e-05,
+      "loss": 0.9341,
+      "step": 8660
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.247909418447096e-05,
+      "loss": 0.9381,
+      "step": 8661
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2477249563949144e-05,
+      "loss": 0.9252,
+      "step": 8662
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.247540475730524e-05,
+      "loss": 0.8631,
+      "step": 8663
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2473559764558874e-05,
+      "loss": 0.9376,
+      "step": 8664
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.247171458572971e-05,
+      "loss": 0.8547,
+      "step": 8665
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2469869220837386e-05,
+      "loss": 0.9416,
+      "step": 8666
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.246802366990157e-05,
+      "loss": 0.9561,
+      "step": 8667
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.246617793294191e-05,
+      "loss": 0.904,
+      "step": 8668
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2464332009978055e-05,
+      "loss": 0.8866,
+      "step": 8669
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2462485901029666e-05,
+      "loss": 0.8548,
+      "step": 8670
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2460639606116406e-05,
+      "loss": 0.9316,
+      "step": 8671
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2458793125257926e-05,
+      "loss": 0.9401,
+      "step": 8672
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.245694645847391e-05,
+      "loss": 0.8266,
+      "step": 8673
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2455099605784e-05,
+      "loss": 0.8812,
+      "step": 8674
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.245325256720789e-05,
+      "loss": 0.9502,
+      "step": 8675
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.245140534276523e-05,
+      "loss": 0.957,
+      "step": 8676
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.24495579324757e-05,
+      "loss": 0.9054,
+      "step": 8677
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.244771033635897e-05,
+      "loss": 0.8973,
+      "step": 8678
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2445862554434715e-05,
+      "loss": 0.8624,
+      "step": 8679
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2444014586722627e-05,
+      "loss": 0.8248,
+      "step": 8680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2442166433242355e-05,
+      "loss": 0.8634,
+      "step": 8681
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.244031809401361e-05,
+      "loss": 0.836,
+      "step": 8682
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.243846956905607e-05,
+      "loss": 0.9115,
+      "step": 8683
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.243662085838942e-05,
+      "loss": 0.8486,
+      "step": 8684
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2434771962033326e-05,
+      "loss": 0.876,
+      "step": 8685
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.243292288000751e-05,
+      "loss": 0.9099,
+      "step": 8686
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2431073612331645e-05,
+      "loss": 0.9015,
+      "step": 8687
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2429224159025424e-05,
+      "loss": 0.9177,
+      "step": 8688
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2427374520108555e-05,
+      "loss": 0.9915,
+      "step": 8689
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2425524695600723e-05,
+      "loss": 0.8927,
+      "step": 8690
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.242367468552164e-05,
+      "loss": 0.945,
+      "step": 8691
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.242182448989099e-05,
+      "loss": 0.8711,
+      "step": 8692
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2419974108728496e-05,
+      "loss": 0.8962,
+      "step": 8693
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2418123542053845e-05,
+      "loss": 0.9365,
+      "step": 8694
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.241627278988676e-05,
+      "loss": 0.9565,
+      "step": 8695
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.241442185224693e-05,
+      "loss": 0.9534,
+      "step": 8696
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.241257072915409e-05,
+      "loss": 0.9425,
+      "step": 8697
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2410719420627945e-05,
+      "loss": 0.9975,
+      "step": 8698
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2408867926688206e-05,
+      "loss": 1.0067,
+      "step": 8699
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.240701624735459e-05,
+      "loss": 0.8828,
+      "step": 8700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.240516438264682e-05,
+      "loss": 0.9296,
+      "step": 8701
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.240331233258462e-05,
+      "loss": 0.8766,
+      "step": 8702
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.240146009718771e-05,
+      "loss": 0.9132,
+      "step": 8703
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.239960767647581e-05,
+      "loss": 0.9746,
+      "step": 8704
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2397755070468655e-05,
+      "loss": 0.9472,
+      "step": 8705
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.239590227918597e-05,
+      "loss": 0.9972,
+      "step": 8706
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.239404930264749e-05,
+      "loss": 0.9203,
+      "step": 8707
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2392196140872936e-05,
+      "loss": 0.8962,
+      "step": 8708
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.239034279388206e-05,
+      "loss": 0.8337,
+      "step": 8709
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.238848926169459e-05,
+      "loss": 0.9342,
+      "step": 8710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.238663554433027e-05,
+      "loss": 0.9551,
+      "step": 8711
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.238478164180883e-05,
+      "loss": 0.8676,
+      "step": 8712
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.238292755415002e-05,
+      "loss": 0.9425,
+      "step": 8713
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2381073281373584e-05,
+      "loss": 0.9995,
+      "step": 8714
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.237921882349928e-05,
+      "loss": 0.9615,
+      "step": 8715
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2377364180546835e-05,
+      "loss": 0.8605,
+      "step": 8716
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.237550935253601e-05,
+      "loss": 0.8258,
+      "step": 8717
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.237365433948657e-05,
+      "loss": 0.9043,
+      "step": 8718
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.237179914141825e-05,
+      "loss": 0.8679,
+      "step": 8719
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.236994375835083e-05,
+      "loss": 1.0011,
+      "step": 8720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.236808819030404e-05,
+      "loss": 0.8879,
+      "step": 8721
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.236623243729767e-05,
+      "loss": 0.9046,
+      "step": 8722
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2364376499351453e-05,
+      "loss": 0.8936,
+      "step": 8723
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2362520376485174e-05,
+      "loss": 0.8736,
+      "step": 8724
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2360664068718594e-05,
+      "loss": 0.9492,
+      "step": 8725
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.235880757607148e-05,
+      "loss": 0.8632,
+      "step": 8726
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.235695089856361e-05,
+      "loss": 0.9776,
+      "step": 8727
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2355094036214745e-05,
+      "loss": 0.9108,
+      "step": 8728
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2353236989044666e-05,
+      "loss": 0.9054,
+      "step": 8729
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2351379757073154e-05,
+      "loss": 0.8704,
+      "step": 8730
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.234952234031997e-05,
+      "loss": 0.9299,
+      "step": 8731
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2347664738804914e-05,
+      "loss": 0.9453,
+      "step": 8732
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.234580695254776e-05,
+      "loss": 0.8943,
+      "step": 8733
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.234394898156829e-05,
+      "loss": 0.9053,
+      "step": 8734
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.234209082588629e-05,
+      "loss": 0.9408,
+      "step": 8735
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.234023248552156e-05,
+      "loss": 0.8522,
+      "step": 8736
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2338373960493885e-05,
+      "loss": 0.9409,
+      "step": 8737
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.233651525082304e-05,
+      "loss": 0.8566,
+      "step": 8738
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.233465635652884e-05,
+      "loss": 0.9437,
+      "step": 8739
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2332797277631075e-05,
+      "loss": 0.9779,
+      "step": 8740
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.233093801414954e-05,
+      "loss": 0.9136,
+      "step": 8741
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.232907856610404e-05,
+      "loss": 0.8469,
+      "step": 8742
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.232721893351437e-05,
+      "loss": 0.9094,
+      "step": 8743
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.232535911640034e-05,
+      "loss": 0.9571,
+      "step": 8744
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.232349911478175e-05,
+      "loss": 0.843,
+      "step": 8745
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.232163892867842e-05,
+      "loss": 0.8815,
+      "step": 8746
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2319778558110145e-05,
+      "loss": 0.8577,
+      "step": 8747
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2317918003096745e-05,
+      "loss": 0.8973,
+      "step": 8748
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2316057263658035e-05,
+      "loss": 0.9034,
+      "step": 8749
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.231419633981383e-05,
+      "loss": 0.9159,
+      "step": 8750
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.231233523158393e-05,
+      "loss": 0.9237,
+      "step": 8751
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2310473938988186e-05,
+      "loss": 0.8683,
+      "step": 8752
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.23086124620464e-05,
+      "loss": 0.9199,
+      "step": 8753
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.230675080077839e-05,
+      "loss": 0.8848,
+      "step": 8754
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.230488895520401e-05,
+      "loss": 0.8711,
+      "step": 8755
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.230302692534306e-05,
+      "loss": 0.847,
+      "step": 8756
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.230116471121538e-05,
+      "loss": 0.8698,
+      "step": 8757
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.229930231284079e-05,
+      "loss": 0.9914,
+      "step": 8758
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2297439730239145e-05,
+      "loss": 0.9309,
+      "step": 8759
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.229557696343026e-05,
+      "loss": 0.9379,
+      "step": 8760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2293714012433986e-05,
+      "loss": 0.8983,
+      "step": 8761
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.229185087727016e-05,
+      "loss": 0.9728,
+      "step": 8762
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.228998755795861e-05,
+      "loss": 0.8288,
+      "step": 8763
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.22881240545192e-05,
+      "loss": 0.9579,
+      "step": 8764
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.228626036697175e-05,
+      "loss": 0.8832,
+      "step": 8765
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.228439649533613e-05,
+      "loss": 0.9345,
+      "step": 8766
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2282532439632185e-05,
+      "loss": 0.9957,
+      "step": 8767
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.228066819987976e-05,
+      "loss": 1.0225,
+      "step": 8768
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.227880377609871e-05,
+      "loss": 0.9422,
+      "step": 8769
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.227693916830889e-05,
+      "loss": 0.8459,
+      "step": 8770
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2275074376530153e-05,
+      "loss": 0.8617,
+      "step": 8771
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2273209400782365e-05,
+      "loss": 0.88,
+      "step": 8772
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.227134424108539e-05,
+      "loss": 0.9011,
+      "step": 8773
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.226947889745907e-05,
+      "loss": 0.938,
+      "step": 8774
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.22676133699233e-05,
+      "loss": 0.9407,
+      "step": 8775
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.226574765849792e-05,
+      "loss": 0.8981,
+      "step": 8776
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.226388176320281e-05,
+      "loss": 0.9476,
+      "step": 8777
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.226201568405784e-05,
+      "loss": 0.9753,
+      "step": 8778
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.226014942108289e-05,
+      "loss": 0.8728,
+      "step": 8779
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.225828297429783e-05,
+      "loss": 0.9535,
+      "step": 8780
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.225641634372253e-05,
+      "loss": 0.9105,
+      "step": 8781
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.225454952937687e-05,
+      "loss": 0.8627,
+      "step": 8782
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.225268253128073e-05,
+      "loss": 0.9507,
+      "step": 8783
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2250815349454e-05,
+      "loss": 0.905,
+      "step": 8784
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.224894798391656e-05,
+      "loss": 0.9104,
+      "step": 8785
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.224708043468829e-05,
+      "loss": 0.914,
+      "step": 8786
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.22452127017891e-05,
+      "loss": 0.8931,
+      "step": 8787
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.224334478523885e-05,
+      "loss": 1.0121,
+      "step": 8788
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2241476685057445e-05,
+      "loss": 0.885,
+      "step": 8789
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.223960840126479e-05,
+      "loss": 0.8608,
+      "step": 8790
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.223773993388077e-05,
+      "loss": 0.889,
+      "step": 8791
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.223587128292529e-05,
+      "loss": 0.8694,
+      "step": 8792
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2234002448418234e-05,
+      "loss": 0.9496,
+      "step": 8793
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2232133430379516e-05,
+      "loss": 0.9419,
+      "step": 8794
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2230264228829043e-05,
+      "loss": 0.8612,
+      "step": 8795
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.222839484378671e-05,
+      "loss": 0.8547,
+      "step": 8796
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.222652527527244e-05,
+      "loss": 0.9171,
+      "step": 8797
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2224655523306136e-05,
+      "loss": 0.9265,
+      "step": 8798
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.22227855879077e-05,
+      "loss": 0.9212,
+      "step": 8799
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.222091546909706e-05,
+      "loss": 0.9775,
+      "step": 8800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.221904516689412e-05,
+      "loss": 0.894,
+      "step": 8801
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2217174681318813e-05,
+      "loss": 0.8563,
+      "step": 8802
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.221530401239104e-05,
+      "loss": 0.9249,
+      "step": 8803
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.221343316013073e-05,
+      "loss": 0.8426,
+      "step": 8804
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.221156212455781e-05,
+      "loss": 0.913,
+      "step": 8805
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.22096909056922e-05,
+      "loss": 0.8991,
+      "step": 8806
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2207819503553834e-05,
+      "loss": 1.0678,
+      "step": 8807
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.220594791816264e-05,
+      "loss": 0.8768,
+      "step": 8808
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.220407614953854e-05,
+      "loss": 0.9429,
+      "step": 8809
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.220220419770148e-05,
+      "loss": 0.8955,
+      "step": 8810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.220033206267138e-05,
+      "loss": 0.8189,
+      "step": 8811
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2198459744468195e-05,
+      "loss": 0.8742,
+      "step": 8812
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.219658724311185e-05,
+      "loss": 0.9071,
+      "step": 8813
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.21947145586223e-05,
+      "loss": 0.8897,
+      "step": 8814
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2192841691019466e-05,
+      "loss": 0.9925,
+      "step": 8815
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.219096864032331e-05,
+      "loss": 0.8442,
+      "step": 8816
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.218909540655378e-05,
+      "loss": 0.9169,
+      "step": 8817
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2187221989730805e-05,
+      "loss": 0.9333,
+      "step": 8818
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.218534838987437e-05,
+      "loss": 0.7992,
+      "step": 8819
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.21834746070044e-05,
+      "loss": 0.9641,
+      "step": 8820
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2181600641140864e-05,
+      "loss": 0.8903,
+      "step": 8821
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.21797264923037e-05,
+      "loss": 0.8355,
+      "step": 8822
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.217785216051289e-05,
+      "loss": 0.8839,
+      "step": 8823
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2175977645788376e-05,
+      "loss": 0.9216,
+      "step": 8824
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2174102948150126e-05,
+      "loss": 0.8842,
+      "step": 8825
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.217222806761811e-05,
+      "loss": 0.8711,
+      "step": 8826
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.217035300421229e-05,
+      "loss": 0.9662,
+      "step": 8827
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.216847775795264e-05,
+      "loss": 0.8535,
+      "step": 8828
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2166602328859115e-05,
+      "loss": 0.8494,
+      "step": 8829
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.216472671695171e-05,
+      "loss": 0.9461,
+      "step": 8830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.216285092225038e-05,
+      "loss": 0.942,
+      "step": 8831
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.216097494477511e-05,
+      "loss": 0.865,
+      "step": 8832
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.215909878454587e-05,
+      "loss": 0.8838,
+      "step": 8833
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2157222441582647e-05,
+      "loss": 0.8989,
+      "step": 8834
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.215534591590542e-05,
+      "loss": 0.8718,
+      "step": 8835
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.215346920753418e-05,
+      "loss": 0.9432,
+      "step": 8836
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.215159231648891e-05,
+      "loss": 0.9408,
+      "step": 8837
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2149715242789575e-05,
+      "loss": 0.8079,
+      "step": 8838
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.214783798645621e-05,
+      "loss": 0.9264,
+      "step": 8839
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2145960547508765e-05,
+      "loss": 0.875,
+      "step": 8840
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2144082925967256e-05,
+      "loss": 0.7987,
+      "step": 8841
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.214220512185168e-05,
+      "loss": 0.897,
+      "step": 8842
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.214032713518201e-05,
+      "loss": 0.8573,
+      "step": 8843
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213844896597827e-05,
+      "loss": 0.8937,
+      "step": 8844
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213657061426045e-05,
+      "loss": 0.8926,
+      "step": 8845
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213469208004856e-05,
+      "loss": 0.8613,
+      "step": 8846
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.21328133633626e-05,
+      "loss": 0.8516,
+      "step": 8847
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213093446422258e-05,
+      "loss": 0.8266,
+      "step": 8848
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2129055382648516e-05,
+      "loss": 0.88,
+      "step": 8849
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.21271761186604e-05,
+      "loss": 0.9329,
+      "step": 8850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.212529667227826e-05,
+      "loss": 0.9556,
+      "step": 8851
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2123417043522115e-05,
+      "loss": 0.8683,
+      "step": 8852
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2121537232411964e-05,
+      "loss": 0.8556,
+      "step": 8853
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.211965723896785e-05,
+      "loss": 0.9347,
+      "step": 8854
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.211777706320976e-05,
+      "loss": 0.9162,
+      "step": 8855
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.211589670515775e-05,
+      "loss": 0.9881,
+      "step": 8856
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2114016164831826e-05,
+      "loss": 0.8488,
+      "step": 8857
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.211213544225202e-05,
+      "loss": 0.9189,
+      "step": 8858
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.211025453743836e-05,
+      "loss": 0.9226,
+      "step": 8859
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2108373450410885e-05,
+      "loss": 0.9312,
+      "step": 8860
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.210649218118962e-05,
+      "loss": 0.8594,
+      "step": 8861
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.210461072979459e-05,
+      "loss": 0.9015,
+      "step": 8862
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.210272909624585e-05,
+      "loss": 0.8344,
+      "step": 8863
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.210084728056342e-05,
+      "loss": 0.9083,
+      "step": 8864
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.209896528276735e-05,
+      "loss": 0.9038,
+      "step": 8865
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.209708310287768e-05,
+      "loss": 0.8814,
+      "step": 8866
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.209520074091446e-05,
+      "loss": 0.8422,
+      "step": 8867
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.209331819689772e-05,
+      "loss": 0.9358,
+      "step": 8868
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.209143547084753e-05,
+      "loss": 0.9644,
+      "step": 8869
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.208955256278393e-05,
+      "loss": 0.9625,
+      "step": 8870
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.208766947272696e-05,
+      "loss": 0.8197,
+      "step": 8871
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2085786200696696e-05,
+      "loss": 0.9247,
+      "step": 8872
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2083902746713175e-05,
+      "loss": 0.8365,
+      "step": 8873
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.208201911079646e-05,
+      "loss": 0.8315,
+      "step": 8874
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.208013529296662e-05,
+      "loss": 0.8293,
+      "step": 8875
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.20782512932437e-05,
+      "loss": 0.9052,
+      "step": 8876
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.207636711164778e-05,
+      "loss": 0.8668,
+      "step": 8877
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.207448274819891e-05,
+      "loss": 1.0445,
+      "step": 8878
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.207259820291717e-05,
+      "loss": 0.8516,
+      "step": 8879
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2070713475822614e-05,
+      "loss": 0.9559,
+      "step": 8880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2068828566935336e-05,
+      "loss": 1.002,
+      "step": 8881
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.206694347627539e-05,
+      "loss": 0.9067,
+      "step": 8882
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.206505820386286e-05,
+      "loss": 0.8514,
+      "step": 8883
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2063172749717824e-05,
+      "loss": 0.8914,
+      "step": 8884
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2061287113860345e-05,
+      "loss": 0.9364,
+      "step": 8885
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.205940129631053e-05,
+      "loss": 0.8178,
+      "step": 8886
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.205751529708845e-05,
+      "loss": 0.9026,
+      "step": 8887
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.205562911621417e-05,
+      "loss": 0.8626,
+      "step": 8888
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2053742753707814e-05,
+      "loss": 0.9019,
+      "step": 8889
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2051856209589435e-05,
+      "loss": 0.8888,
+      "step": 8890
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.204996948387915e-05,
+      "loss": 0.8966,
+      "step": 8891
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2048082576597044e-05,
+      "loss": 0.8804,
+      "step": 8892
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.204619548776321e-05,
+      "loss": 0.8939,
+      "step": 8893
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.204430821739773e-05,
+      "loss": 0.8494,
+      "step": 8894
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.204242076552072e-05,
+      "loss": 0.8829,
+      "step": 8895
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.204053313215228e-05,
+      "loss": 0.8931,
+      "step": 8896
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2038645317312505e-05,
+      "loss": 0.8822,
+      "step": 8897
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.20367573210215e-05,
+      "loss": 0.9306,
+      "step": 8898
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2034869143299375e-05,
+      "loss": 0.8123,
+      "step": 8899
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.203298078416624e-05,
+      "loss": 0.9737,
+      "step": 8900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.20310922436422e-05,
+      "loss": 0.9752,
+      "step": 8901
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.202920352174736e-05,
+      "loss": 0.8763,
+      "step": 8902
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.202731461850185e-05,
+      "loss": 0.8565,
+      "step": 8903
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.202542553392577e-05,
+      "loss": 0.9301,
+      "step": 8904
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2023536268039246e-05,
+      "loss": 0.9389,
+      "step": 8905
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.202164682086239e-05,
+      "loss": 0.9045,
+      "step": 8906
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.201975719241534e-05,
+      "loss": 0.9192,
+      "step": 8907
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.20178673827182e-05,
+      "loss": 0.9444,
+      "step": 8908
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2015977391791115e-05,
+      "loss": 0.9187,
+      "step": 8909
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.201408721965419e-05,
+      "loss": 0.9985,
+      "step": 8910
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.201219686632757e-05,
+      "loss": 0.8909,
+      "step": 8911
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.201030633183139e-05,
+      "loss": 0.9119,
+      "step": 8912
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.200841561618577e-05,
+      "loss": 0.8991,
+      "step": 8913
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.200652471941084e-05,
+      "loss": 0.9736,
+      "step": 8914
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.200463364152676e-05,
+      "loss": 0.9478,
+      "step": 8915
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.200274238255365e-05,
+      "loss": 0.9291,
+      "step": 8916
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.200085094251165e-05,
+      "loss": 0.7998,
+      "step": 8917
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.199895932142092e-05,
+      "loss": 0.9631,
+      "step": 8918
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.199706751930158e-05,
+      "loss": 0.9123,
+      "step": 8919
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1995175536173806e-05,
+      "loss": 0.887,
+      "step": 8920
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.199328337205772e-05,
+      "loss": 0.8203,
+      "step": 8921
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.199139102697349e-05,
+      "loss": 0.8964,
+      "step": 8922
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.198949850094125e-05,
+      "loss": 0.8473,
+      "step": 8923
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.198760579398118e-05,
+      "loss": 0.8178,
+      "step": 8924
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.198571290611341e-05,
+      "loss": 0.9144,
+      "step": 8925
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.198381983735811e-05,
+      "loss": 0.8964,
+      "step": 8926
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1981926587735443e-05,
+      "loss": 0.8079,
+      "step": 8927
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1980033157265576e-05,
+      "loss": 0.9555,
+      "step": 8928
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1978139545968656e-05,
+      "loss": 0.932,
+      "step": 8929
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.197624575386485e-05,
+      "loss": 0.9122,
+      "step": 8930
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.197435178097435e-05,
+      "loss": 0.918,
+      "step": 8931
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1972457627317304e-05,
+      "loss": 0.9036,
+      "step": 8932
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.197056329291388e-05,
+      "loss": 0.8935,
+      "step": 8933
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.196866877778427e-05,
+      "loss": 0.9399,
+      "step": 8934
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.196677408194864e-05,
+      "loss": 0.9221,
+      "step": 8935
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.196487920542715e-05,
+      "loss": 0.8901,
+      "step": 8936
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1962984148240016e-05,
+      "loss": 0.9141,
+      "step": 8937
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1961088910407384e-05,
+      "loss": 0.8899,
+      "step": 8938
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.195919349194946e-05,
+      "loss": 1.0642,
+      "step": 8939
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.195729789288642e-05,
+      "loss": 1.0027,
+      "step": 8940
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.195540211323846e-05,
+      "loss": 0.8357,
+      "step": 8941
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.195350615302576e-05,
+      "loss": 0.8955,
+      "step": 8942
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.19516100122685e-05,
+      "loss": 0.8668,
+      "step": 8943
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.194971369098689e-05,
+      "loss": 0.8859,
+      "step": 8944
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1947817189201124e-05,
+      "loss": 0.9031,
+      "step": 8945
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.194592050693139e-05,
+      "loss": 0.8988,
+      "step": 8946
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.194402364419789e-05,
+      "loss": 0.9219,
+      "step": 8947
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.194212660102083e-05,
+      "loss": 0.9169,
+      "step": 8948
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.194022937742041e-05,
+      "loss": 0.8898,
+      "step": 8949
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1938331973416825e-05,
+      "loss": 0.8573,
+      "step": 8950
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1936434389030286e-05,
+      "loss": 0.9344,
+      "step": 8951
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.193453662428101e-05,
+      "loss": 0.8595,
+      "step": 8952
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.19326386791892e-05,
+      "loss": 0.9255,
+      "step": 8953
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.193074055377506e-05,
+      "loss": 0.8913,
+      "step": 8954
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.192884224805882e-05,
+      "loss": 1.0084,
+      "step": 8955
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.192694376206069e-05,
+      "loss": 0.844,
+      "step": 8956
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.192504509580088e-05,
+      "loss": 0.8811,
+      "step": 8957
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1923146249299616e-05,
+      "loss": 0.8245,
+      "step": 8958
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1921247222577114e-05,
+      "loss": 0.8936,
+      "step": 8959
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.191934801565361e-05,
+      "loss": 0.8261,
+      "step": 8960
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1917448628549314e-05,
+      "loss": 0.8856,
+      "step": 8961
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.191554906128447e-05,
+      "loss": 0.8883,
+      "step": 8962
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.191364931387929e-05,
+      "loss": 0.803,
+      "step": 8963
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.191174938635402e-05,
+      "loss": 0.9392,
+      "step": 8964
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.190984927872889e-05,
+      "loss": 0.8966,
+      "step": 8965
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1907948991024114e-05,
+      "loss": 0.9106,
+      "step": 8966
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.190604852325997e-05,
+      "loss": 0.8685,
+      "step": 8967
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1904147875456655e-05,
+      "loss": 0.9089,
+      "step": 8968
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.190224704763443e-05,
+      "loss": 0.9845,
+      "step": 8969
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1900346039813545e-05,
+      "loss": 0.8981,
+      "step": 8970
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.189844485201422e-05,
+      "loss": 0.8718,
+      "step": 8971
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.189654348425673e-05,
+      "loss": 0.8752,
+      "step": 8972
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.189464193656131e-05,
+      "loss": 0.939,
+      "step": 8973
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.18927402089482e-05,
+      "loss": 0.8989,
+      "step": 8974
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.189083830143767e-05,
+      "loss": 0.9306,
+      "step": 8975
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1888936214049966e-05,
+      "loss": 1.0063,
+      "step": 8976
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.188703394680534e-05,
+      "loss": 0.7742,
+      "step": 8977
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.188513149972406e-05,
+      "loss": 0.8371,
+      "step": 8978
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.188322887282638e-05,
+      "loss": 0.9538,
+      "step": 8979
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1881326066132564e-05,
+      "loss": 0.8693,
+      "step": 8980
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.187942307966287e-05,
+      "loss": 0.8112,
+      "step": 8981
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.187751991343757e-05,
+      "loss": 0.9143,
+      "step": 8982
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.187561656747693e-05,
+      "loss": 0.8577,
+      "step": 8983
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.187371304180122e-05,
+      "loss": 0.9069,
+      "step": 8984
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.187180933643071e-05,
+      "loss": 0.8875,
+      "step": 8985
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.186990545138567e-05,
+      "loss": 0.8378,
+      "step": 8986
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1868001386686386e-05,
+      "loss": 0.819,
+      "step": 8987
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1866097142353123e-05,
+      "loss": 0.898,
+      "step": 8988
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1864192718406166e-05,
+      "loss": 0.848,
+      "step": 8989
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.18622881148658e-05,
+      "loss": 0.9236,
+      "step": 8990
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1860383331752295e-05,
+      "loss": 0.8962,
+      "step": 8991
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.185847836908595e-05,
+      "loss": 0.8874,
+      "step": 8992
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.185657322688704e-05,
+      "loss": 0.802,
+      "step": 8993
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.185466790517586e-05,
+      "loss": 0.8387,
+      "step": 8994
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.18527624039727e-05,
+      "loss": 0.8367,
+      "step": 8995
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.185085672329785e-05,
+      "loss": 1.0749,
+      "step": 8996
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.184895086317161e-05,
+      "loss": 0.8559,
+      "step": 8997
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.184704482361427e-05,
+      "loss": 0.9727,
+      "step": 8998
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1845138604646135e-05,
+      "loss": 0.9095,
+      "step": 8999
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1843232206287495e-05,
+      "loss": 0.932,
+      "step": 9000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1841325628558665e-05,
+      "loss": 0.9657,
+      "step": 9001
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.183941887147994e-05,
+      "loss": 0.9213,
+      "step": 9002
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1837511935071625e-05,
+      "loss": 0.8499,
+      "step": 9003
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.183560481935403e-05,
+      "loss": 0.9035,
+      "step": 9004
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1833697524347456e-05,
+      "loss": 0.8556,
+      "step": 9005
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.183179005007224e-05,
+      "loss": 0.8713,
+      "step": 9006
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.182988239654867e-05,
+      "loss": 0.8567,
+      "step": 9007
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1827974563797065e-05,
+      "loss": 0.9179,
+      "step": 9008
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.182606655183776e-05,
+      "loss": 1.013,
+      "step": 9009
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1824158360691054e-05,
+      "loss": 0.9403,
+      "step": 9010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1822249990377276e-05,
+      "loss": 0.9329,
+      "step": 9011
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1820341440916755e-05,
+      "loss": 0.8507,
+      "step": 9012
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.18184327123298e-05,
+      "loss": 0.9693,
+      "step": 9013
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.181652380463675e-05,
+      "loss": 0.85,
+      "step": 9014
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.181461471785794e-05,
+      "loss": 0.8968,
+      "step": 9015
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.181270545201368e-05,
+      "loss": 0.9106,
+      "step": 9016
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.181079600712432e-05,
+      "loss": 0.8552,
+      "step": 9017
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.180888638321018e-05,
+      "loss": 0.8879,
+      "step": 9018
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1806976580291616e-05,
+      "loss": 0.8952,
+      "step": 9019
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.180506659838895e-05,
+      "loss": 0.8578,
+      "step": 9020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.180315643752253e-05,
+      "loss": 0.9499,
+      "step": 9021
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.180124609771269e-05,
+      "loss": 0.8827,
+      "step": 9022
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.179933557897979e-05,
+      "loss": 0.9396,
+      "step": 9023
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.179742488134416e-05,
+      "loss": 0.9055,
+      "step": 9024
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1795514004826145e-05,
+      "loss": 0.9317,
+      "step": 9025
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1793602949446106e-05,
+      "loss": 0.8593,
+      "step": 9026
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.17916917152244e-05,
+      "loss": 0.8376,
+      "step": 9027
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1789780302181365e-05,
+      "loss": 0.8536,
+      "step": 9028
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.178786871033736e-05,
+      "loss": 0.9067,
+      "step": 9029
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1785956939712754e-05,
+      "loss": 0.8973,
+      "step": 9030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.178404499032789e-05,
+      "loss": 0.8442,
+      "step": 9031
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.178213286220314e-05,
+      "loss": 0.9513,
+      "step": 9032
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1780220555358864e-05,
+      "loss": 0.8965,
+      "step": 9033
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.177830806981543e-05,
+      "loss": 0.9154,
+      "step": 9034
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.17763954055932e-05,
+      "loss": 0.887,
+      "step": 9035
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.177448256271254e-05,
+      "loss": 0.9329,
+      "step": 9036
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1772569541193815e-05,
+      "loss": 0.816,
+      "step": 9037
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1770656341057426e-05,
+      "loss": 0.9055,
+      "step": 9038
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.176874296232372e-05,
+      "loss": 0.8891,
+      "step": 9039
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.176682940501309e-05,
+      "loss": 0.8893,
+      "step": 9040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.176491566914589e-05,
+      "loss": 0.9175,
+      "step": 9041
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1763001754742523e-05,
+      "loss": 0.8872,
+      "step": 9042
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.176108766182337e-05,
+      "loss": 0.8451,
+      "step": 9043
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.175917339040881e-05,
+      "loss": 0.9378,
+      "step": 9044
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.175725894051922e-05,
+      "loss": 0.8887,
+      "step": 9045
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1755344312175006e-05,
+      "loss": 0.8959,
+      "step": 9046
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.175342950539654e-05,
+      "loss": 0.9387,
+      "step": 9047
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.175151452020422e-05,
+      "loss": 0.9249,
+      "step": 9048
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.174959935661844e-05,
+      "loss": 0.8425,
+      "step": 9049
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.17476840146596e-05,
+      "loss": 0.8591,
+      "step": 9050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1745768494348095e-05,
+      "loss": 0.9639,
+      "step": 9051
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1743852795704317e-05,
+      "loss": 0.8421,
+      "step": 9052
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.174193691874868e-05,
+      "loss": 0.8787,
+      "step": 9053
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.174002086350157e-05,
+      "loss": 0.8514,
+      "step": 9054
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.173810462998341e-05,
+      "loss": 0.8464,
+      "step": 9055
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.173618821821459e-05,
+      "loss": 0.9007,
+      "step": 9056
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.173427162821553e-05,
+      "loss": 0.8726,
+      "step": 9057
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.173235486000663e-05,
+      "loss": 0.8405,
+      "step": 9058
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.173043791360832e-05,
+      "loss": 0.8652,
+      "step": 9059
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.172852078904099e-05,
+      "loss": 0.8529,
+      "step": 9060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.172660348632508e-05,
+      "loss": 0.9396,
+      "step": 9061
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.172468600548099e-05,
+      "loss": 0.879,
+      "step": 9062
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.172276834652916e-05,
+      "loss": 0.8883,
+      "step": 9063
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.172085050949e-05,
+      "loss": 0.9353,
+      "step": 9064
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1718932494383924e-05,
+      "loss": 0.9482,
+      "step": 9065
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.171701430123137e-05,
+      "loss": 0.8451,
+      "step": 9066
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1715095930052763e-05,
+      "loss": 0.9614,
+      "step": 9067
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.171317738086853e-05,
+      "loss": 0.8644,
+      "step": 9068
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.17112586536991e-05,
+      "loss": 0.8861,
+      "step": 9069
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1709339748564924e-05,
+      "loss": 0.9196,
+      "step": 9070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.170742066548642e-05,
+      "loss": 0.9594,
+      "step": 9071
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.170550140448403e-05,
+      "loss": 0.8668,
+      "step": 9072
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.170358196557819e-05,
+      "loss": 0.9084,
+      "step": 9073
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.170166234878934e-05,
+      "loss": 0.8697,
+      "step": 9074
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1699742554137924e-05,
+      "loss": 0.9578,
+      "step": 9075
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.16978225816444e-05,
+      "loss": 0.9696,
+      "step": 9076
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1695902431329194e-05,
+      "loss": 0.8791,
+      "step": 9077
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1693982103212755e-05,
+      "loss": 0.8919,
+      "step": 9078
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.169206159731556e-05,
+      "loss": 0.8536,
+      "step": 9079
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.169014091365803e-05,
+      "loss": 0.9473,
+      "step": 9080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.168822005226064e-05,
+      "loss": 0.8468,
+      "step": 9081
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.168629901314383e-05,
+      "loss": 0.8018,
+      "step": 9082
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.168437779632808e-05,
+      "loss": 0.8877,
+      "step": 9083
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1682456401833825e-05,
+      "loss": 0.996,
+      "step": 9084
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.168053482968154e-05,
+      "loss": 0.8762,
+      "step": 9085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.167861307989168e-05,
+      "loss": 0.8735,
+      "step": 9086
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1676691152484723e-05,
+      "loss": 0.8899,
+      "step": 9087
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1674769047481135e-05,
+      "loss": 0.8755,
+      "step": 9088
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.167284676490137e-05,
+      "loss": 0.8857,
+      "step": 9089
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.167092430476591e-05,
+      "loss": 0.8852,
+      "step": 9090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.166900166709523e-05,
+      "loss": 0.908,
+      "step": 9091
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.16670788519098e-05,
+      "loss": 0.8553,
+      "step": 9092
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.166515585923011e-05,
+      "loss": 0.811,
+      "step": 9093
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1663232689076616e-05,
+      "loss": 0.8269,
+      "step": 9094
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.166130934146982e-05,
+      "loss": 0.8527,
+      "step": 9095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1659385816430184e-05,
+      "loss": 0.8179,
+      "step": 9096
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1657462113978205e-05,
+      "loss": 0.9058,
+      "step": 9097
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.165553823413437e-05,
+      "loss": 0.9367,
+      "step": 9098
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1653614176919164e-05,
+      "loss": 0.928,
+      "step": 9099
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.165168994235308e-05,
+      "loss": 0.8654,
+      "step": 9100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1649765530456606e-05,
+      "loss": 0.9032,
+      "step": 9101
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.164784094125024e-05,
+      "loss": 0.9096,
+      "step": 9102
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1645916174754475e-05,
+      "loss": 0.8446,
+      "step": 9103
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1643991230989807e-05,
+      "loss": 0.8617,
+      "step": 9104
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.164206610997674e-05,
+      "loss": 0.9362,
+      "step": 9105
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.164014081173576e-05,
+      "loss": 0.9228,
+      "step": 9106
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.16382153362874e-05,
+      "loss": 0.943,
+      "step": 9107
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.163628968365214e-05,
+      "loss": 0.8967,
+      "step": 9108
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1634363853850486e-05,
+      "loss": 0.8259,
+      "step": 9109
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.163243784690297e-05,
+      "loss": 0.8826,
+      "step": 9110
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1630511662830085e-05,
+      "loss": 0.826,
+      "step": 9111
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.162858530165235e-05,
+      "loss": 0.8913,
+      "step": 9112
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.162665876339027e-05,
+      "loss": 0.9443,
+      "step": 9113
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.162473204806437e-05,
+      "loss": 0.8411,
+      "step": 9114
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.162280515569517e-05,
+      "loss": 0.8981,
+      "step": 9115
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1620878086303184e-05,
+      "loss": 0.8446,
+      "step": 9116
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1618950839908945e-05,
+      "loss": 0.9189,
+      "step": 9117
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1617023416532966e-05,
+      "loss": 0.9097,
+      "step": 9118
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1615095816195765e-05,
+      "loss": 0.8768,
+      "step": 9119
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.161316803891789e-05,
+      "loss": 0.9558,
+      "step": 9120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.161124008471986e-05,
+      "loss": 0.939,
+      "step": 9121
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.160931195362221e-05,
+      "loss": 0.8765,
+      "step": 9122
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.160738364564547e-05,
+      "loss": 0.8277,
+      "step": 9123
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.160545516081018e-05,
+      "loss": 0.8607,
+      "step": 9124
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.160352649913687e-05,
+      "loss": 0.891,
+      "step": 9125
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.160159766064608e-05,
+      "loss": 0.8424,
+      "step": 9126
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.159966864535836e-05,
+      "loss": 0.891,
+      "step": 9127
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1597739453294245e-05,
+      "loss": 0.8989,
+      "step": 9128
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.159581008447428e-05,
+      "loss": 0.9026,
+      "step": 9129
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1593880538919014e-05,
+      "loss": 0.9416,
+      "step": 9130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1591950816648995e-05,
+      "loss": 0.8678,
+      "step": 9131
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1590020917684766e-05,
+      "loss": 0.8524,
+      "step": 9132
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.15880908420469e-05,
+      "loss": 0.8607,
+      "step": 9133
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.158616058975593e-05,
+      "loss": 0.8387,
+      "step": 9134
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1584230160832417e-05,
+      "loss": 0.8498,
+      "step": 9135
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1582299555296925e-05,
+      "loss": 0.8842,
+      "step": 9136
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1580368773170006e-05,
+      "loss": 0.8638,
+      "step": 9137
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.157843781447223e-05,
+      "loss": 0.8338,
+      "step": 9138
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.157650667922416e-05,
+      "loss": 0.8561,
+      "step": 9139
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.157457536744635e-05,
+      "loss": 0.8181,
+      "step": 9140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1572643879159376e-05,
+      "loss": 0.9395,
+      "step": 9141
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.157071221438381e-05,
+      "loss": 0.913,
+      "step": 9142
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.156878037314021e-05,
+      "loss": 0.9066,
+      "step": 9143
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.156684835544917e-05,
+      "loss": 0.9231,
+      "step": 9144
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.156491616133126e-05,
+      "loss": 0.8426,
+      "step": 9145
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.156298379080703e-05,
+      "loss": 0.8826,
+      "step": 9146
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.156105124389709e-05,
+      "loss": 0.9022,
+      "step": 9147
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1559118520622e-05,
+      "loss": 0.8682,
+      "step": 9148
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1557185621002356e-05,
+      "loss": 0.8078,
+      "step": 9149
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.155525254505874e-05,
+      "loss": 0.8424,
+      "step": 9150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1553319292811734e-05,
+      "loss": 0.8568,
+      "step": 9151
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.155138586428192e-05,
+      "loss": 0.8247,
+      "step": 9152
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.154945225948991e-05,
+      "loss": 0.8722,
+      "step": 9153
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.154751847845627e-05,
+      "loss": 0.8732,
+      "step": 9154
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1545584521201606e-05,
+      "loss": 0.875,
+      "step": 9155
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.154365038774651e-05,
+      "loss": 0.9268,
+      "step": 9156
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.154171607811159e-05,
+      "loss": 0.807,
+      "step": 9157
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.153978159231743e-05,
+      "loss": 0.8593,
+      "step": 9158
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1537846930384635e-05,
+      "loss": 0.8928,
+      "step": 9159
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.153591209233381e-05,
+      "loss": 0.862,
+      "step": 9160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.153397707818556e-05,
+      "loss": 0.822,
+      "step": 9161
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.15320418879605e-05,
+      "loss": 0.8986,
+      "step": 9162
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.153010652167923e-05,
+      "loss": 0.8345,
+      "step": 9163
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1528170979362355e-05,
+      "loss": 0.9521,
+      "step": 9164
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.15262352610305e-05,
+      "loss": 0.868,
+      "step": 9165
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.152429936670427e-05,
+      "loss": 0.9264,
+      "step": 9166
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1522363296404285e-05,
+      "loss": 0.8713,
+      "step": 9167
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.152042705015116e-05,
+      "loss": 0.8369,
+      "step": 9168
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.151849062796551e-05,
+      "loss": 0.8783,
+      "step": 9169
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.151655402986797e-05,
+      "loss": 0.8204,
+      "step": 9170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1514617255879164e-05,
+      "loss": 0.8628,
+      "step": 9171
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.15126803060197e-05,
+      "loss": 1.0304,
+      "step": 9172
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.151074318031022e-05,
+      "loss": 0.8332,
+      "step": 9173
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.150880587877135e-05,
+      "loss": 0.7966,
+      "step": 9174
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.150686840142371e-05,
+      "loss": 0.7653,
+      "step": 9175
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1504930748287966e-05,
+      "loss": 0.9067,
+      "step": 9176
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.150299291938471e-05,
+      "loss": 0.9084,
+      "step": 9177
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1501054914734606e-05,
+      "loss": 0.9098,
+      "step": 9178
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.149911673435828e-05,
+      "loss": 0.83,
+      "step": 9179
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.149717837827638e-05,
+      "loss": 0.8404,
+      "step": 9180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.149523984650955e-05,
+      "loss": 0.9233,
+      "step": 9181
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1493301139078426e-05,
+      "loss": 0.8965,
+      "step": 9182
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.149136225600366e-05,
+      "loss": 0.9065,
+      "step": 9183
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1489423197305896e-05,
+      "loss": 0.9313,
+      "step": 9184
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.148748396300579e-05,
+      "loss": 0.8099,
+      "step": 9185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.148554455312399e-05,
+      "loss": 0.9597,
+      "step": 9186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1483604967681144e-05,
+      "loss": 0.8771,
+      "step": 9187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.148166520669791e-05,
+      "loss": 0.8547,
+      "step": 9188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1479725270194955e-05,
+      "loss": 0.8149,
+      "step": 9189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.147778515819293e-05,
+      "loss": 0.7891,
+      "step": 9190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.147584487071249e-05,
+      "loss": 0.8912,
+      "step": 9191
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.147390440777431e-05,
+      "loss": 0.8767,
+      "step": 9192
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1471963769399055e-05,
+      "loss": 0.8974,
+      "step": 9193
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.147002295560738e-05,
+      "loss": 0.907,
+      "step": 9194
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.146808196641996e-05,
+      "loss": 0.8963,
+      "step": 9195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.146614080185747e-05,
+      "loss": 0.8948,
+      "step": 9196
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.146419946194057e-05,
+      "loss": 0.8901,
+      "step": 9197
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1462257946689945e-05,
+      "loss": 0.9038,
+      "step": 9198
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.146031625612627e-05,
+      "loss": 0.9202,
+      "step": 9199
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.145837439027022e-05,
+      "loss": 0.8871,
+      "step": 9200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.145643234914247e-05,
+      "loss": 0.9192,
+      "step": 9201
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.145449013276371e-05,
+      "loss": 0.884,
+      "step": 9202
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1452547741154613e-05,
+      "loss": 0.895,
+      "step": 9203
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.145060517433588e-05,
+      "loss": 0.8556,
+      "step": 9204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.144866243232818e-05,
+      "loss": 0.9671,
+      "step": 9205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.144671951515222e-05,
+      "loss": 0.879,
+      "step": 9206
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.144477642282868e-05,
+      "loss": 0.8756,
+      "step": 9207
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.144283315537826e-05,
+      "loss": 0.9241,
+      "step": 9208
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.144088971282164e-05,
+      "loss": 0.9399,
+      "step": 9209
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1438946095179534e-05,
+      "loss": 0.9305,
+      "step": 9210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.143700230247262e-05,
+      "loss": 0.9229,
+      "step": 9211
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.143505833472162e-05,
+      "loss": 0.9275,
+      "step": 9212
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.143311419194723e-05,
+      "loss": 0.8947,
+      "step": 9213
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.143116987417015e-05,
+      "loss": 0.9492,
+      "step": 9214
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.142922538141108e-05,
+      "loss": 0.9047,
+      "step": 9215
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.142728071369073e-05,
+      "loss": 0.9019,
+      "step": 9216
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1425335871029825e-05,
+      "loss": 0.8206,
+      "step": 9217
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.142339085344906e-05,
+      "loss": 0.8927,
+      "step": 9218
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.142144566096915e-05,
+      "loss": 0.8739,
+      "step": 9219
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.141950029361082e-05,
+      "loss": 0.9071,
+      "step": 9220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.141755475139478e-05,
+      "loss": 0.8966,
+      "step": 9221
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.141560903434174e-05,
+      "loss": 0.9218,
+      "step": 9222
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.141366314247244e-05,
+      "loss": 0.8652,
+      "step": 9223
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1411717075807594e-05,
+      "loss": 0.849,
+      "step": 9224
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1409770834367917e-05,
+      "loss": 0.9269,
+      "step": 9225
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.140782441817415e-05,
+      "loss": 0.7786,
+      "step": 9226
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.140587782724701e-05,
+      "loss": 0.9078,
+      "step": 9227
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1403931061607224e-05,
+      "loss": 0.94,
+      "step": 9228
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1401984121275545e-05,
+      "loss": 0.9216,
+      "step": 9229
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1400037006272683e-05,
+      "loss": 0.9222,
+      "step": 9230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.139808971661939e-05,
+      "loss": 0.8843,
+      "step": 9231
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1396142252336404e-05,
+      "loss": 0.9048,
+      "step": 9232
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1394194613444436e-05,
+      "loss": 0.9203,
+      "step": 9233
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.139224679996426e-05,
+      "loss": 0.887,
+      "step": 9234
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.139029881191662e-05,
+      "loss": 0.8401,
+      "step": 9235
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.138835064932223e-05,
+      "loss": 0.8693,
+      "step": 9236
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.138640231220186e-05,
+      "loss": 0.8377,
+      "step": 9237
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.138445380057626e-05,
+      "loss": 0.8053,
+      "step": 9238
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1382505114466175e-05,
+      "loss": 0.957,
+      "step": 9239
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.138055625389235e-05,
+      "loss": 0.8607,
+      "step": 9240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.137860721887554e-05,
+      "loss": 0.8975,
+      "step": 9241
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.137665800943652e-05,
+      "loss": 0.8371,
+      "step": 9242
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.137470862559603e-05,
+      "loss": 0.847,
+      "step": 9243
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1372759067374836e-05,
+      "loss": 0.8605,
+      "step": 9244
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1370809334793694e-05,
+      "loss": 0.8601,
+      "step": 9245
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.136885942787337e-05,
+      "loss": 0.9052,
+      "step": 9246
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.136690934663464e-05,
+      "loss": 0.8805,
+      "step": 9247
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1364959091098254e-05,
+      "loss": 0.8796,
+      "step": 9248
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.136300866128499e-05,
+      "loss": 0.828,
+      "step": 9249
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1361058057215617e-05,
+      "loss": 0.8986,
+      "step": 9250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1359107278910915e-05,
+      "loss": 0.8704,
+      "step": 9251
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.135715632639164e-05,
+      "loss": 0.9152,
+      "step": 9252
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.135520519967859e-05,
+      "loss": 0.8669,
+      "step": 9253
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1353253898792536e-05,
+      "loss": 0.8931,
+      "step": 9254
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1351302423754257e-05,
+      "loss": 0.922,
+      "step": 9255
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.134935077458453e-05,
+      "loss": 0.8271,
+      "step": 9256
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1347398951304144e-05,
+      "loss": 0.8206,
+      "step": 9257
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1345446953933884e-05,
+      "loss": 0.8614,
+      "step": 9258
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.134349478249454e-05,
+      "loss": 0.9517,
+      "step": 9259
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.134154243700689e-05,
+      "loss": 0.8466,
+      "step": 9260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1339589917491746e-05,
+      "loss": 0.9335,
+      "step": 9261
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.133763722396988e-05,
+      "loss": 0.8258,
+      "step": 9262
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.13356843564621e-05,
+      "loss": 0.8468,
+      "step": 9263
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.13337313149892e-05,
+      "loss": 0.8401,
+      "step": 9264
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1331778099571985e-05,
+      "loss": 0.8283,
+      "step": 9265
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1329824710231235e-05,
+      "loss": 0.8408,
+      "step": 9266
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.132787114698777e-05,
+      "loss": 0.8416,
+      "step": 9267
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1325917409862394e-05,
+      "loss": 0.8722,
+      "step": 9268
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.132396349887591e-05,
+      "loss": 0.8343,
+      "step": 9269
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1322009414049125e-05,
+      "loss": 0.9744,
+      "step": 9270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1320055155402846e-05,
+      "loss": 0.8723,
+      "step": 9271
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.131810072295789e-05,
+      "loss": 0.9023,
+      "step": 9272
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.131614611673505e-05,
+      "loss": 0.8545,
+      "step": 9273
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.131419133675519e-05,
+      "loss": 0.8175,
+      "step": 9274
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1312236383039086e-05,
+      "loss": 0.9246,
+      "step": 9275
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1310281255607565e-05,
+      "loss": 0.8453,
+      "step": 9276
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.130832595448144e-05,
+      "loss": 0.9232,
+      "step": 9277
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.130637047968157e-05,
+      "loss": 0.8443,
+      "step": 9278
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.130441483122873e-05,
+      "loss": 0.9237,
+      "step": 9279
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1302459009143776e-05,
+      "loss": 1.0017,
+      "step": 9280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1300503013447546e-05,
+      "loss": 0.9303,
+      "step": 9281
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.129854684416083e-05,
+      "loss": 0.8046,
+      "step": 9282
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.129659050130451e-05,
+      "loss": 0.8524,
+      "step": 9283
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1294633984899384e-05,
+      "loss": 0.9503,
+      "step": 9284
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1292677294966304e-05,
+      "loss": 0.9154,
+      "step": 9285
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.12907204315261e-05,
+      "loss": 0.8982,
+      "step": 9286
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1288763394599605e-05,
+      "loss": 0.9389,
+      "step": 9287
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.128680618420768e-05,
+      "loss": 0.8827,
+      "step": 9288
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.128484880037115e-05,
+      "loss": 0.9453,
+      "step": 9289
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.128289124311087e-05,
+      "loss": 0.9048,
+      "step": 9290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1280933512447686e-05,
+      "loss": 0.9006,
+      "step": 9291
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1278975608402435e-05,
+      "loss": 0.9178,
+      "step": 9292
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.127701753099599e-05,
+      "loss": 0.9245,
+      "step": 9293
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.127505928024919e-05,
+      "loss": 0.8905,
+      "step": 9294
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.127310085618288e-05,
+      "loss": 0.8949,
+      "step": 9295
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.127114225881793e-05,
+      "loss": 0.8921,
+      "step": 9296
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.126918348817519e-05,
+      "loss": 0.9155,
+      "step": 9297
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.126722454427553e-05,
+      "loss": 0.9215,
+      "step": 9298
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.12652654271398e-05,
+      "loss": 0.8884,
+      "step": 9299
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1263306136788873e-05,
+      "loss": 0.8374,
+      "step": 9300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1261346673243596e-05,
+      "loss": 0.8595,
+      "step": 9301
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.125938703652486e-05,
+      "loss": 0.8652,
+      "step": 9302
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1257427226653524e-05,
+      "loss": 0.8301,
+      "step": 9303
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.125546724365045e-05,
+      "loss": 0.8446,
+      "step": 9304
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.125350708753654e-05,
+      "loss": 0.8287,
+      "step": 9305
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.125154675833263e-05,
+      "loss": 0.9176,
+      "step": 9306
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.124958625605961e-05,
+      "loss": 0.8416,
+      "step": 9307
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.124762558073837e-05,
+      "loss": 0.9147,
+      "step": 9308
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1245664732389774e-05,
+      "loss": 0.7853,
+      "step": 9309
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.124370371103472e-05,
+      "loss": 0.8997,
+      "step": 9310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.124174251669408e-05,
+      "loss": 0.813,
+      "step": 9311
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.123978114938874e-05,
+      "loss": 0.9207,
+      "step": 9312
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1237819609139596e-05,
+      "loss": 0.8362,
+      "step": 9313
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.123585789596752e-05,
+      "loss": 0.8649,
+      "step": 9314
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.123389600989343e-05,
+      "loss": 0.8507,
+      "step": 9315
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.123193395093819e-05,
+      "loss": 0.8556,
+      "step": 9316
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1229971719122716e-05,
+      "loss": 0.8744,
+      "step": 9317
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.12280093144679e-05,
+      "loss": 0.8313,
+      "step": 9318
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1226046736994627e-05,
+      "loss": 0.9006,
+      "step": 9319
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1224083986723816e-05,
+      "loss": 0.9436,
+      "step": 9320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1222121063676356e-05,
+      "loss": 0.9154,
+      "step": 9321
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.122015796787315e-05,
+      "loss": 0.7701,
+      "step": 9322
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.121819469933512e-05,
+      "loss": 0.8665,
+      "step": 9323
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.121623125808316e-05,
+      "loss": 0.874,
+      "step": 9324
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1214267644138184e-05,
+      "loss": 0.8433,
+      "step": 9325
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.121230385752109e-05,
+      "loss": 0.8889,
+      "step": 9326
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1210339898252814e-05,
+      "loss": 0.8759,
+      "step": 9327
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.120837576635426e-05,
+      "loss": 0.9104,
+      "step": 9328
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1206411461846337e-05,
+      "loss": 0.8768,
+      "step": 9329
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1204446984749975e-05,
+      "loss": 0.8547,
+      "step": 9330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.120248233508609e-05,
+      "loss": 0.8108,
+      "step": 9331
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.12005175128756e-05,
+      "loss": 0.8881,
+      "step": 9332
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.119855251813944e-05,
+      "loss": 0.894,
+      "step": 9333
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.119658735089853e-05,
+      "loss": 0.8599,
+      "step": 9334
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.11946220111738e-05,
+      "loss": 0.8366,
+      "step": 9335
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.119265649898617e-05,
+      "loss": 0.8424,
+      "step": 9336
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.119069081435659e-05,
+      "loss": 0.9146,
+      "step": 9337
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.118872495730597e-05,
+      "loss": 0.934,
+      "step": 9338
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.118675892785527e-05,
+      "loss": 0.8962,
+      "step": 9339
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1184792726025404e-05,
+      "loss": 0.8867,
+      "step": 9340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1182826351837326e-05,
+      "loss": 0.8839,
+      "step": 9341
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.118085980531197e-05,
+      "loss": 0.9224,
+      "step": 9342
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.117889308647028e-05,
+      "loss": 0.8505,
+      "step": 9343
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.117692619533321e-05,
+      "loss": 0.8469,
+      "step": 9344
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.117495913192169e-05,
+      "loss": 0.883,
+      "step": 9345
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.117299189625668e-05,
+      "loss": 0.9339,
+      "step": 9346
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1171024488359124e-05,
+      "loss": 0.8575,
+      "step": 9347
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.116905690824997e-05,
+      "loss": 0.7922,
+      "step": 9348
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.116708915595018e-05,
+      "loss": 0.8099,
+      "step": 9349
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.11651212314807e-05,
+      "loss": 0.9375,
+      "step": 9350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1163153134862496e-05,
+      "loss": 0.8437,
+      "step": 9351
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.116118486611652e-05,
+      "loss": 0.8413,
+      "step": 9352
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.115921642526375e-05,
+      "loss": 0.7952,
+      "step": 9353
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.115724781232513e-05,
+      "loss": 0.8705,
+      "step": 9354
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.115527902732162e-05,
+      "loss": 0.8565,
+      "step": 9355
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.115331007027421e-05,
+      "loss": 0.9005,
+      "step": 9356
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.115134094120385e-05,
+      "loss": 0.8322,
+      "step": 9357
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.114937164013152e-05,
+      "loss": 0.816,
+      "step": 9358
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1147402167078174e-05,
+      "loss": 0.8395,
+      "step": 9359
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.114543252206481e-05,
+      "loss": 0.852,
+      "step": 9360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.114346270511238e-05,
+      "loss": 0.8107,
+      "step": 9361
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.114149271624188e-05,
+      "loss": 0.8361,
+      "step": 9362
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.113952255547429e-05,
+      "loss": 0.8666,
+      "step": 9363
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.113755222283057e-05,
+      "loss": 0.8844,
+      "step": 9364
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.113558171833172e-05,
+      "loss": 0.7994,
+      "step": 9365
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1133611041998725e-05,
+      "loss": 0.889,
+      "step": 9366
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1131640193852564e-05,
+      "loss": 0.9426,
+      "step": 9367
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.112966917391423e-05,
+      "loss": 0.8371,
+      "step": 9368
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.112769798220471e-05,
+      "loss": 0.7746,
+      "step": 9369
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1125726618745e-05,
+      "loss": 0.8931,
+      "step": 9370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1123755083556085e-05,
+      "loss": 0.8516,
+      "step": 9371
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1121783376658974e-05,
+      "loss": 0.8511,
+      "step": 9372
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1119811498074654e-05,
+      "loss": 0.9115,
+      "step": 9373
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.111783944782414e-05,
+      "loss": 0.93,
+      "step": 9374
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.111586722592841e-05,
+      "loss": 0.84,
+      "step": 9375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1113894832408486e-05,
+      "loss": 0.8793,
+      "step": 9376
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1111922267285355e-05,
+      "loss": 0.9373,
+      "step": 9377
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.110994953058004e-05,
+      "loss": 0.7702,
+      "step": 9378
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1107976622313535e-05,
+      "loss": 0.7988,
+      "step": 9379
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1106003542506867e-05,
+      "loss": 0.8825,
+      "step": 9380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.110403029118104e-05,
+      "loss": 0.8817,
+      "step": 9381
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.110205686835706e-05,
+      "loss": 0.8338,
+      "step": 9382
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1100083274055955e-05,
+      "loss": 0.8994,
+      "step": 9383
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.109810950829873e-05,
+      "loss": 0.8444,
+      "step": 9384
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1096135571106425e-05,
+      "loss": 0.8652,
+      "step": 9385
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1094161462500044e-05,
+      "loss": 0.9061,
+      "step": 9386
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.109218718250062e-05,
+      "loss": 0.8959,
+      "step": 9387
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.109021273112916e-05,
+      "loss": 0.959,
+      "step": 9388
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.108823810840671e-05,
+      "loss": 0.8933,
+      "step": 9389
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.108626331435429e-05,
+      "loss": 1.0919,
+      "step": 9390
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1084288348992925e-05,
+      "loss": 0.7637,
+      "step": 9391
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.108231321234367e-05,
+      "loss": 0.8716,
+      "step": 9392
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.108033790442753e-05,
+      "loss": 0.9076,
+      "step": 9393
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.107836242526556e-05,
+      "loss": 0.9347,
+      "step": 9394
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.107638677487879e-05,
+      "loss": 0.88,
+      "step": 9395
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.107441095328826e-05,
+      "loss": 0.9058,
+      "step": 9396
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1072434960515016e-05,
+      "loss": 0.9154,
+      "step": 9397
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.107045879658009e-05,
+      "loss": 0.9153,
+      "step": 9398
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1068482461504545e-05,
+      "loss": 0.9249,
+      "step": 9399
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.106650595530941e-05,
+      "loss": 0.7774,
+      "step": 9400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.106452927801574e-05,
+      "loss": 0.8549,
+      "step": 9401
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1062552429644585e-05,
+      "loss": 0.9044,
+      "step": 9402
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.106057541021701e-05,
+      "loss": 0.7608,
+      "step": 9403
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.105859821975404e-05,
+      "loss": 0.9171,
+      "step": 9404
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1056620858276765e-05,
+      "loss": 0.8686,
+      "step": 9405
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.105464332580622e-05,
+      "loss": 0.9262,
+      "step": 9406
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.105266562236346e-05,
+      "loss": 0.8577,
+      "step": 9407
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1050687747969565e-05,
+      "loss": 0.9042,
+      "step": 9408
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1048709702645594e-05,
+      "loss": 0.9671,
+      "step": 9409
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1046731486412606e-05,
+      "loss": 0.8005,
+      "step": 9410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1044753099291664e-05,
+      "loss": 0.8479,
+      "step": 9411
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1042774541303854e-05,
+      "loss": 0.9559,
+      "step": 9412
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.104079581247023e-05,
+      "loss": 0.9323,
+      "step": 9413
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.103881691281187e-05,
+      "loss": 0.8557,
+      "step": 9414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.103683784234984e-05,
+      "loss": 0.86,
+      "step": 9415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.103485860110523e-05,
+      "loss": 0.9566,
+      "step": 9416
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.103287918909911e-05,
+      "loss": 0.8975,
+      "step": 9417
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.103089960635256e-05,
+      "loss": 0.9346,
+      "step": 9418
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.102891985288666e-05,
+      "loss": 0.7968,
+      "step": 9419
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.10269399287225e-05,
+      "loss": 0.8381,
+      "step": 9420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.102495983388116e-05,
+      "loss": 0.8995,
+      "step": 9421
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.102297956838371e-05,
+      "loss": 0.8157,
+      "step": 9422
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.102099913225127e-05,
+      "loss": 0.8483,
+      "step": 9423
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1019018525504916e-05,
+      "loss": 0.9286,
+      "step": 9424
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.101703774816574e-05,
+      "loss": 0.959,
+      "step": 9425
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1015056800254834e-05,
+      "loss": 0.9451,
+      "step": 9426
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.10130756817933e-05,
+      "loss": 0.8481,
+      "step": 9427
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.101109439280223e-05,
+      "loss": 0.943,
+      "step": 9428
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.100911293330272e-05,
+      "loss": 0.8732,
+      "step": 9429
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1007131303315885e-05,
+      "loss": 0.8888,
+      "step": 9430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.100514950286281e-05,
+      "loss": 0.8903,
+      "step": 9431
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.100316753196461e-05,
+      "loss": 0.8678,
+      "step": 9432
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.10011853906424e-05,
+      "loss": 0.9163,
+      "step": 9433
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.099920307891728e-05,
+      "loss": 0.8966,
+      "step": 9434
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.099722059681036e-05,
+      "loss": 0.8298,
+      "step": 9435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0995237944342746e-05,
+      "loss": 0.8452,
+      "step": 9436
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.099325512153556e-05,
+      "loss": 0.9,
+      "step": 9437
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.099127212840992e-05,
+      "loss": 0.8456,
+      "step": 9438
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.098928896498694e-05,
+      "loss": 0.9801,
+      "step": 9439
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.098730563128774e-05,
+      "loss": 0.8876,
+      "step": 9440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.098532212733344e-05,
+      "loss": 0.9147,
+      "step": 9441
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0983338453145166e-05,
+      "loss": 0.9293,
+      "step": 9442
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.098135460874404e-05,
+      "loss": 0.8221,
+      "step": 9443
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.097937059415119e-05,
+      "loss": 0.9419,
+      "step": 9444
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.097738640938774e-05,
+      "loss": 0.8254,
+      "step": 9445
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.097540205447483e-05,
+      "loss": 0.8837,
+      "step": 9446
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.097341752943359e-05,
+      "loss": 0.9103,
+      "step": 9447
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.097143283428515e-05,
+      "loss": 0.8597,
+      "step": 9448
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.096944796905064e-05,
+      "loss": 0.8605,
+      "step": 9449
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.096746293375121e-05,
+      "loss": 0.8758,
+      "step": 9450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0965477728408e-05,
+      "loss": 0.897,
+      "step": 9451
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.096349235304213e-05,
+      "loss": 0.9602,
+      "step": 9452
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.096150680767476e-05,
+      "loss": 0.8349,
+      "step": 9453
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.095952109232705e-05,
+      "loss": 0.9491,
+      "step": 9454
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0957535207020115e-05,
+      "loss": 0.8606,
+      "step": 9455
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0955549151775116e-05,
+      "loss": 0.8678,
+      "step": 9456
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.095356292661322e-05,
+      "loss": 0.8861,
+      "step": 9457
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.095157653155555e-05,
+      "loss": 0.8849,
+      "step": 9458
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.094958996662328e-05,
+      "loss": 0.8326,
+      "step": 9459
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.094760323183756e-05,
+      "loss": 0.9347,
+      "step": 9460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.094561632721955e-05,
+      "loss": 0.8946,
+      "step": 9461
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.094362925279041e-05,
+      "loss": 0.9043,
+      "step": 9462
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.094164200857128e-05,
+      "loss": 0.7448,
+      "step": 9463
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0939654594583354e-05,
+      "loss": 0.9327,
+      "step": 9464
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0937667010847784e-05,
+      "loss": 0.9144,
+      "step": 9465
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.093567925738574e-05,
+      "loss": 0.8494,
+      "step": 9466
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0933691334218384e-05,
+      "loss": 0.849,
+      "step": 9467
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0931703241366885e-05,
+      "loss": 0.9006,
+      "step": 9468
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0929714978852416e-05,
+      "loss": 0.8161,
+      "step": 9469
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.092772654669617e-05,
+      "loss": 0.8906,
+      "step": 9470
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.092573794491928e-05,
+      "loss": 0.9067,
+      "step": 9471
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0923749173542966e-05,
+      "loss": 0.8857,
+      "step": 9472
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.09217602325884e-05,
+      "loss": 0.9145,
+      "step": 9473
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0919771122076734e-05,
+      "loss": 0.8407,
+      "step": 9474
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0917781842029176e-05,
+      "loss": 0.8865,
+      "step": 9475
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.091579239246692e-05,
+      "loss": 0.8743,
+      "step": 9476
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.091380277341113e-05,
+      "loss": 0.8984,
+      "step": 9477
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.091181298488299e-05,
+      "loss": 0.8477,
+      "step": 9478
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.090982302690371e-05,
+      "loss": 0.946,
+      "step": 9479
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.090783289949448e-05,
+      "loss": 0.8646,
+      "step": 9480
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0905842602676484e-05,
+      "loss": 0.8653,
+      "step": 9481
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.090385213647091e-05,
+      "loss": 0.8013,
+      "step": 9482
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.090186150089898e-05,
+      "loss": 0.8805,
+      "step": 9483
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0899870695981876e-05,
+      "loss": 0.8393,
+      "step": 9484
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.089787972174079e-05,
+      "loss": 0.9215,
+      "step": 9485
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.089588857819696e-05,
+      "loss": 0.9003,
+      "step": 9486
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0893897265371557e-05,
+      "loss": 0.9233,
+      "step": 9487
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.089190578328579e-05,
+      "loss": 0.8707,
+      "step": 9488
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0889914131960884e-05,
+      "loss": 0.9163,
+      "step": 9489
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.088792231141804e-05,
+      "loss": 0.8935,
+      "step": 9490
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.088593032167846e-05,
+      "loss": 0.8835,
+      "step": 9491
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.088393816276337e-05,
+      "loss": 0.8432,
+      "step": 9492
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0881945834693977e-05,
+      "loss": 0.7845,
+      "step": 9493
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.087995333749151e-05,
+      "loss": 0.8075,
+      "step": 9494
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0877960671177186e-05,
+      "loss": 0.9417,
+      "step": 9495
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.087596783577221e-05,
+      "loss": 0.8562,
+      "step": 9496
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0873974831297824e-05,
+      "loss": 0.9082,
+      "step": 9497
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.087198165777524e-05,
+      "loss": 0.8946,
+      "step": 9498
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0869988315225684e-05,
+      "loss": 0.878,
+      "step": 9499
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.086799480367039e-05,
+      "loss": 0.7567,
+      "step": 9500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0866001123130585e-05,
+      "loss": 0.8837,
+      "step": 9501
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.08640072736275e-05,
+      "loss": 0.814,
+      "step": 9502
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.086201325518237e-05,
+      "loss": 0.9718,
+      "step": 9503
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.086001906781642e-05,
+      "loss": 0.8289,
+      "step": 9504
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0858024711550915e-05,
+      "loss": 0.7887,
+      "step": 9505
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0856030186407054e-05,
+      "loss": 0.7463,
+      "step": 9506
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.085403549240612e-05,
+      "loss": 0.835,
+      "step": 9507
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.085204062956932e-05,
+      "loss": 0.9265,
+      "step": 9508
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.085004559791791e-05,
+      "loss": 0.8311,
+      "step": 9509
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0848050397473134e-05,
+      "loss": 0.8625,
+      "step": 9510
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0846055028256245e-05,
+      "loss": 0.8223,
+      "step": 9511
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.08440594902885e-05,
+      "loss": 0.8227,
+      "step": 9512
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0842063783591134e-05,
+      "loss": 0.8353,
+      "step": 9513
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0840067908185404e-05,
+      "loss": 0.9617,
+      "step": 9514
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.083807186409257e-05,
+      "loss": 0.8489,
+      "step": 9515
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0836075651333884e-05,
+      "loss": 0.8411,
+      "step": 9516
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0834079269930615e-05,
+      "loss": 0.9282,
+      "step": 9517
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.083208271990401e-05,
+      "loss": 0.9216,
+      "step": 9518
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.083008600127533e-05,
+      "loss": 0.8047,
+      "step": 9519
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.082808911406585e-05,
+      "loss": 0.9158,
+      "step": 9520
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.082609205829683e-05,
+      "loss": 0.8504,
+      "step": 9521
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.082409483398953e-05,
+      "loss": 0.856,
+      "step": 9522
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.082209744116524e-05,
+      "loss": 0.8971,
+      "step": 9523
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.082009987984521e-05,
+      "loss": 0.8523,
+      "step": 9524
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.081810215005073e-05,
+      "loss": 0.9336,
+      "step": 9525
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0816104251803056e-05,
+      "loss": 0.9076,
+      "step": 9526
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.081410618512348e-05,
+      "loss": 0.8462,
+      "step": 9527
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.081210795003327e-05,
+      "loss": 0.805,
+      "step": 9528
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.081010954655371e-05,
+      "loss": 0.8533,
+      "step": 9529
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.080811097470608e-05,
+      "loss": 0.8929,
+      "step": 9530
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.080611223451167e-05,
+      "loss": 0.9025,
+      "step": 9531
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.080411332599176e-05,
+      "loss": 0.8363,
+      "step": 9532
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.080211424916764e-05,
+      "loss": 0.8271,
+      "step": 9533
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.080011500406059e-05,
+      "loss": 0.9458,
+      "step": 9534
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.079811559069192e-05,
+      "loss": 0.8927,
+      "step": 9535
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0796116009082895e-05,
+      "loss": 0.8363,
+      "step": 9536
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.079411625925484e-05,
+      "loss": 0.9304,
+      "step": 9537
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.079211634122903e-05,
+      "loss": 0.8404,
+      "step": 9538
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.079011625502676e-05,
+      "loss": 0.8496,
+      "step": 9539
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0788116000669346e-05,
+      "loss": 0.8236,
+      "step": 9540
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.078611557817808e-05,
+      "loss": 0.9525,
+      "step": 9541
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0784114987574266e-05,
+      "loss": 0.9106,
+      "step": 9542
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.078211422887921e-05,
+      "loss": 0.9456,
+      "step": 9543
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.078011330211422e-05,
+      "loss": 0.9274,
+      "step": 9544
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.077811220730061e-05,
+      "loss": 0.8226,
+      "step": 9545
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.077611094445968e-05,
+      "loss": 0.9124,
+      "step": 9546
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0774109513612744e-05,
+      "loss": 0.8798,
+      "step": 9547
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.077210791478112e-05,
+      "loss": 0.8503,
+      "step": 9548
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.077010614798612e-05,
+      "loss": 0.8793,
+      "step": 9549
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.076810421324906e-05,
+      "loss": 0.8543,
+      "step": 9550
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.076610211059127e-05,
+      "loss": 0.9172,
+      "step": 9551
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0764099840034065e-05,
+      "loss": 0.955,
+      "step": 9552
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.076209740159876e-05,
+      "loss": 0.8599,
+      "step": 9553
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.076009479530669e-05,
+      "loss": 0.838,
+      "step": 9554
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.075809202117918e-05,
+      "loss": 0.8256,
+      "step": 9555
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.075608907923756e-05,
+      "loss": 0.8673,
+      "step": 9556
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.075408596950315e-05,
+      "loss": 0.859,
+      "step": 9557
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.07520826919973e-05,
+      "loss": 0.8569,
+      "step": 9558
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0750079246741325e-05,
+      "loss": 0.9312,
+      "step": 9559
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.074807563375657e-05,
+      "loss": 0.9762,
+      "step": 9560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.074607185306437e-05,
+      "loss": 0.8463,
+      "step": 9561
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.074406790468607e-05,
+      "loss": 0.8298,
+      "step": 9562
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.074206378864299e-05,
+      "loss": 0.8603,
+      "step": 9563
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.07400595049565e-05,
+      "loss": 0.8757,
+      "step": 9564
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.073805505364793e-05,
+      "loss": 0.8007,
+      "step": 9565
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0736050434738624e-05,
+      "loss": 0.8658,
+      "step": 9566
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0734045648249944e-05,
+      "loss": 0.8991,
+      "step": 9567
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0732040694203225e-05,
+      "loss": 0.9088,
+      "step": 9568
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0730035572619826e-05,
+      "loss": 0.8687,
+      "step": 9569
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.072803028352109e-05,
+      "loss": 1.0259,
+      "step": 9570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.072602482692839e-05,
+      "loss": 0.8901,
+      "step": 9571
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0724019202863066e-05,
+      "loss": 0.8774,
+      "step": 9572
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0722013411346483e-05,
+      "loss": 0.8757,
+      "step": 9573
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.072000745240001e-05,
+      "loss": 0.8416,
+      "step": 9574
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0718001326045e-05,
+      "loss": 0.9127,
+      "step": 9575
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.071599503230281e-05,
+      "loss": 0.8067,
+      "step": 9576
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0713988571194815e-05,
+      "loss": 0.9072,
+      "step": 9577
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.071198194274239e-05,
+      "loss": 0.8602,
+      "step": 9578
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.07099751469669e-05,
+      "loss": 0.9187,
+      "step": 9579
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0707968183889704e-05,
+      "loss": 0.9612,
+      "step": 9580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0705961053532184e-05,
+      "loss": 0.781,
+      "step": 9581
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0703953755915716e-05,
+      "loss": 0.8883,
+      "step": 9582
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.070194629106168e-05,
+      "loss": 0.8093,
+      "step": 9583
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.069993865899144e-05,
+      "loss": 0.8556,
+      "step": 9584
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0697930859726384e-05,
+      "loss": 0.8218,
+      "step": 9585
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0695922893287897e-05,
+      "loss": 0.8916,
+      "step": 9586
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0693914759697364e-05,
+      "loss": 0.8279,
+      "step": 9587
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.069190645897617e-05,
+      "loss": 0.8857,
+      "step": 9588
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.068989799114569e-05,
+      "loss": 0.8097,
+      "step": 9589
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0687889356227326e-05,
+      "loss": 0.8112,
+      "step": 9590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0685880554242465e-05,
+      "loss": 0.8381,
+      "step": 9591
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0683871585212494e-05,
+      "loss": 0.9057,
+      "step": 9592
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.068186244915882e-05,
+      "loss": 0.8328,
+      "step": 9593
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.067985314610283e-05,
+      "loss": 0.842,
+      "step": 9594
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0677843676065916e-05,
+      "loss": 0.7981,
+      "step": 9595
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.067583403906949e-05,
+      "loss": 0.8685,
+      "step": 9596
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.067382423513495e-05,
+      "loss": 0.8999,
+      "step": 9597
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.06718142642837e-05,
+      "loss": 0.9102,
+      "step": 9598
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0669804126537147e-05,
+      "loss": 0.8429,
+      "step": 9599
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.066779382191669e-05,
+      "loss": 0.8746,
+      "step": 9600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0665783350443734e-05,
+      "loss": 0.8761,
+      "step": 9601
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0663772712139696e-05,
+      "loss": 0.9151,
+      "step": 9602
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0661761907026e-05,
+      "loss": 0.9416,
+      "step": 9603
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.065975093512404e-05,
+      "loss": 0.834,
+      "step": 9604
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.065773979645523e-05,
+      "loss": 0.8897,
+      "step": 9605
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.065572849104101e-05,
+      "loss": 0.8153,
+      "step": 9606
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.065371701890278e-05,
+      "loss": 0.8364,
+      "step": 9607
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.065170538006198e-05,
+      "loss": 0.9208,
+      "step": 9608
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.064969357454001e-05,
+      "loss": 0.9292,
+      "step": 9609
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.06476816023583e-05,
+      "loss": 0.8348,
+      "step": 9610
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.06456694635383e-05,
+      "loss": 0.8001,
+      "step": 9611
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.06436571581014e-05,
+      "loss": 0.8976,
+      "step": 9612
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.064164468606905e-05,
+      "loss": 0.8163,
+      "step": 9613
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.063963204746268e-05,
+      "loss": 0.8599,
+      "step": 9614
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.063761924230373e-05,
+      "loss": 0.8393,
+      "step": 9615
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.063560627061363e-05,
+      "loss": 0.9189,
+      "step": 9616
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.063359313241382e-05,
+      "loss": 0.8237,
+      "step": 9617
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.063157982772573e-05,
+      "loss": 0.7922,
+      "step": 9618
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.062956635657079e-05,
+      "loss": 0.9858,
+      "step": 9619
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0627552718970476e-05,
+      "loss": 0.8454,
+      "step": 9620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.06255389149462e-05,
+      "loss": 0.859,
+      "step": 9621
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0623524944519433e-05,
+      "loss": 0.8833,
+      "step": 9622
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.06215108077116e-05,
+      "loss": 0.8371,
+      "step": 9623
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0619496504544176e-05,
+      "loss": 0.8313,
+      "step": 9624
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.061748203503858e-05,
+      "loss": 0.8343,
+      "step": 9625
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.061546739921629e-05,
+      "loss": 0.9463,
+      "step": 9626
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0613452597098755e-05,
+      "loss": 0.8759,
+      "step": 9627
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.061143762870743e-05,
+      "loss": 0.8529,
+      "step": 9628
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.060942249406377e-05,
+      "loss": 0.8251,
+      "step": 9629
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0607407193189234e-05,
+      "loss": 0.8831,
+      "step": 9630
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.060539172610529e-05,
+      "loss": 0.9004,
+      "step": 9631
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0603376092833394e-05,
+      "loss": 0.8211,
+      "step": 9632
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.060136029339502e-05,
+      "loss": 0.8244,
+      "step": 9633
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.059934432781163e-05,
+      "loss": 0.8196,
+      "step": 9634
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.059732819610469e-05,
+      "loss": 0.9072,
+      "step": 9635
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.059531189829568e-05,
+      "loss": 0.8653,
+      "step": 9636
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.059329543440607e-05,
+      "loss": 0.9118,
+      "step": 9637
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.059127880445732e-05,
+      "loss": 0.7859,
+      "step": 9638
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0589262008470916e-05,
+      "loss": 0.8321,
+      "step": 9639
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.058724504646834e-05,
+      "loss": 0.7792,
+      "step": 9640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.058522791847107e-05,
+      "loss": 0.8992,
+      "step": 9641
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0583210624500575e-05,
+      "loss": 0.9011,
+      "step": 9642
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.058119316457836e-05,
+      "loss": 0.9245,
+      "step": 9643
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0579175538725895e-05,
+      "loss": 0.8657,
+      "step": 9644
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.057715774696467e-05,
+      "loss": 0.7272,
+      "step": 9645
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.057513978931616e-05,
+      "loss": 0.8513,
+      "step": 9646
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0573121665801876e-05,
+      "loss": 0.8149,
+      "step": 9647
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.057110337644331e-05,
+      "loss": 0.9279,
+      "step": 9648
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0569084921261935e-05,
+      "loss": 0.7694,
+      "step": 9649
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.056706630027926e-05,
+      "loss": 0.876,
+      "step": 9650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.056504751351677e-05,
+      "loss": 0.8038,
+      "step": 9651
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.056302856099599e-05,
+      "loss": 0.8732,
+      "step": 9652
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0561009442738394e-05,
+      "loss": 0.9351,
+      "step": 9653
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0558990158765506e-05,
+      "loss": 0.9679,
+      "step": 9654
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0556970709098805e-05,
+      "loss": 0.8935,
+      "step": 9655
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.055495109375982e-05,
+      "loss": 0.7498,
+      "step": 9656
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.055293131277005e-05,
+      "loss": 0.7945,
+      "step": 9657
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.055091136615099e-05,
+      "loss": 0.7761,
+      "step": 9658
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0548891253924183e-05,
+      "loss": 0.8209,
+      "step": 9659
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.054687097611112e-05,
+      "loss": 0.8224,
+      "step": 9660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.054485053273332e-05,
+      "loss": 0.7903,
+      "step": 9661
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.054282992381229e-05,
+      "loss": 0.9072,
+      "step": 9662
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0540809149369565e-05,
+      "loss": 0.8412,
+      "step": 9663
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.053878820942666e-05,
+      "loss": 0.8783,
+      "step": 9664
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0536767104005086e-05,
+      "loss": 0.9038,
+      "step": 9665
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.053474583312638e-05,
+      "loss": 0.8736,
+      "step": 9666
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0532724396812064e-05,
+      "loss": 0.8397,
+      "step": 9667
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.053070279508366e-05,
+      "loss": 0.9675,
+      "step": 9668
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.05286810279627e-05,
+      "loss": 0.9007,
+      "step": 9669
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.052665909547072e-05,
+      "loss": 1.0249,
+      "step": 9670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.052463699762924e-05,
+      "loss": 0.8494,
+      "step": 9671
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0522614734459804e-05,
+      "loss": 0.8171,
+      "step": 9672
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.052059230598395e-05,
+      "loss": 0.848,
+      "step": 9673
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0518569712223206e-05,
+      "loss": 0.9733,
+      "step": 9674
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.051654695319911e-05,
+      "loss": 0.815,
+      "step": 9675
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.051452402893322e-05,
+      "loss": 0.9053,
+      "step": 9676
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.051250093944706e-05,
+      "loss": 0.8708,
+      "step": 9677
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.051047768476219e-05,
+      "loss": 0.9164,
+      "step": 9678
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.050845426490015e-05,
+      "loss": 0.8848,
+      "step": 9679
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0506430679882485e-05,
+      "loss": 0.8761,
+      "step": 9680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.050440692973074e-05,
+      "loss": 0.8499,
+      "step": 9681
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.050238301446648e-05,
+      "loss": 0.8585,
+      "step": 9682
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.050035893411126e-05,
+      "loss": 0.9514,
+      "step": 9683
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.049833468868662e-05,
+      "loss": 0.9022,
+      "step": 9684
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0496310278214126e-05,
+      "loss": 0.7999,
+      "step": 9685
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0494285702715327e-05,
+      "loss": 0.8781,
+      "step": 9686
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0492260962211806e-05,
+      "loss": 0.8827,
+      "step": 9687
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.04902360567251e-05,
+      "loss": 0.8315,
+      "step": 9688
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.04882109862768e-05,
+      "loss": 0.8957,
+      "step": 9689
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.048618575088844e-05,
+      "loss": 0.8386,
+      "step": 9690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0484160350581616e-05,
+      "loss": 0.8509,
+      "step": 9691
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.048213478537787e-05,
+      "loss": 0.7961,
+      "step": 9692
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.04801090552988e-05,
+      "loss": 0.9272,
+      "step": 9693
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0478083160365964e-05,
+      "loss": 0.9099,
+      "step": 9694
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.047605710060094e-05,
+      "loss": 0.9007,
+      "step": 9695
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.04740308760253e-05,
+      "loss": 0.8043,
+      "step": 9696
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0472004486660624e-05,
+      "loss": 0.8726,
+      "step": 9697
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.04699779325285e-05,
+      "loss": 0.8307,
+      "step": 9698
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.04679512136505e-05,
+      "loss": 0.8288,
+      "step": 9699
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.046592433004821e-05,
+      "loss": 0.8524,
+      "step": 9700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.046389728174322e-05,
+      "loss": 0.8012,
+      "step": 9701
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0461870068757104e-05,
+      "loss": 0.9186,
+      "step": 9702
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0459842691111465e-05,
+      "loss": 0.7866,
+      "step": 9703
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.045781514882788e-05,
+      "loss": 0.7878,
+      "step": 9704
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.045578744192796e-05,
+      "loss": 0.8587,
+      "step": 9705
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.045375957043328e-05,
+      "loss": 0.9488,
+      "step": 9706
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.045173153436544e-05,
+      "loss": 0.8339,
+      "step": 9707
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0449703333746046e-05,
+      "loss": 0.8135,
+      "step": 9708
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0447674968596695e-05,
+      "loss": 0.8245,
+      "step": 9709
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0445646438938975e-05,
+      "loss": 0.9494,
+      "step": 9710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0443617744794505e-05,
+      "loss": 0.8752,
+      "step": 9711
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.044158888618488e-05,
+      "loss": 0.8421,
+      "step": 9712
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.04395598631317e-05,
+      "loss": 0.858,
+      "step": 9713
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.043753067565659e-05,
+      "loss": 0.8706,
+      "step": 9714
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.043550132378115e-05,
+      "loss": 0.8486,
+      "step": 9715
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.043347180752699e-05,
+      "loss": 0.8285,
+      "step": 9716
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.043144212691572e-05,
+      "loss": 0.8429,
+      "step": 9717
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.042941228196897e-05,
+      "loss": 1.0005,
+      "step": 9718
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0427382272708345e-05,
+      "loss": 0.8022,
+      "step": 9719
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.042535209915546e-05,
+      "loss": 0.8155,
+      "step": 9720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.042332176133194e-05,
+      "loss": 0.8311,
+      "step": 9721
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.042129125925941e-05,
+      "loss": 0.8062,
+      "step": 9722
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.041926059295949e-05,
+      "loss": 0.9424,
+      "step": 9723
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.04172297624538e-05,
+      "loss": 0.97,
+      "step": 9724
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0415198767763986e-05,
+      "loss": 0.8551,
+      "step": 9725
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.041316760891165e-05,
+      "loss": 0.7618,
+      "step": 9726
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0411136285918447e-05,
+      "loss": 0.8621,
+      "step": 9727
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0409104798806e-05,
+      "loss": 0.9069,
+      "step": 9728
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.040707314759593e-05,
+      "loss": 0.8837,
+      "step": 9729
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.040504133230989e-05,
+      "loss": 0.8022,
+      "step": 9730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0403009352969516e-05,
+      "loss": 0.8194,
+      "step": 9731
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0400977209596445e-05,
+      "loss": 0.8478,
+      "step": 9732
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.039894490221231e-05,
+      "loss": 0.8063,
+      "step": 9733
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.039691243083876e-05,
+      "loss": 0.9351,
+      "step": 9734
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0394879795497455e-05,
+      "loss": 0.7515,
+      "step": 9735
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.039284699621001e-05,
+      "loss": 0.9937,
+      "step": 9736
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.03908140329981e-05,
+      "loss": 0.8121,
+      "step": 9737
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.038878090588336e-05,
+      "loss": 0.837,
+      "step": 9738
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.038674761488745e-05,
+      "loss": 0.9367,
+      "step": 9739
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.038471416003202e-05,
+      "loss": 0.9742,
+      "step": 9740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.038268054133871e-05,
+      "loss": 0.8728,
+      "step": 9741
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.038064675882921e-05,
+      "loss": 0.8735,
+      "step": 9742
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.037861281252515e-05,
+      "loss": 0.856,
+      "step": 9743
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.03765787024482e-05,
+      "loss": 0.8756,
+      "step": 9744
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.037454442862003e-05,
+      "loss": 0.8526,
+      "step": 9745
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0372509991062285e-05,
+      "loss": 0.9499,
+      "step": 9746
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.037047538979665e-05,
+      "loss": 0.8656,
+      "step": 9747
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0368440624844774e-05,
+      "loss": 0.9175,
+      "step": 9748
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.036640569622835e-05,
+      "loss": 0.8876,
+      "step": 9749
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0364370603969024e-05,
+      "loss": 0.8826,
+      "step": 9750
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0362335348088475e-05,
+      "loss": 0.8305,
+      "step": 9751
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.036029992860839e-05,
+      "loss": 0.8735,
+      "step": 9752
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0358264345550436e-05,
+      "loss": 0.8365,
+      "step": 9753
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.035622859893629e-05,
+      "loss": 0.8626,
+      "step": 9754
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.035419268878763e-05,
+      "loss": 0.7764,
+      "step": 9755
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0352156615126144e-05,
+      "loss": 0.8464,
+      "step": 9756
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0350120377973504e-05,
+      "loss": 0.9117,
+      "step": 9757
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.03480839773514e-05,
+      "loss": 0.8713,
+      "step": 9758
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0346047413281536e-05,
+      "loss": 0.7616,
+      "step": 9759
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0344010685785575e-05,
+      "loss": 0.896,
+      "step": 9760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.034197379488521e-05,
+      "loss": 0.7784,
+      "step": 9761
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0339936740602134e-05,
+      "loss": 0.7976,
+      "step": 9762
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0337899522958056e-05,
+      "loss": 0.7607,
+      "step": 9763
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.033586214197466e-05,
+      "loss": 0.906,
+      "step": 9764
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.033382459767364e-05,
+      "loss": 0.908,
+      "step": 9765
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0331786890076696e-05,
+      "loss": 0.9445,
+      "step": 9766
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0329749019205525e-05,
+      "loss": 0.8619,
+      "step": 9767
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.032771098508184e-05,
+      "loss": 0.9423,
+      "step": 9768
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0325672787727335e-05,
+      "loss": 0.8573,
+      "step": 9769
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0323634427163724e-05,
+      "loss": 0.8009,
+      "step": 9770
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.032159590341271e-05,
+      "loss": 0.8538,
+      "step": 9771
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0319557216496e-05,
+      "loss": 0.8804,
+      "step": 9772
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.03175183664353e-05,
+      "loss": 0.845,
+      "step": 9773
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.031547935325234e-05,
+      "loss": 0.8918,
+      "step": 9774
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.031344017696882e-05,
+      "loss": 0.8415,
+      "step": 9775
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.031140083760645e-05,
+      "loss": 0.803,
+      "step": 9776
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.030936133518697e-05,
+      "loss": 0.9146,
+      "step": 9777
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.030732166973208e-05,
+      "loss": 0.89,
+      "step": 9778
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0305281841263504e-05,
+      "loss": 0.8666,
+      "step": 9779
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.030324184980298e-05,
+      "loss": 0.814,
+      "step": 9780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.030120169537221e-05,
+      "loss": 0.8651,
+      "step": 9781
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0299161377992934e-05,
+      "loss": 0.9867,
+      "step": 9782
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.029712089768688e-05,
+      "loss": 0.8162,
+      "step": 9783
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.029508025447577e-05,
+      "loss": 0.875,
+      "step": 9784
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0293039448381345e-05,
+      "loss": 0.8652,
+      "step": 9785
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0290998479425336e-05,
+      "loss": 0.8835,
+      "step": 9786
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028895734762947e-05,
+      "loss": 0.8206,
+      "step": 9787
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.02869160530155e-05,
+      "loss": 0.8833,
+      "step": 9788
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028487459560515e-05,
+      "loss": 0.8767,
+      "step": 9789
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028283297542017e-05,
+      "loss": 0.8396,
+      "step": 9790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028079119248229e-05,
+      "loss": 0.8525,
+      "step": 9791
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.027874924681326e-05,
+      "loss": 0.8926,
+      "step": 9792
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0276707138434836e-05,
+      "loss": 0.8105,
+      "step": 9793
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.027466486736875e-05,
+      "loss": 0.8422,
+      "step": 9794
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0272622433636754e-05,
+      "loss": 0.7884,
+      "step": 9795
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.02705798372606e-05,
+      "loss": 0.9067,
+      "step": 9796
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0268537078262047e-05,
+      "loss": 0.9036,
+      "step": 9797
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.026649415666285e-05,
+      "loss": 0.8489,
+      "step": 9798
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.026445107248474e-05,
+      "loss": 0.9013,
+      "step": 9799
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.026240782574952e-05,
+      "loss": 0.8609,
+      "step": 9800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0260364416478904e-05,
+      "loss": 0.9325,
+      "step": 9801
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025832084469468e-05,
+      "loss": 0.8935,
+      "step": 9802
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025627711041859e-05,
+      "loss": 0.8251,
+      "step": 9803
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025423321367242e-05,
+      "loss": 0.845,
+      "step": 9804
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025218915447793e-05,
+      "loss": 0.8133,
+      "step": 9805
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025014493285687e-05,
+      "loss": 0.828,
+      "step": 9806
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.024810054883104e-05,
+      "loss": 0.885,
+      "step": 9807
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0246056002422196e-05,
+      "loss": 0.8709,
+      "step": 9808
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0244011293652106e-05,
+      "loss": 0.7992,
+      "step": 9809
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0241966422542554e-05,
+      "loss": 1.0063,
+      "step": 9810
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023992138911531e-05,
+      "loss": 0.8488,
+      "step": 9811
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023787619339217e-05,
+      "loss": 0.8352,
+      "step": 9812
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0235830835394885e-05,
+      "loss": 0.887,
+      "step": 9813
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023378531514525e-05,
+      "loss": 0.8284,
+      "step": 9814
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023173963266505e-05,
+      "loss": 0.8289,
+      "step": 9815
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.022969378797608e-05,
+      "loss": 0.8092,
+      "step": 9816
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.022764778110011e-05,
+      "loss": 0.8311,
+      "step": 9817
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.022560161205894e-05,
+      "loss": 0.845,
+      "step": 9818
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.022355528087435e-05,
+      "loss": 0.8166,
+      "step": 9819
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.022150878756815e-05,
+      "loss": 0.8254,
+      "step": 9820
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.021946213216211e-05,
+      "loss": 0.8934,
+      "step": 9821
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0217415314678046e-05,
+      "loss": 0.883,
+      "step": 9822
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.021536833513775e-05,
+      "loss": 0.9015,
+      "step": 9823
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0213321193563014e-05,
+      "loss": 0.8088,
+      "step": 9824
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.021127388997564e-05,
+      "loss": 0.8051,
+      "step": 9825
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020922642439744e-05,
+      "loss": 0.8697,
+      "step": 9826
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0207178796850206e-05,
+      "loss": 0.9338,
+      "step": 9827
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020513100735575e-05,
+      "loss": 0.8583,
+      "step": 9828
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0203083055935896e-05,
+      "loss": 0.9465,
+      "step": 9829
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0201034942612426e-05,
+      "loss": 0.8478,
+      "step": 9830
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0198986667407164e-05,
+      "loss": 0.8805,
+      "step": 9831
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.019693823034192e-05,
+      "loss": 0.7724,
+      "step": 9832
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.019488963143851e-05,
+      "loss": 0.8623,
+      "step": 9833
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.019284087071875e-05,
+      "loss": 0.8207,
+      "step": 9834
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.019079194820446e-05,
+      "loss": 1.0018,
+      "step": 9835
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0188742863917457e-05,
+      "loss": 0.8713,
+      "step": 9836
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.018669361787957e-05,
+      "loss": 0.8558,
+      "step": 9837
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.01846442101126e-05,
+      "loss": 0.8236,
+      "step": 9838
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.01825946406384e-05,
+      "loss": 0.7934,
+      "step": 9839
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.018054490947879e-05,
+      "loss": 0.8413,
+      "step": 9840
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.017849501665558e-05,
+      "loss": 0.828,
+      "step": 9841
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0176444962190625e-05,
+      "loss": 0.9,
+      "step": 9842
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0174394746105736e-05,
+      "loss": 0.9482,
+      "step": 9843
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.017234436842276e-05,
+      "loss": 0.8548,
+      "step": 9844
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.017029382916353e-05,
+      "loss": 0.8792,
+      "step": 9845
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0168243128349876e-05,
+      "loss": 0.8244,
+      "step": 9846
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.016619226600364e-05,
+      "loss": 0.8411,
+      "step": 9847
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.016414124214667e-05,
+      "loss": 0.8978,
+      "step": 9848
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.01620900568008e-05,
+      "loss": 0.8523,
+      "step": 9849
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.016003870998788e-05,
+      "loss": 0.7529,
+      "step": 9850
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.015798720172974e-05,
+      "loss": 0.8157,
+      "step": 9851
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.015593553204825e-05,
+      "loss": 0.848,
+      "step": 9852
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.015388370096524e-05,
+      "loss": 0.8255,
+      "step": 9853
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0151831708502574e-05,
+      "loss": 0.8356,
+      "step": 9854
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0149779554682096e-05,
+      "loss": 0.8893,
+      "step": 9855
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0147727239525667e-05,
+      "loss": 0.8757,
+      "step": 9856
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.014567476305514e-05,
+      "loss": 0.8176,
+      "step": 9857
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.014362212529237e-05,
+      "loss": 0.7939,
+      "step": 9858
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.014156932625921e-05,
+      "loss": 0.8584,
+      "step": 9859
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.013951636597755e-05,
+      "loss": 0.8104,
+      "step": 9860
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.013746324446921e-05,
+      "loss": 0.8516,
+      "step": 9861
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0135409961756086e-05,
+      "loss": 0.8525,
+      "step": 9862
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.013335651786004e-05,
+      "loss": 0.818,
+      "step": 9863
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0131302912802926e-05,
+      "loss": 0.8034,
+      "step": 9864
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.012924914660663e-05,
+      "loss": 0.8223,
+      "step": 9865
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.012719521929301e-05,
+      "loss": 0.8134,
+      "step": 9866
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.012514113088395e-05,
+      "loss": 0.8269,
+      "step": 9867
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0123086881401306e-05,
+      "loss": 0.7647,
+      "step": 9868
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.012103247086698e-05,
+      "loss": 0.8311,
+      "step": 9869
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0118977899302846e-05,
+      "loss": 0.8643,
+      "step": 9870
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.011692316673076e-05,
+      "loss": 0.8668,
+      "step": 9871
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.011486827317263e-05,
+      "loss": 0.8713,
+      "step": 9872
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.011281321865032e-05,
+      "loss": 0.7975,
+      "step": 9873
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.011075800318573e-05,
+      "loss": 0.8503,
+      "step": 9874
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0108702626800745e-05,
+      "loss": 0.8845,
+      "step": 9875
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.010664708951725e-05,
+      "loss": 0.8088,
+      "step": 9876
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0104591391357124e-05,
+      "loss": 0.8255,
+      "step": 9877
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.010253553234229e-05,
+      "loss": 0.7475,
+      "step": 9878
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0100479512494604e-05,
+      "loss": 0.8548,
+      "step": 9879
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0098423331835984e-05,
+      "loss": 0.8536,
+      "step": 9880
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0096366990388325e-05,
+      "loss": 0.8745,
+      "step": 9881
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.009431048817352e-05,
+      "loss": 0.8145,
+      "step": 9882
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0092253825213474e-05,
+      "loss": 0.8951,
+      "step": 9883
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.009019700153008e-05,
+      "loss": 0.9253,
+      "step": 9884
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.008814001714526e-05,
+      "loss": 0.8939,
+      "step": 9885
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.00860828720809e-05,
+      "loss": 0.8671,
+      "step": 9886
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.008402556635892e-05,
+      "loss": 0.8822,
+      "step": 9887
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0081968100001224e-05,
+      "loss": 0.8924,
+      "step": 9888
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.007991047302973e-05,
+      "loss": 0.8509,
+      "step": 9889
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.007785268546634e-05,
+      "loss": 0.8678,
+      "step": 9890
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0075794737332975e-05,
+      "loss": 0.7884,
+      "step": 9891
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.007373662865155e-05,
+      "loss": 0.8117,
+      "step": 9892
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.007167835944398e-05,
+      "loss": 0.8755,
+      "step": 9893
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.006961992973218e-05,
+      "loss": 0.8599,
+      "step": 9894
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0067561339538086e-05,
+      "loss": 0.9436,
+      "step": 9895
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.006550258888361e-05,
+      "loss": 0.7996,
+      "step": 9896
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0063443677790666e-05,
+      "loss": 0.8966,
+      "step": 9897
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0061384606281205e-05,
+      "loss": 0.8475,
+      "step": 9898
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.005932537437713e-05,
+      "loss": 0.9032,
+      "step": 9899
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.005726598210039e-05,
+      "loss": 0.7987,
+      "step": 9900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0055206429472916e-05,
+      "loss": 0.7956,
+      "step": 9901
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.005314671651663e-05,
+      "loss": 0.907,
+      "step": 9902
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.005108684325347e-05,
+      "loss": 0.8489,
+      "step": 9903
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0049026809705364e-05,
+      "loss": 0.8237,
+      "step": 9904
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.004696661589426e-05,
+      "loss": 0.7963,
+      "step": 9905
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0044906261842106e-05,
+      "loss": 0.8946,
+      "step": 9906
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0042845747570834e-05,
+      "loss": 0.8338,
+      "step": 9907
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.004078507310238e-05,
+      "loss": 0.9128,
+      "step": 9908
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.00387242384587e-05,
+      "loss": 0.8521,
+      "step": 9909
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0036663243661734e-05,
+      "loss": 0.8555,
+      "step": 9910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.003460208873344e-05,
+      "loss": 0.9247,
+      "step": 9911
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.003254077369576e-05,
+      "loss": 0.8816,
+      "step": 9912
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.003047929857064e-05,
+      "loss": 0.8355,
+      "step": 9913
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0028417663380054e-05,
+      "loss": 0.9417,
+      "step": 9914
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.002635586814594e-05,
+      "loss": 0.8431,
+      "step": 9915
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.002429391289025e-05,
+      "loss": 0.8516,
+      "step": 9916
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.002223179763496e-05,
+      "loss": 0.8448,
+      "step": 9917
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0020169522402016e-05,
+      "loss": 0.8772,
+      "step": 9918
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.001810708721339e-05,
+      "loss": 0.8381,
+      "step": 9919
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.001604449209103e-05,
+      "loss": 0.8725,
+      "step": 9920
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.001398173705693e-05,
+      "loss": 0.8879,
+      "step": 9921
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.001191882213303e-05,
+      "loss": 0.9818,
+      "step": 9922
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0009855747341304e-05,
+      "loss": 0.9428,
+      "step": 9923
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0007792512703735e-05,
+      "loss": 0.8063,
+      "step": 9924
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0005729118242284e-05,
+      "loss": 0.8552,
+      "step": 9925
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0003665563978926e-05,
+      "loss": 0.8167,
+      "step": 9926
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.000160184993564e-05,
+      "loss": 0.8555,
+      "step": 9927
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9999537976134406e-05,
+      "loss": 0.9871,
+      "step": 9928
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9997473942597194e-05,
+      "loss": 0.9376,
+      "step": 9929
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.999540974934599e-05,
+      "loss": 0.9112,
+      "step": 9930
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9993345396402785e-05,
+      "loss": 0.9499,
+      "step": 9931
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9991280883789545e-05,
+      "loss": 0.8824,
+      "step": 9932
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.998921621152827e-05,
+      "loss": 0.9001,
+      "step": 9933
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9987151379640934e-05,
+      "loss": 0.7993,
+      "step": 9934
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.998508638814954e-05,
+      "loss": 0.8106,
+      "step": 9935
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9983021237076076e-05,
+      "loss": 0.9278,
+      "step": 9936
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.998095592644253e-05,
+      "loss": 0.9271,
+      "step": 9937
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.99788904562709e-05,
+      "loss": 0.7977,
+      "step": 9938
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.997682482658318e-05,
+      "loss": 0.7799,
+      "step": 9939
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.997475903740137e-05,
+      "loss": 0.8101,
+      "step": 9940
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.997269308874746e-05,
+      "loss": 0.8211,
+      "step": 9941
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.997062698064347e-05,
+      "loss": 0.8801,
+      "step": 9942
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.996856071311138e-05,
+      "loss": 0.8936,
+      "step": 9943
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.996649428617322e-05,
+      "loss": 0.8752,
+      "step": 9944
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.996442769985097e-05,
+      "loss": 0.8465,
+      "step": 9945
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.996236095416665e-05,
+      "loss": 0.8978,
+      "step": 9946
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.996029404914228e-05,
+      "loss": 0.8451,
+      "step": 9947
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.995822698479985e-05,
+      "loss": 0.8881,
+      "step": 9948
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.995615976116138e-05,
+      "loss": 0.7884,
+      "step": 9949
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.99540923782489e-05,
+      "loss": 0.8337,
+      "step": 9950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.995202483608442e-05,
+      "loss": 0.8251,
+      "step": 9951
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.994995713468994e-05,
+      "loss": 0.7907,
+      "step": 9952
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.99478892740875e-05,
+      "loss": 0.7511,
+      "step": 9953
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9945821254299115e-05,
+      "loss": 0.869,
+      "step": 9954
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.99437530753468e-05,
+      "loss": 0.8653,
+      "step": 9955
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.994168473725259e-05,
+      "loss": 0.9304,
+      "step": 9956
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9939616240038515e-05,
+      "loss": 0.9211,
+      "step": 9957
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.99375475837266e-05,
+      "loss": 0.8686,
+      "step": 9958
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.993547876833886e-05,
+      "loss": 0.9363,
+      "step": 9959
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.993340979389735e-05,
+      "loss": 0.7581,
+      "step": 9960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9931340660424085e-05,
+      "loss": 0.955,
+      "step": 9961
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.992927136794111e-05,
+      "loss": 0.927,
+      "step": 9962
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9927201916470454e-05,
+      "loss": 0.9101,
+      "step": 9963
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9925132306034165e-05,
+      "loss": 0.8241,
+      "step": 9964
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.992306253665427e-05,
+      "loss": 0.8767,
+      "step": 9965
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.992099260835282e-05,
+      "loss": 0.823,
+      "step": 9966
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.991892252115187e-05,
+      "loss": 0.7872,
+      "step": 9967
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.991685227507345e-05,
+      "loss": 0.8089,
+      "step": 9968
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9914781870139593e-05,
+      "loss": 0.8275,
+      "step": 9969
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.991271130637238e-05,
+      "loss": 0.8687,
+      "step": 9970
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9910640583793843e-05,
+      "loss": 0.881,
+      "step": 9971
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9908569702426034e-05,
+      "loss": 0.7963,
+      "step": 9972
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.990649866229101e-05,
+      "loss": 0.8974,
+      "step": 9973
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.990442746341083e-05,
+      "loss": 0.7366,
+      "step": 9974
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.990235610580754e-05,
+      "loss": 0.8088,
+      "step": 9975
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.99002845895032e-05,
+      "loss": 0.8554,
+      "step": 9976
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.989821291451988e-05,
+      "loss": 0.8687,
+      "step": 9977
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.989614108087964e-05,
+      "loss": 0.8639,
+      "step": 9978
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9894069088604535e-05,
+      "loss": 0.8367,
+      "step": 9979
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9891996937716646e-05,
+      "loss": 0.8507,
+      "step": 9980
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9889924628238026e-05,
+      "loss": 0.8025,
+      "step": 9981
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9887852160190746e-05,
+      "loss": 0.8201,
+      "step": 9982
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.988577953359688e-05,
+      "loss": 0.8583,
+      "step": 9983
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9883706748478504e-05,
+      "loss": 0.8605,
+      "step": 9984
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.988163380485769e-05,
+      "loss": 0.8129,
+      "step": 9985
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.98795607027565e-05,
+      "loss": 0.8563,
+      "step": 9986
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.987748744219702e-05,
+      "loss": 0.8109,
+      "step": 9987
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.987541402320134e-05,
+      "loss": 0.9056,
+      "step": 9988
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9873340445791535e-05,
+      "loss": 0.8779,
+      "step": 9989
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.987126670998968e-05,
+      "loss": 0.8865,
+      "step": 9990
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.986919281581786e-05,
+      "loss": 0.8314,
+      "step": 9991
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.986711876329817e-05,
+      "loss": 0.8049,
+      "step": 9992
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.986504455245268e-05,
+      "loss": 0.8284,
+      "step": 9993
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9862970183303506e-05,
+      "loss": 0.8858,
+      "step": 9994
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.986089565587271e-05,
+      "loss": 0.8296,
+      "step": 9995
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.98588209701824e-05,
+      "loss": 0.8616,
+      "step": 9996
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.985674612625467e-05,
+      "loss": 0.8057,
+      "step": 9997
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.985467112411161e-05,
+      "loss": 0.9228,
+      "step": 9998
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.985259596377533e-05,
+      "loss": 0.8225,
+      "step": 9999
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9850520645267906e-05,
+      "loss": 0.765,
+      "step": 10000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.984844516861147e-05,
+      "loss": 0.8957,
+      "step": 10001
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9846369533828096e-05,
+      "loss": 0.8515,
+      "step": 10002
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.984429374093991e-05,
+      "loss": 0.847,
+      "step": 10003
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.984221778996899e-05,
+      "loss": 0.861,
+      "step": 10004
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.984014168093748e-05,
+      "loss": 0.8161,
+      "step": 10005
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9838065413867466e-05,
+      "loss": 0.8535,
+      "step": 10006
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.983598898878106e-05,
+      "loss": 0.9294,
+      "step": 10007
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.983391240570037e-05,
+      "loss": 0.922,
+      "step": 10008
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.983183566464753e-05,
+      "loss": 0.8228,
+      "step": 10009
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.982975876564464e-05,
+      "loss": 0.8464,
+      "step": 10010
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.982768170871383e-05,
+      "loss": 0.7754,
+      "step": 10011
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.982560449387721e-05,
+      "loss": 0.8202,
+      "step": 10012
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.98235271211569e-05,
+      "loss": 0.8642,
+      "step": 10013
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9821449590575025e-05,
+      "loss": 0.8993,
+      "step": 10014
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.981937190215371e-05,
+      "loss": 0.8593,
+      "step": 10015
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.981729405591509e-05,
+      "loss": 0.8352,
+      "step": 10016
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.981521605188127e-05,
+      "loss": 0.8625,
+      "step": 10017
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9813137890074406e-05,
+      "loss": 0.7615,
+      "step": 10018
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9811059570516606e-05,
+      "loss": 0.8844,
+      "step": 10019
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.980898109323002e-05,
+      "loss": 0.9337,
+      "step": 10020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.980690245823677e-05,
+      "loss": 0.8984,
+      "step": 10021
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.980482366555901e-05,
+      "loss": 0.846,
+      "step": 10022
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.980274471521886e-05,
+      "loss": 0.9298,
+      "step": 10023
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9800665607238466e-05,
+      "loss": 0.832,
+      "step": 10024
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.979858634163996e-05,
+      "loss": 0.8662,
+      "step": 10025
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.979650691844551e-05,
+      "loss": 0.8002,
+      "step": 10026
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9794427337677233e-05,
+      "loss": 0.9224,
+      "step": 10027
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9792347599357285e-05,
+      "loss": 0.8756,
+      "step": 10028
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.979026770350782e-05,
+      "loss": 0.9207,
+      "step": 10029
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9788187650150985e-05,
+      "loss": 0.788,
+      "step": 10030
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.978610743930893e-05,
+      "loss": 0.8341,
+      "step": 10031
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.978402707100381e-05,
+      "loss": 0.8719,
+      "step": 10032
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.978194654525777e-05,
+      "loss": 0.8108,
+      "step": 10033
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.977986586209298e-05,
+      "loss": 0.8634,
+      "step": 10034
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.977778502153159e-05,
+      "loss": 0.8656,
+      "step": 10035
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.977570402359576e-05,
+      "loss": 0.8223,
+      "step": 10036
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.977362286830765e-05,
+      "loss": 0.8577,
+      "step": 10037
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.977154155568943e-05,
+      "loss": 0.9454,
+      "step": 10038
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.976946008576326e-05,
+      "loss": 0.8407,
+      "step": 10039
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.976737845855131e-05,
+      "loss": 0.7912,
+      "step": 10040
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.976529667407574e-05,
+      "loss": 0.8672,
+      "step": 10041
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9763214732358724e-05,
+      "loss": 0.8607,
+      "step": 10042
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9761132633422434e-05,
+      "loss": 0.9207,
+      "step": 10043
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.975905037728904e-05,
+      "loss": 0.854,
+      "step": 10044
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.975696796398073e-05,
+      "loss": 0.9036,
+      "step": 10045
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.975488539351967e-05,
+      "loss": 0.7995,
+      "step": 10046
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.975280266592803e-05,
+      "loss": 0.887,
+      "step": 10047
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9750719781228005e-05,
+      "loss": 0.8377,
+      "step": 10048
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.974863673944176e-05,
+      "loss": 0.9152,
+      "step": 10049
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.974655354059149e-05,
+      "loss": 0.9133,
+      "step": 10050
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9744470184699386e-05,
+      "loss": 0.8393,
+      "step": 10051
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.974238667178763e-05,
+      "loss": 0.91,
+      "step": 10052
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.97403030018784e-05,
+      "loss": 0.8975,
+      "step": 10053
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.973821917499389e-05,
+      "loss": 0.8645,
+      "step": 10054
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9736135191156284e-05,
+      "loss": 0.7828,
+      "step": 10055
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.97340510503878e-05,
+      "loss": 0.9222,
+      "step": 10056
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9731966752710616e-05,
+      "loss": 0.7847,
+      "step": 10057
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.972988229814693e-05,
+      "loss": 0.8596,
+      "step": 10058
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9727797686718926e-05,
+      "loss": 0.8306,
+      "step": 10059
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9725712918448834e-05,
+      "loss": 0.8605,
+      "step": 10060
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.972362799335883e-05,
+      "loss": 0.8795,
+      "step": 10061
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.972154291147113e-05,
+      "loss": 0.7669,
+      "step": 10062
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.971945767280794e-05,
+      "loss": 0.8607,
+      "step": 10063
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.971737227739146e-05,
+      "loss": 0.8279,
+      "step": 10064
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.97152867252439e-05,
+      "loss": 0.8205,
+      "step": 10065
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9713201016387466e-05,
+      "loss": 0.8706,
+      "step": 10066
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.971111515084438e-05,
+      "loss": 0.9049,
+      "step": 10067
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.970902912863684e-05,
+      "loss": 0.9139,
+      "step": 10068
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.970694294978708e-05,
+      "loss": 0.8517,
+      "step": 10069
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9704856614317306e-05,
+      "loss": 0.8316,
+      "step": 10070
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.970277012224973e-05,
+      "loss": 0.7886,
+      "step": 10071
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9700683473606585e-05,
+      "loss": 0.8949,
+      "step": 10072
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.969859666841008e-05,
+      "loss": 0.7725,
+      "step": 10073
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.969650970668245e-05,
+      "loss": 0.878,
+      "step": 10074
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.969442258844591e-05,
+      "loss": 0.9516,
+      "step": 10075
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9692335313722696e-05,
+      "loss": 0.8442,
+      "step": 10076
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.969024788253502e-05,
+      "loss": 0.8381,
+      "step": 10077
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9688160294905133e-05,
+      "loss": 0.7911,
+      "step": 10078
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.968607255085525e-05,
+      "loss": 0.8269,
+      "step": 10079
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9683984650407616e-05,
+      "loss": 0.8831,
+      "step": 10080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.968189659358445e-05,
+      "loss": 0.8329,
+      "step": 10081
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9679808380408014e-05,
+      "loss": 0.9021,
+      "step": 10082
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9677720010900524e-05,
+      "loss": 0.877,
+      "step": 10083
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.967563148508422e-05,
+      "loss": 0.8864,
+      "step": 10084
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.967354280298136e-05,
+      "loss": 0.8394,
+      "step": 10085
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.967145396461418e-05,
+      "loss": 0.8365,
+      "step": 10086
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9669364970004904e-05,
+      "loss": 0.8509,
+      "step": 10087
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.966727581917582e-05,
+      "loss": 0.8844,
+      "step": 10088
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9665186512149137e-05,
+      "loss": 0.8494,
+      "step": 10089
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.966309704894713e-05,
+      "loss": 0.8541,
+      "step": 10090
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9661007429592035e-05,
+      "loss": 0.953,
+      "step": 10091
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.965891765410612e-05,
+      "loss": 0.8825,
+      "step": 10092
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9656827722511626e-05,
+      "loss": 0.8027,
+      "step": 10093
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.965473763483082e-05,
+      "loss": 0.8503,
+      "step": 10094
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.965264739108595e-05,
+      "loss": 0.8547,
+      "step": 10095
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.965055699129928e-05,
+      "loss": 0.8513,
+      "step": 10096
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.964846643549308e-05,
+      "loss": 0.8084,
+      "step": 10097
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.96463757236896e-05,
+      "loss": 0.8457,
+      "step": 10098
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9644284855911105e-05,
+      "loss": 0.8735,
+      "step": 10099
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.964219383217988e-05,
+      "loss": 0.8914,
+      "step": 10100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.964010265251817e-05,
+      "loss": 0.833,
+      "step": 10101
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.963801131694826e-05,
+      "loss": 0.8962,
+      "step": 10102
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9635919825492415e-05,
+      "loss": 0.7823,
+      "step": 10103
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9633828178172916e-05,
+      "loss": 0.9091,
+      "step": 10104
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.963173637501203e-05,
+      "loss": 0.8628,
+      "step": 10105
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.962964441603202e-05,
+      "loss": 0.8442,
+      "step": 10106
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.962755230125519e-05,
+      "loss": 0.7781,
+      "step": 10107
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.962546003070381e-05,
+      "loss": 0.8427,
+      "step": 10108
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.962336760440015e-05,
+      "loss": 0.7634,
+      "step": 10109
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.962127502236652e-05,
+      "loss": 0.8832,
+      "step": 10110
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961918228462518e-05,
+      "loss": 0.8923,
+      "step": 10111
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961708939119841e-05,
+      "loss": 0.8325,
+      "step": 10112
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961499634210853e-05,
+      "loss": 0.8894,
+      "step": 10113
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961290313737781e-05,
+      "loss": 0.863,
+      "step": 10114
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961080977702854e-05,
+      "loss": 0.8471,
+      "step": 10115
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9608716261083015e-05,
+      "loss": 0.8192,
+      "step": 10116
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9606622589563535e-05,
+      "loss": 0.8653,
+      "step": 10117
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9604528762492384e-05,
+      "loss": 0.8291,
+      "step": 10118
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.960243477989187e-05,
+      "loss": 0.842,
+      "step": 10119
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9600340641784305e-05,
+      "loss": 0.8155,
+      "step": 10120
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.959824634819196e-05,
+      "loss": 0.8748,
+      "step": 10121
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.959615189913716e-05,
+      "loss": 0.8852,
+      "step": 10122
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.95940572946422e-05,
+      "loss": 0.9156,
+      "step": 10123
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9591962534729405e-05,
+      "loss": 0.8563,
+      "step": 10124
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9589867619421054e-05,
+      "loss": 0.8603,
+      "step": 10125
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.958777254873948e-05,
+      "loss": 0.7783,
+      "step": 10126
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.958567732270697e-05,
+      "loss": 0.7711,
+      "step": 10127
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.958358194134586e-05,
+      "loss": 0.8672,
+      "step": 10128
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9581486404678456e-05,
+      "loss": 0.877,
+      "step": 10129
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.957939071272708e-05,
+      "loss": 0.8511,
+      "step": 10130
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9577294865514036e-05,
+      "loss": 0.8872,
+      "step": 10131
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.957519886306166e-05,
+      "loss": 0.9285,
+      "step": 10132
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.957310270539226e-05,
+      "loss": 0.8858,
+      "step": 10133
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.957100639252817e-05,
+      "loss": 0.905,
+      "step": 10134
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9568909924491696e-05,
+      "loss": 0.8137,
+      "step": 10135
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9566813301305175e-05,
+      "loss": 0.9733,
+      "step": 10136
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9564716522990954e-05,
+      "loss": 0.885,
+      "step": 10137
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9562619589571325e-05,
+      "loss": 0.8453,
+      "step": 10138
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.956052250106865e-05,
+      "loss": 0.9378,
+      "step": 10139
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.955842525750525e-05,
+      "loss": 0.9113,
+      "step": 10140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.955632785890345e-05,
+      "loss": 0.9174,
+      "step": 10141
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9554230305285606e-05,
+      "loss": 0.8412,
+      "step": 10142
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.955213259667404e-05,
+      "loss": 0.8685,
+      "step": 10143
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9550034733091104e-05,
+      "loss": 0.8337,
+      "step": 10144
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.954793671455912e-05,
+      "loss": 0.8259,
+      "step": 10145
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9545838541100456e-05,
+      "loss": 0.7642,
+      "step": 10146
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9543740212737436e-05,
+      "loss": 0.8275,
+      "step": 10147
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.954164172949241e-05,
+      "loss": 0.9025,
+      "step": 10148
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.953954309138773e-05,
+      "loss": 0.8513,
+      "step": 10149
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.953744429844574e-05,
+      "loss": 0.8782,
+      "step": 10150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9535345350688795e-05,
+      "loss": 0.8317,
+      "step": 10151
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.953324624813925e-05,
+      "loss": 0.8843,
+      "step": 10152
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.953114699081946e-05,
+      "loss": 0.7909,
+      "step": 10153
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9529047578751764e-05,
+      "loss": 0.7777,
+      "step": 10154
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.952694801195854e-05,
+      "loss": 0.8494,
+      "step": 10155
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.952484829046214e-05,
+      "loss": 0.9191,
+      "step": 10156
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9522748414284924e-05,
+      "loss": 0.8357,
+      "step": 10157
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9520648383449254e-05,
+      "loss": 0.8709,
+      "step": 10158
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.951854819797749e-05,
+      "loss": 0.8414,
+      "step": 10159
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9516447857892e-05,
+      "loss": 0.8034,
+      "step": 10160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9514347363215165e-05,
+      "loss": 0.8936,
+      "step": 10161
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.951224671396933e-05,
+      "loss": 0.791,
+      "step": 10162
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.951014591017689e-05,
+      "loss": 0.9138,
+      "step": 10163
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.95080449518602e-05,
+      "loss": 0.7455,
+      "step": 10164
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.950594383904164e-05,
+      "loss": 0.828,
+      "step": 10165
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.950384257174359e-05,
+      "loss": 0.9078,
+      "step": 10166
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9501741149988414e-05,
+      "loss": 0.8288,
+      "step": 10167
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.94996395737985e-05,
+      "loss": 0.8866,
+      "step": 10168
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.949753784319623e-05,
+      "loss": 0.8242,
+      "step": 10169
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.949543595820397e-05,
+      "loss": 0.8846,
+      "step": 10170
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.949333391884413e-05,
+      "loss": 0.8485,
+      "step": 10171
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.949123172513909e-05,
+      "loss": 0.9152,
+      "step": 10172
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9489129377111225e-05,
+      "loss": 0.7522,
+      "step": 10173
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.948702687478293e-05,
+      "loss": 0.8912,
+      "step": 10174
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.948492421817658e-05,
+      "loss": 0.8381,
+      "step": 10175
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9482821407314595e-05,
+      "loss": 0.8891,
+      "step": 10176
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.948071844221935e-05,
+      "loss": 0.8474,
+      "step": 10177
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9478615322913245e-05,
+      "loss": 0.8983,
+      "step": 10178
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.947651204941868e-05,
+      "loss": 0.882,
+      "step": 10179
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.947440862175805e-05,
+      "loss": 0.7568,
+      "step": 10180
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.947230503995375e-05,
+      "loss": 0.8579,
+      "step": 10181
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9470201304028195e-05,
+      "loss": 0.9113,
+      "step": 10182
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.946809741400378e-05,
+      "loss": 0.9021,
+      "step": 10183
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.946599336990291e-05,
+      "loss": 0.7869,
+      "step": 10184
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9463889171747995e-05,
+      "loss": 0.8508,
+      "step": 10185
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.946178481956144e-05,
+      "loss": 0.8567,
+      "step": 10186
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9459680313365656e-05,
+      "loss": 0.8631,
+      "step": 10187
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.945757565318305e-05,
+      "loss": 0.7796,
+      "step": 10188
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.945547083903605e-05,
+      "loss": 0.8796,
+      "step": 10189
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9453365870947054e-05,
+      "loss": 0.7535,
+      "step": 10190
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.945126074893849e-05,
+      "loss": 0.8373,
+      "step": 10191
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.944915547303277e-05,
+      "loss": 0.7929,
+      "step": 10192
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.944705004325232e-05,
+      "loss": 0.7855,
+      "step": 10193
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.944494445961956e-05,
+      "loss": 0.8804,
+      "step": 10194
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9442838722156904e-05,
+      "loss": 0.8518,
+      "step": 10195
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9440732830886796e-05,
+      "loss": 0.8474,
+      "step": 10196
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.943862678583164e-05,
+      "loss": 0.839,
+      "step": 10197
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.943652058701387e-05,
+      "loss": 0.8029,
+      "step": 10198
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9434414234455936e-05,
+      "loss": 0.8781,
+      "step": 10199
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.943230772818024e-05,
+      "loss": 0.8222,
+      "step": 10200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9430201068209236e-05,
+      "loss": 1.0109,
+      "step": 10201
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.942809425456535e-05,
+      "loss": 0.8258,
+      "step": 10202
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9425987287271024e-05,
+      "loss": 0.8404,
+      "step": 10203
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9423880166348684e-05,
+      "loss": 0.9657,
+      "step": 10204
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.942177289182078e-05,
+      "loss": 0.8612,
+      "step": 10205
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9419665463709755e-05,
+      "loss": 0.8295,
+      "step": 10206
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.941755788203804e-05,
+      "loss": 0.847,
+      "step": 10207
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.941545014682809e-05,
+      "loss": 0.9,
+      "step": 10208
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.941334225810235e-05,
+      "loss": 0.8704,
+      "step": 10209
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9411234215883266e-05,
+      "loss": 0.8033,
+      "step": 10210
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.940912602019328e-05,
+      "loss": 0.8401,
+      "step": 10211
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.940701767105485e-05,
+      "loss": 0.8915,
+      "step": 10212
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9404909168490436e-05,
+      "loss": 0.871,
+      "step": 10213
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.940280051252248e-05,
+      "loss": 0.8262,
+      "step": 10214
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.940069170317344e-05,
+      "loss": 0.8728,
+      "step": 10215
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9398582740465784e-05,
+      "loss": 0.8197,
+      "step": 10216
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9396473624421956e-05,
+      "loss": 0.8467,
+      "step": 10217
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9394364355064426e-05,
+      "loss": 0.8287,
+      "step": 10218
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9392254932415664e-05,
+      "loss": 0.8927,
+      "step": 10219
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9390145356498106e-05,
+      "loss": 0.8403,
+      "step": 10220
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.938803562733425e-05,
+      "loss": 0.9167,
+      "step": 10221
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.938592574494655e-05,
+      "loss": 0.9709,
+      "step": 10222
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9383815709357474e-05,
+      "loss": 0.9073,
+      "step": 10223
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9381705520589495e-05,
+      "loss": 0.8815,
+      "step": 10224
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.937959517866507e-05,
+      "loss": 0.8571,
+      "step": 10225
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9377484683606705e-05,
+      "loss": 0.8897,
+      "step": 10226
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.937537403543684e-05,
+      "loss": 0.7679,
+      "step": 10227
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.937326323417798e-05,
+      "loss": 0.822,
+      "step": 10228
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.937115227985259e-05,
+      "loss": 0.7835,
+      "step": 10229
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9369041172483156e-05,
+      "loss": 0.811,
+      "step": 10230
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9366929912092156e-05,
+      "loss": 0.7856,
+      "step": 10231
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9364818498702075e-05,
+      "loss": 0.7898,
+      "step": 10232
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.936270693233539e-05,
+      "loss": 0.747,
+      "step": 10233
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.936059521301461e-05,
+      "loss": 0.8277,
+      "step": 10234
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9358483340762194e-05,
+      "loss": 0.842,
+      "step": 10235
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9356371315600657e-05,
+      "loss": 0.865,
+      "step": 10236
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.935425913755248e-05,
+      "loss": 0.7614,
+      "step": 10237
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.935214680664015e-05,
+      "loss": 0.8645,
+      "step": 10238
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.935003432288619e-05,
+      "loss": 0.8392,
+      "step": 10239
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9347921686313054e-05,
+      "loss": 0.894,
+      "step": 10240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.934580889694327e-05,
+      "loss": 0.8269,
+      "step": 10241
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.934369595479934e-05,
+      "loss": 0.8875,
+      "step": 10242
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.934158285990375e-05,
+      "loss": 0.8203,
+      "step": 10243
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.933946961227901e-05,
+      "loss": 0.851,
+      "step": 10244
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.933735621194763e-05,
+      "loss": 0.8034,
+      "step": 10245
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.93352426589321e-05,
+      "loss": 0.8422,
+      "step": 10246
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.933312895325494e-05,
+      "loss": 0.8612,
+      "step": 10247
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.933101509493866e-05,
+      "loss": 0.8256,
+      "step": 10248
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.932890108400578e-05,
+      "loss": 0.8245,
+      "step": 10249
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.932678692047879e-05,
+      "loss": 0.8651,
+      "step": 10250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.932467260438022e-05,
+      "loss": 0.7813,
+      "step": 10251
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9322558135732585e-05,
+      "loss": 0.9186,
+      "step": 10252
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.93204435145584e-05,
+      "loss": 0.817,
+      "step": 10253
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.931832874088019e-05,
+      "loss": 0.7653,
+      "step": 10254
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9316213814720474e-05,
+      "loss": 0.978,
+      "step": 10255
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.931409873610177e-05,
+      "loss": 0.7677,
+      "step": 10256
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.931198350504659e-05,
+      "loss": 0.884,
+      "step": 10257
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.930986812157749e-05,
+      "loss": 0.8409,
+      "step": 10258
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9307752585716987e-05,
+      "loss": 0.8191,
+      "step": 10259
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.930563689748759e-05,
+      "loss": 0.8311,
+      "step": 10260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.930352105691185e-05,
+      "loss": 0.7978,
+      "step": 10261
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9301405064012296e-05,
+      "loss": 0.8324,
+      "step": 10262
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.929928891881147e-05,
+      "loss": 0.7631,
+      "step": 10263
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.929717262133189e-05,
+      "loss": 0.8143,
+      "step": 10264
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9295056171596104e-05,
+      "loss": 0.8944,
+      "step": 10265
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.929293956962664e-05,
+      "loss": 0.8344,
+      "step": 10266
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.929082281544606e-05,
+      "loss": 0.7996,
+      "step": 10267
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.928870590907689e-05,
+      "loss": 0.9381,
+      "step": 10268
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.928658885054167e-05,
+      "loss": 0.7696,
+      "step": 10269
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.928447163986295e-05,
+      "loss": 0.8758,
+      "step": 10270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.928235427706329e-05,
+      "loss": 0.8513,
+      "step": 10271
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.928023676216522e-05,
+      "loss": 0.9242,
+      "step": 10272
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.92781190951913e-05,
+      "loss": 0.8777,
+      "step": 10273
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9276001276164085e-05,
+      "loss": 0.8613,
+      "step": 10274
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.927388330510612e-05,
+      "loss": 0.8614,
+      "step": 10275
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.927176518203997e-05,
+      "loss": 0.8149,
+      "step": 10276
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9269646906988175e-05,
+      "loss": 0.883,
+      "step": 10277
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9267528479973306e-05,
+      "loss": 0.8062,
+      "step": 10278
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9265409901017924e-05,
+      "loss": 0.8496,
+      "step": 10279
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.926329117014459e-05,
+      "loss": 0.8005,
+      "step": 10280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9261172287375856e-05,
+      "loss": 0.8447,
+      "step": 10281
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.92590532527343e-05,
+      "loss": 0.7937,
+      "step": 10282
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.925693406624248e-05,
+      "loss": 0.7861,
+      "step": 10283
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9254814727922985e-05,
+      "loss": 0.7856,
+      "step": 10284
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.925269523779835e-05,
+      "loss": 0.8578,
+      "step": 10285
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9250575595891166e-05,
+      "loss": 0.8305,
+      "step": 10286
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9248455802224e-05,
+      "loss": 0.8343,
+      "step": 10287
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.924633585681944e-05,
+      "loss": 0.7729,
+      "step": 10288
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.924421575970004e-05,
+      "loss": 0.837,
+      "step": 10289
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9242095510888405e-05,
+      "loss": 0.8704,
+      "step": 10290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.923997511040709e-05,
+      "loss": 0.8199,
+      "step": 10291
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9237854558278684e-05,
+      "loss": 0.7399,
+      "step": 10292
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.923573385452578e-05,
+      "loss": 0.8156,
+      "step": 10293
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.923361299917095e-05,
+      "loss": 0.8556,
+      "step": 10294
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.923149199223678e-05,
+      "loss": 0.8211,
+      "step": 10295
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.922937083374586e-05,
+      "loss": 0.8411,
+      "step": 10296
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.922724952372078e-05,
+      "loss": 0.8313,
+      "step": 10297
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.922512806218413e-05,
+      "loss": 0.8335,
+      "step": 10298
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.922300644915851e-05,
+      "loss": 0.7961,
+      "step": 10299
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9220884684666495e-05,
+      "loss": 0.85,
+      "step": 10300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9218762768730695e-05,
+      "loss": 0.883,
+      "step": 10301
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.92166407013737e-05,
+      "loss": 0.88,
+      "step": 10302
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.921451848261812e-05,
+      "loss": 0.8565,
+      "step": 10303
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.921239611248655e-05,
+      "loss": 0.8475,
+      "step": 10304
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.921027359100159e-05,
+      "loss": 0.8015,
+      "step": 10305
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.920815091818584e-05,
+      "loss": 0.766,
+      "step": 10306
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9206028094061905e-05,
+      "loss": 0.8831,
+      "step": 10307
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9203905118652395e-05,
+      "loss": 0.8535,
+      "step": 10308
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9201781991979925e-05,
+      "loss": 0.8812,
+      "step": 10309
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.919965871406709e-05,
+      "loss": 0.9117,
+      "step": 10310
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.919753528493652e-05,
+      "loss": 0.8983,
+      "step": 10311
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.919541170461082e-05,
+      "loss": 0.8164,
+      "step": 10312
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9193287973112596e-05,
+      "loss": 0.8757,
+      "step": 10313
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9191164090464474e-05,
+      "loss": 0.8032,
+      "step": 10314
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.918904005668907e-05,
+      "loss": 0.9324,
+      "step": 10315
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9186915871809e-05,
+      "loss": 0.9135,
+      "step": 10316
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.918479153584691e-05,
+      "loss": 0.8064,
+      "step": 10317
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.918266704882538e-05,
+      "loss": 0.9539,
+      "step": 10318
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.918054241076706e-05,
+      "loss": 0.8591,
+      "step": 10319
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9178417621694574e-05,
+      "loss": 0.8053,
+      "step": 10320
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.917629268163055e-05,
+      "loss": 0.7916,
+      "step": 10321
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.917416759059762e-05,
+      "loss": 0.8608,
+      "step": 10322
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.91720423486184e-05,
+      "loss": 0.8206,
+      "step": 10323
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.916991695571554e-05,
+      "loss": 0.8178,
+      "step": 10324
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.916779141191167e-05,
+      "loss": 0.8409,
+      "step": 10325
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.916566571722942e-05,
+      "loss": 0.8035,
+      "step": 10326
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9163539871691424e-05,
+      "loss": 0.9103,
+      "step": 10327
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.916141387532033e-05,
+      "loss": 0.8384,
+      "step": 10328
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.915928772813878e-05,
+      "loss": 0.8647,
+      "step": 10329
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.915716143016941e-05,
+      "loss": 0.8137,
+      "step": 10330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.915503498143487e-05,
+      "loss": 0.8613,
+      "step": 10331
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.915290838195779e-05,
+      "loss": 0.8865,
+      "step": 10332
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.915078163176084e-05,
+      "loss": 0.8033,
+      "step": 10333
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.914865473086665e-05,
+      "loss": 0.815,
+      "step": 10334
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.914652767929788e-05,
+      "loss": 0.9275,
+      "step": 10335
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9144400477077176e-05,
+      "loss": 0.8324,
+      "step": 10336
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.91422731242272e-05,
+      "loss": 0.8419,
+      "step": 10337
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.91401456207706e-05,
+      "loss": 0.7332,
+      "step": 10338
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.913801796673002e-05,
+      "loss": 0.8502,
+      "step": 10339
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.913589016212815e-05,
+      "loss": 0.8326,
+      "step": 10340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.913376220698762e-05,
+      "loss": 0.8438,
+      "step": 10341
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.913163410133111e-05,
+      "loss": 0.7883,
+      "step": 10342
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.912950584518127e-05,
+      "loss": 0.7932,
+      "step": 10343
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.912737743856077e-05,
+      "loss": 0.7883,
+      "step": 10344
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9125248881492295e-05,
+      "loss": 0.7984,
+      "step": 10345
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.912312017399848e-05,
+      "loss": 0.9159,
+      "step": 10346
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.912099131610202e-05,
+      "loss": 0.8504,
+      "step": 10347
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.911886230782556e-05,
+      "loss": 0.8169,
+      "step": 10348
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.91167331491918e-05,
+      "loss": 0.8494,
+      "step": 10349
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.911460384022339e-05,
+      "loss": 0.8583,
+      "step": 10350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.911247438094303e-05,
+      "loss": 0.8229,
+      "step": 10351
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.911034477137339e-05,
+      "loss": 0.9016,
+      "step": 10352
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.910821501153713e-05,
+      "loss": 0.841,
+      "step": 10353
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.910608510145695e-05,
+      "loss": 0.7911,
+      "step": 10354
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9103955041155536e-05,
+      "loss": 0.8381,
+      "step": 10355
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9101824830655556e-05,
+      "loss": 0.88,
+      "step": 10356
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.909969446997971e-05,
+      "loss": 0.8538,
+      "step": 10357
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.909756395915067e-05,
+      "loss": 0.7671,
+      "step": 10358
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9095433298191134e-05,
+      "loss": 0.9043,
+      "step": 10359
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9093302487123796e-05,
+      "loss": 0.8562,
+      "step": 10360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.909117152597134e-05,
+      "loss": 0.8094,
+      "step": 10361
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.908904041475645e-05,
+      "loss": 0.8589,
+      "step": 10362
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.908690915350185e-05,
+      "loss": 0.8211,
+      "step": 10363
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.908477774223022e-05,
+      "loss": 0.8284,
+      "step": 10364
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.908264618096425e-05,
+      "loss": 0.8602,
+      "step": 10365
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9080514469726645e-05,
+      "loss": 0.8707,
+      "step": 10366
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.907838260854012e-05,
+      "loss": 0.8878,
+      "step": 10367
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.907625059742736e-05,
+      "loss": 0.8057,
+      "step": 10368
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.907411843641108e-05,
+      "loss": 0.9137,
+      "step": 10369
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.907198612551398e-05,
+      "loss": 0.8067,
+      "step": 10370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.906985366475878e-05,
+      "loss": 0.8237,
+      "step": 10371
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.906772105416817e-05,
+      "loss": 0.8581,
+      "step": 10372
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.906558829376488e-05,
+      "loss": 0.8425,
+      "step": 10373
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9063455383571606e-05,
+      "loss": 0.9164,
+      "step": 10374
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.906132232361108e-05,
+      "loss": 0.7717,
+      "step": 10375
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9059189113906006e-05,
+      "loss": 0.8825,
+      "step": 10376
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.905705575447909e-05,
+      "loss": 0.8116,
+      "step": 10377
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.905492224535308e-05,
+      "loss": 0.8703,
+      "step": 10378
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.905278858655067e-05,
+      "loss": 0.8264,
+      "step": 10379
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.90506547780946e-05,
+      "loss": 0.8389,
+      "step": 10380
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.904852082000758e-05,
+      "loss": 0.8587,
+      "step": 10381
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.904638671231236e-05,
+      "loss": 0.8784,
+      "step": 10382
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9044252455031626e-05,
+      "loss": 0.8914,
+      "step": 10383
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.904211804818815e-05,
+      "loss": 0.8331,
+      "step": 10384
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9039983491804635e-05,
+      "loss": 0.8407,
+      "step": 10385
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.903784878590381e-05,
+      "loss": 0.8538,
+      "step": 10386
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.903571393050843e-05,
+      "loss": 0.8167,
+      "step": 10387
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9033578925641215e-05,
+      "loss": 0.8414,
+      "step": 10388
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9031443771324896e-05,
+      "loss": 0.8508,
+      "step": 10389
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.902930846758223e-05,
+      "loss": 0.9255,
+      "step": 10390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.902717301443594e-05,
+      "loss": 0.8285,
+      "step": 10391
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9025037411908776e-05,
+      "loss": 0.8966,
+      "step": 10392
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.902290166002348e-05,
+      "loss": 0.8052,
+      "step": 10393
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.902076575880279e-05,
+      "loss": 0.7964,
+      "step": 10394
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.901862970826946e-05,
+      "loss": 0.8158,
+      "step": 10395
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.901649350844624e-05,
+      "loss": 0.7563,
+      "step": 10396
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.901435715935587e-05,
+      "loss": 0.7665,
+      "step": 10397
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9012220661021104e-05,
+      "loss": 0.7979,
+      "step": 10398
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.901008401346469e-05,
+      "loss": 0.9629,
+      "step": 10399
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9007947216709387e-05,
+      "loss": 0.9208,
+      "step": 10400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.900581027077796e-05,
+      "loss": 0.809,
+      "step": 10401
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.900367317569315e-05,
+      "loss": 0.7791,
+      "step": 10402
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9001535931477726e-05,
+      "loss": 0.9098,
+      "step": 10403
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.899939853815444e-05,
+      "loss": 0.8352,
+      "step": 10404
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8997260995746065e-05,
+      "loss": 0.8715,
+      "step": 10405
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8995123304275355e-05,
+      "loss": 0.8172,
+      "step": 10406
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.899298546376508e-05,
+      "loss": 0.8312,
+      "step": 10407
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8990847474238e-05,
+      "loss": 0.7886,
+      "step": 10408
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.898870933571689e-05,
+      "loss": 0.828,
+      "step": 10409
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.898657104822452e-05,
+      "loss": 0.7736,
+      "step": 10410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.898443261178366e-05,
+      "loss": 0.848,
+      "step": 10411
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8982294026417085e-05,
+      "loss": 0.7931,
+      "step": 10412
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8980155292147566e-05,
+      "loss": 0.8312,
+      "step": 10413
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.897801640899788e-05,
+      "loss": 0.9607,
+      "step": 10414
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.89758773769908e-05,
+      "loss": 0.7866,
+      "step": 10415
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.897373819614912e-05,
+      "loss": 0.7958,
+      "step": 10416
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.897159886649561e-05,
+      "loss": 0.8582,
+      "step": 10417
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8969459388053055e-05,
+      "loss": 0.8171,
+      "step": 10418
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8967319760844235e-05,
+      "loss": 0.8233,
+      "step": 10419
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8965179984891944e-05,
+      "loss": 0.7705,
+      "step": 10420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8963040060218956e-05,
+      "loss": 0.9132,
+      "step": 10421
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8960899986848074e-05,
+      "loss": 0.9514,
+      "step": 10422
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8958759764802086e-05,
+      "loss": 0.8831,
+      "step": 10423
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8956619394103774e-05,
+      "loss": 0.8088,
+      "step": 10424
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.895447887477595e-05,
+      "loss": 0.7895,
+      "step": 10425
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.895233820684139e-05,
+      "loss": 0.8696,
+      "step": 10426
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8950197390322896e-05,
+      "loss": 0.7405,
+      "step": 10427
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8948056425243276e-05,
+      "loss": 0.7834,
+      "step": 10428
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8945915311625316e-05,
+      "loss": 0.7651,
+      "step": 10429
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8943774049491824e-05,
+      "loss": 0.8385,
+      "step": 10430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.894163263886561e-05,
+      "loss": 0.7874,
+      "step": 10431
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8939491079769474e-05,
+      "loss": 0.8808,
+      "step": 10432
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8937349372226214e-05,
+      "loss": 0.8206,
+      "step": 10433
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.893520751625865e-05,
+      "loss": 0.9247,
+      "step": 10434
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.893306551188958e-05,
+      "loss": 0.8682,
+      "step": 10435
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8930923359141834e-05,
+      "loss": 0.8372,
+      "step": 10436
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8928781058038197e-05,
+      "loss": 0.9112,
+      "step": 10437
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8926638608601506e-05,
+      "loss": 0.8975,
+      "step": 10438
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.892449601085457e-05,
+      "loss": 0.7998,
+      "step": 10439
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.89223532648202e-05,
+      "loss": 0.801,
+      "step": 10440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.892021037052123e-05,
+      "loss": 0.8145,
+      "step": 10441
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.891806732798045e-05,
+      "loss": 0.849,
+      "step": 10442
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.891592413722073e-05,
+      "loss": 0.8608,
+      "step": 10443
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8913780798264855e-05,
+      "loss": 0.7911,
+      "step": 10444
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.891163731113565e-05,
+      "loss": 0.8362,
+      "step": 10445
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.890949367585597e-05,
+      "loss": 0.8011,
+      "step": 10446
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.890734989244862e-05,
+      "loss": 0.9055,
+      "step": 10447
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8905205960936444e-05,
+      "loss": 0.822,
+      "step": 10448
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.890306188134225e-05,
+      "loss": 0.8518,
+      "step": 10449
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.890091765368891e-05,
+      "loss": 0.8076,
+      "step": 10450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8898773277999216e-05,
+      "loss": 0.8482,
+      "step": 10451
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.889662875429603e-05,
+      "loss": 0.7643,
+      "step": 10452
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.889448408260219e-05,
+      "loss": 0.8192,
+      "step": 10453
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.889233926294052e-05,
+      "loss": 0.7453,
+      "step": 10454
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8890194295333886e-05,
+      "loss": 0.8711,
+      "step": 10455
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.888804917980511e-05,
+      "loss": 0.891,
+      "step": 10456
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.888590391637704e-05,
+      "loss": 0.8858,
+      "step": 10457
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8883758505072526e-05,
+      "loss": 0.9423,
+      "step": 10458
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.888161294591441e-05,
+      "loss": 0.7273,
+      "step": 10459
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.887946723892555e-05,
+      "loss": 0.8679,
+      "step": 10460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.887732138412878e-05,
+      "loss": 0.8496,
+      "step": 10461
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.887517538154697e-05,
+      "loss": 0.797,
+      "step": 10462
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.887302923120296e-05,
+      "loss": 0.8352,
+      "step": 10463
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8870882933119614e-05,
+      "loss": 0.8482,
+      "step": 10464
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.886873648731979e-05,
+      "loss": 0.8272,
+      "step": 10465
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.886658989382634e-05,
+      "loss": 0.8739,
+      "step": 10466
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.886444315266213e-05,
+      "loss": 0.8122,
+      "step": 10467
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.886229626385001e-05,
+      "loss": 0.8261,
+      "step": 10468
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8860149227412854e-05,
+      "loss": 0.8807,
+      "step": 10469
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.885800204337352e-05,
+      "loss": 0.8192,
+      "step": 10470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.885585471175488e-05,
+      "loss": 0.7664,
+      "step": 10471
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.88537072325798e-05,
+      "loss": 0.8185,
+      "step": 10472
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8851559605871155e-05,
+      "loss": 0.8536,
+      "step": 10473
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.88494118316518e-05,
+      "loss": 0.8345,
+      "step": 10474
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.884726390994462e-05,
+      "loss": 0.83,
+      "step": 10475
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.884511584077249e-05,
+      "loss": 0.8582,
+      "step": 10476
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8842967624158274e-05,
+      "loss": 0.8009,
+      "step": 10477
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.884081926012487e-05,
+      "loss": 0.7842,
+      "step": 10478
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8838670748695135e-05,
+      "loss": 0.8976,
+      "step": 10479
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.883652208989196e-05,
+      "loss": 0.8391,
+      "step": 10480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8834373283738224e-05,
+      "loss": 0.843,
+      "step": 10481
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.883222433025681e-05,
+      "loss": 0.7904,
+      "step": 10482
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.883007522947061e-05,
+      "loss": 0.8439,
+      "step": 10483
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8827925981402496e-05,
+      "loss": 0.8272,
+      "step": 10484
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.882577658607537e-05,
+      "loss": 0.7812,
+      "step": 10485
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.882362704351212e-05,
+      "loss": 0.8735,
+      "step": 10486
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8821477353735634e-05,
+      "loss": 0.8551,
+      "step": 10487
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.881932751676881e-05,
+      "loss": 0.8709,
+      "step": 10488
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.881717753263453e-05,
+      "loss": 0.8063,
+      "step": 10489
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8815027401355694e-05,
+      "loss": 0.769,
+      "step": 10490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8812877122955214e-05,
+      "loss": 0.8121,
+      "step": 10491
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.881072669745597e-05,
+      "loss": 0.8868,
+      "step": 10492
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.880857612488088e-05,
+      "loss": 0.9819,
+      "step": 10493
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8806425405252834e-05,
+      "loss": 0.8471,
+      "step": 10494
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.880427453859473e-05,
+      "loss": 0.8711,
+      "step": 10495
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.88021235249295e-05,
+      "loss": 0.8381,
+      "step": 10496
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.879997236428002e-05,
+      "loss": 0.8221,
+      "step": 10497
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.879782105666923e-05,
+      "loss": 0.9537,
+      "step": 10498
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.879566960212e-05,
+      "loss": 0.8347,
+      "step": 10499
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.879351800065528e-05,
+      "loss": 0.8445,
+      "step": 10500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8791366252297955e-05,
+      "loss": 0.7655,
+      "step": 10501
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8789214357070963e-05,
+      "loss": 0.7576,
+      "step": 10502
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.878706231499721e-05,
+      "loss": 0.8088,
+      "step": 10503
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.878491012609961e-05,
+      "loss": 0.7476,
+      "step": 10504
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.878275779040108e-05,
+      "loss": 0.8071,
+      "step": 10505
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8780605307924564e-05,
+      "loss": 0.8345,
+      "step": 10506
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.877845267869296e-05,
+      "loss": 0.8044,
+      "step": 10507
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.877629990272919e-05,
+      "loss": 0.8355,
+      "step": 10508
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.87741469800562e-05,
+      "loss": 0.8452,
+      "step": 10509
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.87719939106969e-05,
+      "loss": 0.7682,
+      "step": 10510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.876984069467423e-05,
+      "loss": 0.7985,
+      "step": 10511
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.876768733201112e-05,
+      "loss": 0.7652,
+      "step": 10512
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.876553382273049e-05,
+      "loss": 0.8785,
+      "step": 10513
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.876338016685529e-05,
+      "loss": 0.7455,
+      "step": 10514
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.876122636440845e-05,
+      "loss": 0.8758,
+      "step": 10515
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.87590724154129e-05,
+      "loss": 0.9354,
+      "step": 10516
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.875691831989158e-05,
+      "loss": 0.8352,
+      "step": 10517
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.875476407786743e-05,
+      "loss": 0.809,
+      "step": 10518
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.87526096893634e-05,
+      "loss": 0.8053,
+      "step": 10519
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.875045515440242e-05,
+      "loss": 0.8186,
+      "step": 10520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.874830047300744e-05,
+      "loss": 0.8171,
+      "step": 10521
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8746145645201416e-05,
+      "loss": 0.8695,
+      "step": 10522
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.874399067100728e-05,
+      "loss": 0.8477,
+      "step": 10523
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.874183555044798e-05,
+      "loss": 0.8597,
+      "step": 10524
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8739680283546484e-05,
+      "loss": 0.8088,
+      "step": 10525
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8737524870325735e-05,
+      "loss": 0.8119,
+      "step": 10526
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.873536931080869e-05,
+      "loss": 0.8003,
+      "step": 10527
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.873321360501829e-05,
+      "loss": 1.0092,
+      "step": 10528
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.873105775297751e-05,
+      "loss": 0.7705,
+      "step": 10529
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8728901754709294e-05,
+      "loss": 0.7901,
+      "step": 10530
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.872674561023661e-05,
+      "loss": 0.8007,
+      "step": 10531
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.872458931958243e-05,
+      "loss": 0.7931,
+      "step": 10532
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.87224328827697e-05,
+      "loss": 0.8194,
+      "step": 10533
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.872027629982138e-05,
+      "loss": 0.8059,
+      "step": 10534
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.871811957076046e-05,
+      "loss": 0.8271,
+      "step": 10535
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.87159626956099e-05,
+      "loss": 0.8109,
+      "step": 10536
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.871380567439265e-05,
+      "loss": 0.9511,
+      "step": 10537
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8711648507131706e-05,
+      "loss": 0.8329,
+      "step": 10538
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8709491193850024e-05,
+      "loss": 0.8175,
+      "step": 10539
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.870733373457059e-05,
+      "loss": 0.9386,
+      "step": 10540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.870517612931637e-05,
+      "loss": 0.9198,
+      "step": 10541
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8703018378110345e-05,
+      "loss": 0.8879,
+      "step": 10542
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.870086048097549e-05,
+      "loss": 0.8125,
+      "step": 10543
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8698702437934785e-05,
+      "loss": 0.8114,
+      "step": 10544
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.869654424901123e-05,
+      "loss": 0.8731,
+      "step": 10545
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8694385914227785e-05,
+      "loss": 0.8855,
+      "step": 10546
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.869222743360745e-05,
+      "loss": 0.8626,
+      "step": 10547
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.86900688071732e-05,
+      "loss": 0.8232,
+      "step": 10548
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.868791003494803e-05,
+      "loss": 0.8871,
+      "step": 10549
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.868575111695493e-05,
+      "loss": 0.7896,
+      "step": 10550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.868359205321688e-05,
+      "loss": 0.7606,
+      "step": 10551
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8681432843756894e-05,
+      "loss": 0.9176,
+      "step": 10552
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.867927348859795e-05,
+      "loss": 0.8891,
+      "step": 10553
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8677113987763045e-05,
+      "loss": 0.8923,
+      "step": 10554
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.867495434127519e-05,
+      "loss": 0.8438,
+      "step": 10555
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.867279454915736e-05,
+      "loss": 0.8576,
+      "step": 10556
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8670634611432585e-05,
+      "loss": 0.8199,
+      "step": 10557
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8668474528123836e-05,
+      "loss": 0.8669,
+      "step": 10558
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8666314299254136e-05,
+      "loss": 0.814,
+      "step": 10559
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.866415392484649e-05,
+      "loss": 0.8646,
+      "step": 10560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8661993404923895e-05,
+      "loss": 0.9147,
+      "step": 10561
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.865983273950937e-05,
+      "loss": 0.8834,
+      "step": 10562
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.865767192862592e-05,
+      "loss": 0.8079,
+      "step": 10563
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.865551097229655e-05,
+      "loss": 0.7371,
+      "step": 10564
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.865334987054428e-05,
+      "loss": 0.9475,
+      "step": 10565
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.865118862339212e-05,
+      "loss": 0.8645,
+      "step": 10566
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.86490272308631e-05,
+      "loss": 0.9421,
+      "step": 10567
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8646865692980214e-05,
+      "loss": 0.87,
+      "step": 10568
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8644704009766504e-05,
+      "loss": 0.8083,
+      "step": 10569
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8642542181244976e-05,
+      "loss": 0.8594,
+      "step": 10570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8640380207438656e-05,
+      "loss": 0.8226,
+      "step": 10571
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.863821808837056e-05,
+      "loss": 0.8809,
+      "step": 10572
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8636055824063736e-05,
+      "loss": 0.8437,
+      "step": 10573
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.863389341454119e-05,
+      "loss": 0.8445,
+      "step": 10574
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.863173085982596e-05,
+      "loss": 0.8372,
+      "step": 10575
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.862956815994106e-05,
+      "loss": 0.8821,
+      "step": 10576
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8627405314909546e-05,
+      "loss": 0.8017,
+      "step": 10577
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.862524232475443e-05,
+      "loss": 0.8118,
+      "step": 10578
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.862307918949876e-05,
+      "loss": 0.7915,
+      "step": 10579
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.862091590916557e-05,
+      "loss": 0.8661,
+      "step": 10580
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8618752483777886e-05,
+      "loss": 0.8608,
+      "step": 10581
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8616588913358764e-05,
+      "loss": 0.787,
+      "step": 10582
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.861442519793124e-05,
+      "loss": 0.8218,
+      "step": 10583
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8612261337518346e-05,
+      "loss": 0.8747,
+      "step": 10584
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.861009733214314e-05,
+      "loss": 0.8485,
+      "step": 10585
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.860793318182865e-05,
+      "loss": 0.8225,
+      "step": 10586
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.860576888659794e-05,
+      "loss": 0.8175,
+      "step": 10587
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.860360444647405e-05,
+      "loss": 0.818,
+      "step": 10588
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8601439861480036e-05,
+      "loss": 0.8331,
+      "step": 10589
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8599275131638935e-05,
+      "loss": 0.742,
+      "step": 10590
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.859711025697382e-05,
+      "loss": 0.8618,
+      "step": 10591
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.859494523750773e-05,
+      "loss": 0.8056,
+      "step": 10592
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.859278007326373e-05,
+      "loss": 0.7852,
+      "step": 10593
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.859061476426487e-05,
+      "loss": 0.8424,
+      "step": 10594
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8588449310534224e-05,
+      "loss": 0.8506,
+      "step": 10595
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.858628371209483e-05,
+      "loss": 0.8949,
+      "step": 10596
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.858411796896977e-05,
+      "loss": 0.8778,
+      "step": 10597
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8581952081182104e-05,
+      "loss": 0.8392,
+      "step": 10598
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.857978604875489e-05,
+      "loss": 0.8269,
+      "step": 10599
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.85776198717112e-05,
+      "loss": 0.8052,
+      "step": 10600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.85754535500741e-05,
+      "loss": 0.839,
+      "step": 10601
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.857328708386666e-05,
+      "loss": 0.8818,
+      "step": 10602
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8571120473111956e-05,
+      "loss": 0.8157,
+      "step": 10603
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8568953717833054e-05,
+      "loss": 0.8118,
+      "step": 10604
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8566786818053036e-05,
+      "loss": 0.7733,
+      "step": 10605
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8564619773794974e-05,
+      "loss": 0.7808,
+      "step": 10606
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8562452585081955e-05,
+      "loss": 0.8887,
+      "step": 10607
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8560285251937044e-05,
+      "loss": 0.8185,
+      "step": 10608
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.855811777438333e-05,
+      "loss": 0.8664,
+      "step": 10609
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8555950152443885e-05,
+      "loss": 0.8795,
+      "step": 10610
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.855378238614181e-05,
+      "loss": 0.8333,
+      "step": 10611
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.855161447550018e-05,
+      "loss": 0.8746,
+      "step": 10612
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8549446420542066e-05,
+      "loss": 0.8044,
+      "step": 10613
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.854727822129059e-05,
+      "loss": 0.8388,
+      "step": 10614
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.854510987776883e-05,
+      "loss": 0.9059,
+      "step": 10615
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8542941389999866e-05,
+      "loss": 0.8363,
+      "step": 10616
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8540772758006795e-05,
+      "loss": 0.8239,
+      "step": 10617
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.853860398181273e-05,
+      "loss": 0.856,
+      "step": 10618
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8536435061440736e-05,
+      "loss": 0.8158,
+      "step": 10619
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.853426599691393e-05,
+      "loss": 0.76,
+      "step": 10620
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.853209678825541e-05,
+      "loss": 0.7891,
+      "step": 10621
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8529927435488274e-05,
+      "loss": 0.8343,
+      "step": 10622
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.852775793863562e-05,
+      "loss": 0.8087,
+      "step": 10623
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8525588297720564e-05,
+      "loss": 0.8437,
+      "step": 10624
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.85234185127662e-05,
+      "loss": 0.8114,
+      "step": 10625
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8521248583795644e-05,
+      "loss": 0.8289,
+      "step": 10626
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8519078510832e-05,
+      "loss": 0.7838,
+      "step": 10627
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.851690829389838e-05,
+      "loss": 0.9596,
+      "step": 10628
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8514737933017884e-05,
+      "loss": 0.848,
+      "step": 10629
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8512567428213635e-05,
+      "loss": 0.8639,
+      "step": 10630
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8510396779508753e-05,
+      "loss": 0.8459,
+      "step": 10631
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.850822598692634e-05,
+      "loss": 0.7929,
+      "step": 10632
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8506055050489524e-05,
+      "loss": 0.831,
+      "step": 10633
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8503883970221424e-05,
+      "loss": 0.7751,
+      "step": 10634
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.850171274614516e-05,
+      "loss": 0.8115,
+      "step": 10635
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.849954137828384e-05,
+      "loss": 0.8829,
+      "step": 10636
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8497369866660615e-05,
+      "loss": 0.8398,
+      "step": 10637
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.849519821129859e-05,
+      "loss": 0.8271,
+      "step": 10638
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.84930264122209e-05,
+      "loss": 0.8186,
+      "step": 10639
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8490854469450666e-05,
+      "loss": 0.8473,
+      "step": 10640
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.848868238301102e-05,
+      "loss": 0.828,
+      "step": 10641
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.84865101529251e-05,
+      "loss": 0.8328,
+      "step": 10642
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8484337779216035e-05,
+      "loss": 0.8757,
+      "step": 10643
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.848216526190696e-05,
+      "loss": 0.8385,
+      "step": 10644
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.847999260102101e-05,
+      "loss": 0.7781,
+      "step": 10645
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.847781979658132e-05,
+      "loss": 0.8515,
+      "step": 10646
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.847564684861103e-05,
+      "loss": 0.8276,
+      "step": 10647
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8473473757133285e-05,
+      "loss": 0.7279,
+      "step": 10648
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8471300522171226e-05,
+      "loss": 0.8717,
+      "step": 10649
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8469127143748e-05,
+      "loss": 0.8534,
+      "step": 10650
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.846695362188674e-05,
+      "loss": 0.7884,
+      "step": 10651
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.84647799566106e-05,
+      "loss": 0.9125,
+      "step": 10652
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8462606147942723e-05,
+      "loss": 0.8156,
+      "step": 10653
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.846043219590627e-05,
+      "loss": 0.7708,
+      "step": 10654
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.845825810052439e-05,
+      "loss": 0.856,
+      "step": 10655
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8456083861820224e-05,
+      "loss": 0.8031,
+      "step": 10656
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8453909479816944e-05,
+      "loss": 0.7912,
+      "step": 10657
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.845173495453769e-05,
+      "loss": 0.7621,
+      "step": 10658
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8449560286005624e-05,
+      "loss": 0.8239,
+      "step": 10659
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.844738547424391e-05,
+      "loss": 0.8,
+      "step": 10660
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.84452105192757e-05,
+      "loss": 0.8516,
+      "step": 10661
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.844303542112416e-05,
+      "loss": 0.8156,
+      "step": 10662
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.844086017981246e-05,
+      "loss": 0.8339,
+      "step": 10663
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.843868479536376e-05,
+      "loss": 0.7584,
+      "step": 10664
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8436509267801225e-05,
+      "loss": 0.8621,
+      "step": 10665
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.843433359714802e-05,
+      "loss": 0.8125,
+      "step": 10666
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.843215778342732e-05,
+      "loss": 0.7953,
+      "step": 10667
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.84299818266623e-05,
+      "loss": 0.8457,
+      "step": 10668
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.842780572687612e-05,
+      "loss": 0.7945,
+      "step": 10669
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.842562948409196e-05,
+      "loss": 0.8048,
+      "step": 10670
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8423453098333004e-05,
+      "loss": 0.7902,
+      "step": 10671
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.842127656962241e-05,
+      "loss": 0.7871,
+      "step": 10672
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.841909989798338e-05,
+      "loss": 0.7747,
+      "step": 10673
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8416923083439086e-05,
+      "loss": 0.8598,
+      "step": 10674
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.841474612601269e-05,
+      "loss": 0.7805,
+      "step": 10675
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.841256902572741e-05,
+      "loss": 0.7304,
+      "step": 10676
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8410391782606404e-05,
+      "loss": 0.8757,
+      "step": 10677
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.840821439667287e-05,
+      "loss": 0.8403,
+      "step": 10678
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.840603686794999e-05,
+      "loss": 0.835,
+      "step": 10679
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.840385919646096e-05,
+      "loss": 0.8556,
+      "step": 10680
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.840168138222896e-05,
+      "loss": 0.8664,
+      "step": 10681
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.83995034252772e-05,
+      "loss": 0.8309,
+      "step": 10682
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8397325325628854e-05,
+      "loss": 0.8613,
+      "step": 10683
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.839514708330713e-05,
+      "loss": 0.7921,
+      "step": 10684
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.839296869833523e-05,
+      "loss": 0.9222,
+      "step": 10685
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8390790170736334e-05,
+      "loss": 0.809,
+      "step": 10686
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.838861150053365e-05,
+      "loss": 0.836,
+      "step": 10687
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.83864326877504e-05,
+      "loss": 0.8735,
+      "step": 10688
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8384253732409754e-05,
+      "loss": 0.9182,
+      "step": 10689
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8382074634534934e-05,
+      "loss": 0.8687,
+      "step": 10690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8379895394149145e-05,
+      "loss": 0.9204,
+      "step": 10691
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.837771601127559e-05,
+      "loss": 0.9229,
+      "step": 10692
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8375536485937486e-05,
+      "loss": 0.8189,
+      "step": 10693
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8373356818158035e-05,
+      "loss": 0.8289,
+      "step": 10694
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.837117700796045e-05,
+      "loss": 0.8291,
+      "step": 10695
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.836899705536796e-05,
+      "loss": 0.7637,
+      "step": 10696
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.836681696040376e-05,
+      "loss": 0.7685,
+      "step": 10697
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8364636723091075e-05,
+      "loss": 0.8192,
+      "step": 10698
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.836245634345312e-05,
+      "loss": 0.8569,
+      "step": 10699
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.836027582151312e-05,
+      "loss": 0.8569,
+      "step": 10700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.83580951572943e-05,
+      "loss": 0.8379,
+      "step": 10701
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8355914350819866e-05,
+      "loss": 0.8348,
+      "step": 10702
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.835373340211306e-05,
+      "loss": 0.8062,
+      "step": 10703
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8351552311197095e-05,
+      "loss": 0.8558,
+      "step": 10704
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.83493710780952e-05,
+      "loss": 0.897,
+      "step": 10705
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8347189702830624e-05,
+      "loss": 0.8289,
+      "step": 10706
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.834500818542656e-05,
+      "loss": 0.8745,
+      "step": 10707
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.834282652590628e-05,
+      "loss": 0.7603,
+      "step": 10708
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8340644724292985e-05,
+      "loss": 0.865,
+      "step": 10709
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8338462780609926e-05,
+      "loss": 0.9343,
+      "step": 10710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8336280694880344e-05,
+      "loss": 0.8101,
+      "step": 10711
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.833409846712746e-05,
+      "loss": 0.7385,
+      "step": 10712
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.833191609737452e-05,
+      "loss": 0.8233,
+      "step": 10713
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.832973358564478e-05,
+      "loss": 0.8259,
+      "step": 10714
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.832755093196146e-05,
+      "loss": 0.7646,
+      "step": 10715
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.832536813634782e-05,
+      "loss": 0.7635,
+      "step": 10716
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.83231851988271e-05,
+      "loss": 0.8198,
+      "step": 10717
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8321002119422545e-05,
+      "loss": 0.849,
+      "step": 10718
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.83188188981574e-05,
+      "loss": 0.8502,
+      "step": 10719
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8316635535054925e-05,
+      "loss": 0.8005,
+      "step": 10720
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8314452030138356e-05,
+      "loss": 0.8097,
+      "step": 10721
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.831226838343097e-05,
+      "loss": 0.7926,
+      "step": 10722
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.831008459495601e-05,
+      "loss": 0.8396,
+      "step": 10723
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.830790066473671e-05,
+      "loss": 0.9068,
+      "step": 10724
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8305716592796366e-05,
+      "loss": 0.7871,
+      "step": 10725
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8303532379158216e-05,
+      "loss": 0.7979,
+      "step": 10726
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8301348023845506e-05,
+      "loss": 0.8113,
+      "step": 10727
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.829916352688154e-05,
+      "loss": 0.8492,
+      "step": 10728
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.829697888828954e-05,
+      "loss": 0.9169,
+      "step": 10729
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.829479410809279e-05,
+      "loss": 0.8302,
+      "step": 10730
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.829260918631456e-05,
+      "loss": 0.843,
+      "step": 10731
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8290424122978106e-05,
+      "loss": 0.8053,
+      "step": 10732
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8288238918106704e-05,
+      "loss": 0.8663,
+      "step": 10733
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.828605357172363e-05,
+      "loss": 0.863,
+      "step": 10734
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.828386808385215e-05,
+      "loss": 0.7699,
+      "step": 10735
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8281682454515535e-05,
+      "loss": 0.8787,
+      "step": 10736
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.827949668373707e-05,
+      "loss": 0.7848,
+      "step": 10737
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.827731077154002e-05,
+      "loss": 0.809,
+      "step": 10738
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.827512471794769e-05,
+      "loss": 0.8265,
+      "step": 10739
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8272938522983324e-05,
+      "loss": 0.8214,
+      "step": 10740
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.827075218667022e-05,
+      "loss": 0.8132,
+      "step": 10741
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8268565709031675e-05,
+      "loss": 0.8198,
+      "step": 10742
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.826637909009095e-05,
+      "loss": 0.8457,
+      "step": 10743
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.826419232987135e-05,
+      "loss": 0.9481,
+      "step": 10744
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.826200542839615e-05,
+      "loss": 0.778,
+      "step": 10745
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.825981838568864e-05,
+      "loss": 0.7555,
+      "step": 10746
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8257631201772114e-05,
+      "loss": 0.8277,
+      "step": 10747
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8255443876669875e-05,
+      "loss": 0.7534,
+      "step": 10748
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8253256410405205e-05,
+      "loss": 0.7356,
+      "step": 10749
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8251068803001386e-05,
+      "loss": 0.7907,
+      "step": 10750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.824888105448174e-05,
+      "loss": 0.8051,
+      "step": 10751
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.824669316486955e-05,
+      "loss": 0.8271,
+      "step": 10752
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.824450513418813e-05,
+      "loss": 0.7711,
+      "step": 10753
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8242316962460765e-05,
+      "loss": 0.8897,
+      "step": 10754
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.824012864971076e-05,
+      "loss": 0.8929,
+      "step": 10755
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.823794019596143e-05,
+      "loss": 0.8411,
+      "step": 10756
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.823575160123607e-05,
+      "loss": 0.8178,
+      "step": 10757
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8233562865557994e-05,
+      "loss": 0.8463,
+      "step": 10758
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8231373988950506e-05,
+      "loss": 0.8595,
+      "step": 10759
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.822918497143691e-05,
+      "loss": 0.8566,
+      "step": 10760
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.822699581304053e-05,
+      "loss": 0.8862,
+      "step": 10761
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8224806513784683e-05,
+      "loss": 0.8595,
+      "step": 10762
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8222617073692675e-05,
+      "loss": 0.766,
+      "step": 10763
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8220427492787815e-05,
+      "loss": 0.8954,
+      "step": 10764
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8218237771093436e-05,
+      "loss": 0.8672,
+      "step": 10765
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8216047908632846e-05,
+      "loss": 0.8865,
+      "step": 10766
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.821385790542936e-05,
+      "loss": 0.8586,
+      "step": 10767
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.821166776150632e-05,
+      "loss": 0.7761,
+      "step": 10768
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.820947747688704e-05,
+      "loss": 0.807,
+      "step": 10769
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8207287051594844e-05,
+      "loss": 0.8529,
+      "step": 10770
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8205096485653056e-05,
+      "loss": 0.7659,
+      "step": 10771
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8202905779085015e-05,
+      "loss": 0.8132,
+      "step": 10772
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.820071493191403e-05,
+      "loss": 0.8368,
+      "step": 10773
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8198523944163455e-05,
+      "loss": 0.7882,
+      "step": 10774
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.819633281585662e-05,
+      "loss": 0.8126,
+      "step": 10775
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.819414154701683e-05,
+      "loss": 0.9836,
+      "step": 10776
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.819195013766746e-05,
+      "loss": 0.9556,
+      "step": 10777
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.818975858783182e-05,
+      "loss": 0.8062,
+      "step": 10778
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8187566897533276e-05,
+      "loss": 0.8084,
+      "step": 10779
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8185375066795135e-05,
+      "loss": 0.7924,
+      "step": 10780
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.818318309564076e-05,
+      "loss": 0.9525,
+      "step": 10781
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.818099098409349e-05,
+      "loss": 0.865,
+      "step": 10782
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8178798732176664e-05,
+      "loss": 0.8361,
+      "step": 10783
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.817660633991364e-05,
+      "loss": 0.8193,
+      "step": 10784
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.817441380732775e-05,
+      "loss": 0.7854,
+      "step": 10785
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.817222113444235e-05,
+      "loss": 0.8236,
+      "step": 10786
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8170028321280795e-05,
+      "loss": 0.8618,
+      "step": 10787
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8167835367866436e-05,
+      "loss": 0.8126,
+      "step": 10788
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.816564227422262e-05,
+      "loss": 0.7504,
+      "step": 10789
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.816344904037271e-05,
+      "loss": 0.801,
+      "step": 10790
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.816125566634007e-05,
+      "loss": 0.8029,
+      "step": 10791
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.815906215214804e-05,
+      "loss": 0.8667,
+      "step": 10792
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.815686849781997e-05,
+      "loss": 0.8271,
+      "step": 10793
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.815467470337926e-05,
+      "loss": 0.9096,
+      "step": 10794
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.815248076884925e-05,
+      "loss": 0.863,
+      "step": 10795
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.815028669425329e-05,
+      "loss": 0.8903,
+      "step": 10796
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8148092479614774e-05,
+      "loss": 0.75,
+      "step": 10797
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.814589812495705e-05,
+      "loss": 0.8306,
+      "step": 10798
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8143703630303495e-05,
+      "loss": 0.804,
+      "step": 10799
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8141508995677474e-05,
+      "loss": 0.8617,
+      "step": 10800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.813931422110236e-05,
+      "loss": 0.7748,
+      "step": 10801
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.813711930660153e-05,
+      "loss": 0.8004,
+      "step": 10802
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8134924252198346e-05,
+      "loss": 0.8963,
+      "step": 10803
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.81327290579162e-05,
+      "loss": 0.8746,
+      "step": 10804
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.813053372377846e-05,
+      "loss": 0.8309,
+      "step": 10805
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.812833824980852e-05,
+      "loss": 0.8114,
+      "step": 10806
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8126142636029724e-05,
+      "loss": 0.8277,
+      "step": 10807
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8123946882465496e-05,
+      "loss": 0.7533,
+      "step": 10808
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8121750989139205e-05,
+      "loss": 0.7636,
+      "step": 10809
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8119554956074224e-05,
+      "loss": 0.8231,
+      "step": 10810
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.811735878329394e-05,
+      "loss": 0.8379,
+      "step": 10811
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8115162470821766e-05,
+      "loss": 0.8546,
+      "step": 10812
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8112966018681054e-05,
+      "loss": 0.7903,
+      "step": 10813
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.811076942689523e-05,
+      "loss": 0.8009,
+      "step": 10814
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.810857269548767e-05,
+      "loss": 0.8115,
+      "step": 10815
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.810637582448176e-05,
+      "loss": 0.8641,
+      "step": 10816
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8104178813900915e-05,
+      "loss": 0.8988,
+      "step": 10817
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8101981663768516e-05,
+      "loss": 0.8661,
+      "step": 10818
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.809978437410797e-05,
+      "loss": 0.8083,
+      "step": 10819
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.809758694494268e-05,
+      "loss": 0.8451,
+      "step": 10820
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.809538937629602e-05,
+      "loss": 0.8512,
+      "step": 10821
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.809319166819143e-05,
+      "loss": 0.7616,
+      "step": 10822
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8090993820652295e-05,
+      "loss": 0.8648,
+      "step": 10823
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8088795833702024e-05,
+      "loss": 0.8208,
+      "step": 10824
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.808659770736401e-05,
+      "loss": 0.8306,
+      "step": 10825
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.80843994416617e-05,
+      "loss": 0.7388,
+      "step": 10826
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.808220103661846e-05,
+      "loss": 0.7909,
+      "step": 10827
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8080002492257724e-05,
+      "loss": 0.8337,
+      "step": 10828
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.807780380860292e-05,
+      "loss": 0.8038,
+      "step": 10829
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8075604985677424e-05,
+      "loss": 0.8569,
+      "step": 10830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.807340602350469e-05,
+      "loss": 0.9329,
+      "step": 10831
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8071206922108105e-05,
+      "loss": 0.7778,
+      "step": 10832
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8069007681511105e-05,
+      "loss": 0.8298,
+      "step": 10833
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.806680830173711e-05,
+      "loss": 0.8499,
+      "step": 10834
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8064608782809544e-05,
+      "loss": 0.8564,
+      "step": 10835
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.806240912475182e-05,
+      "loss": 0.8246,
+      "step": 10836
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.806020932758737e-05,
+      "loss": 0.8855,
+      "step": 10837
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.805800939133962e-05,
+      "loss": 0.8525,
+      "step": 10838
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8055809316032e-05,
+      "loss": 0.8372,
+      "step": 10839
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8053609101687934e-05,
+      "loss": 0.8543,
+      "step": 10840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.805140874833086e-05,
+      "loss": 0.7795,
+      "step": 10841
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.80492082559842e-05,
+      "loss": 0.8881,
+      "step": 10842
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8047007624671406e-05,
+      "loss": 0.8027,
+      "step": 10843
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.804480685441589e-05,
+      "loss": 0.8421,
+      "step": 10844
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8042605945241106e-05,
+      "loss": 0.8132,
+      "step": 10845
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.804040489717049e-05,
+      "loss": 0.8087,
+      "step": 10846
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.803820371022747e-05,
+      "loss": 0.8648,
+      "step": 10847
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.80360023844355e-05,
+      "loss": 0.8957,
+      "step": 10848
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8033800919818016e-05,
+      "loss": 0.7711,
+      "step": 10849
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.803159931639848e-05,
+      "loss": 0.8141,
+      "step": 10850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.802939757420031e-05,
+      "loss": 0.9109,
+      "step": 10851
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.802719569324697e-05,
+      "loss": 0.8078,
+      "step": 10852
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8024993673561904e-05,
+      "loss": 0.7417,
+      "step": 10853
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.802279151516856e-05,
+      "loss": 0.9154,
+      "step": 10854
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8020589218090394e-05,
+      "loss": 0.8298,
+      "step": 10855
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.801838678235086e-05,
+      "loss": 0.883,
+      "step": 10856
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.80161842079734e-05,
+      "loss": 0.7773,
+      "step": 10857
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.80139814949815e-05,
+      "loss": 0.8678,
+      "step": 10858
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.801177864339859e-05,
+      "loss": 0.7978,
+      "step": 10859
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8009575653248134e-05,
+      "loss": 0.8966,
+      "step": 10860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8007372524553594e-05,
+      "loss": 0.8715,
+      "step": 10861
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.800516925733844e-05,
+      "loss": 0.8746,
+      "step": 10862
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.800296585162613e-05,
+      "loss": 0.7695,
+      "step": 10863
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.800076230744012e-05,
+      "loss": 0.847,
+      "step": 10864
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.799855862480388e-05,
+      "loss": 0.8896,
+      "step": 10865
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.79963548037409e-05,
+      "loss": 0.8438,
+      "step": 10866
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7994150844274614e-05,
+      "loss": 0.8366,
+      "step": 10867
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.799194674642852e-05,
+      "loss": 0.7369,
+      "step": 10868
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.798974251022608e-05,
+      "loss": 0.8739,
+      "step": 10869
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.798753813569077e-05,
+      "loss": 0.8345,
+      "step": 10870
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.798533362284606e-05,
+      "loss": 0.7731,
+      "step": 10871
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7983128971715434e-05,
+      "loss": 0.8093,
+      "step": 10872
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.798092418232236e-05,
+      "loss": 0.8947,
+      "step": 10873
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.797871925469033e-05,
+      "loss": 0.8013,
+      "step": 10874
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7976514188842814e-05,
+      "loss": 0.8231,
+      "step": 10875
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7974308984803305e-05,
+      "loss": 0.8672,
+      "step": 10876
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.797210364259528e-05,
+      "loss": 0.8706,
+      "step": 10877
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.796989816224223e-05,
+      "loss": 0.8119,
+      "step": 10878
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7967692543767644e-05,
+      "loss": 0.8384,
+      "step": 10879
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.796548678719499e-05,
+      "loss": 0.8283,
+      "step": 10880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.796328089254779e-05,
+      "loss": 0.7538,
+      "step": 10881
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.79610748598495e-05,
+      "loss": 0.8543,
+      "step": 10882
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.795886868912365e-05,
+      "loss": 0.8599,
+      "step": 10883
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7956662380393706e-05,
+      "loss": 0.8173,
+      "step": 10884
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.795445593368317e-05,
+      "loss": 0.8452,
+      "step": 10885
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.795224934901555e-05,
+      "loss": 0.8781,
+      "step": 10886
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.795004262641434e-05,
+      "loss": 0.8355,
+      "step": 10887
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7947835765903026e-05,
+      "loss": 0.7854,
+      "step": 10888
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7945628767505134e-05,
+      "loss": 0.7673,
+      "step": 10889
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.794342163124416e-05,
+      "loss": 0.8702,
+      "step": 10890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7941214357143595e-05,
+      "loss": 0.8185,
+      "step": 10891
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.793900694522695e-05,
+      "loss": 0.7836,
+      "step": 10892
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.793679939551775e-05,
+      "loss": 0.8549,
+      "step": 10893
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.793459170803947e-05,
+      "loss": 0.8283,
+      "step": 10894
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.793238388281566e-05,
+      "loss": 0.811,
+      "step": 10895
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.79301759198698e-05,
+      "loss": 0.792,
+      "step": 10896
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.792796781922542e-05,
+      "loss": 0.8085,
+      "step": 10897
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.792575958090604e-05,
+      "loss": 0.9088,
+      "step": 10898
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.792355120493516e-05,
+      "loss": 0.867,
+      "step": 10899
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.792134269133631e-05,
+      "loss": 0.8211,
+      "step": 10900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.791913404013301e-05,
+      "loss": 0.9079,
+      "step": 10901
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.791692525134876e-05,
+      "loss": 0.877,
+      "step": 10902
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7914716325007106e-05,
+      "loss": 0.8612,
+      "step": 10903
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.791250726113157e-05,
+      "loss": 0.7681,
+      "step": 10904
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7910298059745655e-05,
+      "loss": 0.8528,
+      "step": 10905
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.790808872087292e-05,
+      "loss": 0.8644,
+      "step": 10906
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.790587924453687e-05,
+      "loss": 0.891,
+      "step": 10907
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.790366963076104e-05,
+      "loss": 0.7894,
+      "step": 10908
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.790145987956897e-05,
+      "loss": 0.877,
+      "step": 10909
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.789924999098418e-05,
+      "loss": 0.873,
+      "step": 10910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.789703996503021e-05,
+      "loss": 0.7926,
+      "step": 10911
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.789482980173059e-05,
+      "loss": 0.8211,
+      "step": 10912
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.789261950110887e-05,
+      "loss": 0.726,
+      "step": 10913
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.789040906318857e-05,
+      "loss": 0.8514,
+      "step": 10914
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.788819848799324e-05,
+      "loss": 0.7953,
+      "step": 10915
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.788598777554643e-05,
+      "loss": 0.7811,
+      "step": 10916
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.788377692587166e-05,
+      "loss": 0.8094,
+      "step": 10917
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.78815659389925e-05,
+      "loss": 0.7881,
+      "step": 10918
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7879354814932475e-05,
+      "loss": 0.7782,
+      "step": 10919
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787714355371514e-05,
+      "loss": 0.8472,
+      "step": 10920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787493215536406e-05,
+      "loss": 0.8801,
+      "step": 10921
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7872720619902745e-05,
+      "loss": 0.8034,
+      "step": 10922
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787050894735479e-05,
+      "loss": 0.8361,
+      "step": 10923
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.786829713774372e-05,
+      "loss": 0.8243,
+      "step": 10924
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.78660851910931e-05,
+      "loss": 0.8256,
+      "step": 10925
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7863873107426486e-05,
+      "loss": 0.8277,
+      "step": 10926
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7861660886767426e-05,
+      "loss": 1.1501,
+      "step": 10927
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7859448529139487e-05,
+      "loss": 0.7731,
+      "step": 10928
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7857236034566235e-05,
+      "loss": 0.8764,
+      "step": 10929
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7855023403071225e-05,
+      "loss": 0.9226,
+      "step": 10930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.785281063467802e-05,
+      "loss": 0.7703,
+      "step": 10931
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.785059772941018e-05,
+      "loss": 0.8727,
+      "step": 10932
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7848384687291286e-05,
+      "loss": 0.7889,
+      "step": 10933
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7846171508344876e-05,
+      "loss": 0.805,
+      "step": 10934
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7843958192594554e-05,
+      "loss": 0.879,
+      "step": 10935
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.784174474006387e-05,
+      "loss": 0.8708,
+      "step": 10936
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.78395311507764e-05,
+      "loss": 0.8427,
+      "step": 10937
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.783731742475573e-05,
+      "loss": 0.8061,
+      "step": 10938
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.783510356202541e-05,
+      "loss": 0.7227,
+      "step": 10939
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.783288956260903e-05,
+      "loss": 0.8438,
+      "step": 10940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.783067542653017e-05,
+      "loss": 0.7718,
+      "step": 10941
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7828461153812413e-05,
+      "loss": 0.7192,
+      "step": 10942
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.782624674447932e-05,
+      "loss": 0.8104,
+      "step": 10943
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7824032198554494e-05,
+      "loss": 0.7905,
+      "step": 10944
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.782181751606151e-05,
+      "loss": 0.7459,
+      "step": 10945
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7819602697023946e-05,
+      "loss": 0.8022,
+      "step": 10946
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.781738774146541e-05,
+      "loss": 0.7813,
+      "step": 10947
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.781517264940947e-05,
+      "loss": 0.7545,
+      "step": 10948
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7812957420879715e-05,
+      "loss": 0.8412,
+      "step": 10949
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.781074205589975e-05,
+      "loss": 0.8236,
+      "step": 10950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7808526554493165e-05,
+      "loss": 0.8585,
+      "step": 10951
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.780631091668354e-05,
+      "loss": 0.8149,
+      "step": 10952
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7804095142494475e-05,
+      "loss": 0.7814,
+      "step": 10953
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7801879231949575e-05,
+      "loss": 0.8509,
+      "step": 10954
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7799663185072446e-05,
+      "loss": 0.7803,
+      "step": 10955
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7797447001886654e-05,
+      "loss": 0.8095,
+      "step": 10956
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.779523068241583e-05,
+      "loss": 0.839,
+      "step": 10957
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.779301422668357e-05,
+      "loss": 0.8022,
+      "step": 10958
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.779079763471347e-05,
+      "loss": 0.7673,
+      "step": 10959
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.778858090652914e-05,
+      "loss": 0.7638,
+      "step": 10960
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.77863640421542e-05,
+      "loss": 0.8766,
+      "step": 10961
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.778414704161224e-05,
+      "loss": 0.8695,
+      "step": 10962
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.778192990492687e-05,
+      "loss": 0.7944,
+      "step": 10963
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.777971263212171e-05,
+      "loss": 0.8618,
+      "step": 10964
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.777749522322036e-05,
+      "loss": 0.8204,
+      "step": 10965
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.777527767824645e-05,
+      "loss": 0.8726,
+      "step": 10966
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.777305999722359e-05,
+      "loss": 0.7499,
+      "step": 10967
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.77708421801754e-05,
+      "loss": 0.7762,
+      "step": 10968
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.776862422712549e-05,
+      "loss": 0.7622,
+      "step": 10969
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.776640613809748e-05,
+      "loss": 0.8617,
+      "step": 10970
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7764187913115e-05,
+      "loss": 0.788,
+      "step": 10971
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7761969552201673e-05,
+      "loss": 0.8029,
+      "step": 10972
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.775975105538111e-05,
+      "loss": 0.843,
+      "step": 10973
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.775753242267694e-05,
+      "loss": 0.8634,
+      "step": 10974
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7755313654112814e-05,
+      "loss": 0.8407,
+      "step": 10975
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7753094749712324e-05,
+      "loss": 0.8811,
+      "step": 10976
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.775087570949912e-05,
+      "loss": 0.7856,
+      "step": 10977
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.774865653349684e-05,
+      "loss": 0.8232,
+      "step": 10978
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7746437221729105e-05,
+      "loss": 0.7712,
+      "step": 10979
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7744217774219546e-05,
+      "loss": 0.7901,
+      "step": 10980
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7741998190991815e-05,
+      "loss": 0.7793,
+      "step": 10981
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7739778472069534e-05,
+      "loss": 0.8425,
+      "step": 10982
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7737558617476334e-05,
+      "loss": 0.8197,
+      "step": 10983
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.773533862723588e-05,
+      "loss": 0.839,
+      "step": 10984
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.77331185013718e-05,
+      "loss": 0.8183,
+      "step": 10985
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.773089823990774e-05,
+      "loss": 0.8629,
+      "step": 10986
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.772867784286734e-05,
+      "loss": 0.7948,
+      "step": 10987
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7726457310274255e-05,
+      "loss": 0.8208,
+      "step": 10988
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7724236642152115e-05,
+      "loss": 0.8772,
+      "step": 10989
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7722015838524586e-05,
+      "loss": 0.7749,
+      "step": 10990
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7719794899415316e-05,
+      "loss": 0.8147,
+      "step": 10991
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7717573824847954e-05,
+      "loss": 0.7357,
+      "step": 10992
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.771535261484614e-05,
+      "loss": 0.9187,
+      "step": 10993
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.771313126943355e-05,
+      "loss": 0.8256,
+      "step": 10994
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.771090978863383e-05,
+      "loss": 0.836,
+      "step": 10995
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.770868817247063e-05,
+      "loss": 0.7384,
+      "step": 10996
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7706466420967625e-05,
+      "loss": 0.7841,
+      "step": 10997
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.770424453414846e-05,
+      "loss": 0.8306,
+      "step": 10998
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.770202251203681e-05,
+      "loss": 0.9061,
+      "step": 10999
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7699800354656335e-05,
+      "loss": 0.7076,
+      "step": 11000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7697578062030685e-05,
+      "loss": 0.8926,
+      "step": 11001
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.769535563418355e-05,
+      "loss": 0.8353,
+      "step": 11002
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7693133071138574e-05,
+      "loss": 0.8405,
+      "step": 11003
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7690910372919444e-05,
+      "loss": 0.8421,
+      "step": 11004
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.768868753954982e-05,
+      "loss": 0.8132,
+      "step": 11005
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.768646457105337e-05,
+      "loss": 0.8075,
+      "step": 11006
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7684241467453787e-05,
+      "loss": 0.8113,
+      "step": 11007
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.768201822877473e-05,
+      "loss": 0.8324,
+      "step": 11008
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.767979485503988e-05,
+      "loss": 0.8429,
+      "step": 11009
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.76775713462729e-05,
+      "loss": 0.8797,
+      "step": 11010
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.76753477024975e-05,
+      "loss": 0.8186,
+      "step": 11011
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.767312392373733e-05,
+      "loss": 0.8252,
+      "step": 11012
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.767090001001609e-05,
+      "loss": 0.8615,
+      "step": 11013
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.766867596135746e-05,
+      "loss": 0.8405,
+      "step": 11014
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.766645177778511e-05,
+      "loss": 0.7627,
+      "step": 11015
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.766422745932275e-05,
+      "loss": 0.8066,
+      "step": 11016
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7662003005994054e-05,
+      "loss": 0.7856,
+      "step": 11017
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7659778417822706e-05,
+      "loss": 0.7585,
+      "step": 11018
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.765755369483242e-05,
+      "loss": 0.8165,
+      "step": 11019
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.765532883704686e-05,
+      "loss": 0.7826,
+      "step": 11020
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.765310384448973e-05,
+      "loss": 0.8244,
+      "step": 11021
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.765087871718473e-05,
+      "loss": 0.8052,
+      "step": 11022
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.764865345515555e-05,
+      "loss": 0.8054,
+      "step": 11023
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.764642805842589e-05,
+      "loss": 0.7852,
+      "step": 11024
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.764420252701945e-05,
+      "loss": 0.8262,
+      "step": 11025
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.764197686095993e-05,
+      "loss": 0.8476,
+      "step": 11026
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.763975106027102e-05,
+      "loss": 0.8652,
+      "step": 11027
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7637525124976454e-05,
+      "loss": 0.9272,
+      "step": 11028
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7635299055099906e-05,
+      "loss": 0.8247,
+      "step": 11029
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.76330728506651e-05,
+      "loss": 0.8559,
+      "step": 11030
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.763084651169574e-05,
+      "loss": 0.8271,
+      "step": 11031
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.762862003821552e-05,
+      "loss": 0.8044,
+      "step": 11032
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7626393430248173e-05,
+      "loss": 0.8392,
+      "step": 11033
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.762416668781741e-05,
+      "loss": 0.7668,
+      "step": 11034
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7621939810946924e-05,
+      "loss": 0.8168,
+      "step": 11035
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.761971279966044e-05,
+      "loss": 0.7892,
+      "step": 11036
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7617485653981686e-05,
+      "loss": 0.7546,
+      "step": 11037
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.761525837393436e-05,
+      "loss": 0.7565,
+      "step": 11038
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.76130309595422e-05,
+      "loss": 0.7644,
+      "step": 11039
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7610803410828913e-05,
+      "loss": 0.7605,
+      "step": 11040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7608575727818226e-05,
+      "loss": 0.8393,
+      "step": 11041
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.760634791053387e-05,
+      "loss": 0.8655,
+      "step": 11042
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.760411995899955e-05,
+      "loss": 0.827,
+      "step": 11043
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7601891873239014e-05,
+      "loss": 0.8022,
+      "step": 11044
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.759966365327598e-05,
+      "loss": 0.7179,
+      "step": 11045
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.759743529913417e-05,
+      "loss": 0.748,
+      "step": 11046
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7595206810837324e-05,
+      "loss": 0.8672,
+      "step": 11047
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.759297818840918e-05,
+      "loss": 0.7885,
+      "step": 11048
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.759074943187346e-05,
+      "loss": 0.8313,
+      "step": 11049
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.75885205412539e-05,
+      "loss": 0.8215,
+      "step": 11050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.758629151657425e-05,
+      "loss": 0.8903,
+      "step": 11051
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.758406235785822e-05,
+      "loss": 0.7651,
+      "step": 11052
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7581833065129566e-05,
+      "loss": 0.8049,
+      "step": 11053
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.757960363841204e-05,
+      "loss": 0.7731,
+      "step": 11054
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.757737407772937e-05,
+      "loss": 0.9158,
+      "step": 11055
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.757514438310529e-05,
+      "loss": 0.8036,
+      "step": 11056
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.757291455456357e-05,
+      "loss": 0.8066,
+      "step": 11057
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7570684592127935e-05,
+      "loss": 0.8052,
+      "step": 11058
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7568454495822146e-05,
+      "loss": 0.8406,
+      "step": 11059
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7566224265669944e-05,
+      "loss": 0.9023,
+      "step": 11060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.756399390169509e-05,
+      "loss": 0.9261,
+      "step": 11061
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7561763403921324e-05,
+      "loss": 0.9395,
+      "step": 11062
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7559532772372407e-05,
+      "loss": 0.811,
+      "step": 11063
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.755730200707209e-05,
+      "loss": 0.7956,
+      "step": 11064
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7555071108044124e-05,
+      "loss": 0.8168,
+      "step": 11065
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.755284007531228e-05,
+      "loss": 0.7713,
+      "step": 11066
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.755060890890031e-05,
+      "loss": 0.8931,
+      "step": 11067
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7548377608831966e-05,
+      "loss": 0.8774,
+      "step": 11068
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.754614617513103e-05,
+      "loss": 0.8326,
+      "step": 11069
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7543914607821244e-05,
+      "loss": 0.8341,
+      "step": 11070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.754168290692639e-05,
+      "loss": 0.8754,
+      "step": 11071
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7539451072470236e-05,
+      "loss": 0.8473,
+      "step": 11072
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.753721910447653e-05,
+      "loss": 0.7992,
+      "step": 11073
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7534987002969044e-05,
+      "loss": 0.8206,
+      "step": 11074
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7532754767971565e-05,
+      "loss": 0.8847,
+      "step": 11075
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.753052239950786e-05,
+      "loss": 0.8317,
+      "step": 11076
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7528289897601696e-05,
+      "loss": 0.8547,
+      "step": 11077
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.752605726227685e-05,
+      "loss": 0.7869,
+      "step": 11078
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.75238244935571e-05,
+      "loss": 0.8123,
+      "step": 11079
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.752159159146622e-05,
+      "loss": 0.8671,
+      "step": 11080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.751935855602799e-05,
+      "loss": 0.7655,
+      "step": 11081
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.75171253872662e-05,
+      "loss": 0.772,
+      "step": 11082
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7514892085204623e-05,
+      "loss": 0.853,
+      "step": 11083
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.751265864986704e-05,
+      "loss": 0.8041,
+      "step": 11084
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.751042508127724e-05,
+      "loss": 0.8888,
+      "step": 11085
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.750819137945901e-05,
+      "loss": 0.7996,
+      "step": 11086
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7505957544436124e-05,
+      "loss": 0.8797,
+      "step": 11087
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7503723576232395e-05,
+      "loss": 0.8137,
+      "step": 11088
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.75014894748716e-05,
+      "loss": 0.7755,
+      "step": 11089
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.749925524037753e-05,
+      "loss": 0.8699,
+      "step": 11090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.749702087277398e-05,
+      "loss": 0.8374,
+      "step": 11091
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.749478637208474e-05,
+      "loss": 0.8407,
+      "step": 11092
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.749255173833362e-05,
+      "loss": 0.8202,
+      "step": 11093
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7490316971544405e-05,
+      "loss": 0.7887,
+      "step": 11094
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7488082071740894e-05,
+      "loss": 0.8305,
+      "step": 11095
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7485847038946895e-05,
+      "loss": 0.7416,
+      "step": 11096
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7483611873186206e-05,
+      "loss": 0.8102,
+      "step": 11097
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.748137657448262e-05,
+      "loss": 0.895,
+      "step": 11098
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.747914114285996e-05,
+      "loss": 0.837,
+      "step": 11099
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.747690557834202e-05,
+      "loss": 0.789,
+      "step": 11100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.747466988095261e-05,
+      "loss": 0.8363,
+      "step": 11101
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7472434050715546e-05,
+      "loss": 0.9513,
+      "step": 11102
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.747019808765463e-05,
+      "loss": 0.8575,
+      "step": 11103
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7467961991793664e-05,
+      "loss": 0.7847,
+      "step": 11104
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7465725763156476e-05,
+      "loss": 0.857,
+      "step": 11105
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7463489401766885e-05,
+      "loss": 0.8443,
+      "step": 11106
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7461252907648694e-05,
+      "loss": 0.8005,
+      "step": 11107
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.745901628082573e-05,
+      "loss": 0.7506,
+      "step": 11108
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.74567795213218e-05,
+      "loss": 0.824,
+      "step": 11109
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7454542629160725e-05,
+      "loss": 0.8963,
+      "step": 11110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7452305604366344e-05,
+      "loss": 0.815,
+      "step": 11111
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.745006844696247e-05,
+      "loss": 0.8854,
+      "step": 11112
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.744783115697292e-05,
+      "loss": 0.7791,
+      "step": 11113
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.744559373442152e-05,
+      "loss": 0.8028,
+      "step": 11114
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.744335617933211e-05,
+      "loss": 0.8483,
+      "step": 11115
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.744111849172851e-05,
+      "loss": 0.7885,
+      "step": 11116
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7438880671634545e-05,
+      "loss": 0.7092,
+      "step": 11117
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.743664271907407e-05,
+      "loss": 0.8348,
+      "step": 11118
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.743440463407088e-05,
+      "loss": 0.7548,
+      "step": 11119
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7432166416648836e-05,
+      "loss": 0.8377,
+      "step": 11120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7429928066831776e-05,
+      "loss": 0.7955,
+      "step": 11121
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7427689584643523e-05,
+      "loss": 0.8355,
+      "step": 11122
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7425450970107925e-05,
+      "loss": 0.8177,
+      "step": 11123
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7423212223248805e-05,
+      "loss": 0.8569,
+      "step": 11124
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7420973344090024e-05,
+      "loss": 0.8912,
+      "step": 11125
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.741873433265543e-05,
+      "loss": 0.8534,
+      "step": 11126
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.741649518896883e-05,
+      "loss": 0.757,
+      "step": 11127
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.741425591305411e-05,
+      "loss": 0.783,
+      "step": 11128
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.741201650493511e-05,
+      "loss": 0.765,
+      "step": 11129
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.740977696463565e-05,
+      "loss": 0.8573,
+      "step": 11130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.740753729217961e-05,
+      "loss": 0.783,
+      "step": 11131
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7405297487590834e-05,
+      "loss": 0.8501,
+      "step": 11132
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.740305755089317e-05,
+      "loss": 0.8422,
+      "step": 11133
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.740081748211047e-05,
+      "loss": 0.8403,
+      "step": 11134
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.739857728126659e-05,
+      "loss": 0.8631,
+      "step": 11135
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.739633694838539e-05,
+      "loss": 0.9153,
+      "step": 11136
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7394096483490734e-05,
+      "loss": 0.8569,
+      "step": 11137
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.739185588660646e-05,
+      "loss": 0.7728,
+      "step": 11138
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.738961515775646e-05,
+      "loss": 0.8622,
+      "step": 11139
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.738737429696457e-05,
+      "loss": 0.6431,
+      "step": 11140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7385133304254674e-05,
+      "loss": 0.7983,
+      "step": 11141
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.738289217965062e-05,
+      "loss": 0.7991,
+      "step": 11142
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.738065092317629e-05,
+      "loss": 0.8728,
+      "step": 11143
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.737840953485554e-05,
+      "loss": 0.7866,
+      "step": 11144
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7376168014712246e-05,
+      "loss": 0.8609,
+      "step": 11145
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7373926362770275e-05,
+      "loss": 0.8806,
+      "step": 11146
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.737168457905349e-05,
+      "loss": 0.8575,
+      "step": 11147
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.736944266358579e-05,
+      "loss": 0.8554,
+      "step": 11148
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.736720061639103e-05,
+      "loss": 0.8277,
+      "step": 11149
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7364958437493084e-05,
+      "loss": 0.806,
+      "step": 11150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.736271612691584e-05,
+      "loss": 0.7476,
+      "step": 11151
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.736047368468319e-05,
+      "loss": 0.7923,
+      "step": 11152
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.735823111081898e-05,
+      "loss": 0.7404,
+      "step": 11153
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.735598840534713e-05,
+      "loss": 0.7266,
+      "step": 11154
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.73537455682915e-05,
+      "loss": 0.769,
+      "step": 11155
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.735150259967597e-05,
+      "loss": 0.8849,
+      "step": 11156
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.734925949952444e-05,
+      "loss": 0.8743,
+      "step": 11157
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.73470162678608e-05,
+      "loss": 0.8832,
+      "step": 11158
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.734477290470892e-05,
+      "loss": 0.8395,
+      "step": 11159
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.734252941009271e-05,
+      "loss": 0.8907,
+      "step": 11160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.734028578403606e-05,
+      "loss": 0.8202,
+      "step": 11161
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7338042026562854e-05,
+      "loss": 0.8678,
+      "step": 11162
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7335798137697e-05,
+      "loss": 0.8643,
+      "step": 11163
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.733355411746237e-05,
+      "loss": 0.8433,
+      "step": 11164
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.733130996588288e-05,
+      "loss": 0.7651,
+      "step": 11165
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.732906568298243e-05,
+      "loss": 0.7952,
+      "step": 11166
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.732682126878492e-05,
+      "loss": 0.8001,
+      "step": 11167
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.732457672331424e-05,
+      "loss": 0.9704,
+      "step": 11168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.73223320465943e-05,
+      "loss": 0.7769,
+      "step": 11169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.732008723864901e-05,
+      "loss": 0.7797,
+      "step": 11170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7317842299502256e-05,
+      "loss": 0.7966,
+      "step": 11171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.731559722917798e-05,
+      "loss": 0.7891,
+      "step": 11172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.731335202770006e-05,
+      "loss": 0.773,
+      "step": 11173
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.731110669509243e-05,
+      "loss": 0.8074,
+      "step": 11174
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.730886123137897e-05,
+      "loss": 0.8089,
+      "step": 11175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7306615636583614e-05,
+      "loss": 0.7663,
+      "step": 11176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7304369910730276e-05,
+      "loss": 0.8711,
+      "step": 11177
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.730212405384287e-05,
+      "loss": 0.8029,
+      "step": 11178
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.72998780659453e-05,
+      "loss": 0.786,
+      "step": 11179
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.729763194706151e-05,
+      "loss": 0.8274,
+      "step": 11180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7295385697215405e-05,
+      "loss": 0.8144,
+      "step": 11181
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.72931393164309e-05,
+      "loss": 0.8104,
+      "step": 11182
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7290892804731936e-05,
+      "loss": 0.7598,
+      "step": 11183
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.728864616214242e-05,
+      "loss": 0.7981,
+      "step": 11184
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7286399388686275e-05,
+      "loss": 0.8528,
+      "step": 11185
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7284152484387446e-05,
+      "loss": 0.9508,
+      "step": 11186
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7281905449269846e-05,
+      "loss": 0.8683,
+      "step": 11187
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.727965828335741e-05,
+      "loss": 0.8783,
+      "step": 11188
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.727741098667407e-05,
+      "loss": 0.8674,
+      "step": 11189
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7275163559243756e-05,
+      "loss": 0.874,
+      "step": 11190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.72729160010904e-05,
+      "loss": 0.7502,
+      "step": 11191
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.727066831223794e-05,
+      "loss": 0.78,
+      "step": 11192
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.726842049271032e-05,
+      "loss": 0.793,
+      "step": 11193
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.726617254253146e-05,
+      "loss": 0.9282,
+      "step": 11194
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7263924461725316e-05,
+      "loss": 0.7832,
+      "step": 11195
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.726167625031582e-05,
+      "loss": 0.8225,
+      "step": 11196
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.725942790832691e-05,
+      "loss": 0.7873,
+      "step": 11197
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.725717943578255e-05,
+      "loss": 0.8301,
+      "step": 11198
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7254930832706656e-05,
+      "loss": 0.7794,
+      "step": 11199
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.72526820991232e-05,
+      "loss": 0.9028,
+      "step": 11200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.725043323505612e-05,
+      "loss": 0.7756,
+      "step": 11201
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.724818424052935e-05,
+      "loss": 0.8591,
+      "step": 11202
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.724593511556687e-05,
+      "loss": 0.8078,
+      "step": 11203
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.72436858601926e-05,
+      "loss": 0.7877,
+      "step": 11204
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.724143647443051e-05,
+      "loss": 0.7949,
+      "step": 11205
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7239186958304565e-05,
+      "loss": 0.8161,
+      "step": 11206
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.723693731183871e-05,
+      "loss": 0.8428,
+      "step": 11207
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7234687535056885e-05,
+      "loss": 0.7852,
+      "step": 11208
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.723243762798307e-05,
+      "loss": 0.8157,
+      "step": 11209
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.723018759064123e-05,
+      "loss": 0.7497,
+      "step": 11210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.722793742305531e-05,
+      "loss": 0.7974,
+      "step": 11211
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.722568712524929e-05,
+      "loss": 0.8122,
+      "step": 11212
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7223436697247116e-05,
+      "loss": 0.9115,
+      "step": 11213
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7221186139072764e-05,
+      "loss": 0.8786,
+      "step": 11214
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.72189354507502e-05,
+      "loss": 0.7493,
+      "step": 11215
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7216684632303395e-05,
+      "loss": 0.861,
+      "step": 11216
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.721443368375631e-05,
+      "loss": 0.7267,
+      "step": 11217
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.721218260513293e-05,
+      "loss": 0.8226,
+      "step": 11218
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.720993139645721e-05,
+      "loss": 0.7964,
+      "step": 11219
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.720768005775314e-05,
+      "loss": 0.9083,
+      "step": 11220
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.720542858904468e-05,
+      "loss": 0.7669,
+      "step": 11221
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7203176990355825e-05,
+      "loss": 0.8141,
+      "step": 11222
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.720092526171054e-05,
+      "loss": 0.8237,
+      "step": 11223
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.719867340313281e-05,
+      "loss": 0.8348,
+      "step": 11224
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.719642141464661e-05,
+      "loss": 0.8909,
+      "step": 11225
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7194169296275935e-05,
+      "loss": 0.878,
+      "step": 11226
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.719191704804476e-05,
+      "loss": 0.8117,
+      "step": 11227
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7189664669977056e-05,
+      "loss": 0.8311,
+      "step": 11228
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.718741216209683e-05,
+      "loss": 0.7966,
+      "step": 11229
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.718515952442807e-05,
+      "loss": 0.8108,
+      "step": 11230
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.718290675699475e-05,
+      "loss": 0.854,
+      "step": 11231
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.718065385982088e-05,
+      "loss": 0.8345,
+      "step": 11232
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7178400832930434e-05,
+      "loss": 0.8289,
+      "step": 11233
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.717614767634742e-05,
+      "loss": 0.7126,
+      "step": 11234
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.717389439009582e-05,
+      "loss": 0.8304,
+      "step": 11235
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.717164097419964e-05,
+      "loss": 0.8443,
+      "step": 11236
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.716938742868287e-05,
+      "loss": 0.8901,
+      "step": 11237
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7167133753569505e-05,
+      "loss": 0.7612,
+      "step": 11238
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.716487994888356e-05,
+      "loss": 0.7527,
+      "step": 11239
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.716262601464903e-05,
+      "loss": 0.84,
+      "step": 11240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.716037195088991e-05,
+      "loss": 0.818,
+      "step": 11241
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.715811775763022e-05,
+      "loss": 0.8097,
+      "step": 11242
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7155863434893956e-05,
+      "loss": 0.7963,
+      "step": 11243
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7153608982705126e-05,
+      "loss": 0.7703,
+      "step": 11244
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.715135440108773e-05,
+      "loss": 0.804,
+      "step": 11245
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.71490996900658e-05,
+      "loss": 0.8293,
+      "step": 11246
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7146844849663325e-05,
+      "loss": 0.7791,
+      "step": 11247
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.714458987990433e-05,
+      "loss": 0.8392,
+      "step": 11248
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7142334780812824e-05,
+      "loss": 0.8139,
+      "step": 11249
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7140079552412834e-05,
+      "loss": 0.8407,
+      "step": 11250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7137824194728356e-05,
+      "loss": 0.7922,
+      "step": 11251
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.713556870778342e-05,
+      "loss": 0.8546,
+      "step": 11252
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7133313091602054e-05,
+      "loss": 0.7939,
+      "step": 11253
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.713105734620826e-05,
+      "loss": 0.8788,
+      "step": 11254
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7128801471626073e-05,
+      "loss": 0.8682,
+      "step": 11255
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.712654546787952e-05,
+      "loss": 0.8612,
+      "step": 11256
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.712428933499261e-05,
+      "loss": 0.8085,
+      "step": 11257
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.712203307298939e-05,
+      "loss": 0.8171,
+      "step": 11258
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.711977668189387e-05,
+      "loss": 0.7893,
+      "step": 11259
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7117520161730085e-05,
+      "loss": 0.8526,
+      "step": 11260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.711526351252207e-05,
+      "loss": 0.7931,
+      "step": 11261
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7113006734293856e-05,
+      "loss": 0.7939,
+      "step": 11262
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.711074982706947e-05,
+      "loss": 0.8599,
+      "step": 11263
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.710849279087296e-05,
+      "loss": 0.9611,
+      "step": 11264
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7106235625728334e-05,
+      "loss": 0.8235,
+      "step": 11265
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.710397833165967e-05,
+      "loss": 0.8726,
+      "step": 11266
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7101720908690975e-05,
+      "loss": 0.805,
+      "step": 11267
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.70994633568463e-05,
+      "loss": 0.8365,
+      "step": 11268
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.709720567614968e-05,
+      "loss": 0.7725,
+      "step": 11269
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7094947866625177e-05,
+      "loss": 0.7689,
+      "step": 11270
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.709268992829681e-05,
+      "loss": 0.7983,
+      "step": 11271
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.709043186118865e-05,
+      "loss": 0.8158,
+      "step": 11272
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.708817366532472e-05,
+      "loss": 0.8912,
+      "step": 11273
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7085915340729084e-05,
+      "loss": 0.8228,
+      "step": 11274
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7083656887425785e-05,
+      "loss": 0.8712,
+      "step": 11275
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.708139830543888e-05,
+      "loss": 0.77,
+      "step": 11276
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.707913959479241e-05,
+      "loss": 0.8113,
+      "step": 11277
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7076880755510444e-05,
+      "loss": 0.8062,
+      "step": 11278
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.707462178761703e-05,
+      "loss": 0.9002,
+      "step": 11279
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.707236269113622e-05,
+      "loss": 0.7799,
+      "step": 11280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7070103466092074e-05,
+      "loss": 0.7557,
+      "step": 11281
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.706784411250867e-05,
+      "loss": 0.8326,
+      "step": 11282
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7065584630410033e-05,
+      "loss": 0.8286,
+      "step": 11283
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.706332501982026e-05,
+      "loss": 0.7929,
+      "step": 11284
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.706106528076339e-05,
+      "loss": 0.7917,
+      "step": 11285
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7058805413263504e-05,
+      "loss": 0.7412,
+      "step": 11286
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7056545417344656e-05,
+      "loss": 0.8019,
+      "step": 11287
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.705428529303092e-05,
+      "loss": 0.8098,
+      "step": 11288
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.705202504034636e-05,
+      "loss": 0.9994,
+      "step": 11289
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7049764659315064e-05,
+      "loss": 0.8014,
+      "step": 11290
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7047504149961076e-05,
+      "loss": 0.745,
+      "step": 11291
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7045243512308484e-05,
+      "loss": 0.7918,
+      "step": 11292
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.704298274638137e-05,
+      "loss": 0.8169,
+      "step": 11293
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7040721852203795e-05,
+      "loss": 0.817,
+      "step": 11294
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.703846082979985e-05,
+      "loss": 0.8164,
+      "step": 11295
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.70361996791936e-05,
+      "loss": 0.7643,
+      "step": 11296
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.703393840040913e-05,
+      "loss": 0.8638,
+      "step": 11297
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.703167699347052e-05,
+      "loss": 0.7417,
+      "step": 11298
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.702941545840186e-05,
+      "loss": 0.8568,
+      "step": 11299
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.702715379522722e-05,
+      "loss": 0.8127,
+      "step": 11300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7024892003970704e-05,
+      "loss": 0.7751,
+      "step": 11301
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7022630084656385e-05,
+      "loss": 0.7826,
+      "step": 11302
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7020368037308354e-05,
+      "loss": 0.7373,
+      "step": 11303
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.70181058619507e-05,
+      "loss": 0.7665,
+      "step": 11304
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.701584355860752e-05,
+      "loss": 0.7828,
+      "step": 11305
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.70135811273029e-05,
+      "loss": 0.8629,
+      "step": 11306
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.701131856806092e-05,
+      "loss": 0.8097,
+      "step": 11307
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7009055880905715e-05,
+      "loss": 0.9244,
+      "step": 11308
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.700679306586133e-05,
+      "loss": 0.8123,
+      "step": 11309
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.700453012295191e-05,
+      "loss": 0.7838,
+      "step": 11310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.700226705220152e-05,
+      "loss": 0.8308,
+      "step": 11311
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7000003853634274e-05,
+      "loss": 0.8864,
+      "step": 11312
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.699774052727428e-05,
+      "loss": 0.842,
+      "step": 11313
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6995477073145624e-05,
+      "loss": 0.8509,
+      "step": 11314
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.699321349127243e-05,
+      "loss": 0.8346,
+      "step": 11315
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6990949781678784e-05,
+      "loss": 0.8499,
+      "step": 11316
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.69886859443888e-05,
+      "loss": 0.772,
+      "step": 11317
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.69864219794266e-05,
+      "loss": 0.8298,
+      "step": 11318
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6984157886816275e-05,
+      "loss": 0.8359,
+      "step": 11319
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.698189366658195e-05,
+      "loss": 0.8162,
+      "step": 11320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.697962931874772e-05,
+      "loss": 0.8468,
+      "step": 11321
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6977364843337726e-05,
+      "loss": 0.7901,
+      "step": 11322
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.697510024037606e-05,
+      "loss": 0.884,
+      "step": 11323
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6972835509886846e-05,
+      "loss": 0.8346,
+      "step": 11324
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.697057065189421e-05,
+      "loss": 0.7633,
+      "step": 11325
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.696830566642226e-05,
+      "loss": 0.7786,
+      "step": 11326
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.696604055349512e-05,
+      "loss": 0.8106,
+      "step": 11327
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.696377531313691e-05,
+      "loss": 0.7924,
+      "step": 11328
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.696150994537176e-05,
+      "loss": 0.8027,
+      "step": 11329
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.695924445022379e-05,
+      "loss": 0.8219,
+      "step": 11330
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6956978827717125e-05,
+      "loss": 0.7332,
+      "step": 11331
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.695471307787589e-05,
+      "loss": 0.7413,
+      "step": 11332
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6952447200724234e-05,
+      "loss": 0.8434,
+      "step": 11333
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6950181196286264e-05,
+      "loss": 0.9085,
+      "step": 11334
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.694791506458613e-05,
+      "loss": 0.7736,
+      "step": 11335
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.694564880564794e-05,
+      "loss": 0.7924,
+      "step": 11336
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.694338241949584e-05,
+      "loss": 0.8418,
+      "step": 11337
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6941115906153975e-05,
+      "loss": 0.8071,
+      "step": 11338
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.693884926564647e-05,
+      "loss": 0.8485,
+      "step": 11339
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.693658249799748e-05,
+      "loss": 0.7499,
+      "step": 11340
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6934315603231126e-05,
+      "loss": 0.8403,
+      "step": 11341
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6932048581371556e-05,
+      "loss": 0.7775,
+      "step": 11342
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6929781432442915e-05,
+      "loss": 0.7701,
+      "step": 11343
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.692751415646934e-05,
+      "loss": 0.8792,
+      "step": 11344
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.692524675347498e-05,
+      "loss": 0.9079,
+      "step": 11345
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6922979223483986e-05,
+      "loss": 0.8394,
+      "step": 11346
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.692071156652051e-05,
+      "loss": 0.8219,
+      "step": 11347
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.691844378260868e-05,
+      "loss": 0.8086,
+      "step": 11348
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.691617587177266e-05,
+      "loss": 0.7721,
+      "step": 11349
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.691390783403661e-05,
+      "loss": 0.8447,
+      "step": 11350
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.691163966942467e-05,
+      "loss": 0.7091,
+      "step": 11351
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.690937137796099e-05,
+      "loss": 0.8869,
+      "step": 11352
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6907102959669745e-05,
+      "loss": 0.7574,
+      "step": 11353
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.690483441457509e-05,
+      "loss": 0.7883,
+      "step": 11354
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.690256574270117e-05,
+      "loss": 0.8747,
+      "step": 11355
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6900296944072144e-05,
+      "loss": 0.9392,
+      "step": 11356
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6898028018712186e-05,
+      "loss": 0.8018,
+      "step": 11357
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6895758966645457e-05,
+      "loss": 0.8118,
+      "step": 11358
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.689348978789611e-05,
+      "loss": 0.8346,
+      "step": 11359
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.689122048248832e-05,
+      "loss": 0.7769,
+      "step": 11360
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.688895105044625e-05,
+      "loss": 0.8011,
+      "step": 11361
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.688668149179407e-05,
+      "loss": 0.8993,
+      "step": 11362
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.688441180655595e-05,
+      "loss": 0.8178,
+      "step": 11363
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.688214199475606e-05,
+      "loss": 0.7834,
+      "step": 11364
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.687987205641858e-05,
+      "loss": 0.7797,
+      "step": 11365
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.687760199156766e-05,
+      "loss": 0.8108,
+      "step": 11366
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.687533180022749e-05,
+      "loss": 0.866,
+      "step": 11367
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.687306148242225e-05,
+      "loss": 0.7205,
+      "step": 11368
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6870791038176115e-05,
+      "loss": 0.7887,
+      "step": 11369
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6868520467513265e-05,
+      "loss": 0.9113,
+      "step": 11370
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.686624977045787e-05,
+      "loss": 0.7845,
+      "step": 11371
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.686397894703413e-05,
+      "loss": 0.8776,
+      "step": 11372
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6861707997266206e-05,
+      "loss": 0.7848,
+      "step": 11373
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.68594369211783e-05,
+      "loss": 0.802,
+      "step": 11374
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.685716571879459e-05,
+      "loss": 0.7878,
+      "step": 11375
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.685489439013926e-05,
+      "loss": 0.9667,
+      "step": 11376
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.685262293523651e-05,
+      "loss": 0.8325,
+      "step": 11377
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6850351354110514e-05,
+      "loss": 0.8089,
+      "step": 11378
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6848079646785475e-05,
+      "loss": 0.7853,
+      "step": 11379
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6845807813285574e-05,
+      "loss": 0.8888,
+      "step": 11380
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6843535853635016e-05,
+      "loss": 0.8897,
+      "step": 11381
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6841263767857994e-05,
+      "loss": 0.891,
+      "step": 11382
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6838991555978704e-05,
+      "loss": 0.7211,
+      "step": 11383
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.683671921802133e-05,
+      "loss": 0.8128,
+      "step": 11384
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.68344467540101e-05,
+      "loss": 0.8817,
+      "step": 11385
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.683217416396918e-05,
+      "loss": 0.7886,
+      "step": 11386
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.68299014479228e-05,
+      "loss": 0.8309,
+      "step": 11387
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6827628605895145e-05,
+      "loss": 0.8591,
+      "step": 11388
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6825355637910425e-05,
+      "loss": 0.8318,
+      "step": 11389
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.682308254399286e-05,
+      "loss": 0.8093,
+      "step": 11390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.682080932416663e-05,
+      "loss": 0.7587,
+      "step": 11391
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.681853597845596e-05,
+      "loss": 0.7695,
+      "step": 11392
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.681626250688506e-05,
+      "loss": 0.8277,
+      "step": 11393
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.681398890947813e-05,
+      "loss": 0.808,
+      "step": 11394
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.68117151862594e-05,
+      "loss": 0.8534,
+      "step": 11395
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.680944133725307e-05,
+      "loss": 0.8385,
+      "step": 11396
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6807167362483364e-05,
+      "loss": 0.7466,
+      "step": 11397
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.680489326197449e-05,
+      "loss": 0.7792,
+      "step": 11398
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.680261903575067e-05,
+      "loss": 0.8497,
+      "step": 11399
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.680034468383613e-05,
+      "loss": 0.7297,
+      "step": 11400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6798070206255064e-05,
+      "loss": 0.7429,
+      "step": 11401
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.679579560303173e-05,
+      "loss": 0.8222,
+      "step": 11402
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.679352087419033e-05,
+      "loss": 0.8004,
+      "step": 11403
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6791246019755085e-05,
+      "loss": 0.8528,
+      "step": 11404
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.678897103975025e-05,
+      "loss": 0.8079,
+      "step": 11405
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6786695934200014e-05,
+      "loss": 0.8929,
+      "step": 11406
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.678442070312862e-05,
+      "loss": 0.825,
+      "step": 11407
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.678214534656031e-05,
+      "loss": 0.9269,
+      "step": 11408
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6779869864519304e-05,
+      "loss": 0.8272,
+      "step": 11409
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6777594257029826e-05,
+      "loss": 0.8305,
+      "step": 11410
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6775318524116126e-05,
+      "loss": 0.9012,
+      "step": 11411
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.677304266580243e-05,
+      "loss": 0.8034,
+      "step": 11412
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.677076668211298e-05,
+      "loss": 0.8826,
+      "step": 11413
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6768490573072004e-05,
+      "loss": 0.808,
+      "step": 11414
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.676621433870376e-05,
+      "loss": 0.8371,
+      "step": 11415
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.676393797903247e-05,
+      "loss": 0.7682,
+      "step": 11416
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6761661494082386e-05,
+      "loss": 0.7839,
+      "step": 11417
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6759384883877744e-05,
+      "loss": 0.8451,
+      "step": 11418
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.675710814844279e-05,
+      "loss": 0.7629,
+      "step": 11419
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6754831287801786e-05,
+      "loss": 0.8379,
+      "step": 11420
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.675255430197895e-05,
+      "loss": 0.6757,
+      "step": 11421
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.675027719099855e-05,
+      "loss": 0.7644,
+      "step": 11422
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.674799995488483e-05,
+      "loss": 0.8077,
+      "step": 11423
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6745722593662045e-05,
+      "loss": 0.8188,
+      "step": 11424
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6743445107354454e-05,
+      "loss": 0.7488,
+      "step": 11425
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.67411674959863e-05,
+      "loss": 0.7323,
+      "step": 11426
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.673888975958182e-05,
+      "loss": 0.8748,
+      "step": 11427
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.673661189816531e-05,
+      "loss": 0.7538,
+      "step": 11428
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.673433391176101e-05,
+      "loss": 0.8278,
+      "step": 11429
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.673205580039317e-05,
+      "loss": 0.7387,
+      "step": 11430
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6729777564086056e-05,
+      "loss": 0.7833,
+      "step": 11431
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.672749920286394e-05,
+      "loss": 0.8069,
+      "step": 11432
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.672522071675107e-05,
+      "loss": 0.8531,
+      "step": 11433
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.672294210577172e-05,
+      "loss": 0.9027,
+      "step": 11434
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.672066336995016e-05,
+      "loss": 0.8363,
+      "step": 11435
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.671838450931065e-05,
+      "loss": 0.8427,
+      "step": 11436
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6716105523877456e-05,
+      "loss": 0.7754,
+      "step": 11437
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6713826413674844e-05,
+      "loss": 0.8674,
+      "step": 11438
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.671154717872711e-05,
+      "loss": 0.7568,
+      "step": 11439
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.67092678190585e-05,
+      "loss": 0.7971,
+      "step": 11440
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6706988334693285e-05,
+      "loss": 0.8153,
+      "step": 11441
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6704708725655765e-05,
+      "loss": 0.7695,
+      "step": 11442
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.67024289919702e-05,
+      "loss": 0.8801,
+      "step": 11443
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.670014913366087e-05,
+      "loss": 0.7329,
+      "step": 11444
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.669786915075206e-05,
+      "loss": 0.8053,
+      "step": 11445
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.669558904326804e-05,
+      "loss": 0.7694,
+      "step": 11446
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.66933088112331e-05,
+      "loss": 0.7655,
+      "step": 11447
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.669102845467152e-05,
+      "loss": 0.887,
+      "step": 11448
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.668874797360758e-05,
+      "loss": 0.7966,
+      "step": 11449
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6686467368065575e-05,
+      "loss": 0.7472,
+      "step": 11450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.668418663806979e-05,
+      "loss": 0.807,
+      "step": 11451
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.668190578364451e-05,
+      "loss": 0.7496,
+      "step": 11452
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.667962480481402e-05,
+      "loss": 0.7882,
+      "step": 11453
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.667734370160262e-05,
+      "loss": 0.8376,
+      "step": 11454
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6675062474034596e-05,
+      "loss": 0.8013,
+      "step": 11455
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.667278112213425e-05,
+      "loss": 0.8035,
+      "step": 11456
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.667049964592587e-05,
+      "loss": 0.7403,
+      "step": 11457
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.666821804543376e-05,
+      "loss": 0.8551,
+      "step": 11458
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6665936320682204e-05,
+      "loss": 0.7647,
+      "step": 11459
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6663654471695505e-05,
+      "loss": 0.7352,
+      "step": 11460
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.666137249849798e-05,
+      "loss": 0.7364,
+      "step": 11461
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6659090401113904e-05,
+      "loss": 0.7386,
+      "step": 11462
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.66568081795676e-05,
+      "loss": 0.9127,
+      "step": 11463
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.665452583388337e-05,
+      "loss": 0.8156,
+      "step": 11464
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.665224336408552e-05,
+      "loss": 0.8502,
+      "step": 11465
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.664996077019834e-05,
+      "loss": 0.7281,
+      "step": 11466
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.664767805224615e-05,
+      "loss": 0.9034,
+      "step": 11467
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.664539521025326e-05,
+      "loss": 0.7929,
+      "step": 11468
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.664311224424399e-05,
+      "loss": 0.8602,
+      "step": 11469
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6640829154242636e-05,
+      "loss": 0.7989,
+      "step": 11470
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.663854594027352e-05,
+      "loss": 0.9264,
+      "step": 11471
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.663626260236096e-05,
+      "loss": 0.7921,
+      "step": 11472
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.663397914052926e-05,
+      "loss": 0.8916,
+      "step": 11473
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6631695554802755e-05,
+      "loss": 0.8121,
+      "step": 11474
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.662941184520575e-05,
+      "loss": 0.7746,
+      "step": 11475
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.662712801176257e-05,
+      "loss": 0.8258,
+      "step": 11476
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.662484405449754e-05,
+      "loss": 0.8593,
+      "step": 11477
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6622559973434975e-05,
+      "loss": 0.851,
+      "step": 11478
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.662027576859919e-05,
+      "loss": 0.7905,
+      "step": 11479
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.661799144001454e-05,
+      "loss": 0.7655,
+      "step": 11480
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6615706987705325e-05,
+      "loss": 0.8335,
+      "step": 11481
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.661342241169588e-05,
+      "loss": 0.8215,
+      "step": 11482
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.661113771201054e-05,
+      "loss": 0.7902,
+      "step": 11483
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6608852888673646e-05,
+      "loss": 0.8395,
+      "step": 11484
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6606567941709505e-05,
+      "loss": 0.8056,
+      "step": 11485
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.660428287114246e-05,
+      "loss": 0.8458,
+      "step": 11486
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6601997676996854e-05,
+      "loss": 0.8374,
+      "step": 11487
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.659971235929701e-05,
+      "loss": 0.7875,
+      "step": 11488
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.659742691806728e-05,
+      "loss": 0.7262,
+      "step": 11489
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6595141353331984e-05,
+      "loss": 0.811,
+      "step": 11490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6592855665115475e-05,
+      "loss": 0.766,
+      "step": 11491
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.65905698534421e-05,
+      "loss": 0.8149,
+      "step": 11492
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.658828391833618e-05,
+      "loss": 0.7897,
+      "step": 11493
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.658599785982208e-05,
+      "loss": 0.8205,
+      "step": 11494
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.658371167792414e-05,
+      "loss": 0.7935,
+      "step": 11495
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.65814253726667e-05,
+      "loss": 0.8009,
+      "step": 11496
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.657913894407411e-05,
+      "loss": 0.8445,
+      "step": 11497
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.657685239217072e-05,
+      "loss": 0.8671,
+      "step": 11498
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6574565716980875e-05,
+      "loss": 0.7325,
+      "step": 11499
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6572278918528934e-05,
+      "loss": 0.7848,
+      "step": 11500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6569991996839255e-05,
+      "loss": 0.8598,
+      "step": 11501
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.656770495193618e-05,
+      "loss": 0.7186,
+      "step": 11502
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6565417783844066e-05,
+      "loss": 0.8233,
+      "step": 11503
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.656313049258729e-05,
+      "loss": 0.7776,
+      "step": 11504
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6560843078190175e-05,
+      "loss": 0.8523,
+      "step": 11505
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.655855554067712e-05,
+      "loss": 0.7984,
+      "step": 11506
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.655626788007245e-05,
+      "loss": 0.7986,
+      "step": 11507
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.655398009640055e-05,
+      "loss": 0.8394,
+      "step": 11508
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6551692189685774e-05,
+      "loss": 0.8558,
+      "step": 11509
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6549404159952486e-05,
+      "loss": 0.823,
+      "step": 11510
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.654711600722506e-05,
+      "loss": 0.7339,
+      "step": 11511
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6544827731527865e-05,
+      "loss": 0.8089,
+      "step": 11512
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6542539332885254e-05,
+      "loss": 0.7432,
+      "step": 11513
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.654025081132161e-05,
+      "loss": 0.821,
+      "step": 11514
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.653796216686131e-05,
+      "loss": 0.7996,
+      "step": 11515
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.65356733995287e-05,
+      "loss": 0.7853,
+      "step": 11516
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.653338450934819e-05,
+      "loss": 0.8141,
+      "step": 11517
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.653109549634412e-05,
+      "loss": 0.7615,
+      "step": 11518
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6528806360540893e-05,
+      "loss": 0.783,
+      "step": 11519
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6526517101962875e-05,
+      "loss": 0.8219,
+      "step": 11520
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6524227720634455e-05,
+      "loss": 0.8112,
+      "step": 11521
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.652193821658e-05,
+      "loss": 0.832,
+      "step": 11522
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.65196485898239e-05,
+      "loss": 0.7572,
+      "step": 11523
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.651735884039053e-05,
+      "loss": 0.8063,
+      "step": 11524
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6515068968304285e-05,
+      "loss": 0.8075,
+      "step": 11525
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.651277897358955e-05,
+      "loss": 0.8435,
+      "step": 11526
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6510488856270705e-05,
+      "loss": 0.8464,
+      "step": 11527
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.650819861637214e-05,
+      "loss": 0.828,
+      "step": 11528
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.650590825391825e-05,
+      "loss": 0.7413,
+      "step": 11529
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.650361776893343e-05,
+      "loss": 0.772,
+      "step": 11530
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.650132716144204e-05,
+      "loss": 0.8655,
+      "step": 11531
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.649903643146852e-05,
+      "loss": 0.7925,
+      "step": 11532
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.649674557903724e-05,
+      "loss": 0.8744,
+      "step": 11533
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.64944546041726e-05,
+      "loss": 0.8724,
+      "step": 11534
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.649216350689899e-05,
+      "loss": 0.9269,
+      "step": 11535
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.648987228724082e-05,
+      "loss": 0.8457,
+      "step": 11536
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.648758094522249e-05,
+      "loss": 0.9351,
+      "step": 11537
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.648528948086839e-05,
+      "loss": 0.8598,
+      "step": 11538
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.648299789420293e-05,
+      "loss": 0.8149,
+      "step": 11539
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.648070618525052e-05,
+      "loss": 0.9001,
+      "step": 11540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.647841435403556e-05,
+      "loss": 0.7965,
+      "step": 11541
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.647612240058245e-05,
+      "loss": 0.7937,
+      "step": 11542
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.64738303249156e-05,
+      "loss": 0.7369,
+      "step": 11543
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.647153812705943e-05,
+      "loss": 0.7966,
+      "step": 11544
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.646924580703834e-05,
+      "loss": 0.7522,
+      "step": 11545
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6466953364876755e-05,
+      "loss": 0.8171,
+      "step": 11546
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.646466080059907e-05,
+      "loss": 0.8171,
+      "step": 11547
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.646236811422971e-05,
+      "loss": 0.903,
+      "step": 11548
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.646007530579308e-05,
+      "loss": 0.8978,
+      "step": 11549
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.645778237531362e-05,
+      "loss": 0.7908,
+      "step": 11550
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6455489322815715e-05,
+      "loss": 0.8706,
+      "step": 11551
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.645319614832382e-05,
+      "loss": 0.8734,
+      "step": 11552
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6450902851862335e-05,
+      "loss": 0.8096,
+      "step": 11553
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.644860943345568e-05,
+      "loss": 0.7919,
+      "step": 11554
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6446315893128295e-05,
+      "loss": 0.7728,
+      "step": 11555
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6444022230904596e-05,
+      "loss": 0.822,
+      "step": 11556
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6441728446809e-05,
+      "loss": 0.7549,
+      "step": 11557
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6439434540865945e-05,
+      "loss": 0.7767,
+      "step": 11558
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.643714051309986e-05,
+      "loss": 0.7513,
+      "step": 11559
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.643484636353517e-05,
+      "loss": 0.8359,
+      "step": 11560
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.643255209219631e-05,
+      "loss": 0.7652,
+      "step": 11561
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.643025769910771e-05,
+      "loss": 0.7481,
+      "step": 11562
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.64279631842938e-05,
+      "loss": 0.7862,
+      "step": 11563
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.642566854777903e-05,
+      "loss": 0.7298,
+      "step": 11564
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6423373789587824e-05,
+      "loss": 0.7397,
+      "step": 11565
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.642107890974461e-05,
+      "loss": 0.8328,
+      "step": 11566
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6418783908273854e-05,
+      "loss": 0.7709,
+      "step": 11567
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6416488785199976e-05,
+      "loss": 0.7736,
+      "step": 11568
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6414193540547427e-05,
+      "loss": 0.7986,
+      "step": 11569
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.641189817434065e-05,
+      "loss": 0.8428,
+      "step": 11570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.640960268660407e-05,
+      "loss": 0.7946,
+      "step": 11571
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.640730707736216e-05,
+      "loss": 0.7734,
+      "step": 11572
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.640501134663936e-05,
+      "loss": 0.8381,
+      "step": 11573
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.64027154944601e-05,
+      "loss": 0.7804,
+      "step": 11574
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.640041952084885e-05,
+      "loss": 0.7789,
+      "step": 11575
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6398123425830055e-05,
+      "loss": 0.8337,
+      "step": 11576
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6395827209428166e-05,
+      "loss": 0.8588,
+      "step": 11577
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.639353087166764e-05,
+      "loss": 0.8634,
+      "step": 11578
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6391234412572914e-05,
+      "loss": 0.6988,
+      "step": 11579
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.638893783216846e-05,
+      "loss": 0.8023,
+      "step": 11580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6386641130478745e-05,
+      "loss": 0.7878,
+      "step": 11581
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6384344307528204e-05,
+      "loss": 0.8868,
+      "step": 11582
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.638204736334131e-05,
+      "loss": 0.7582,
+      "step": 11583
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.637975029794252e-05,
+      "loss": 0.8612,
+      "step": 11584
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6377453111356305e-05,
+      "loss": 0.8304,
+      "step": 11585
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.637515580360712e-05,
+      "loss": 0.8011,
+      "step": 11586
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6372858374719424e-05,
+      "loss": 0.7964,
+      "step": 11587
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.637056082471769e-05,
+      "loss": 0.8007,
+      "step": 11588
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.63682631536264e-05,
+      "loss": 0.7774,
+      "step": 11589
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.636596536147e-05,
+      "loss": 0.7158,
+      "step": 11590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6363667448272965e-05,
+      "loss": 0.7645,
+      "step": 11591
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.636136941405978e-05,
+      "loss": 0.7912,
+      "step": 11592
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6359071258854905e-05,
+      "loss": 0.7945,
+      "step": 11593
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.635677298268282e-05,
+      "loss": 0.7422,
+      "step": 11594
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.635447458556799e-05,
+      "loss": 0.7973,
+      "step": 11595
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.635217606753491e-05,
+      "loss": 0.682,
+      "step": 11596
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6349877428608036e-05,
+      "loss": 0.7768,
+      "step": 11597
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6347578668811866e-05,
+      "loss": 0.7737,
+      "step": 11598
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6345279788170874e-05,
+      "loss": 0.8134,
+      "step": 11599
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.634298078670953e-05,
+      "loss": 0.7477,
+      "step": 11600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6340681664452336e-05,
+      "loss": 0.9034,
+      "step": 11601
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.633838242142377e-05,
+      "loss": 0.8317,
+      "step": 11602
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6336083057648296e-05,
+      "loss": 0.8327,
+      "step": 11603
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6333783573150435e-05,
+      "loss": 0.8045,
+      "step": 11604
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.633148396795466e-05,
+      "loss": 0.8731,
+      "step": 11605
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.632918424208546e-05,
+      "loss": 0.8875,
+      "step": 11606
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.632688439556732e-05,
+      "loss": 0.7966,
+      "step": 11607
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.632458442842475e-05,
+      "loss": 0.8471,
+      "step": 11608
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6322284340682214e-05,
+      "loss": 0.8354,
+      "step": 11609
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.631998413236424e-05,
+      "loss": 0.817,
+      "step": 11610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.63176838034953e-05,
+      "loss": 0.7469,
+      "step": 11611
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.631538335409989e-05,
+      "loss": 0.7727,
+      "step": 11612
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.631308278420253e-05,
+      "loss": 0.8377,
+      "step": 11613
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.63107820938277e-05,
+      "loss": 0.7675,
+      "step": 11614
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.630848128299991e-05,
+      "loss": 0.8484,
+      "step": 11615
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6306180351743665e-05,
+      "loss": 0.7962,
+      "step": 11616
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.630387930008346e-05,
+      "loss": 0.7607,
+      "step": 11617
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.63015781280438e-05,
+      "loss": 0.7916,
+      "step": 11618
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.62992768356492e-05,
+      "loss": 0.762,
+      "step": 11619
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6296975422924154e-05,
+      "loss": 0.8739,
+      "step": 11620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6294673889893176e-05,
+      "loss": 0.8263,
+      "step": 11621
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.629237223658078e-05,
+      "loss": 0.7976,
+      "step": 11622
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.629007046301148e-05,
+      "loss": 0.8197,
+      "step": 11623
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.628776856920979e-05,
+      "loss": 0.7893,
+      "step": 11624
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.628546655520021e-05,
+      "loss": 0.8419,
+      "step": 11625
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.628316442100726e-05,
+      "loss": 0.8303,
+      "step": 11626
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.628086216665547e-05,
+      "loss": 0.8125,
+      "step": 11627
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.627855979216934e-05,
+      "loss": 0.7587,
+      "step": 11628
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6276257297573394e-05,
+      "loss": 0.869,
+      "step": 11629
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.627395468289215e-05,
+      "loss": 0.8857,
+      "step": 11630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.627165194815014e-05,
+      "loss": 0.7741,
+      "step": 11631
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.626934909337188e-05,
+      "loss": 0.7863,
+      "step": 11632
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.626704611858189e-05,
+      "loss": 0.8189,
+      "step": 11633
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6264743023804705e-05,
+      "loss": 0.7957,
+      "step": 11634
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.626243980906484e-05,
+      "loss": 0.7744,
+      "step": 11635
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.626013647438683e-05,
+      "loss": 0.79,
+      "step": 11636
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.625783301979521e-05,
+      "loss": 0.8588,
+      "step": 11637
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.62555294453145e-05,
+      "loss": 0.7958,
+      "step": 11638
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.625322575096923e-05,
+      "loss": 0.8488,
+      "step": 11639
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6250921936783935e-05,
+      "loss": 0.8098,
+      "step": 11640
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6248618002783154e-05,
+      "loss": 0.7537,
+      "step": 11641
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.624631394899142e-05,
+      "loss": 0.9054,
+      "step": 11642
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.624400977543327e-05,
+      "loss": 0.7683,
+      "step": 11643
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.624170548213325e-05,
+      "loss": 0.7979,
+      "step": 11644
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.623940106911588e-05,
+      "loss": 0.7677,
+      "step": 11645
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6237096536405726e-05,
+      "loss": 0.815,
+      "step": 11646
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.623479188402731e-05,
+      "loss": 0.8428,
+      "step": 11647
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6232487112005186e-05,
+      "loss": 0.8524,
+      "step": 11648
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.623018222036388e-05,
+      "loss": 0.7869,
+      "step": 11649
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6227877209127965e-05,
+      "loss": 0.7822,
+      "step": 11650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.622557207832197e-05,
+      "loss": 0.7688,
+      "step": 11651
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.622326682797044e-05,
+      "loss": 0.8128,
+      "step": 11652
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.622096145809794e-05,
+      "loss": 0.7634,
+      "step": 11653
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6218655968729006e-05,
+      "loss": 0.8016,
+      "step": 11654
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.621635035988821e-05,
+      "loss": 0.7602,
+      "step": 11655
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.621404463160008e-05,
+      "loss": 0.7964,
+      "step": 11656
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.621173878388919e-05,
+      "loss": 0.7835,
+      "step": 11657
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.620943281678009e-05,
+      "loss": 0.8031,
+      "step": 11658
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.620712673029733e-05,
+      "loss": 0.7785,
+      "step": 11659
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.620482052446547e-05,
+      "loss": 0.964,
+      "step": 11660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6202514199309085e-05,
+      "loss": 0.8215,
+      "step": 11661
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.620020775485272e-05,
+      "loss": 0.8689,
+      "step": 11662
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.619790119112094e-05,
+      "loss": 0.8444,
+      "step": 11663
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.619559450813832e-05,
+      "loss": 0.7957,
+      "step": 11664
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6193287705929404e-05,
+      "loss": 0.7029,
+      "step": 11665
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.619098078451877e-05,
+      "loss": 0.8645,
+      "step": 11666
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.618867374393099e-05,
+      "loss": 0.7243,
+      "step": 11667
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.618636658419063e-05,
+      "loss": 0.7398,
+      "step": 11668
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6184059305322254e-05,
+      "loss": 0.883,
+      "step": 11669
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6181751907350434e-05,
+      "loss": 0.8195,
+      "step": 11670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.617944439029975e-05,
+      "loss": 0.7744,
+      "step": 11671
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.617713675419477e-05,
+      "loss": 0.8686,
+      "step": 11672
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.617482899906007e-05,
+      "loss": 0.7489,
+      "step": 11673
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.617252112492022e-05,
+      "loss": 0.8429,
+      "step": 11674
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.61702131317998e-05,
+      "loss": 0.7672,
+      "step": 11675
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.61679050197234e-05,
+      "loss": 0.9054,
+      "step": 11676
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.616559678871559e-05,
+      "loss": 0.8165,
+      "step": 11677
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.616328843880095e-05,
+      "loss": 0.833,
+      "step": 11678
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.616097997000407e-05,
+      "loss": 0.7351,
+      "step": 11679
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6158671382349524e-05,
+      "loss": 0.8219,
+      "step": 11680
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.615636267586191e-05,
+      "loss": 0.8793,
+      "step": 11681
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.61540538505658e-05,
+      "loss": 0.8458,
+      "step": 11682
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.61517449064858e-05,
+      "loss": 0.8268,
+      "step": 11683
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.614943584364647e-05,
+      "loss": 0.8245,
+      "step": 11684
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.614712666207242e-05,
+      "loss": 0.7439,
+      "step": 11685
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.614481736178825e-05,
+      "loss": 0.7835,
+      "step": 11686
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.614250794281853e-05,
+      "loss": 0.794,
+      "step": 11687
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.614019840518788e-05,
+      "loss": 0.9076,
+      "step": 11688
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.613788874892086e-05,
+      "loss": 0.8519,
+      "step": 11689
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.61355789740421e-05,
+      "loss": 0.8579,
+      "step": 11690
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.613326908057618e-05,
+      "loss": 0.742,
+      "step": 11691
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6130959068547705e-05,
+      "loss": 0.8472,
+      "step": 11692
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6128648937981277e-05,
+      "loss": 0.8544,
+      "step": 11693
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.612633868890148e-05,
+      "loss": 0.7937,
+      "step": 11694
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6124028321332945e-05,
+      "loss": 0.7818,
+      "step": 11695
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6121717835300264e-05,
+      "loss": 0.8,
+      "step": 11696
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.611940723082804e-05,
+      "loss": 0.7626,
+      "step": 11697
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.611709650794087e-05,
+      "loss": 0.8117,
+      "step": 11698
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6114785666663374e-05,
+      "loss": 0.856,
+      "step": 11699
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.611247470702016e-05,
+      "loss": 0.8333,
+      "step": 11700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6110163629035836e-05,
+      "loss": 0.8027,
+      "step": 11701
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.610785243273502e-05,
+      "loss": 0.7847,
+      "step": 11702
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.610554111814232e-05,
+      "loss": 0.7759,
+      "step": 11703
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.610322968528234e-05,
+      "loss": 0.8325,
+      "step": 11704
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.61009181341797e-05,
+      "loss": 0.8116,
+      "step": 11705
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6098606464859044e-05,
+      "loss": 0.7674,
+      "step": 11706
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6096294677344945e-05,
+      "loss": 0.8011,
+      "step": 11707
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.609398277166206e-05,
+      "loss": 0.7301,
+      "step": 11708
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6091670747834984e-05,
+      "loss": 0.8567,
+      "step": 11709
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.608935860588836e-05,
+      "loss": 0.8114,
+      "step": 11710
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6087046345846786e-05,
+      "loss": 0.8373,
+      "step": 11711
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.60847339677349e-05,
+      "loss": 0.8062,
+      "step": 11712
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.608242147157733e-05,
+      "loss": 0.7931,
+      "step": 11713
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.608010885739871e-05,
+      "loss": 0.7171,
+      "step": 11714
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.607779612522365e-05,
+      "loss": 0.7812,
+      "step": 11715
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.607548327507678e-05,
+      "loss": 0.7595,
+      "step": 11716
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.607317030698275e-05,
+      "loss": 0.7898,
+      "step": 11717
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.607085722096617e-05,
+      "loss": 0.7444,
+      "step": 11718
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6068544017051684e-05,
+      "loss": 0.8249,
+      "step": 11719
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.606623069526393e-05,
+      "loss": 0.8006,
+      "step": 11720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.606391725562752e-05,
+      "loss": 0.7933,
+      "step": 11721
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6061603698167126e-05,
+      "loss": 0.8245,
+      "step": 11722
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6059290022907364e-05,
+      "loss": 0.7185,
+      "step": 11723
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.605697622987287e-05,
+      "loss": 0.7306,
+      "step": 11724
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.60546623190883e-05,
+      "loss": 0.8303,
+      "step": 11725
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.605234829057828e-05,
+      "loss": 0.8152,
+      "step": 11726
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6050034144367464e-05,
+      "loss": 0.7686,
+      "step": 11727
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.60477198804805e-05,
+      "loss": 0.8641,
+      "step": 11728
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.604540549894202e-05,
+      "loss": 0.7916,
+      "step": 11729
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.604309099977668e-05,
+      "loss": 0.7827,
+      "step": 11730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6040776383009125e-05,
+      "loss": 0.7616,
+      "step": 11731
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6038461648664e-05,
+      "loss": 0.8284,
+      "step": 11732
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.603614679676597e-05,
+      "loss": 0.8001,
+      "step": 11733
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.603383182733967e-05,
+      "loss": 0.784,
+      "step": 11734
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6031516740409754e-05,
+      "loss": 0.7401,
+      "step": 11735
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.602920153600089e-05,
+      "loss": 0.799,
+      "step": 11736
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.602688621413773e-05,
+      "loss": 0.8532,
+      "step": 11737
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6024570774844915e-05,
+      "loss": 0.7275,
+      "step": 11738
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.602225521814713e-05,
+      "loss": 0.7837,
+      "step": 11739
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6019939544069005e-05,
+      "loss": 0.823,
+      "step": 11740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.601762375263521e-05,
+      "loss": 0.7241,
+      "step": 11741
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6015307843870424e-05,
+      "loss": 0.8692,
+      "step": 11742
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.601299181779929e-05,
+      "loss": 0.8075,
+      "step": 11743
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.601067567444648e-05,
+      "loss": 0.9312,
+      "step": 11744
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.600835941383666e-05,
+      "loss": 0.8952,
+      "step": 11745
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6006043035994494e-05,
+      "loss": 0.7707,
+      "step": 11746
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6003726540944656e-05,
+      "loss": 0.7316,
+      "step": 11747
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.600140992871181e-05,
+      "loss": 0.6992,
+      "step": 11748
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.599909319932063e-05,
+      "loss": 0.8344,
+      "step": 11749
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.599677635279578e-05,
+      "loss": 0.8368,
+      "step": 11750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.599445938916194e-05,
+      "loss": 0.7643,
+      "step": 11751
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.599214230844378e-05,
+      "loss": 0.8657,
+      "step": 11752
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.598982511066597e-05,
+      "loss": 0.7877,
+      "step": 11753
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.598750779585321e-05,
+      "loss": 0.8958,
+      "step": 11754
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.598519036403015e-05,
+      "loss": 0.8012,
+      "step": 11755
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5982872815221494e-05,
+      "loss": 0.8131,
+      "step": 11756
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.59805551494519e-05,
+      "loss": 0.737,
+      "step": 11757
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.597823736674607e-05,
+      "loss": 0.7641,
+      "step": 11758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.597591946712867e-05,
+      "loss": 0.7842,
+      "step": 11759
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5973601450624394e-05,
+      "loss": 0.8737,
+      "step": 11760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.597128331725792e-05,
+      "loss": 0.9028,
+      "step": 11761
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5968965067053945e-05,
+      "loss": 0.8085,
+      "step": 11762
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.596664670003716e-05,
+      "loss": 0.7989,
+      "step": 11763
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5964328216232225e-05,
+      "loss": 0.7702,
+      "step": 11764
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.596200961566386e-05,
+      "loss": 0.7338,
+      "step": 11765
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5959690898356754e-05,
+      "loss": 0.8114,
+      "step": 11766
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.595737206433558e-05,
+      "loss": 0.8551,
+      "step": 11767
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.595505311362507e-05,
+      "loss": 0.8222,
+      "step": 11768
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5952734046249874e-05,
+      "loss": 0.8747,
+      "step": 11769
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5950414862234714e-05,
+      "loss": 0.7447,
+      "step": 11770
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5948095561604286e-05,
+      "loss": 0.7907,
+      "step": 11771
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.594577614438329e-05,
+      "loss": 0.8569,
+      "step": 11772
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.594345661059642e-05,
+      "loss": 0.8337,
+      "step": 11773
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5941136960268376e-05,
+      "loss": 0.8251,
+      "step": 11774
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.593881719342388e-05,
+      "loss": 0.8143,
+      "step": 11775
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.593649731008761e-05,
+      "loss": 0.8597,
+      "step": 11776
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5934177310284285e-05,
+      "loss": 0.8562,
+      "step": 11777
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5931857194038616e-05,
+      "loss": 0.8317,
+      "step": 11778
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.59295369613753e-05,
+      "loss": 0.7865,
+      "step": 11779
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5927216612319055e-05,
+      "loss": 0.84,
+      "step": 11780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5924896146894576e-05,
+      "loss": 0.8853,
+      "step": 11781
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.592257556512659e-05,
+      "loss": 0.7846,
+      "step": 11782
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5920254867039814e-05,
+      "loss": 0.7879,
+      "step": 11783
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5917934052658944e-05,
+      "loss": 0.721,
+      "step": 11784
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5915613122008714e-05,
+      "loss": 0.8274,
+      "step": 11785
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.591329207511382e-05,
+      "loss": 0.8405,
+      "step": 11786
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5910970911999e-05,
+      "loss": 0.7527,
+      "step": 11787
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.590864963268897e-05,
+      "loss": 0.8758,
+      "step": 11788
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5906328237208435e-05,
+      "loss": 0.8041,
+      "step": 11789
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.590400672558212e-05,
+      "loss": 0.7831,
+      "step": 11790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5901685097834766e-05,
+      "loss": 0.857,
+      "step": 11791
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.589936335399108e-05,
+      "loss": 0.8145,
+      "step": 11792
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5897041494075786e-05,
+      "loss": 0.7741,
+      "step": 11793
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.589471951811362e-05,
+      "loss": 0.733,
+      "step": 11794
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5892397426129307e-05,
+      "loss": 0.7806,
+      "step": 11795
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5890075218147564e-05,
+      "loss": 0.7449,
+      "step": 11796
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.588775289419314e-05,
+      "loss": 0.8556,
+      "step": 11797
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5885430454290755e-05,
+      "loss": 0.7753,
+      "step": 11798
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.588310789846515e-05,
+      "loss": 0.7641,
+      "step": 11799
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.588078522674105e-05,
+      "loss": 0.709,
+      "step": 11800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.587846243914319e-05,
+      "loss": 0.8198,
+      "step": 11801
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.587613953569632e-05,
+      "loss": 0.7867,
+      "step": 11802
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.587381651642516e-05,
+      "loss": 0.767,
+      "step": 11803
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.587149338135445e-05,
+      "loss": 0.8535,
+      "step": 11804
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5869170130508936e-05,
+      "loss": 0.7781,
+      "step": 11805
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.586684676391337e-05,
+      "loss": 0.7784,
+      "step": 11806
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5864523281592476e-05,
+      "loss": 0.7715,
+      "step": 11807
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5862199683571005e-05,
+      "loss": 0.8053,
+      "step": 11808
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5859875969873714e-05,
+      "loss": 0.7228,
+      "step": 11809
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5857552140525316e-05,
+      "loss": 0.7699,
+      "step": 11810
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.585522819555059e-05,
+      "loss": 0.946,
+      "step": 11811
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.585290413497428e-05,
+      "loss": 0.7292,
+      "step": 11812
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.585057995882112e-05,
+      "loss": 0.8243,
+      "step": 11813
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5848255667115875e-05,
+      "loss": 0.7882,
+      "step": 11814
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5845931259883295e-05,
+      "loss": 0.7183,
+      "step": 11815
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.584360673714813e-05,
+      "loss": 0.785,
+      "step": 11816
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.584128209893514e-05,
+      "loss": 0.7898,
+      "step": 11817
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.583895734526907e-05,
+      "loss": 0.7814,
+      "step": 11818
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.583663247617469e-05,
+      "loss": 0.7451,
+      "step": 11819
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.583430749167676e-05,
+      "loss": 0.7118,
+      "step": 11820
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5831982391800025e-05,
+      "loss": 0.8598,
+      "step": 11821
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5829657176569245e-05,
+      "loss": 0.7615,
+      "step": 11822
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.58273318460092e-05,
+      "loss": 0.7969,
+      "step": 11823
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5825006400144646e-05,
+      "loss": 0.8168,
+      "step": 11824
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.582268083900034e-05,
+      "loss": 0.8062,
+      "step": 11825
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.582035516260106e-05,
+      "loss": 0.8063,
+      "step": 11826
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5818029370971565e-05,
+      "loss": 0.8187,
+      "step": 11827
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.581570346413662e-05,
+      "loss": 0.7588,
+      "step": 11828
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.581337744212101e-05,
+      "loss": 0.8753,
+      "step": 11829
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5811051304949485e-05,
+      "loss": 0.8389,
+      "step": 11830
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.580872505264683e-05,
+      "loss": 0.7507,
+      "step": 11831
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.580639868523782e-05,
+      "loss": 0.8544,
+      "step": 11832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.580407220274722e-05,
+      "loss": 0.7616,
+      "step": 11833
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.580174560519981e-05,
+      "loss": 0.8925,
+      "step": 11834
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.579941889262037e-05,
+      "loss": 0.7581,
+      "step": 11835
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.579709206503367e-05,
+      "loss": 0.8467,
+      "step": 11836
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.57947651224645e-05,
+      "loss": 0.7968,
+      "step": 11837
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.579243806493764e-05,
+      "loss": 0.8374,
+      "step": 11838
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.579011089247786e-05,
+      "loss": 0.8136,
+      "step": 11839
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5787783605109946e-05,
+      "loss": 0.7994,
+      "step": 11840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.578545620285869e-05,
+      "loss": 0.7559,
+      "step": 11841
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.578312868574887e-05,
+      "loss": 0.7511,
+      "step": 11842
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5780801053805285e-05,
+      "loss": 0.836,
+      "step": 11843
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.577847330705271e-05,
+      "loss": 0.8757,
+      "step": 11844
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.577614544551593e-05,
+      "loss": 0.7943,
+      "step": 11845
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5773817469219755e-05,
+      "loss": 0.8461,
+      "step": 11846
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.577148937818896e-05,
+      "loss": 0.7183,
+      "step": 11847
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.576916117244834e-05,
+      "loss": 0.8103,
+      "step": 11848
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.57668328520227e-05,
+      "loss": 0.7752,
+      "step": 11849
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.576450441693682e-05,
+      "loss": 0.8162,
+      "step": 11850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5762175867215506e-05,
+      "loss": 0.811,
+      "step": 11851
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.575984720288355e-05,
+      "loss": 0.7583,
+      "step": 11852
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5757518423965754e-05,
+      "loss": 0.8245,
+      "step": 11853
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.575518953048692e-05,
+      "loss": 0.832,
+      "step": 11854
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.575286052247185e-05,
+      "loss": 0.759,
+      "step": 11855
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5750531399945336e-05,
+      "loss": 0.8464,
+      "step": 11856
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.574820216293219e-05,
+      "loss": 0.8092,
+      "step": 11857
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.574587281145722e-05,
+      "loss": 0.7786,
+      "step": 11858
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.574354334554523e-05,
+      "loss": 0.8326,
+      "step": 11859
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.574121376522103e-05,
+      "loss": 0.8235,
+      "step": 11860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5738884070509415e-05,
+      "loss": 0.8447,
+      "step": 11861
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.573655426143521e-05,
+      "loss": 0.7353,
+      "step": 11862
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5734224338023215e-05,
+      "loss": 0.7908,
+      "step": 11863
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.573189430029825e-05,
+      "loss": 0.8372,
+      "step": 11864
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5729564148285124e-05,
+      "loss": 0.8303,
+      "step": 11865
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.572723388200866e-05,
+      "loss": 0.7471,
+      "step": 11866
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5724903501493665e-05,
+      "loss": 0.7307,
+      "step": 11867
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5722573006764956e-05,
+      "loss": 0.7851,
+      "step": 11868
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.572024239784736e-05,
+      "loss": 0.8316,
+      "step": 11869
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.571791167476569e-05,
+      "loss": 0.8163,
+      "step": 11870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.571558083754477e-05,
+      "loss": 0.784,
+      "step": 11871
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.571324988620941e-05,
+      "loss": 0.8234,
+      "step": 11872
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.571091882078445e-05,
+      "loss": 0.7682,
+      "step": 11873
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5708587641294706e-05,
+      "loss": 0.8527,
+      "step": 11874
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5706256347765e-05,
+      "loss": 0.8011,
+      "step": 11875
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.570392494022017e-05,
+      "loss": 0.7192,
+      "step": 11876
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5701593418685036e-05,
+      "loss": 0.7795,
+      "step": 11877
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5699261783184436e-05,
+      "loss": 0.7449,
+      "step": 11878
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.569693003374318e-05,
+      "loss": 0.8569,
+      "step": 11879
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.569459817038613e-05,
+      "loss": 0.7241,
+      "step": 11880
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.569226619313809e-05,
+      "loss": 0.9166,
+      "step": 11881
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5689934102023904e-05,
+      "loss": 0.7739,
+      "step": 11882
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.568760189706842e-05,
+      "loss": 0.865,
+      "step": 11883
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.568526957829646e-05,
+      "loss": 0.8005,
+      "step": 11884
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5682937145732864e-05,
+      "loss": 0.7998,
+      "step": 11885
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5680604599402475e-05,
+      "loss": 0.708,
+      "step": 11886
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.567827193933013e-05,
+      "loss": 0.7767,
+      "step": 11887
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.567593916554067e-05,
+      "loss": 0.8061,
+      "step": 11888
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5673606278058955e-05,
+      "loss": 0.7489,
+      "step": 11889
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.56712732769098e-05,
+      "loss": 0.7689,
+      "step": 11890
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.566894016211806e-05,
+      "loss": 0.7526,
+      "step": 11891
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5666606933708594e-05,
+      "loss": 0.792,
+      "step": 11892
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.566427359170623e-05,
+      "loss": 0.8045,
+      "step": 11893
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.566194013613583e-05,
+      "loss": 0.7388,
+      "step": 11894
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.565960656702224e-05,
+      "loss": 0.7944,
+      "step": 11895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.565727288439031e-05,
+      "loss": 0.7382,
+      "step": 11896
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.56549390882649e-05,
+      "loss": 0.7938,
+      "step": 11897
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.565260517867086e-05,
+      "loss": 0.812,
+      "step": 11898
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.565027115563303e-05,
+      "loss": 0.795,
+      "step": 11899
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.564793701917629e-05,
+      "loss": 0.7071,
+      "step": 11900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.564560276932548e-05,
+      "loss": 0.8637,
+      "step": 11901
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5643268406105465e-05,
+      "loss": 0.9363,
+      "step": 11902
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5640933929541095e-05,
+      "loss": 0.8124,
+      "step": 11903
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5638599339657256e-05,
+      "loss": 0.7686,
+      "step": 11904
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5636264636478776e-05,
+      "loss": 0.757,
+      "step": 11905
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5633929820030544e-05,
+      "loss": 0.8138,
+      "step": 11906
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5631594890337415e-05,
+      "loss": 0.8054,
+      "step": 11907
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.562925984742426e-05,
+      "loss": 0.7142,
+      "step": 11908
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.562692469131593e-05,
+      "loss": 0.7399,
+      "step": 11909
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.562458942203731e-05,
+      "loss": 0.734,
+      "step": 11910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.562225403961326e-05,
+      "loss": 0.8512,
+      "step": 11911
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.561991854406865e-05,
+      "loss": 0.9115,
+      "step": 11912
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.561758293542837e-05,
+      "loss": 0.7987,
+      "step": 11913
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.561524721371726e-05,
+      "loss": 0.7986,
+      "step": 11914
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5612911378960226e-05,
+      "loss": 0.8283,
+      "step": 11915
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.561057543118212e-05,
+      "loss": 0.8165,
+      "step": 11916
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.560823937040783e-05,
+      "loss": 0.8341,
+      "step": 11917
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5605903196662225e-05,
+      "loss": 0.7975,
+      "step": 11918
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.56035669099702e-05,
+      "loss": 0.9032,
+      "step": 11919
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.560123051035662e-05,
+      "loss": 0.7905,
+      "step": 11920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5598893997846374e-05,
+      "loss": 0.7935,
+      "step": 11921
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.559655737246435e-05,
+      "loss": 0.8025,
+      "step": 11922
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.55942206342354e-05,
+      "loss": 0.8352,
+      "step": 11923
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.559188378318445e-05,
+      "loss": 0.8204,
+      "step": 11924
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.558954681933636e-05,
+      "loss": 0.7647,
+      "step": 11925
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5587209742716024e-05,
+      "loss": 0.7603,
+      "step": 11926
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5584872553348344e-05,
+      "loss": 0.7945,
+      "step": 11927
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.558253525125819e-05,
+      "loss": 0.7949,
+      "step": 11928
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.558019783647046e-05,
+      "loss": 0.7839,
+      "step": 11929
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5577860309010055e-05,
+      "loss": 0.7842,
+      "step": 11930
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.557552266890184e-05,
+      "loss": 0.8044,
+      "step": 11931
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.557318491617074e-05,
+      "loss": 0.842,
+      "step": 11932
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.557084705084165e-05,
+      "loss": 0.7783,
+      "step": 11933
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5568509072939445e-05,
+      "loss": 0.7813,
+      "step": 11934
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5566170982489036e-05,
+      "loss": 0.9454,
+      "step": 11935
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.556383277951532e-05,
+      "loss": 0.8377,
+      "step": 11936
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.55614944640432e-05,
+      "loss": 0.8567,
+      "step": 11937
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.555915603609758e-05,
+      "loss": 0.8328,
+      "step": 11938
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.555681749570336e-05,
+      "loss": 0.8949,
+      "step": 11939
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.555447884288543e-05,
+      "loss": 0.79,
+      "step": 11940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5552140077668724e-05,
+      "loss": 0.8588,
+      "step": 11941
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5549801200078126e-05,
+      "loss": 0.8313,
+      "step": 11942
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.554746221013854e-05,
+      "loss": 0.7533,
+      "step": 11943
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.55451231078749e-05,
+      "loss": 0.7622,
+      "step": 11944
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.554278389331209e-05,
+      "loss": 0.7618,
+      "step": 11945
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.554044456647504e-05,
+      "loss": 0.9041,
+      "step": 11946
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.553810512738865e-05,
+      "loss": 0.756,
+      "step": 11947
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.553576557607784e-05,
+      "loss": 0.7915,
+      "step": 11948
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5533425912567516e-05,
+      "loss": 0.7944,
+      "step": 11949
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5531086136882616e-05,
+      "loss": 0.8289,
+      "step": 11950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.552874624904803e-05,
+      "loss": 0.7857,
+      "step": 11951
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5526406249088696e-05,
+      "loss": 0.8315,
+      "step": 11952
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.552406613702953e-05,
+      "loss": 0.7804,
+      "step": 11953
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5521725912895435e-05,
+      "loss": 0.8169,
+      "step": 11954
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.551938557671136e-05,
+      "loss": 0.8405,
+      "step": 11955
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5517045128502205e-05,
+      "loss": 0.8705,
+      "step": 11956
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.551470456829291e-05,
+      "loss": 0.8439,
+      "step": 11957
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.551236389610839e-05,
+      "loss": 0.8027,
+      "step": 11958
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.551002311197358e-05,
+      "loss": 0.7916,
+      "step": 11959
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5507682215913404e-05,
+      "loss": 0.7762,
+      "step": 11960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5505341207952795e-05,
+      "loss": 0.8485,
+      "step": 11961
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.550300008811668e-05,
+      "loss": 0.8586,
+      "step": 11962
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.550065885642999e-05,
+      "loss": 0.7983,
+      "step": 11963
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5498317512917655e-05,
+      "loss": 0.823,
+      "step": 11964
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.549597605760461e-05,
+      "loss": 0.7982,
+      "step": 11965
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.549363449051579e-05,
+      "loss": 0.8954,
+      "step": 11966
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.549129281167615e-05,
+      "loss": 0.8164,
+      "step": 11967
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.548895102111059e-05,
+      "loss": 0.8189,
+      "step": 11968
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.548660911884408e-05,
+      "loss": 0.8706,
+      "step": 11969
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.548426710490156e-05,
+      "loss": 0.8418,
+      "step": 11970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5481924979307946e-05,
+      "loss": 0.7489,
+      "step": 11971
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5479582742088203e-05,
+      "loss": 0.7804,
+      "step": 11972
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.547724039326726e-05,
+      "loss": 0.8503,
+      "step": 11973
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5474897932870073e-05,
+      "loss": 0.805,
+      "step": 11974
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.547255536092157e-05,
+      "loss": 0.8135,
+      "step": 11975
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.547021267744672e-05,
+      "loss": 0.8721,
+      "step": 11976
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.546786988247047e-05,
+      "loss": 0.7782,
+      "step": 11977
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.546552697601774e-05,
+      "loss": 0.7531,
+      "step": 11978
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.546318395811352e-05,
+      "loss": 0.8486,
+      "step": 11979
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5460840828782735e-05,
+      "loss": 0.8702,
+      "step": 11980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.545849758805035e-05,
+      "loss": 0.8283,
+      "step": 11981
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.545615423594132e-05,
+      "loss": 0.8681,
+      "step": 11982
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.545381077248058e-05,
+      "loss": 0.8212,
+      "step": 11983
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.545146719769311e-05,
+      "loss": 0.7629,
+      "step": 11984
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5449123511603864e-05,
+      "loss": 0.7592,
+      "step": 11985
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.544677971423779e-05,
+      "loss": 0.8512,
+      "step": 11986
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.544443580561986e-05,
+      "loss": 0.8471,
+      "step": 11987
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.544209178577502e-05,
+      "loss": 0.7734,
+      "step": 11988
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.543974765472825e-05,
+      "loss": 0.8477,
+      "step": 11989
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.543740341250451e-05,
+      "loss": 0.7421,
+      "step": 11990
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.543505905912876e-05,
+      "loss": 0.8305,
+      "step": 11991
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.543271459462595e-05,
+      "loss": 0.7692,
+      "step": 11992
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.543037001902108e-05,
+      "loss": 0.8095,
+      "step": 11993
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.542802533233909e-05,
+      "loss": 0.8445,
+      "step": 11994
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.542568053460497e-05,
+      "loss": 0.8001,
+      "step": 11995
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.542333562584368e-05,
+      "loss": 0.7223,
+      "step": 11996
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.542099060608019e-05,
+      "loss": 0.7959,
+      "step": 11997
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.541864547533948e-05,
+      "loss": 0.6952,
+      "step": 11998
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.541630023364652e-05,
+      "loss": 0.75,
+      "step": 11999
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.541395488102629e-05,
+      "loss": 0.7457,
+      "step": 12000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.541160941750376e-05,
+      "loss": 0.7949,
+      "step": 12001
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.540926384310391e-05,
+      "loss": 0.8262,
+      "step": 12002
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.540691815785172e-05,
+      "loss": 0.8669,
+      "step": 12003
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.540457236177217e-05,
+      "loss": 0.766,
+      "step": 12004
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5402226454890244e-05,
+      "loss": 0.7607,
+      "step": 12005
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.539988043723092e-05,
+      "loss": 0.8129,
+      "step": 12006
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5397534308819174e-05,
+      "loss": 0.8623,
+      "step": 12007
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.539518806968001e-05,
+      "loss": 0.7798,
+      "step": 12008
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.53928417198384e-05,
+      "loss": 0.8293,
+      "step": 12009
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5390495259319346e-05,
+      "loss": 0.846,
+      "step": 12010
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5388148688147814e-05,
+      "loss": 0.8046,
+      "step": 12011
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5385802006348805e-05,
+      "loss": 0.8496,
+      "step": 12012
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.538345521394732e-05,
+      "loss": 0.7589,
+      "step": 12013
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5381108310968334e-05,
+      "loss": 0.8213,
+      "step": 12014
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.537876129743685e-05,
+      "loss": 0.8025,
+      "step": 12015
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.537641417337785e-05,
+      "loss": 0.8435,
+      "step": 12016
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5374066938816344e-05,
+      "loss": 0.8637,
+      "step": 12017
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5371719593777324e-05,
+      "loss": 0.7446,
+      "step": 12018
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5369372138285797e-05,
+      "loss": 0.7844,
+      "step": 12019
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.536702457236673e-05,
+      "loss": 0.8418,
+      "step": 12020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5364676896045165e-05,
+      "loss": 0.8389,
+      "step": 12021
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.536232910934608e-05,
+      "loss": 0.8177,
+      "step": 12022
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.535998121229448e-05,
+      "loss": 0.7194,
+      "step": 12023
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.535763320491536e-05,
+      "loss": 0.759,
+      "step": 12024
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5355285087233744e-05,
+      "loss": 0.7109,
+      "step": 12025
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.535293685927463e-05,
+      "loss": 0.7861,
+      "step": 12026
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5350588521063024e-05,
+      "loss": 0.6985,
+      "step": 12027
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.534824007262393e-05,
+      "loss": 0.8176,
+      "step": 12028
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.534589151398236e-05,
+      "loss": 0.783,
+      "step": 12029
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.534354284516333e-05,
+      "loss": 0.7638,
+      "step": 12030
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.534119406619186e-05,
+      "loss": 0.7457,
+      "step": 12031
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.533884517709293e-05,
+      "loss": 0.8067,
+      "step": 12032
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.533649617789159e-05,
+      "loss": 0.8384,
+      "step": 12033
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.533414706861284e-05,
+      "loss": 0.8246,
+      "step": 12034
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.533179784928169e-05,
+      "loss": 0.7092,
+      "step": 12035
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.532944851992317e-05,
+      "loss": 0.7122,
+      "step": 12036
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.532709908056229e-05,
+      "loss": 0.8319,
+      "step": 12037
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.532474953122408e-05,
+      "loss": 0.7803,
+      "step": 12038
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.532239987193355e-05,
+      "loss": 0.8982,
+      "step": 12039
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.532005010271574e-05,
+      "loss": 0.8503,
+      "step": 12040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.531770022359565e-05,
+      "loss": 0.8385,
+      "step": 12041
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5315350234598325e-05,
+      "loss": 0.8564,
+      "step": 12042
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.531300013574878e-05,
+      "loss": 0.813,
+      "step": 12043
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.531064992707204e-05,
+      "loss": 0.8354,
+      "step": 12044
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5308299608593145e-05,
+      "loss": 0.8183,
+      "step": 12045
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5305949180337114e-05,
+      "loss": 0.7437,
+      "step": 12046
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.530359864232898e-05,
+      "loss": 0.8592,
+      "step": 12047
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.530124799459378e-05,
+      "loss": 0.8058,
+      "step": 12048
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5298897237156534e-05,
+      "loss": 0.7729,
+      "step": 12049
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.52965463700423e-05,
+      "loss": 0.7561,
+      "step": 12050
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.529419539327609e-05,
+      "loss": 0.8507,
+      "step": 12051
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5291844306882944e-05,
+      "loss": 0.7863,
+      "step": 12052
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5289493110887904e-05,
+      "loss": 0.9054,
+      "step": 12053
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.528714180531602e-05,
+      "loss": 0.7943,
+      "step": 12054
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.528479039019231e-05,
+      "loss": 0.7989,
+      "step": 12055
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5282438865541834e-05,
+      "loss": 0.7584,
+      "step": 12056
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.528008723138963e-05,
+      "loss": 0.7528,
+      "step": 12057
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.527773548776073e-05,
+      "loss": 0.8923,
+      "step": 12058
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5275383634680186e-05,
+      "loss": 0.8473,
+      "step": 12059
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.527303167217305e-05,
+      "loss": 0.8204,
+      "step": 12060
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.527067960026436e-05,
+      "loss": 0.7673,
+      "step": 12061
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.526832741897917e-05,
+      "loss": 0.8348,
+      "step": 12062
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.526597512834253e-05,
+      "loss": 0.8208,
+      "step": 12063
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.526362272837948e-05,
+      "loss": 0.7926,
+      "step": 12064
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.526127021911508e-05,
+      "loss": 0.7476,
+      "step": 12065
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5258917600574395e-05,
+      "loss": 0.7268,
+      "step": 12066
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.525656487278245e-05,
+      "loss": 0.8069,
+      "step": 12067
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5254212035764324e-05,
+      "loss": 0.8713,
+      "step": 12068
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.525185908954506e-05,
+      "loss": 0.8727,
+      "step": 12069
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5249506034149724e-05,
+      "loss": 0.7767,
+      "step": 12070
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.524715286960337e-05,
+      "loss": 0.8513,
+      "step": 12071
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.524479959593106e-05,
+      "loss": 0.8802,
+      "step": 12072
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.524244621315786e-05,
+      "loss": 0.8469,
+      "step": 12073
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.524009272130882e-05,
+      "loss": 0.8116,
+      "step": 12074
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.523773912040901e-05,
+      "loss": 0.781,
+      "step": 12075
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.523538541048349e-05,
+      "loss": 0.7948,
+      "step": 12076
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.523303159155734e-05,
+      "loss": 0.9158,
+      "step": 12077
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.52306776636556e-05,
+      "loss": 0.8652,
+      "step": 12078
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5228323626803364e-05,
+      "loss": 0.7642,
+      "step": 12079
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5225969481025694e-05,
+      "loss": 0.7948,
+      "step": 12080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5223615226347655e-05,
+      "loss": 0.7941,
+      "step": 12081
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5221260862794325e-05,
+      "loss": 0.8302,
+      "step": 12082
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.521890639039077e-05,
+      "loss": 0.7741,
+      "step": 12083
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.521655180916206e-05,
+      "loss": 0.807,
+      "step": 12084
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.521419711913328e-05,
+      "loss": 0.8285,
+      "step": 12085
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.52118423203295e-05,
+      "loss": 0.7241,
+      "step": 12086
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.520948741277579e-05,
+      "loss": 0.7515,
+      "step": 12087
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.520713239649725e-05,
+      "loss": 0.7996,
+      "step": 12088
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.520477727151895e-05,
+      "loss": 0.7352,
+      "step": 12089
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.520242203786596e-05,
+      "loss": 0.7341,
+      "step": 12090
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5200066695563374e-05,
+      "loss": 0.7255,
+      "step": 12091
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.519771124463627e-05,
+      "loss": 0.9245,
+      "step": 12092
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.519535568510972e-05,
+      "loss": 0.7702,
+      "step": 12093
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5193000017008835e-05,
+      "loss": 0.8138,
+      "step": 12094
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5190644240358686e-05,
+      "loss": 0.7959,
+      "step": 12095
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.518828835518436e-05,
+      "loss": 0.895,
+      "step": 12096
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.518593236151095e-05,
+      "loss": 0.7632,
+      "step": 12097
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.518357625936354e-05,
+      "loss": 0.8342,
+      "step": 12098
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5181220048767233e-05,
+      "loss": 0.8001,
+      "step": 12099
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.517886372974711e-05,
+      "loss": 0.8437,
+      "step": 12100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.517650730232827e-05,
+      "loss": 0.8126,
+      "step": 12101
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5174150766535804e-05,
+      "loss": 0.8304,
+      "step": 12102
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.517179412239481e-05,
+      "loss": 0.8646,
+      "step": 12103
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.516943736993038e-05,
+      "loss": 0.7316,
+      "step": 12104
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.516708050916762e-05,
+      "loss": 0.7712,
+      "step": 12105
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.516472354013163e-05,
+      "loss": 0.797,
+      "step": 12106
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.516236646284749e-05,
+      "loss": 0.8881,
+      "step": 12107
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5160009277340325e-05,
+      "loss": 0.8558,
+      "step": 12108
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5157651983635235e-05,
+      "loss": 0.8047,
+      "step": 12109
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5155294581757314e-05,
+      "loss": 0.7701,
+      "step": 12110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.515293707173168e-05,
+      "loss": 0.7113,
+      "step": 12111
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5150579453583424e-05,
+      "loss": 0.8193,
+      "step": 12112
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.514822172733765e-05,
+      "loss": 0.8072,
+      "step": 12113
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.514586389301948e-05,
+      "loss": 0.7329,
+      "step": 12114
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5143505950654025e-05,
+      "loss": 0.7869,
+      "step": 12115
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.514114790026638e-05,
+      "loss": 0.7571,
+      "step": 12116
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.513878974188168e-05,
+      "loss": 0.7385,
+      "step": 12117
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5136431475525024e-05,
+      "loss": 0.8225,
+      "step": 12118
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.513407310122151e-05,
+      "loss": 0.7806,
+      "step": 12119
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.513171461899629e-05,
+      "loss": 0.8424,
+      "step": 12120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.512935602887445e-05,
+      "loss": 0.8556,
+      "step": 12121
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.512699733088112e-05,
+      "loss": 0.7512,
+      "step": 12122
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.512463852504141e-05,
+      "loss": 0.8404,
+      "step": 12123
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5122279611380457e-05,
+      "loss": 0.7751,
+      "step": 12124
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5119920589923364e-05,
+      "loss": 0.7711,
+      "step": 12125
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.511756146069527e-05,
+      "loss": 0.7315,
+      "step": 12126
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.511520222372128e-05,
+      "loss": 0.8686,
+      "step": 12127
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.511284287902653e-05,
+      "loss": 0.7289,
+      "step": 12128
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5110483426636146e-05,
+      "loss": 0.8385,
+      "step": 12129
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5108123866575244e-05,
+      "loss": 0.8418,
+      "step": 12130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5105764198868964e-05,
+      "loss": 0.745,
+      "step": 12131
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.510340442354244e-05,
+      "loss": 0.8914,
+      "step": 12132
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.510104454062077e-05,
+      "loss": 0.7855,
+      "step": 12133
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.509868455012913e-05,
+      "loss": 0.7613,
+      "step": 12134
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5096324452092625e-05,
+      "loss": 0.7783,
+      "step": 12135
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.509396424653639e-05,
+      "loss": 0.7836,
+      "step": 12136
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5091603933485564e-05,
+      "loss": 0.8059,
+      "step": 12137
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.508924351296529e-05,
+      "loss": 0.7478,
+      "step": 12138
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5086882985000684e-05,
+      "loss": 0.7685,
+      "step": 12139
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.508452234961691e-05,
+      "loss": 0.8991,
+      "step": 12140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.508216160683909e-05,
+      "loss": 0.7666,
+      "step": 12141
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.507980075669238e-05,
+      "loss": 0.7992,
+      "step": 12142
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.507743979920191e-05,
+      "loss": 0.7619,
+      "step": 12143
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.507507873439281e-05,
+      "loss": 0.8118,
+      "step": 12144
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5072717562290246e-05,
+      "loss": 0.7555,
+      "step": 12145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5070356282919354e-05,
+      "loss": 0.8241,
+      "step": 12146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.506799489630528e-05,
+      "loss": 0.8027,
+      "step": 12147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5065633402473175e-05,
+      "loss": 0.8518,
+      "step": 12148
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.506327180144819e-05,
+      "loss": 0.788,
+      "step": 12149
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5060910093255464e-05,
+      "loss": 0.9006,
+      "step": 12150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.505854827792016e-05,
+      "loss": 0.8573,
+      "step": 12151
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.505618635546742e-05,
+      "loss": 0.7711,
+      "step": 12152
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.505382432592239e-05,
+      "loss": 0.8374,
+      "step": 12153
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5051462189310255e-05,
+      "loss": 0.7467,
+      "step": 12154
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.504909994565614e-05,
+      "loss": 0.7441,
+      "step": 12155
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.50467375949852e-05,
+      "loss": 0.7913,
+      "step": 12156
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.504437513732262e-05,
+      "loss": 0.8577,
+      "step": 12157
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.504201257269354e-05,
+      "loss": 0.8028,
+      "step": 12158
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.503964990112312e-05,
+      "loss": 0.8075,
+      "step": 12159
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.503728712263652e-05,
+      "loss": 0.7708,
+      "step": 12160
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5034924237258914e-05,
+      "loss": 0.8136,
+      "step": 12161
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5032561245015455e-05,
+      "loss": 0.7829,
+      "step": 12162
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.503019814593131e-05,
+      "loss": 0.7736,
+      "step": 12163
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5027834940031646e-05,
+      "loss": 0.801,
+      "step": 12164
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5025471627341616e-05,
+      "loss": 0.6834,
+      "step": 12165
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.502310820788641e-05,
+      "loss": 0.7718,
+      "step": 12166
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.502074468169119e-05,
+      "loss": 0.7756,
+      "step": 12167
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.501838104878111e-05,
+      "loss": 0.8423,
+      "step": 12168
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.501601730918136e-05,
+      "loss": 0.8533,
+      "step": 12169
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.501365346291712e-05,
+      "loss": 0.7704,
+      "step": 12170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.501128951001353e-05,
+      "loss": 0.7411,
+      "step": 12171
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.50089254504958e-05,
+      "loss": 0.7491,
+      "step": 12172
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.500656128438908e-05,
+      "loss": 0.7764,
+      "step": 12173
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.500419701171856e-05,
+      "loss": 0.7253,
+      "step": 12174
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.500183263250942e-05,
+      "loss": 0.7935,
+      "step": 12175
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.499946814678683e-05,
+      "loss": 0.7699,
+      "step": 12176
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4997103554575976e-05,
+      "loss": 0.7907,
+      "step": 12177
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.499473885590204e-05,
+      "loss": 0.768,
+      "step": 12178
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.49923740507902e-05,
+      "loss": 0.7585,
+      "step": 12179
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.499000913926565e-05,
+      "loss": 0.8368,
+      "step": 12180
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.498764412135356e-05,
+      "loss": 0.7355,
+      "step": 12181
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.498527899707913e-05,
+      "loss": 0.8027,
+      "step": 12182
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.498291376646754e-05,
+      "loss": 0.7983,
+      "step": 12183
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.498054842954397e-05,
+      "loss": 0.8327,
+      "step": 12184
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.497818298633362e-05,
+      "loss": 0.8459,
+      "step": 12185
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.497581743686169e-05,
+      "loss": 0.776,
+      "step": 12186
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.497345178115336e-05,
+      "loss": 0.7094,
+      "step": 12187
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.497108601923381e-05,
+      "loss": 0.792,
+      "step": 12188
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.496872015112826e-05,
+      "loss": 0.7288,
+      "step": 12189
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4966354176861894e-05,
+      "loss": 0.7492,
+      "step": 12190
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4963988096459895e-05,
+      "loss": 0.7659,
+      "step": 12191
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.496162190994748e-05,
+      "loss": 0.7137,
+      "step": 12192
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.495925561734984e-05,
+      "loss": 0.7826,
+      "step": 12193
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.495688921869217e-05,
+      "loss": 0.7928,
+      "step": 12194
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.495452271399968e-05,
+      "loss": 0.8363,
+      "step": 12195
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.495215610329756e-05,
+      "loss": 0.8677,
+      "step": 12196
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.494978938661102e-05,
+      "loss": 0.774,
+      "step": 12197
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4947422563965274e-05,
+      "loss": 0.882,
+      "step": 12198
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.494505563538551e-05,
+      "loss": 0.7515,
+      "step": 12199
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.494268860089693e-05,
+      "loss": 0.7885,
+      "step": 12200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.494032146052477e-05,
+      "loss": 0.7952,
+      "step": 12201
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.493795421429422e-05,
+      "loss": 0.7993,
+      "step": 12202
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.493558686223049e-05,
+      "loss": 0.7467,
+      "step": 12203
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.493321940435879e-05,
+      "loss": 0.8063,
+      "step": 12204
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4930851840704324e-05,
+      "loss": 0.7691,
+      "step": 12205
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.492848417129233e-05,
+      "loss": 0.7544,
+      "step": 12206
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4926116396147993e-05,
+      "loss": 0.8834,
+      "step": 12207
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.492374851529655e-05,
+      "loss": 0.865,
+      "step": 12208
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.492138052876322e-05,
+      "loss": 0.7459,
+      "step": 12209
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.49190124365732e-05,
+      "loss": 0.7899,
+      "step": 12210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.491664423875171e-05,
+      "loss": 0.8508,
+      "step": 12211
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4914275935324e-05,
+      "loss": 0.7725,
+      "step": 12212
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.491190752631526e-05,
+      "loss": 0.8039,
+      "step": 12213
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.490953901175072e-05,
+      "loss": 0.7874,
+      "step": 12214
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.490717039165561e-05,
+      "loss": 0.7467,
+      "step": 12215
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.490480166605515e-05,
+      "loss": 0.8324,
+      "step": 12216
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.490243283497456e-05,
+      "loss": 0.8357,
+      "step": 12217
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.490006389843908e-05,
+      "loss": 0.8181,
+      "step": 12218
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.489769485647392e-05,
+      "loss": 0.7707,
+      "step": 12219
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4895325709104324e-05,
+      "loss": 0.7747,
+      "step": 12220
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.489295645635552e-05,
+      "loss": 0.7345,
+      "step": 12221
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.489058709825274e-05,
+      "loss": 0.8406,
+      "step": 12222
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4888217634821205e-05,
+      "loss": 0.7638,
+      "step": 12223
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4885848066086156e-05,
+      "loss": 0.8405,
+      "step": 12224
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.488347839207283e-05,
+      "loss": 0.8037,
+      "step": 12225
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.488110861280646e-05,
+      "loss": 0.7785,
+      "step": 12226
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.487873872831228e-05,
+      "loss": 0.8225,
+      "step": 12227
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4876368738615525e-05,
+      "loss": 0.7462,
+      "step": 12228
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.487399864374144e-05,
+      "loss": 0.7506,
+      "step": 12229
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.487162844371527e-05,
+      "loss": 0.7759,
+      "step": 12230
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4869258138562244e-05,
+      "loss": 0.8194,
+      "step": 12231
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.486688772830762e-05,
+      "loss": 0.832,
+      "step": 12232
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.486451721297663e-05,
+      "loss": 0.79,
+      "step": 12233
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4862146592594515e-05,
+      "loss": 0.7694,
+      "step": 12234
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.485977586718653e-05,
+      "loss": 0.8366,
+      "step": 12235
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.485740503677791e-05,
+      "loss": 0.8511,
+      "step": 12236
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.485503410139392e-05,
+      "loss": 0.7966,
+      "step": 12237
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.48526630610598e-05,
+      "loss": 0.6777,
+      "step": 12238
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.485029191580079e-05,
+      "loss": 0.8626,
+      "step": 12239
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.484792066564216e-05,
+      "loss": 0.832,
+      "step": 12240
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4845549310609155e-05,
+      "loss": 0.822,
+      "step": 12241
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.484317785072703e-05,
+      "loss": 0.7421,
+      "step": 12242
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.484080628602103e-05,
+      "loss": 0.7345,
+      "step": 12243
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.483843461651642e-05,
+      "loss": 0.8665,
+      "step": 12244
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.483606284223845e-05,
+      "loss": 0.8343,
+      "step": 12245
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4833690963212384e-05,
+      "loss": 0.7797,
+      "step": 12246
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.483131897946348e-05,
+      "loss": 0.7984,
+      "step": 12247
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4828946891017e-05,
+      "loss": 0.761,
+      "step": 12248
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.48265746978982e-05,
+      "loss": 0.8601,
+      "step": 12249
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.482420240013235e-05,
+      "loss": 0.8481,
+      "step": 12250
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4821829997744705e-05,
+      "loss": 0.8411,
+      "step": 12251
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.481945749076053e-05,
+      "loss": 0.7668,
+      "step": 12252
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.481708487920511e-05,
+      "loss": 0.8669,
+      "step": 12253
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.481471216310368e-05,
+      "loss": 0.7201,
+      "step": 12254
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.481233934248153e-05,
+      "loss": 0.8878,
+      "step": 12255
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.480996641736392e-05,
+      "loss": 0.8486,
+      "step": 12256
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4807593387776125e-05,
+      "loss": 0.83,
+      "step": 12257
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4805220253743415e-05,
+      "loss": 0.7217,
+      "step": 12258
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.480284701529106e-05,
+      "loss": 0.8286,
+      "step": 12259
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.480047367244434e-05,
+      "loss": 0.7083,
+      "step": 12260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.479810022522852e-05,
+      "loss": 0.8097,
+      "step": 12261
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4795726673668885e-05,
+      "loss": 0.7938,
+      "step": 12262
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.479335301779072e-05,
+      "loss": 0.7516,
+      "step": 12263
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.479097925761927e-05,
+      "loss": 0.777,
+      "step": 12264
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4788605393179854e-05,
+      "loss": 0.7572,
+      "step": 12265
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.478623142449772e-05,
+      "loss": 0.7929,
+      "step": 12266
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4783857351598165e-05,
+      "loss": 0.7035,
+      "step": 12267
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4781483174506466e-05,
+      "loss": 0.8592,
+      "step": 12268
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4779108893247914e-05,
+      "loss": 0.8337,
+      "step": 12269
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4776734507847794e-05,
+      "loss": 0.8556,
+      "step": 12270
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.477436001833138e-05,
+      "loss": 0.7805,
+      "step": 12271
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.477198542472396e-05,
+      "loss": 0.8952,
+      "step": 12272
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.476961072705085e-05,
+      "loss": 0.7498,
+      "step": 12273
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.476723592533729e-05,
+      "loss": 0.7649,
+      "step": 12274
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.476486101960862e-05,
+      "loss": 0.8009,
+      "step": 12275
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.47624860098901e-05,
+      "loss": 0.8759,
+      "step": 12276
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4760110896207025e-05,
+      "loss": 0.8981,
+      "step": 12277
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.47577356785847e-05,
+      "loss": 0.8125,
+      "step": 12278
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4755360357048416e-05,
+      "loss": 0.7462,
+      "step": 12279
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.475298493162346e-05,
+      "loss": 0.7415,
+      "step": 12280
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4750609402335144e-05,
+      "loss": 0.7817,
+      "step": 12281
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.474823376920876e-05,
+      "loss": 0.8774,
+      "step": 12282
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.47458580322696e-05,
+      "loss": 0.7493,
+      "step": 12283
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.474348219154296e-05,
+      "loss": 0.8595,
+      "step": 12284
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4741106247054156e-05,
+      "loss": 0.7289,
+      "step": 12285
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4738730198828484e-05,
+      "loss": 0.6945,
+      "step": 12286
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.473635404689125e-05,
+      "loss": 0.7969,
+      "step": 12287
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.473397779126776e-05,
+      "loss": 0.8693,
+      "step": 12288
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.473160143198331e-05,
+      "loss": 0.7405,
+      "step": 12289
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4729224969063214e-05,
+      "loss": 0.758,
+      "step": 12290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.472684840253278e-05,
+      "loss": 0.7593,
+      "step": 12291
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.472447173241731e-05,
+      "loss": 0.7541,
+      "step": 12292
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4722094958742134e-05,
+      "loss": 0.8119,
+      "step": 12293
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.471971808153254e-05,
+      "loss": 0.7788,
+      "step": 12294
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.471734110081385e-05,
+      "loss": 0.8653,
+      "step": 12295
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.471496401661138e-05,
+      "loss": 0.8145,
+      "step": 12296
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.471258682895043e-05,
+      "loss": 0.8694,
+      "step": 12297
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.471020953785634e-05,
+      "loss": 0.7979,
+      "step": 12298
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.470783214335441e-05,
+      "loss": 0.8562,
+      "step": 12299
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.470545464546996e-05,
+      "loss": 0.7602,
+      "step": 12300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4703077044228313e-05,
+      "loss": 0.8171,
+      "step": 12301
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.470069933965478e-05,
+      "loss": 0.7427,
+      "step": 12302
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4698321531774694e-05,
+      "loss": 0.8254,
+      "step": 12303
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4695943620613375e-05,
+      "loss": 0.7725,
+      "step": 12304
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.469356560619614e-05,
+      "loss": 0.8063,
+      "step": 12305
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.469118748854831e-05,
+      "loss": 0.6837,
+      "step": 12306
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.468880926769522e-05,
+      "loss": 0.8146,
+      "step": 12307
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.46864309436622e-05,
+      "loss": 0.7424,
+      "step": 12308
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4684052516474556e-05,
+      "loss": 0.7382,
+      "step": 12309
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4681673986157644e-05,
+      "loss": 0.7605,
+      "step": 12310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4679295352736786e-05,
+      "loss": 0.6752,
+      "step": 12311
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4676916616237296e-05,
+      "loss": 0.8193,
+      "step": 12312
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.467453777668453e-05,
+      "loss": 0.7407,
+      "step": 12313
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4672158834103804e-05,
+      "loss": 0.6892,
+      "step": 12314
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.466977978852047e-05,
+      "loss": 0.771,
+      "step": 12315
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.466740063995984e-05,
+      "loss": 0.8905,
+      "step": 12316
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.466502138844727e-05,
+      "loss": 0.6854,
+      "step": 12317
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.466264203400808e-05,
+      "loss": 0.7396,
+      "step": 12318
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.466026257666763e-05,
+      "loss": 0.7032,
+      "step": 12319
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4657883016451236e-05,
+      "loss": 0.8198,
+      "step": 12320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4655503353384255e-05,
+      "loss": 0.8796,
+      "step": 12321
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.465312358749203e-05,
+      "loss": 0.8096,
+      "step": 12322
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.46507437187999e-05,
+      "loss": 0.8913,
+      "step": 12323
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.464836374733321e-05,
+      "loss": 0.8092,
+      "step": 12324
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.46459836731173e-05,
+      "loss": 0.8204,
+      "step": 12325
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.464360349617751e-05,
+      "loss": 0.7901,
+      "step": 12326
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.464122321653921e-05,
+      "loss": 0.7929,
+      "step": 12327
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.463884283422773e-05,
+      "loss": 0.7913,
+      "step": 12328
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.463646234926843e-05,
+      "loss": 0.77,
+      "step": 12329
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.463408176168666e-05,
+      "loss": 0.7639,
+      "step": 12330
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.463170107150776e-05,
+      "loss": 0.7845,
+      "step": 12331
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.462932027875709e-05,
+      "loss": 0.7926,
+      "step": 12332
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.462693938346002e-05,
+      "loss": 0.7776,
+      "step": 12333
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.462455838564188e-05,
+      "loss": 0.7748,
+      "step": 12334
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4622177285328035e-05,
+      "loss": 0.807,
+      "step": 12335
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.461979608254384e-05,
+      "loss": 0.8114,
+      "step": 12336
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.461741477731467e-05,
+      "loss": 0.8381,
+      "step": 12337
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.461503336966586e-05,
+      "loss": 0.7448,
+      "step": 12338
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4612651859622785e-05,
+      "loss": 0.7772,
+      "step": 12339
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.461027024721081e-05,
+      "loss": 0.772,
+      "step": 12340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4607888532455277e-05,
+      "loss": 0.7403,
+      "step": 12341
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.460550671538157e-05,
+      "loss": 0.8317,
+      "step": 12342
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.460312479601506e-05,
+      "loss": 0.7167,
+      "step": 12343
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.460074277438109e-05,
+      "loss": 0.776,
+      "step": 12344
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.459836065050505e-05,
+      "loss": 0.799,
+      "step": 12345
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.459597842441228e-05,
+      "loss": 0.7424,
+      "step": 12346
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.459359609612817e-05,
+      "loss": 0.915,
+      "step": 12347
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.459121366567809e-05,
+      "loss": 0.8263,
+      "step": 12348
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4588831133087404e-05,
+      "loss": 0.9033,
+      "step": 12349
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.458644849838149e-05,
+      "loss": 0.7586,
+      "step": 12350
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.458406576158572e-05,
+      "loss": 0.8448,
+      "step": 12351
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4581682922725464e-05,
+      "loss": 0.8001,
+      "step": 12352
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4579299981826105e-05,
+      "loss": 0.7945,
+      "step": 12353
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.457691693891303e-05,
+      "loss": 0.8229,
+      "step": 12354
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.457453379401159e-05,
+      "loss": 0.7822,
+      "step": 12355
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.457215054714718e-05,
+      "loss": 0.7625,
+      "step": 12356
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.456976719834518e-05,
+      "loss": 0.7793,
+      "step": 12357
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.456738374763096e-05,
+      "loss": 0.7939,
+      "step": 12358
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.456500019502992e-05,
+      "loss": 0.8251,
+      "step": 12359
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.456261654056744e-05,
+      "loss": 0.7915,
+      "step": 12360
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4560232784268886e-05,
+      "loss": 0.7521,
+      "step": 12361
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.455784892615966e-05,
+      "loss": 0.8341,
+      "step": 12362
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4555464966265153e-05,
+      "loss": 0.7141,
+      "step": 12363
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.455308090461075e-05,
+      "loss": 0.8298,
+      "step": 12364
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.455069674122182e-05,
+      "loss": 0.7501,
+      "step": 12365
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4548312476123774e-05,
+      "loss": 0.7959,
+      "step": 12366
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4545928109342e-05,
+      "loss": 0.7927,
+      "step": 12367
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4543543640901886e-05,
+      "loss": 0.8502,
+      "step": 12368
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.454115907082882e-05,
+      "loss": 0.7389,
+      "step": 12369
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.453877439914821e-05,
+      "loss": 0.8384,
+      "step": 12370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4536389625885445e-05,
+      "loss": 0.7701,
+      "step": 12371
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.453400475106592e-05,
+      "loss": 0.7351,
+      "step": 12372
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.453161977471502e-05,
+      "loss": 0.7855,
+      "step": 12373
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4529234696858174e-05,
+      "loss": 0.781,
+      "step": 12374
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4526849517520756e-05,
+      "loss": 0.8487,
+      "step": 12375
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.452446423672818e-05,
+      "loss": 0.7142,
+      "step": 12376
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.452207885450584e-05,
+      "loss": 0.9391,
+      "step": 12377
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4519693370879124e-05,
+      "loss": 0.7572,
+      "step": 12378
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.451730778587347e-05,
+      "loss": 0.8229,
+      "step": 12379
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4514922099514266e-05,
+      "loss": 0.7334,
+      "step": 12380
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.451253631182691e-05,
+      "loss": 0.9032,
+      "step": 12381
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4510150422836825e-05,
+      "loss": 0.8075,
+      "step": 12382
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.450776443256941e-05,
+      "loss": 0.7447,
+      "step": 12383
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4505378341050074e-05,
+      "loss": 0.7447,
+      "step": 12384
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.450299214830424e-05,
+      "loss": 0.7506,
+      "step": 12385
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4500605854357295e-05,
+      "loss": 0.7337,
+      "step": 12386
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4498219459234674e-05,
+      "loss": 0.8704,
+      "step": 12387
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4495832962961776e-05,
+      "loss": 0.7699,
+      "step": 12388
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.449344636556403e-05,
+      "loss": 0.7677,
+      "step": 12389
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.449105966706683e-05,
+      "loss": 0.8106,
+      "step": 12390
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.448867286749562e-05,
+      "loss": 0.7722,
+      "step": 12391
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4486285966875805e-05,
+      "loss": 0.7522,
+      "step": 12392
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.448389896523279e-05,
+      "loss": 0.7739,
+      "step": 12393
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.448151186259202e-05,
+      "loss": 0.7262,
+      "step": 12394
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4479124658978906e-05,
+      "loss": 0.8676,
+      "step": 12395
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.447673735441886e-05,
+      "loss": 0.7803,
+      "step": 12396
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.447434994893733e-05,
+      "loss": 0.7932,
+      "step": 12397
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.447196244255971e-05,
+      "loss": 0.8522,
+      "step": 12398
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.446957483531145e-05,
+      "loss": 0.8452,
+      "step": 12399
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4467187127217964e-05,
+      "loss": 0.8267,
+      "step": 12400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.446479931830468e-05,
+      "loss": 0.7754,
+      "step": 12401
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4462411408597026e-05,
+      "loss": 0.8569,
+      "step": 12402
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.446002339812044e-05,
+      "loss": 0.79,
+      "step": 12403
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4457635286900355e-05,
+      "loss": 0.7679,
+      "step": 12404
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4455247074962196e-05,
+      "loss": 0.7628,
+      "step": 12405
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.445285876233139e-05,
+      "loss": 0.8383,
+      "step": 12406
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.445047034903337e-05,
+      "loss": 0.7568,
+      "step": 12407
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.444808183509359e-05,
+      "loss": 0.8493,
+      "step": 12408
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.444569322053747e-05,
+      "loss": 0.7961,
+      "step": 12409
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.444330450539045e-05,
+      "loss": 0.7663,
+      "step": 12410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4440915689677975e-05,
+      "loss": 0.7636,
+      "step": 12411
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4438526773425476e-05,
+      "loss": 0.7558,
+      "step": 12412
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4436137756658396e-05,
+      "loss": 0.7968,
+      "step": 12413
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.443374863940219e-05,
+      "loss": 0.908,
+      "step": 12414
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.443135942168227e-05,
+      "loss": 0.8381,
+      "step": 12415
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.442897010352411e-05,
+      "loss": 0.8847,
+      "step": 12416
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4426580684953144e-05,
+      "loss": 0.752,
+      "step": 12417
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.44241911659948e-05,
+      "loss": 0.7927,
+      "step": 12418
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.442180154667456e-05,
+      "loss": 0.8315,
+      "step": 12419
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4419411827017845e-05,
+      "loss": 0.7677,
+      "step": 12420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.441702200705011e-05,
+      "loss": 0.8194,
+      "step": 12421
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4414632086796816e-05,
+      "loss": 0.7601,
+      "step": 12422
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.441224206628341e-05,
+      "loss": 0.8095,
+      "step": 12423
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4409851945535334e-05,
+      "loss": 0.7938,
+      "step": 12424
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.440746172457804e-05,
+      "loss": 0.7322,
+      "step": 12425
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4405071403437e-05,
+      "loss": 0.7372,
+      "step": 12426
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.440268098213765e-05,
+      "loss": 0.7641,
+      "step": 12427
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4400290460705465e-05,
+      "loss": 0.8002,
+      "step": 12428
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.439789983916589e-05,
+      "loss": 0.7872,
+      "step": 12429
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4395509117544386e-05,
+      "loss": 0.764,
+      "step": 12430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4393118295866423e-05,
+      "loss": 0.7826,
+      "step": 12431
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.439072737415744e-05,
+      "loss": 0.7388,
+      "step": 12432
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.438833635244292e-05,
+      "loss": 0.7533,
+      "step": 12433
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4385945230748324e-05,
+      "loss": 0.8141,
+      "step": 12434
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4383554009099094e-05,
+      "loss": 0.7445,
+      "step": 12435
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.438116268752072e-05,
+      "loss": 0.7764,
+      "step": 12436
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4378771266038653e-05,
+      "loss": 0.7776,
+      "step": 12437
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4376379744678375e-05,
+      "loss": 0.8248,
+      "step": 12438
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.437398812346534e-05,
+      "loss": 0.7988,
+      "step": 12439
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.437159640242502e-05,
+      "loss": 0.8329,
+      "step": 12440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.436920458158289e-05,
+      "loss": 0.7406,
+      "step": 12441
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.436681266096442e-05,
+      "loss": 0.8311,
+      "step": 12442
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.436442064059508e-05,
+      "loss": 0.7477,
+      "step": 12443
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.436202852050035e-05,
+      "loss": 0.7978,
+      "step": 12444
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.43596363007057e-05,
+      "loss": 0.6786,
+      "step": 12445
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4357243981236605e-05,
+      "loss": 0.7642,
+      "step": 12446
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.435485156211853e-05,
+      "loss": 0.8535,
+      "step": 12447
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4352459043376985e-05,
+      "loss": 0.7351,
+      "step": 12448
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.435006642503742e-05,
+      "loss": 0.8312,
+      "step": 12449
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.434767370712532e-05,
+      "loss": 0.7931,
+      "step": 12450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.434528088966618e-05,
+      "loss": 0.789,
+      "step": 12451
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4342887972685464e-05,
+      "loss": 0.7696,
+      "step": 12452
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4340494956208655e-05,
+      "loss": 0.8382,
+      "step": 12453
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.433810184026126e-05,
+      "loss": 0.8209,
+      "step": 12454
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4335708624868754e-05,
+      "loss": 0.8134,
+      "step": 12455
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.433331531005661e-05,
+      "loss": 0.8625,
+      "step": 12456
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.433092189585033e-05,
+      "loss": 0.7896,
+      "step": 12457
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4328528382275386e-05,
+      "loss": 0.7848,
+      "step": 12458
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4326134769357285e-05,
+      "loss": 0.7912,
+      "step": 12459
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4323741057121514e-05,
+      "loss": 0.9914,
+      "step": 12460
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.432134724559355e-05,
+      "loss": 0.7725,
+      "step": 12461
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4318953334798906e-05,
+      "loss": 0.7848,
+      "step": 12462
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.431655932476307e-05,
+      "loss": 0.7359,
+      "step": 12463
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.431416521551152e-05,
+      "loss": 0.761,
+      "step": 12464
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4311771007069784e-05,
+      "loss": 0.867,
+      "step": 12465
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.430937669946332e-05,
+      "loss": 0.8227,
+      "step": 12466
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.430698229271766e-05,
+      "loss": 0.8126,
+      "step": 12467
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4304587786858285e-05,
+      "loss": 0.7433,
+      "step": 12468
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.43021931819107e-05,
+      "loss": 0.7837,
+      "step": 12469
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.42997984779004e-05,
+      "loss": 0.7477,
+      "step": 12470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.42974036748529e-05,
+      "loss": 0.8131,
+      "step": 12471
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.429500877279369e-05,
+      "loss": 0.8016,
+      "step": 12472
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.429261377174828e-05,
+      "loss": 0.8036,
+      "step": 12473
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.429021867174218e-05,
+      "loss": 0.8297,
+      "step": 12474
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.428782347280089e-05,
+      "loss": 0.8867,
+      "step": 12475
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.428542817494991e-05,
+      "loss": 0.8226,
+      "step": 12476
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.428303277821477e-05,
+      "loss": 0.8006,
+      "step": 12477
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.428063728262096e-05,
+      "loss": 0.7777,
+      "step": 12478
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4278241688193994e-05,
+      "loss": 0.7493,
+      "step": 12479
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.427584599495939e-05,
+      "loss": 0.7748,
+      "step": 12480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4273450202942656e-05,
+      "loss": 0.7954,
+      "step": 12481
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.427105431216931e-05,
+      "loss": 0.7499,
+      "step": 12482
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4268658322664866e-05,
+      "loss": 0.7596,
+      "step": 12483
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.426626223445484e-05,
+      "loss": 0.7096,
+      "step": 12484
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.426386604756474e-05,
+      "loss": 0.7459,
+      "step": 12485
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.426146976202009e-05,
+      "loss": 0.7952,
+      "step": 12486
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.425907337784642e-05,
+      "loss": 0.7526,
+      "step": 12487
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.425667689506923e-05,
+      "loss": 0.6759,
+      "step": 12488
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4254280313714055e-05,
+      "loss": 0.7624,
+      "step": 12489
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.425188363380641e-05,
+      "loss": 0.7475,
+      "step": 12490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4249486855371824e-05,
+      "loss": 0.8679,
+      "step": 12491
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.424708997843582e-05,
+      "loss": 0.7893,
+      "step": 12492
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.424469300302392e-05,
+      "loss": 0.8092,
+      "step": 12493
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.424229592916165e-05,
+      "loss": 0.7366,
+      "step": 12494
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.423989875687454e-05,
+      "loss": 0.8566,
+      "step": 12495
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4237501486188124e-05,
+      "loss": 0.7957,
+      "step": 12496
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.423510411712792e-05,
+      "loss": 0.8344,
+      "step": 12497
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.423270664971947e-05,
+      "loss": 0.8241,
+      "step": 12498
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4230309083988285e-05,
+      "loss": 0.7427,
+      "step": 12499
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4227911419959926e-05,
+      "loss": 0.7384,
+      "step": 12500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.422551365765991e-05,
+      "loss": 0.7727,
+      "step": 12501
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.422311579711377e-05,
+      "loss": 0.7648,
+      "step": 12502
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.422071783834705e-05,
+      "loss": 0.7312,
+      "step": 12503
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.421831978138528e-05,
+      "loss": 0.7769,
+      "step": 12504
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4215921626254e-05,
+      "loss": 0.7502,
+      "step": 12505
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4213523372978764e-05,
+      "loss": 0.7562,
+      "step": 12506
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4211125021585086e-05,
+      "loss": 0.7456,
+      "step": 12507
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4208726572098515e-05,
+      "loss": 0.8247,
+      "step": 12508
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.42063280245446e-05,
+      "loss": 0.7585,
+      "step": 12509
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4203929378948885e-05,
+      "loss": 0.7861,
+      "step": 12510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.42015306353369e-05,
+      "loss": 0.7646,
+      "step": 12511
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4199131793734206e-05,
+      "loss": 0.7577,
+      "step": 12512
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4196732854166345e-05,
+      "loss": 0.7128,
+      "step": 12513
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.419433381665886e-05,
+      "loss": 0.7051,
+      "step": 12514
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.419193468123731e-05,
+      "loss": 0.8061,
+      "step": 12515
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.418953544792722e-05,
+      "loss": 0.8021,
+      "step": 12516
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.418713611675417e-05,
+      "loss": 0.8347,
+      "step": 12517
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.41847366877437e-05,
+      "loss": 0.7774,
+      "step": 12518
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.418233716092135e-05,
+      "loss": 0.731,
+      "step": 12519
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.417993753631269e-05,
+      "loss": 0.8679,
+      "step": 12520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.417753781394327e-05,
+      "loss": 0.7579,
+      "step": 12521
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4175137993838636e-05,
+      "loss": 0.7584,
+      "step": 12522
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.417273807602436e-05,
+      "loss": 0.7925,
+      "step": 12523
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.417033806052599e-05,
+      "loss": 0.7102,
+      "step": 12524
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4167937947369085e-05,
+      "loss": 0.7102,
+      "step": 12525
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.416553773657922e-05,
+      "loss": 0.8058,
+      "step": 12526
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.416313742818193e-05,
+      "loss": 0.8092,
+      "step": 12527
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.416073702220279e-05,
+      "loss": 0.7414,
+      "step": 12528
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4158336518667364e-05,
+      "loss": 0.7906,
+      "step": 12529
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.415593591760121e-05,
+      "loss": 0.7709,
+      "step": 12530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4153535219029906e-05,
+      "loss": 0.7852,
+      "step": 12531
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4151134422979006e-05,
+      "loss": 0.8279,
+      "step": 12532
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.414873352947407e-05,
+      "loss": 0.7351,
+      "step": 12533
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4146332538540684e-05,
+      "loss": 0.7104,
+      "step": 12534
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.414393145020442e-05,
+      "loss": 0.7041,
+      "step": 12535
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.414153026449082e-05,
+      "loss": 0.7636,
+      "step": 12536
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.413912898142548e-05,
+      "loss": 0.762,
+      "step": 12537
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.413672760103397e-05,
+      "loss": 0.7146,
+      "step": 12538
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4134326123341855e-05,
+      "loss": 0.9163,
+      "step": 12539
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4131924548374703e-05,
+      "loss": 0.7447,
+      "step": 12540
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4129522876158113e-05,
+      "loss": 0.7847,
+      "step": 12541
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.412712110671764e-05,
+      "loss": 0.7773,
+      "step": 12542
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4124719240078865e-05,
+      "loss": 0.8075,
+      "step": 12543
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4122317276267375e-05,
+      "loss": 0.7951,
+      "step": 12544
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.411991521530875e-05,
+      "loss": 0.7594,
+      "step": 12545
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.411751305722856e-05,
+      "loss": 0.7301,
+      "step": 12546
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.411511080205238e-05,
+      "loss": 0.8,
+      "step": 12547
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.411270844980581e-05,
+      "loss": 0.7936,
+      "step": 12548
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4110306000514434e-05,
+      "loss": 0.7857,
+      "step": 12549
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.410790345420383e-05,
+      "loss": 0.7653,
+      "step": 12550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.410550081089958e-05,
+      "loss": 0.7329,
+      "step": 12551
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.410309807062727e-05,
+      "loss": 0.8053,
+      "step": 12552
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.41006952334125e-05,
+      "loss": 0.7718,
+      "step": 12553
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.409829229928084e-05,
+      "loss": 0.7848,
+      "step": 12554
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4095889268257906e-05,
+      "loss": 0.7517,
+      "step": 12555
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4093486140369266e-05,
+      "loss": 0.807,
+      "step": 12556
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.409108291564052e-05,
+      "loss": 0.7519,
+      "step": 12557
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.408867959409726e-05,
+      "loss": 0.7182,
+      "step": 12558
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.408627617576508e-05,
+      "loss": 0.7855,
+      "step": 12559
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4083872660669576e-05,
+      "loss": 0.7788,
+      "step": 12560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.408146904883634e-05,
+      "loss": 0.7531,
+      "step": 12561
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4079065340290975e-05,
+      "loss": 0.7753,
+      "step": 12562
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4076661535059076e-05,
+      "loss": 0.8017,
+      "step": 12563
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.407425763316624e-05,
+      "loss": 0.7429,
+      "step": 12564
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.407185363463807e-05,
+      "loss": 0.778,
+      "step": 12565
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.406944953950017e-05,
+      "loss": 0.864,
+      "step": 12566
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4067045347778136e-05,
+      "loss": 0.7525,
+      "step": 12567
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.406464105949757e-05,
+      "loss": 0.7107,
+      "step": 12568
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.406223667468409e-05,
+      "loss": 0.7292,
+      "step": 12569
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4059832193363286e-05,
+      "loss": 0.7325,
+      "step": 12570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.405742761556076e-05,
+      "loss": 0.7132,
+      "step": 12571
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4055022941302136e-05,
+      "loss": 0.8065,
+      "step": 12572
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.405261817061302e-05,
+      "loss": 0.6918,
+      "step": 12573
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4050213303519014e-05,
+      "loss": 0.753,
+      "step": 12574
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4047808340045725e-05,
+      "loss": 0.7956,
+      "step": 12575
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4045403280218786e-05,
+      "loss": 0.7321,
+      "step": 12576
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.404299812406378e-05,
+      "loss": 0.7724,
+      "step": 12577
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4040592871606334e-05,
+      "loss": 0.7847,
+      "step": 12578
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.403818752287207e-05,
+      "loss": 0.7563,
+      "step": 12579
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4035782077886585e-05,
+      "loss": 0.7266,
+      "step": 12580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4033376536675514e-05,
+      "loss": 0.8174,
+      "step": 12581
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4030970899264466e-05,
+      "loss": 0.8231,
+      "step": 12582
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.402856516567906e-05,
+      "loss": 0.8121,
+      "step": 12583
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.402615933594492e-05,
+      "loss": 0.7384,
+      "step": 12584
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.402375341008766e-05,
+      "loss": 0.742,
+      "step": 12585
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.4021347388132906e-05,
+      "loss": 0.6992,
+      "step": 12586
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.4018941270106284e-05,
+      "loss": 0.7872,
+      "step": 12587
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.40165350560334e-05,
+      "loss": 0.7593,
+      "step": 12588
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.40141287459399e-05,
+      "loss": 0.8264,
+      "step": 12589
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.40117223398514e-05,
+      "loss": 0.6761,
+      "step": 12590
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.4009315837793526e-05,
+      "loss": 0.7614,
+      "step": 12591
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.4006909239791916e-05,
+      "loss": 0.7727,
+      "step": 12592
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.400450254587219e-05,
+      "loss": 0.7412,
+      "step": 12593
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.4002095756059966e-05,
+      "loss": 0.7705,
+      "step": 12594
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3999688870380895e-05,
+      "loss": 0.7682,
+      "step": 12595
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.399728188886061e-05,
+      "loss": 0.8105,
+      "step": 12596
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.399487481152472e-05,
+      "loss": 0.7153,
+      "step": 12597
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3992467638398886e-05,
+      "loss": 0.7821,
+      "step": 12598
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3990060369508734e-05,
+      "loss": 0.7936,
+      "step": 12599
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3987653004879883e-05,
+      "loss": 0.8787,
+      "step": 12600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.398524554453799e-05,
+      "loss": 0.7405,
+      "step": 12601
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.398283798850868e-05,
+      "loss": 0.6944,
+      "step": 12602
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3980430336817616e-05,
+      "loss": 0.7451,
+      "step": 12603
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.397802258949041e-05,
+      "loss": 0.7146,
+      "step": 12604
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.397561474655272e-05,
+      "loss": 0.8303,
+      "step": 12605
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3973206808030175e-05,
+      "loss": 0.7935,
+      "step": 12606
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.397079877394843e-05,
+      "loss": 0.8093,
+      "step": 12607
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.396839064433313e-05,
+      "loss": 0.735,
+      "step": 12608
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.39659824192099e-05,
+      "loss": 0.8206,
+      "step": 12609
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.39635740986044e-05,
+      "loss": 0.7385,
+      "step": 12610
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.396116568254229e-05,
+      "loss": 0.7608,
+      "step": 12611
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.39587571710492e-05,
+      "loss": 0.7581,
+      "step": 12612
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3956348564150794e-05,
+      "loss": 0.7238,
+      "step": 12613
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.39539398618727e-05,
+      "loss": 0.7782,
+      "step": 12614
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.395153106424059e-05,
+      "loss": 0.7872,
+      "step": 12615
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3949122171280115e-05,
+      "loss": 0.7558,
+      "step": 12616
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.394671318301691e-05,
+      "loss": 0.7789,
+      "step": 12617
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3944304099476644e-05,
+      "loss": 0.7504,
+      "step": 12618
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3941894920684974e-05,
+      "loss": 0.793,
+      "step": 12619
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.393948564666755e-05,
+      "loss": 0.7904,
+      "step": 12620
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3937076277450024e-05,
+      "loss": 0.7736,
+      "step": 12621
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.393466681305807e-05,
+      "loss": 0.7514,
+      "step": 12622
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3932257253517337e-05,
+      "loss": 0.7494,
+      "step": 12623
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.392984759885349e-05,
+      "loss": 0.7642,
+      "step": 12624
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3927437849092184e-05,
+      "loss": 0.8216,
+      "step": 12625
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3925028004259086e-05,
+      "loss": 0.8263,
+      "step": 12626
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.392261806437986e-05,
+      "loss": 0.8193,
+      "step": 12627
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.392020802948016e-05,
+      "loss": 0.7682,
+      "step": 12628
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.391779789958567e-05,
+      "loss": 0.7729,
+      "step": 12629
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.391538767472204e-05,
+      "loss": 0.7678,
+      "step": 12630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3912977354914946e-05,
+      "loss": 0.7167,
+      "step": 12631
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.391056694019005e-05,
+      "loss": 0.7512,
+      "step": 12632
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.390815643057303e-05,
+      "loss": 0.7905,
+      "step": 12633
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3905745826089556e-05,
+      "loss": 0.8608,
+      "step": 12634
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.390333512676529e-05,
+      "loss": 0.7134,
+      "step": 12635
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.390092433262592e-05,
+      "loss": 0.8688,
+      "step": 12636
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.38985134436971e-05,
+      "loss": 0.7671,
+      "step": 12637
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.389610246000452e-05,
+      "loss": 0.8127,
+      "step": 12638
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.389369138157385e-05,
+      "loss": 0.6715,
+      "step": 12639
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3891280208430757e-05,
+      "loss": 0.802,
+      "step": 12640
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.388886894060093e-05,
+      "loss": 0.766,
+      "step": 12641
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.388645757811005e-05,
+      "loss": 0.8522,
+      "step": 12642
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.388404612098379e-05,
+      "loss": 0.7744,
+      "step": 12643
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.388163456924783e-05,
+      "loss": 0.7713,
+      "step": 12644
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3879222922927864e-05,
+      "loss": 0.7736,
+      "step": 12645
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.387681118204956e-05,
+      "loss": 0.825,
+      "step": 12646
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.387439934663861e-05,
+      "loss": 0.74,
+      "step": 12647
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.387198741672068e-05,
+      "loss": 0.7688,
+      "step": 12648
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.386957539232149e-05,
+      "loss": 0.7848,
+      "step": 12649
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.386716327346669e-05,
+      "loss": 0.7577,
+      "step": 12650
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.386475106018199e-05,
+      "loss": 0.7756,
+      "step": 12651
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.386233875249307e-05,
+      "loss": 0.7957,
+      "step": 12652
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.385992635042563e-05,
+      "loss": 0.7712,
+      "step": 12653
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.385751385400535e-05,
+      "loss": 0.8048,
+      "step": 12654
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.385510126325792e-05,
+      "loss": 0.7948,
+      "step": 12655
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.385268857820905e-05,
+      "loss": 0.8092,
+      "step": 12656
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.385027579888441e-05,
+      "loss": 0.7024,
+      "step": 12657
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3847862925309716e-05,
+      "loss": 0.8223,
+      "step": 12658
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.384544995751064e-05,
+      "loss": 0.7876,
+      "step": 12659
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3843036895512905e-05,
+      "loss": 0.8,
+      "step": 12660
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.384062373934219e-05,
+      "loss": 0.7559,
+      "step": 12661
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.38382104890242e-05,
+      "loss": 0.744,
+      "step": 12662
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.383579714458464e-05,
+      "loss": 0.7216,
+      "step": 12663
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3833383706049186e-05,
+      "loss": 0.7187,
+      "step": 12664
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.383097017344357e-05,
+      "loss": 0.8049,
+      "step": 12665
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.382855654679349e-05,
+      "loss": 0.7663,
+      "step": 12666
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3826142826124636e-05,
+      "loss": 0.83,
+      "step": 12667
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3823729011462726e-05,
+      "loss": 0.79,
+      "step": 12668
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.382131510283345e-05,
+      "loss": 0.7783,
+      "step": 12669
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.381890110026252e-05,
+      "loss": 0.753,
+      "step": 12670
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.381648700377566e-05,
+      "loss": 0.7716,
+      "step": 12671
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.381407281339856e-05,
+      "loss": 0.7263,
+      "step": 12672
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.381165852915693e-05,
+      "loss": 0.7486,
+      "step": 12673
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3809244151076493e-05,
+      "loss": 0.7729,
+      "step": 12674
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.380682967918295e-05,
+      "loss": 0.7595,
+      "step": 12675
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.380441511350203e-05,
+      "loss": 0.7277,
+      "step": 12676
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3802000454059424e-05,
+      "loss": 0.7128,
+      "step": 12677
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3799585700880854e-05,
+      "loss": 0.8149,
+      "step": 12678
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.379717085399204e-05,
+      "loss": 0.7902,
+      "step": 12679
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.379475591341871e-05,
+      "loss": 0.7556,
+      "step": 12680
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3792340879186556e-05,
+      "loss": 0.7777,
+      "step": 12681
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.378992575132131e-05,
+      "loss": 0.7537,
+      "step": 12682
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3787510529848696e-05,
+      "loss": 0.7312,
+      "step": 12683
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3785095214794426e-05,
+      "loss": 0.7179,
+      "step": 12684
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.378267980618423e-05,
+      "loss": 0.7933,
+      "step": 12685
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3780264304043826e-05,
+      "loss": 0.7827,
+      "step": 12686
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3777848708398934e-05,
+      "loss": 0.8261,
+      "step": 12687
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3775433019275284e-05,
+      "loss": 0.7267,
+      "step": 12688
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.37730172366986e-05,
+      "loss": 0.7821,
+      "step": 12689
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.377060136069461e-05,
+      "loss": 0.721,
+      "step": 12690
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.376818539128904e-05,
+      "loss": 0.856,
+      "step": 12691
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.376576932850761e-05,
+      "loss": 0.7612,
+      "step": 12692
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3763353172376064e-05,
+      "loss": 0.8691,
+      "step": 12693
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3760936922920136e-05,
+      "loss": 0.7785,
+      "step": 12694
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3758520580165534e-05,
+      "loss": 0.7486,
+      "step": 12695
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3756104144138016e-05,
+      "loss": 0.7748,
+      "step": 12696
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.37536876148633e-05,
+      "loss": 0.7459,
+      "step": 12697
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.375127099236713e-05,
+      "loss": 0.6888,
+      "step": 12698
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3748854276675234e-05,
+      "loss": 0.803,
+      "step": 12699
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3746437467813356e-05,
+      "loss": 0.7771,
+      "step": 12700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.374402056580722e-05,
+      "loss": 0.7614,
+      "step": 12701
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.374160357068258e-05,
+      "loss": 0.7832,
+      "step": 12702
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.373918648246517e-05,
+      "loss": 0.7781,
+      "step": 12703
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3736769301180725e-05,
+      "loss": 0.8431,
+      "step": 12704
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.373435202685499e-05,
+      "loss": 0.7762,
+      "step": 12705
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.373193465951372e-05,
+      "loss": 0.8092,
+      "step": 12706
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.372951719918264e-05,
+      "loss": 0.7097,
+      "step": 12707
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.37270996458875e-05,
+      "loss": 0.7798,
+      "step": 12708
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.372468199965404e-05,
+      "loss": 0.7603,
+      "step": 12709
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.372226426050802e-05,
+      "loss": 0.8846,
+      "step": 12710
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.371984642847519e-05,
+      "loss": 0.7318,
+      "step": 12711
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.371742850358128e-05,
+      "loss": 0.8055,
+      "step": 12712
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.371501048585205e-05,
+      "loss": 0.7616,
+      "step": 12713
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.371259237531325e-05,
+      "loss": 0.7364,
+      "step": 12714
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.371017417199063e-05,
+      "loss": 0.7509,
+      "step": 12715
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.370775587590995e-05,
+      "loss": 0.824,
+      "step": 12716
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3705337487096954e-05,
+      "loss": 0.77,
+      "step": 12717
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.370291900557739e-05,
+      "loss": 0.7577,
+      "step": 12718
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.370050043137703e-05,
+      "loss": 0.8096,
+      "step": 12719
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.369808176452162e-05,
+      "loss": 0.7102,
+      "step": 12720
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3695663005036914e-05,
+      "loss": 0.7146,
+      "step": 12721
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.369324415294869e-05,
+      "loss": 0.7476,
+      "step": 12722
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3690825208282686e-05,
+      "loss": 0.8151,
+      "step": 12723
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.368840617106467e-05,
+      "loss": 0.7358,
+      "step": 12724
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.36859870413204e-05,
+      "loss": 0.6681,
+      "step": 12725
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.368356781907564e-05,
+      "loss": 0.8223,
+      "step": 12726
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.368114850435616e-05,
+      "loss": 0.8053,
+      "step": 12727
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3678729097187726e-05,
+      "loss": 0.6916,
+      "step": 12728
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.367630959759609e-05,
+      "loss": 0.6664,
+      "step": 12729
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3673890005607015e-05,
+      "loss": 0.7201,
+      "step": 12730
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.367147032124629e-05,
+      "loss": 0.7564,
+      "step": 12731
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.366905054453966e-05,
+      "loss": 0.7035,
+      "step": 12732
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.366663067551291e-05,
+      "loss": 0.797,
+      "step": 12733
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.366421071419181e-05,
+      "loss": 0.7287,
+      "step": 12734
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.366179066060212e-05,
+      "loss": 0.7142,
+      "step": 12735
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.365937051476962e-05,
+      "loss": 0.6431,
+      "step": 12736
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.365695027672009e-05,
+      "loss": 0.7776,
+      "step": 12737
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.365452994647928e-05,
+      "loss": 0.7562,
+      "step": 12738
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3652109524072987e-05,
+      "loss": 0.7613,
+      "step": 12739
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.364968900952698e-05,
+      "loss": 0.7528,
+      "step": 12740
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3647268402867035e-05,
+      "loss": 0.7963,
+      "step": 12741
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.364484770411893e-05,
+      "loss": 0.7806,
+      "step": 12742
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.364242691330846e-05,
+      "loss": 0.772,
+      "step": 12743
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.364000603046137e-05,
+      "loss": 0.7887,
+      "step": 12744
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.363758505560347e-05,
+      "loss": 0.782,
+      "step": 12745
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.363516398876053e-05,
+      "loss": 0.7437,
+      "step": 12746
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3632742829958344e-05,
+      "loss": 0.8872,
+      "step": 12747
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3630321579222684e-05,
+      "loss": 0.8215,
+      "step": 12748
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.362790023657934e-05,
+      "loss": 0.7637,
+      "step": 12749
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3625478802054087e-05,
+      "loss": 0.7528,
+      "step": 12750
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.362305727567273e-05,
+      "loss": 0.6751,
+      "step": 12751
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3620635657461055e-05,
+      "loss": 0.7814,
+      "step": 12752
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3618213947444825e-05,
+      "loss": 0.7512,
+      "step": 12753
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.361579214564986e-05,
+      "loss": 0.7792,
+      "step": 12754
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3613370252101945e-05,
+      "loss": 0.6742,
+      "step": 12755
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3610948266826856e-05,
+      "loss": 0.8602,
+      "step": 12756
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.36085261898504e-05,
+      "loss": 0.7962,
+      "step": 12757
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.360610402119837e-05,
+      "loss": 0.7384,
+      "step": 12758
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3603681760896554e-05,
+      "loss": 0.8575,
+      "step": 12759
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.360125940897075e-05,
+      "loss": 0.6903,
+      "step": 12760
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.359883696544675e-05,
+      "loss": 0.7758,
+      "step": 12761
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.359641443035036e-05,
+      "loss": 0.7506,
+      "step": 12762
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.359399180370738e-05,
+      "loss": 0.7607,
+      "step": 12763
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.35915690855436e-05,
+      "loss": 0.9045,
+      "step": 12764
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.358914627588483e-05,
+      "loss": 0.7751,
+      "step": 12765
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.358672337475687e-05,
+      "loss": 0.7955,
+      "step": 12766
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.358430038218551e-05,
+      "loss": 0.7992,
+      "step": 12767
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.358187729819657e-05,
+      "loss": 0.7668,
+      "step": 12768
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.357945412281585e-05,
+      "loss": 0.8133,
+      "step": 12769
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.357703085606914e-05,
+      "loss": 0.7642,
+      "step": 12770
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.357460749798226e-05,
+      "loss": 0.7647,
+      "step": 12771
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3572184048581026e-05,
+      "loss": 0.7962,
+      "step": 12772
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3569760507891224e-05,
+      "loss": 0.717,
+      "step": 12773
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.356733687593868e-05,
+      "loss": 0.8707,
+      "step": 12774
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.35649131527492e-05,
+      "loss": 0.7489,
+      "step": 12775
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.35624893383486e-05,
+      "loss": 0.8108,
+      "step": 12776
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.356006543276268e-05,
+      "loss": 0.6841,
+      "step": 12777
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.355764143601726e-05,
+      "loss": 0.7796,
+      "step": 12778
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3555217348138165e-05,
+      "loss": 0.8395,
+      "step": 12779
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3552793169151195e-05,
+      "loss": 0.7281,
+      "step": 12780
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.355036889908216e-05,
+      "loss": 0.7372,
+      "step": 12781
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.354794453795689e-05,
+      "loss": 0.8279,
+      "step": 12782
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3545520085801206e-05,
+      "loss": 0.7546,
+      "step": 12783
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3543095542640924e-05,
+      "loss": 0.7426,
+      "step": 12784
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.354067090850185e-05,
+      "loss": 0.7406,
+      "step": 12785
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.353824618340983e-05,
+      "loss": 0.8398,
+      "step": 12786
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3535821367390664e-05,
+      "loss": 0.7325,
+      "step": 12787
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.353339646047018e-05,
+      "loss": 0.8021,
+      "step": 12788
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.353097146267421e-05,
+      "loss": 0.812,
+      "step": 12789
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.352854637402857e-05,
+      "loss": 0.7097,
+      "step": 12790
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.352612119455909e-05,
+      "loss": 0.8149,
+      "step": 12791
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.35236959242916e-05,
+      "loss": 0.7974,
+      "step": 12792
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.352127056325191e-05,
+      "loss": 0.747,
+      "step": 12793
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.351884511146587e-05,
+      "loss": 0.7568,
+      "step": 12794
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3516419568959303e-05,
+      "loss": 0.7344,
+      "step": 12795
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3513993935758035e-05,
+      "loss": 0.7498,
+      "step": 12796
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.35115682118879e-05,
+      "loss": 0.7408,
+      "step": 12797
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.350914239737474e-05,
+      "loss": 0.7623,
+      "step": 12798
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3506716492244374e-05,
+      "loss": 0.7493,
+      "step": 12799
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.350429049652264e-05,
+      "loss": 0.7696,
+      "step": 12800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3501864410235377e-05,
+      "loss": 0.8194,
+      "step": 12801
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3499438233408416e-05,
+      "loss": 0.751,
+      "step": 12802
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3497011966067603e-05,
+      "loss": 0.7515,
+      "step": 12803
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.349458560823877e-05,
+      "loss": 0.7673,
+      "step": 12804
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.349215915994776e-05,
+      "loss": 0.7386,
+      "step": 12805
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.348973262122041e-05,
+      "loss": 0.7292,
+      "step": 12806
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.348730599208255e-05,
+      "loss": 0.7404,
+      "step": 12807
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.348487927256004e-05,
+      "loss": 0.7595,
+      "step": 12808
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.348245246267873e-05,
+      "loss": 0.7249,
+      "step": 12809
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3480025562464436e-05,
+      "loss": 0.6783,
+      "step": 12810
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.347759857194303e-05,
+      "loss": 0.8721,
+      "step": 12811
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.347517149114033e-05,
+      "loss": 0.7463,
+      "step": 12812
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.347274432008221e-05,
+      "loss": 0.6864,
+      "step": 12813
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3470317058794505e-05,
+      "loss": 0.8014,
+      "step": 12814
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.346788970730307e-05,
+      "loss": 0.7192,
+      "step": 12815
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.346546226563374e-05,
+      "loss": 0.6523,
+      "step": 12816
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.346303473381238e-05,
+      "loss": 0.7683,
+      "step": 12817
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.346060711186485e-05,
+      "loss": 0.7236,
+      "step": 12818
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.345817939981698e-05,
+      "loss": 0.6969,
+      "step": 12819
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3455751597694635e-05,
+      "loss": 0.7077,
+      "step": 12820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3453323705523676e-05,
+      "loss": 0.7615,
+      "step": 12821
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.345089572332994e-05,
+      "loss": 0.7423,
+      "step": 12822
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3448467651139305e-05,
+      "loss": 0.7966,
+      "step": 12823
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.344603948897762e-05,
+      "loss": 0.7244,
+      "step": 12824
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.344361123687073e-05,
+      "loss": 0.7611,
+      "step": 12825
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.344118289484451e-05,
+      "loss": 0.6873,
+      "step": 12826
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.343875446292483e-05,
+      "loss": 0.7389,
+      "step": 12827
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.343632594113753e-05,
+      "loss": 0.7465,
+      "step": 12828
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.343389732950847e-05,
+      "loss": 0.8745,
+      "step": 12829
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.343146862806353e-05,
+      "loss": 0.6737,
+      "step": 12830
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.342903983682858e-05,
+      "loss": 0.7461,
+      "step": 12831
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.342661095582946e-05,
+      "loss": 0.837,
+      "step": 12832
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.342418198509206e-05,
+      "loss": 0.7117,
+      "step": 12833
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.342175292464223e-05,
+      "loss": 0.7144,
+      "step": 12834
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.341932377450584e-05,
+      "loss": 0.8679,
+      "step": 12835
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3416894534708764e-05,
+      "loss": 0.8123,
+      "step": 12836
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.341446520527687e-05,
+      "loss": 0.7468,
+      "step": 12837
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3412035786236047e-05,
+      "loss": 0.7358,
+      "step": 12838
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.340960627761213e-05,
+      "loss": 0.7227,
+      "step": 12839
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.340717667943102e-05,
+      "loss": 0.7906,
+      "step": 12840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.340474699171859e-05,
+      "loss": 0.7736,
+      "step": 12841
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3402317214500696e-05,
+      "loss": 0.7085,
+      "step": 12842
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3399887347803224e-05,
+      "loss": 0.7713,
+      "step": 12843
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.339745739165206e-05,
+      "loss": 0.8573,
+      "step": 12844
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3395027346073066e-05,
+      "loss": 0.7383,
+      "step": 12845
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.339259721109213e-05,
+      "loss": 0.7552,
+      "step": 12846
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3390166986735126e-05,
+      "loss": 0.7749,
+      "step": 12847
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.338773667302795e-05,
+      "loss": 0.7243,
+      "step": 12848
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.338530626999646e-05,
+      "loss": 0.7528,
+      "step": 12849
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.338287577766655e-05,
+      "loss": 0.724,
+      "step": 12850
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.33804451960641e-05,
+      "loss": 0.7261,
+      "step": 12851
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.337801452521499e-05,
+      "loss": 0.6696,
+      "step": 12852
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.337558376514513e-05,
+      "loss": 0.6985,
+      "step": 12853
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.337315291588037e-05,
+      "loss": 0.7487,
+      "step": 12854
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3370721977446625e-05,
+      "loss": 0.7163,
+      "step": 12855
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.336829094986978e-05,
+      "loss": 0.7597,
+      "step": 12856
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3365859833175705e-05,
+      "loss": 0.9108,
+      "step": 12857
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.336342862739031e-05,
+      "loss": 0.8715,
+      "step": 12858
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.336099733253947e-05,
+      "loss": 0.7786,
+      "step": 12859
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.335856594864909e-05,
+      "loss": 0.707,
+      "step": 12860
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.335613447574506e-05,
+      "loss": 0.7524,
+      "step": 12861
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.335370291385326e-05,
+      "loss": 0.7773,
+      "step": 12862
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.33512712629996e-05,
+      "loss": 0.8661,
+      "step": 12863
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3348839523209976e-05,
+      "loss": 0.8207,
+      "step": 12864
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.334640769451027e-05,
+      "loss": 0.7797,
+      "step": 12865
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.33439757769264e-05,
+      "loss": 0.7869,
+      "step": 12866
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3341543770484246e-05,
+      "loss": 0.7594,
+      "step": 12867
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3339111675209716e-05,
+      "loss": 0.7927,
+      "step": 12868
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3336679491128715e-05,
+      "loss": 0.7538,
+      "step": 12869
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3334247218267135e-05,
+      "loss": 0.7122,
+      "step": 12870
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.333181485665088e-05,
+      "loss": 0.7642,
+      "step": 12871
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.332938240630586e-05,
+      "loss": 0.7291,
+      "step": 12872
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3326949867257975e-05,
+      "loss": 0.7412,
+      "step": 12873
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.332451723953312e-05,
+      "loss": 0.8211,
+      "step": 12874
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3322084523157215e-05,
+      "loss": 0.828,
+      "step": 12875
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.331965171815617e-05,
+      "loss": 0.7943,
+      "step": 12876
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3317218824555874e-05,
+      "loss": 0.7424,
+      "step": 12877
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3314785842382256e-05,
+      "loss": 0.7362,
+      "step": 12878
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3312352771661216e-05,
+      "loss": 0.8065,
+      "step": 12879
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.330991961241866e-05,
+      "loss": 0.8432,
+      "step": 12880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.330748636468051e-05,
+      "loss": 0.714,
+      "step": 12881
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3305053028472666e-05,
+      "loss": 0.6747,
+      "step": 12882
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.330261960382106e-05,
+      "loss": 0.7291,
+      "step": 12883
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.330018609075159e-05,
+      "loss": 0.7353,
+      "step": 12884
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3297752489290175e-05,
+      "loss": 0.7904,
+      "step": 12885
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.329531879946274e-05,
+      "loss": 0.6853,
+      "step": 12886
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.329288502129519e-05,
+      "loss": 0.7498,
+      "step": 12887
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3290451154813454e-05,
+      "loss": 0.7312,
+      "step": 12888
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.328801720004344e-05,
+      "loss": 0.8026,
+      "step": 12889
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.328558315701109e-05,
+      "loss": 0.6879,
+      "step": 12890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.328314902574229e-05,
+      "loss": 0.816,
+      "step": 12891
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3280714806262995e-05,
+      "loss": 0.6937,
+      "step": 12892
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.327828049859911e-05,
+      "loss": 0.766,
+      "step": 12893
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3275846102776555e-05,
+      "loss": 0.7595,
+      "step": 12894
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3273411618821274e-05,
+      "loss": 0.7383,
+      "step": 12895
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3270977046759184e-05,
+      "loss": 0.7453,
+      "step": 12896
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.32685423866162e-05,
+      "loss": 0.7972,
+      "step": 12897
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.326610763841826e-05,
+      "loss": 0.7229,
+      "step": 12898
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3263672802191316e-05,
+      "loss": 0.7251,
+      "step": 12899
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.326123787796125e-05,
+      "loss": 0.6956,
+      "step": 12900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.325880286575402e-05,
+      "loss": 0.7881,
+      "step": 12901
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.325636776559555e-05,
+      "loss": 0.7924,
+      "step": 12902
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3253932577511784e-05,
+      "loss": 0.7811,
+      "step": 12903
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.325149730152864e-05,
+      "loss": 0.7365,
+      "step": 12904
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3249061937672066e-05,
+      "loss": 0.7016,
+      "step": 12905
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.324662648596799e-05,
+      "loss": 0.8073,
+      "step": 12906
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.324419094644234e-05,
+      "loss": 0.7445,
+      "step": 12907
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.324175531912107e-05,
+      "loss": 0.7546,
+      "step": 12908
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.323931960403011e-05,
+      "loss": 0.7934,
+      "step": 12909
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.32368838011954e-05,
+      "loss": 0.7906,
+      "step": 12910
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.323444791064287e-05,
+      "loss": 0.7806,
+      "step": 12911
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.323201193239848e-05,
+      "loss": 0.7813,
+      "step": 12912
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.322957586648815e-05,
+      "loss": 0.7708,
+      "step": 12913
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.322713971293783e-05,
+      "loss": 0.8722,
+      "step": 12914
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3224703471773474e-05,
+      "loss": 0.7665,
+      "step": 12915
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3222267143021024e-05,
+      "loss": 0.7251,
+      "step": 12916
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3219830726706413e-05,
+      "loss": 0.8463,
+      "step": 12917
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3217394222855603e-05,
+      "loss": 0.8239,
+      "step": 12918
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.321495763149453e-05,
+      "loss": 0.7918,
+      "step": 12919
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.321252095264914e-05,
+      "loss": 0.7736,
+      "step": 12920
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.32100841863454e-05,
+      "loss": 0.7131,
+      "step": 12921
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.320764733260924e-05,
+      "loss": 0.8123,
+      "step": 12922
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.320521039146661e-05,
+      "loss": 0.7354,
+      "step": 12923
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.320277336294349e-05,
+      "loss": 0.7594,
+      "step": 12924
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.32003362470658e-05,
+      "loss": 0.7797,
+      "step": 12925
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.31978990438595e-05,
+      "loss": 0.7284,
+      "step": 12926
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.319546175335057e-05,
+      "loss": 0.7195,
+      "step": 12927
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.319302437556494e-05,
+      "loss": 0.8067,
+      "step": 12928
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.319058691052858e-05,
+      "loss": 0.7464,
+      "step": 12929
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.318814935826744e-05,
+      "loss": 0.7791,
+      "step": 12930
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.318571171880748e-05,
+      "loss": 0.7654,
+      "step": 12931
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3183273992174656e-05,
+      "loss": 0.8074,
+      "step": 12932
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3180836178394935e-05,
+      "loss": 0.7779,
+      "step": 12933
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.317839827749427e-05,
+      "loss": 0.7405,
+      "step": 12934
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.317596028949863e-05,
+      "loss": 0.8094,
+      "step": 12935
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3173522214433984e-05,
+      "loss": 0.7808,
+      "step": 12936
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.317108405232628e-05,
+      "loss": 0.7716,
+      "step": 12937
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3168645803201495e-05,
+      "loss": 0.762,
+      "step": 12938
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3166207467085606e-05,
+      "loss": 0.7334,
+      "step": 12939
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3163769044004545e-05,
+      "loss": 0.8874,
+      "step": 12940
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.316133053398431e-05,
+      "loss": 0.7166,
+      "step": 12941
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.315889193705086e-05,
+      "loss": 0.7716,
+      "step": 12942
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3156453253230154e-05,
+      "loss": 0.7083,
+      "step": 12943
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.315401448254818e-05,
+      "loss": 0.756,
+      "step": 12944
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3151575625030905e-05,
+      "loss": 0.7689,
+      "step": 12945
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.314913668070429e-05,
+      "loss": 0.7247,
+      "step": 12946
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.314669764959433e-05,
+      "loss": 0.6449,
+      "step": 12947
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.314425853172697e-05,
+      "loss": 0.7554,
+      "step": 12948
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.314181932712822e-05,
+      "loss": 0.8276,
+      "step": 12949
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.313938003582403e-05,
+      "loss": 0.7074,
+      "step": 12950
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3136940657840373e-05,
+      "loss": 0.7936,
+      "step": 12951
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3134501193203244e-05,
+      "loss": 0.7372,
+      "step": 12952
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.313206164193862e-05,
+      "loss": 0.8131,
+      "step": 12953
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3129622004072476e-05,
+      "loss": 0.7719,
+      "step": 12954
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3127182279630785e-05,
+      "loss": 0.7935,
+      "step": 12955
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3124742468639545e-05,
+      "loss": 0.7785,
+      "step": 12956
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.312230257112473e-05,
+      "loss": 0.75,
+      "step": 12957
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.311986258711232e-05,
+      "loss": 0.764,
+      "step": 12958
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.311742251662831e-05,
+      "loss": 0.7135,
+      "step": 12959
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.311498235969867e-05,
+      "loss": 0.7358,
+      "step": 12960
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.31125421163494e-05,
+      "loss": 0.7061,
+      "step": 12961
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.311010178660649e-05,
+      "loss": 0.7721,
+      "step": 12962
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.31076613704959e-05,
+      "loss": 0.741,
+      "step": 12963
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.310522086804365e-05,
+      "loss": 0.7856,
+      "step": 12964
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.310278027927572e-05,
+      "loss": 0.752,
+      "step": 12965
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.310033960421809e-05,
+      "loss": 0.7808,
+      "step": 12966
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.309789884289677e-05,
+      "loss": 0.8154,
+      "step": 12967
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.309545799533774e-05,
+      "loss": 0.6724,
+      "step": 12968
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3093017061567e-05,
+      "loss": 0.7674,
+      "step": 12969
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.309057604161055e-05,
+      "loss": 0.765,
+      "step": 12970
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3088134935494363e-05,
+      "loss": 0.7492,
+      "step": 12971
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.308569374324445e-05,
+      "loss": 0.7227,
+      "step": 12972
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3083252464886814e-05,
+      "loss": 0.7319,
+      "step": 12973
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.308081110044744e-05,
+      "loss": 0.7852,
+      "step": 12974
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.307836964995234e-05,
+      "loss": 0.7324,
+      "step": 12975
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3075928113427506e-05,
+      "loss": 0.7749,
+      "step": 12976
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3073486490898943e-05,
+      "loss": 0.8263,
+      "step": 12977
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.307104478239264e-05,
+      "loss": 0.7343,
+      "step": 12978
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3068602987934624e-05,
+      "loss": 0.7549,
+      "step": 12979
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3066161107550875e-05,
+      "loss": 0.715,
+      "step": 12980
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3063719141267415e-05,
+      "loss": 0.7,
+      "step": 12981
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3061277089110236e-05,
+      "loss": 0.7449,
+      "step": 12982
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.305883495110535e-05,
+      "loss": 0.7357,
+      "step": 12983
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.305639272727877e-05,
+      "loss": 0.7113,
+      "step": 12984
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.305395041765649e-05,
+      "loss": 0.751,
+      "step": 12985
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3051508022264534e-05,
+      "loss": 0.7083,
+      "step": 12986
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.30490655411289e-05,
+      "loss": 0.7092,
+      "step": 12987
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.304662297427561e-05,
+      "loss": 0.7123,
+      "step": 12988
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3044180321730676e-05,
+      "loss": 0.8254,
+      "step": 12989
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3041737583520096e-05,
+      "loss": 0.7966,
+      "step": 12990
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.30392947596699e-05,
+      "loss": 0.7413,
+      "step": 12991
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.303685185020608e-05,
+      "loss": 0.8155,
+      "step": 12992
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3034408855154684e-05,
+      "loss": 0.7524,
+      "step": 12993
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.303196577454171e-05,
+      "loss": 0.7035,
+      "step": 12994
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3029522608393164e-05,
+      "loss": 0.7243,
+      "step": 12995
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.302707935673509e-05,
+      "loss": 0.7214,
+      "step": 12996
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.302463601959349e-05,
+      "loss": 0.773,
+      "step": 12997
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3022192596994386e-05,
+      "loss": 0.7087,
+      "step": 12998
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.30197490889638e-05,
+      "loss": 0.6969,
+      "step": 12999
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.301730549552776e-05,
+      "loss": 0.7705,
+      "step": 13000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3014861816712286e-05,
+      "loss": 0.784,
+      "step": 13001
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.30124180525434e-05,
+      "loss": 0.8151,
+      "step": 13002
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.300997420304712e-05,
+      "loss": 0.6895,
+      "step": 13003
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3007530268249474e-05,
+      "loss": 0.7237,
+      "step": 13004
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.30050862481765e-05,
+      "loss": 0.7253,
+      "step": 13005
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3002642142854224e-05,
+      "loss": 0.822,
+      "step": 13006
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3000197952308654e-05,
+      "loss": 0.7115,
+      "step": 13007
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.2997753676565834e-05,
+      "loss": 0.7829,
+      "step": 13008
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.29953093156518e-05,
+      "loss": 0.8036,
+      "step": 13009
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.2992864869592574e-05,
+      "loss": 0.744,
+      "step": 13010
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.299042033841419e-05,
+      "loss": 0.6986,
+      "step": 13011
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.298797572214268e-05,
+      "loss": 0.7442,
+      "step": 13012
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.2985531020804074e-05,
+      "loss": 0.8237,
+      "step": 13013
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.298308623442441e-05,
+      "loss": 0.7124,
+      "step": 13014
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.298064136302973e-05,
+      "loss": 0.7208,
+      "step": 13015
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.297819640664607e-05,
+      "loss": 0.8903,
+      "step": 13016
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.297575136529945e-05,
+      "loss": 0.7922,
+      "step": 13017
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.297330623901592e-05,
+      "loss": 0.8315,
+      "step": 13018
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.297086102782153e-05,
+      "loss": 0.7319,
+      "step": 13019
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.2968415731742306e-05,
+      "loss": 0.8138,
+      "step": 13020
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.2965970350804286e-05,
+      "loss": 0.8084,
+      "step": 13021
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.2963524885033525e-05,
+      "loss": 0.8276,
+      "step": 13022
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.2961079334456056e-05,
+      "loss": 0.7739,
+      "step": 13023
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.295863369909792e-05,
+      "loss": 0.7712,
+      "step": 13024
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2956187978985175e-05,
+      "loss": 0.7224,
+      "step": 13025
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.295374217414385e-05,
+      "loss": 0.7771,
+      "step": 13026
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.295129628460001e-05,
+      "loss": 0.7492,
+      "step": 13027
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.294885031037969e-05,
+      "loss": 0.7423,
+      "step": 13028
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2946404251508935e-05,
+      "loss": 0.7886,
+      "step": 13029
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.29439581080138e-05,
+      "loss": 0.7397,
+      "step": 13030
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2941511879920336e-05,
+      "loss": 0.7692,
+      "step": 13031
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2939065567254585e-05,
+      "loss": 0.7585,
+      "step": 13032
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2936619170042614e-05,
+      "loss": 0.7074,
+      "step": 13033
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.293417268831047e-05,
+      "loss": 0.8242,
+      "step": 13034
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2931726122084186e-05,
+      "loss": 0.7268,
+      "step": 13035
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2929279471389853e-05,
+      "loss": 0.7296,
+      "step": 13036
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2926832736253505e-05,
+      "loss": 0.7734,
+      "step": 13037
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.292438591670119e-05,
+      "loss": 0.7203,
+      "step": 13038
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.292193901275899e-05,
+      "loss": 0.7522,
+      "step": 13039
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.291949202445295e-05,
+      "loss": 0.8236,
+      "step": 13040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.291704495180911e-05,
+      "loss": 0.8397,
+      "step": 13041
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2914597794853566e-05,
+      "loss": 0.7241,
+      "step": 13042
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.291215055361235e-05,
+      "loss": 0.7125,
+      "step": 13043
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.290970322811153e-05,
+      "loss": 0.7539,
+      "step": 13044
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.290725581837718e-05,
+      "loss": 0.8305,
+      "step": 13045
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.290480832443536e-05,
+      "loss": 0.8487,
+      "step": 13046
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.290236074631212e-05,
+      "loss": 0.8677,
+      "step": 13047
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2899913084033545e-05,
+      "loss": 0.7684,
+      "step": 13048
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2897465337625685e-05,
+      "loss": 0.7583,
+      "step": 13049
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.289501750711462e-05,
+      "loss": 0.7279,
+      "step": 13050
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2892569592526414e-05,
+      "loss": 0.752,
+      "step": 13051
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2890121593887125e-05,
+      "loss": 0.8979,
+      "step": 13052
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.288767351122284e-05,
+      "loss": 0.7961,
+      "step": 13053
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.288522534455961e-05,
+      "loss": 0.7705,
+      "step": 13054
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.288277709392352e-05,
+      "loss": 0.7325,
+      "step": 13055
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2880328759340644e-05,
+      "loss": 0.78,
+      "step": 13056
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.287788034083705e-05,
+      "loss": 0.8153,
+      "step": 13057
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.287543183843882e-05,
+      "loss": 0.7266,
+      "step": 13058
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.287298325217201e-05,
+      "loss": 0.8123,
+      "step": 13059
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2870534582062714e-05,
+      "loss": 0.7056,
+      "step": 13060
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2868085828137e-05,
+      "loss": 0.7268,
+      "step": 13061
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.286563699042096e-05,
+      "loss": 0.7957,
+      "step": 13062
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2863188068940657e-05,
+      "loss": 0.8391,
+      "step": 13063
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2860739063722155e-05,
+      "loss": 0.7904,
+      "step": 13064
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.285828997479158e-05,
+      "loss": 0.7916,
+      "step": 13065
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.285584080217498e-05,
+      "loss": 0.7704,
+      "step": 13066
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.285339154589843e-05,
+      "loss": 0.7421,
+      "step": 13067
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.285094220598805e-05,
+      "loss": 0.7472,
+      "step": 13068
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.284849278246989e-05,
+      "loss": 0.7621,
+      "step": 13069
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2846043275370046e-05,
+      "loss": 0.766,
+      "step": 13070
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.284359368471461e-05,
+      "loss": 0.779,
+      "step": 13071
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.284114401052966e-05,
+      "loss": 0.8148,
+      "step": 13072
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2838694252841284e-05,
+      "loss": 0.8035,
+      "step": 13073
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.283624441167558e-05,
+      "loss": 0.734,
+      "step": 13074
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.283379448705863e-05,
+      "loss": 0.7408,
+      "step": 13075
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.283134447901652e-05,
+      "loss": 0.7761,
+      "step": 13076
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.282889438757535e-05,
+      "loss": 0.7279,
+      "step": 13077
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.282644421276121e-05,
+      "loss": 0.7839,
+      "step": 13078
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.282399395460017e-05,
+      "loss": 0.7856,
+      "step": 13079
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.282154361311837e-05,
+      "loss": 0.8303,
+      "step": 13080
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.281909318834187e-05,
+      "loss": 0.6955,
+      "step": 13081
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.281664268029677e-05,
+      "loss": 0.7324,
+      "step": 13082
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.281419208900918e-05,
+      "loss": 0.7239,
+      "step": 13083
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.281174141450518e-05,
+      "loss": 0.7271,
+      "step": 13084
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.280929065681088e-05,
+      "loss": 0.7534,
+      "step": 13085
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.280683981595239e-05,
+      "loss": 0.7541,
+      "step": 13086
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.280438889195577e-05,
+      "loss": 0.7673,
+      "step": 13087
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.280193788484716e-05,
+      "loss": 0.7794,
+      "step": 13088
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2799486794652645e-05,
+      "loss": 0.7129,
+      "step": 13089
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.279703562139833e-05,
+      "loss": 0.7008,
+      "step": 13090
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.279458436511033e-05,
+      "loss": 0.7807,
+      "step": 13091
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.279213302581473e-05,
+      "loss": 0.7843,
+      "step": 13092
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2789681603537636e-05,
+      "loss": 0.6935,
+      "step": 13093
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.278723009830518e-05,
+      "loss": 0.8581,
+      "step": 13094
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.278477851014344e-05,
+      "loss": 0.6998,
+      "step": 13095
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.278232683907853e-05,
+      "loss": 0.7864,
+      "step": 13096
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.277987508513657e-05,
+      "loss": 0.7553,
+      "step": 13097
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.277742324834366e-05,
+      "loss": 0.7376,
+      "step": 13098
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.277497132872592e-05,
+      "loss": 0.8398,
+      "step": 13099
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.277251932630946e-05,
+      "loss": 0.8127,
+      "step": 13100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2770067241120386e-05,
+      "loss": 0.7121,
+      "step": 13101
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.27676150731848e-05,
+      "loss": 0.7642,
+      "step": 13102
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2765162822528845e-05,
+      "loss": 0.8022,
+      "step": 13103
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.276271048917861e-05,
+      "loss": 0.745,
+      "step": 13104
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2760258073160225e-05,
+      "loss": 0.7952,
+      "step": 13105
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2757805574499805e-05,
+      "loss": 0.7586,
+      "step": 13106
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.275535299322346e-05,
+      "loss": 0.7679,
+      "step": 13107
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.275290032935732e-05,
+      "loss": 0.832,
+      "step": 13108
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.27504475829275e-05,
+      "loss": 0.7422,
+      "step": 13109
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.27479947539601e-05,
+      "loss": 0.7568,
+      "step": 13110
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.274554184248129e-05,
+      "loss": 0.7323,
+      "step": 13111
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.274308884851715e-05,
+      "loss": 0.8324,
+      "step": 13112
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.27406357720938e-05,
+      "loss": 0.7964,
+      "step": 13113
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.273818261323739e-05,
+      "loss": 0.755,
+      "step": 13114
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2735729371974036e-05,
+      "loss": 0.7209,
+      "step": 13115
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.273327604832985e-05,
+      "loss": 0.7474,
+      "step": 13116
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2730822642330985e-05,
+      "loss": 0.7356,
+      "step": 13117
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.272836915400354e-05,
+      "loss": 0.7479,
+      "step": 13118
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.272591558337365e-05,
+      "loss": 0.7931,
+      "step": 13119
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.272346193046747e-05,
+      "loss": 0.7699,
+      "step": 13120
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2721008195311096e-05,
+      "loss": 0.7258,
+      "step": 13121
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2718554377930675e-05,
+      "loss": 0.787,
+      "step": 13122
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.271610047835233e-05,
+      "loss": 0.7451,
+      "step": 13123
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2713646496602214e-05,
+      "loss": 0.7069,
+      "step": 13124
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2711192432706426e-05,
+      "loss": 0.7965,
+      "step": 13125
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.270873828669113e-05,
+      "loss": 0.7271,
+      "step": 13126
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.270628405858246e-05,
+      "loss": 0.748,
+      "step": 13127
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.270382974840653e-05,
+      "loss": 0.7097,
+      "step": 13128
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2701375356189494e-05,
+      "loss": 0.7222,
+      "step": 13129
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.269892088195749e-05,
+      "loss": 0.8656,
+      "step": 13130
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.269646632573665e-05,
+      "loss": 0.7913,
+      "step": 13131
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.269401168755312e-05,
+      "loss": 0.7402,
+      "step": 13132
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.269155696743303e-05,
+      "loss": 0.775,
+      "step": 13133
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.268910216540253e-05,
+      "loss": 0.7062,
+      "step": 13134
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.268664728148777e-05,
+      "loss": 0.7563,
+      "step": 13135
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.268419231571487e-05,
+      "loss": 0.804,
+      "step": 13136
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2681737268109994e-05,
+      "loss": 0.7303,
+      "step": 13137
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2679282138699275e-05,
+      "loss": 0.7161,
+      "step": 13138
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2676826927508866e-05,
+      "loss": 0.6834,
+      "step": 13139
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.267437163456491e-05,
+      "loss": 0.7202,
+      "step": 13140
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2671916259893566e-05,
+      "loss": 0.7371,
+      "step": 13141
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.266946080352095e-05,
+      "loss": 0.7537,
+      "step": 13142
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.266700526547325e-05,
+      "loss": 0.8231,
+      "step": 13143
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.266454964577659e-05,
+      "loss": 0.7966,
+      "step": 13144
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2662093944457126e-05,
+      "loss": 0.7594,
+      "step": 13145
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.265963816154103e-05,
+      "loss": 0.7361,
+      "step": 13146
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.265718229705442e-05,
+      "loss": 0.7817,
+      "step": 13147
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.265472635102348e-05,
+      "loss": 0.6921,
+      "step": 13148
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2652270323474344e-05,
+      "loss": 0.7476,
+      "step": 13149
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2649814214433186e-05,
+      "loss": 0.8218,
+      "step": 13150
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.264735802392614e-05,
+      "loss": 0.771,
+      "step": 13151
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.264490175197937e-05,
+      "loss": 0.7357,
+      "step": 13152
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2642445398619046e-05,
+      "loss": 0.7728,
+      "step": 13153
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2639988963871316e-05,
+      "loss": 0.7714,
+      "step": 13154
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.263753244776234e-05,
+      "loss": 0.8112,
+      "step": 13155
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.263507585031828e-05,
+      "loss": 0.7395,
+      "step": 13156
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2632619171565306e-05,
+      "loss": 0.7727,
+      "step": 13157
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.263016241152956e-05,
+      "loss": 0.6803,
+      "step": 13158
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.262770557023722e-05,
+      "loss": 0.6808,
+      "step": 13159
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.262524864771446e-05,
+      "loss": 0.7104,
+      "step": 13160
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2622791643987424e-05,
+      "loss": 0.8438,
+      "step": 13161
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.262033455908228e-05,
+      "loss": 0.6698,
+      "step": 13162
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.26178773930252e-05,
+      "loss": 0.759,
+      "step": 13163
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.261542014584235e-05,
+      "loss": 0.7154,
+      "step": 13164
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.26129628175599e-05,
+      "loss": 0.7336,
+      "step": 13165
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.261050540820402e-05,
+      "loss": 0.6807,
+      "step": 13166
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.260804791780087e-05,
+      "loss": 0.6556,
+      "step": 13167
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.260559034637664e-05,
+      "loss": 0.8005,
+      "step": 13168
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.260313269395748e-05,
+      "loss": 0.7429,
+      "step": 13169
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.260067496056958e-05,
+      "loss": 0.7418,
+      "step": 13170
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.25982171462391e-05,
+      "loss": 0.7202,
+      "step": 13171
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.259575925099223e-05,
+      "loss": 0.7217,
+      "step": 13172
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2593301274855125e-05,
+      "loss": 0.8154,
+      "step": 13173
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.259084321785398e-05,
+      "loss": 0.805,
+      "step": 13174
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.258838508001496e-05,
+      "loss": 0.8665,
+      "step": 13175
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2585926861364244e-05,
+      "loss": 0.7101,
+      "step": 13176
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.258346856192801e-05,
+      "loss": 0.8368,
+      "step": 13177
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.258101018173245e-05,
+      "loss": 0.7698,
+      "step": 13178
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2578551720803726e-05,
+      "loss": 0.668,
+      "step": 13179
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.257609317916802e-05,
+      "loss": 0.8669,
+      "step": 13180
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.257363455685154e-05,
+      "loss": 0.7481,
+      "step": 13181
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.257117585388044e-05,
+      "loss": 0.7241,
+      "step": 13182
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.256871707028091e-05,
+      "loss": 0.7107,
+      "step": 13183
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.256625820607915e-05,
+      "loss": 0.7298,
+      "step": 13184
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.256379926130132e-05,
+      "loss": 0.748,
+      "step": 13185
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.256134023597363e-05,
+      "loss": 0.7853,
+      "step": 13186
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.255888113012225e-05,
+      "loss": 0.7376,
+      "step": 13187
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2556421943773376e-05,
+      "loss": 0.752,
+      "step": 13188
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.25539626769532e-05,
+      "loss": 0.8158,
+      "step": 13189
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2551503329687906e-05,
+      "loss": 0.7299,
+      "step": 13190
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2549043902003685e-05,
+      "loss": 0.8049,
+      "step": 13191
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.254658439392673e-05,
+      "loss": 0.7281,
+      "step": 13192
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2544124805483235e-05,
+      "loss": 0.7092,
+      "step": 13193
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2541665136699386e-05,
+      "loss": 0.7588,
+      "step": 13194
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2539205387601384e-05,
+      "loss": 0.7733,
+      "step": 13195
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.253674555821543e-05,
+      "loss": 0.7871,
+      "step": 13196
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.25342856485677e-05,
+      "loss": 0.7618,
+      "step": 13197
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.25318256586844e-05,
+      "loss": 0.8669,
+      "step": 13198
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.252936558859174e-05,
+      "loss": 0.7241,
+      "step": 13199
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.25269054383159e-05,
+      "loss": 0.8146,
+      "step": 13200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2524445207883095e-05,
+      "loss": 0.7548,
+      "step": 13201
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2521984897319505e-05,
+      "loss": 0.7626,
+      "step": 13202
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.251952450665135e-05,
+      "loss": 0.7828,
+      "step": 13203
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2517064035904834e-05,
+      "loss": 0.8038,
+      "step": 13204
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2514603485106133e-05,
+      "loss": 0.7382,
+      "step": 13205
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2512142854281474e-05,
+      "loss": 0.7088,
+      "step": 13206
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.250968214345705e-05,
+      "loss": 0.7887,
+      "step": 13207
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2507221352659074e-05,
+      "loss": 0.7375,
+      "step": 13208
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.250476048191375e-05,
+      "loss": 0.8002,
+      "step": 13209
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.250229953124728e-05,
+      "loss": 0.7976,
+      "step": 13210
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.249983850068588e-05,
+      "loss": 0.7891,
+      "step": 13211
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.249737739025575e-05,
+      "loss": 0.7523,
+      "step": 13212
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.249491619998311e-05,
+      "loss": 0.7226,
+      "step": 13213
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2492454929894144e-05,
+      "loss": 0.7664,
+      "step": 13214
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.24899935800151e-05,
+      "loss": 0.7156,
+      "step": 13215
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2487532150372164e-05,
+      "loss": 0.7425,
+      "step": 13216
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2485070640991556e-05,
+      "loss": 0.76,
+      "step": 13217
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.248260905189949e-05,
+      "loss": 0.7078,
+      "step": 13218
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.248014738312219e-05,
+      "loss": 0.7202,
+      "step": 13219
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.247768563468584e-05,
+      "loss": 0.7562,
+      "step": 13220
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2475223806616696e-05,
+      "loss": 0.9522,
+      "step": 13221
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2472761898940954e-05,
+      "loss": 0.826,
+      "step": 13222
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.247029991168483e-05,
+      "loss": 0.7995,
+      "step": 13223
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.246783784487455e-05,
+      "loss": 0.7206,
+      "step": 13224
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.246537569853633e-05,
+      "loss": 0.7821,
+      "step": 13225
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2462913472696385e-05,
+      "loss": 0.8053,
+      "step": 13226
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.246045116738095e-05,
+      "loss": 0.7957,
+      "step": 13227
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2457988782616234e-05,
+      "loss": 0.7975,
+      "step": 13228
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2455526318428476e-05,
+      "loss": 0.6692,
+      "step": 13229
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.245306377484388e-05,
+      "loss": 0.7843,
+      "step": 13230
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.245060115188868e-05,
+      "loss": 0.7704,
+      "step": 13231
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2448138449589106e-05,
+      "loss": 0.7559,
+      "step": 13232
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.244567566797138e-05,
+      "loss": 0.7468,
+      "step": 13233
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.244321280706172e-05,
+      "loss": 0.7631,
+      "step": 13234
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.244074986688637e-05,
+      "loss": 0.7012,
+      "step": 13235
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.243828684747155e-05,
+      "loss": 0.7312,
+      "step": 13236
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.243582374884349e-05,
+      "loss": 0.7112,
+      "step": 13237
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.243336057102842e-05,
+      "loss": 0.7817,
+      "step": 13238
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2430897314052576e-05,
+      "loss": 0.7118,
+      "step": 13239
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.242843397794218e-05,
+      "loss": 0.7326,
+      "step": 13240
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.242597056272348e-05,
+      "loss": 0.7824,
+      "step": 13241
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.24235070684227e-05,
+      "loss": 0.8368,
+      "step": 13242
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2421043495066076e-05,
+      "loss": 0.7021,
+      "step": 13243
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2418579842679846e-05,
+      "loss": 0.7477,
+      "step": 13244
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.241611611129024e-05,
+      "loss": 0.948,
+      "step": 13245
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2413652300923504e-05,
+      "loss": 0.7394,
+      "step": 13246
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.241118841160587e-05,
+      "loss": 0.7742,
+      "step": 13247
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.240872444336358e-05,
+      "loss": 0.6848,
+      "step": 13248
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.240626039622286e-05,
+      "loss": 0.7305,
+      "step": 13249
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.240379627020999e-05,
+      "loss": 0.7967,
+      "step": 13250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2401332065351165e-05,
+      "loss": 0.725,
+      "step": 13251
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2398867781672655e-05,
+      "loss": 0.7161,
+      "step": 13252
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2396403419200694e-05,
+      "loss": 0.8128,
+      "step": 13253
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.239393897796152e-05,
+      "loss": 0.7507,
+      "step": 13254
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.239147445798139e-05,
+      "loss": 0.7693,
+      "step": 13255
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.238900985928655e-05,
+      "loss": 0.7774,
+      "step": 13256
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.238654518190323e-05,
+      "loss": 0.753,
+      "step": 13257
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2384080425857685e-05,
+      "loss": 0.7214,
+      "step": 13258
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2381615591176186e-05,
+      "loss": 0.8538,
+      "step": 13259
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.237915067788495e-05,
+      "loss": 0.7767,
+      "step": 13260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.237668568601023e-05,
+      "loss": 0.7686,
+      "step": 13261
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.237422061557831e-05,
+      "loss": 0.7125,
+      "step": 13262
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.23717554666154e-05,
+      "loss": 0.6931,
+      "step": 13263
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2369290239147773e-05,
+      "loss": 0.7557,
+      "step": 13264
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2366824933201686e-05,
+      "loss": 0.8071,
+      "step": 13265
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2364359548803366e-05,
+      "loss": 0.673,
+      "step": 13266
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2361894085979105e-05,
+      "loss": 0.7178,
+      "step": 13267
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.235942854475514e-05,
+      "loss": 0.8147,
+      "step": 13268
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.235696292515772e-05,
+      "loss": 0.7925,
+      "step": 13269
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.235449722721313e-05,
+      "loss": 0.8009,
+      "step": 13270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.23520314509476e-05,
+      "loss": 0.7913,
+      "step": 13271
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.234956559638739e-05,
+      "loss": 0.8234,
+      "step": 13272
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.234709966355878e-05,
+      "loss": 0.7285,
+      "step": 13273
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.234463365248801e-05,
+      "loss": 0.7566,
+      "step": 13274
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2342167563201354e-05,
+      "loss": 0.7122,
+      "step": 13275
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.233970139572507e-05,
+      "loss": 0.8593,
+      "step": 13276
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2337235150085424e-05,
+      "loss": 0.8739,
+      "step": 13277
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.233476882630868e-05,
+      "loss": 0.7594,
+      "step": 13278
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.23323024244211e-05,
+      "loss": 0.7813,
+      "step": 13279
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.232983594444895e-05,
+      "loss": 0.7594,
+      "step": 13280
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.232736938641849e-05,
+      "loss": 0.6715,
+      "step": 13281
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.232490275035601e-05,
+      "loss": 0.7113,
+      "step": 13282
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.232243603628775e-05,
+      "loss": 0.8449,
+      "step": 13283
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.231996924424e-05,
+      "loss": 0.8177,
+      "step": 13284
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.231750237423902e-05,
+      "loss": 0.7591,
+      "step": 13285
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.231503542631108e-05,
+      "loss": 0.8053,
+      "step": 13286
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.231256840048245e-05,
+      "loss": 0.7764,
+      "step": 13287
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2310101296779417e-05,
+      "loss": 0.7926,
+      "step": 13288
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.230763411522823e-05,
+      "loss": 0.7404,
+      "step": 13289
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2305166855855186e-05,
+      "loss": 0.7237,
+      "step": 13290
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.230269951868654e-05,
+      "loss": 0.7943,
+      "step": 13291
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.230023210374859e-05,
+      "loss": 0.8383,
+      "step": 13292
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2297764611067596e-05,
+      "loss": 0.7775,
+      "step": 13293
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2295297040669834e-05,
+      "loss": 0.756,
+      "step": 13294
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.229282939258159e-05,
+      "loss": 0.7945,
+      "step": 13295
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2290361666829146e-05,
+      "loss": 0.7288,
+      "step": 13296
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.228789386343877e-05,
+      "loss": 0.7431,
+      "step": 13297
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.228542598243675e-05,
+      "loss": 0.8122,
+      "step": 13298
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.228295802384937e-05,
+      "loss": 0.7757,
+      "step": 13299
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2280489987702914e-05,
+      "loss": 0.7838,
+      "step": 13300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.227802187402365e-05,
+      "loss": 0.6996,
+      "step": 13301
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.227555368283787e-05,
+      "loss": 0.8053,
+      "step": 13302
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2273085414171864e-05,
+      "loss": 0.799,
+      "step": 13303
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.227061706805191e-05,
+      "loss": 0.742,
+      "step": 13304
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.226814864450431e-05,
+      "loss": 0.8162,
+      "step": 13305
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2265680143555324e-05,
+      "loss": 0.7152,
+      "step": 13306
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.226321156523127e-05,
+      "loss": 0.7341,
+      "step": 13307
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.226074290955841e-05,
+      "loss": 0.7173,
+      "step": 13308
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2258274176563054e-05,
+      "loss": 0.7105,
+      "step": 13309
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.225580536627148e-05,
+      "loss": 0.7012,
+      "step": 13310
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.225333647870999e-05,
+      "loss": 0.7123,
+      "step": 13311
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.225086751390487e-05,
+      "loss": 0.7228,
+      "step": 13312
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.224839847188241e-05,
+      "loss": 0.8213,
+      "step": 13313
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2245929352668905e-05,
+      "loss": 0.7063,
+      "step": 13314
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.224346015629065e-05,
+      "loss": 0.7193,
+      "step": 13315
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.224099088277395e-05,
+      "loss": 0.7351,
+      "step": 13316
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2238521532145085e-05,
+      "loss": 0.766,
+      "step": 13317
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2236052104430357e-05,
+      "loss": 0.7189,
+      "step": 13318
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.223358259965608e-05,
+      "loss": 0.8084,
+      "step": 13319
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2231113017848535e-05,
+      "loss": 0.7107,
+      "step": 13320
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.222864335903402e-05,
+      "loss": 0.7652,
+      "step": 13321
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2226173623238846e-05,
+      "loss": 0.738,
+      "step": 13322
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.222370381048931e-05,
+      "loss": 0.7947,
+      "step": 13323
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.222123392081172e-05,
+      "loss": 0.8369,
+      "step": 13324
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2218763954232376e-05,
+      "loss": 0.8265,
+      "step": 13325
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.221629391077756e-05,
+      "loss": 0.7654,
+      "step": 13326
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2213823790473606e-05,
+      "loss": 0.808,
+      "step": 13327
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.221135359334681e-05,
+      "loss": 0.736,
+      "step": 13328
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.220888331942347e-05,
+      "loss": 0.7715,
+      "step": 13329
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.22064129687299e-05,
+      "loss": 0.7522,
+      "step": 13330
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.220394254129242e-05,
+      "loss": 0.6944,
+      "step": 13331
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.220147203713731e-05,
+      "loss": 0.6898,
+      "step": 13332
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2199001456290906e-05,
+      "loss": 0.8742,
+      "step": 13333
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2196530798779504e-05,
+      "loss": 0.7607,
+      "step": 13334
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2194060064629415e-05,
+      "loss": 0.8128,
+      "step": 13335
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.219158925386696e-05,
+      "loss": 0.7653,
+      "step": 13336
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2189118366518436e-05,
+      "loss": 0.7774,
+      "step": 13337
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2186647402610165e-05,
+      "loss": 0.7607,
+      "step": 13338
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.218417636216847e-05,
+      "loss": 0.7349,
+      "step": 13339
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.218170524521966e-05,
+      "loss": 0.7972,
+      "step": 13340
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.217923405179004e-05,
+      "loss": 0.8633,
+      "step": 13341
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.217676278190594e-05,
+      "loss": 0.7437,
+      "step": 13342
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2174291435593676e-05,
+      "loss": 0.7798,
+      "step": 13343
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.217182001287957e-05,
+      "loss": 0.7402,
+      "step": 13344
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2169348513789927e-05,
+      "loss": 0.719,
+      "step": 13345
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.216687693835107e-05,
+      "loss": 0.7483,
+      "step": 13346
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2164405286589336e-05,
+      "loss": 0.6631,
+      "step": 13347
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.216193355853103e-05,
+      "loss": 0.7599,
+      "step": 13348
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.215946175420248e-05,
+      "loss": 0.7661,
+      "step": 13349
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2156989873630006e-05,
+      "loss": 0.7919,
+      "step": 13350
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.215451791683993e-05,
+      "loss": 0.7769,
+      "step": 13351
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2152045883858595e-05,
+      "loss": 0.7639,
+      "step": 13352
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.214957377471231e-05,
+      "loss": 0.7458,
+      "step": 13353
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2147101589427404e-05,
+      "loss": 0.7005,
+      "step": 13354
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.21446293280302e-05,
+      "loss": 0.7974,
+      "step": 13355
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.214215699054703e-05,
+      "loss": 0.6931,
+      "step": 13356
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2139684577004234e-05,
+      "loss": 0.8391,
+      "step": 13357
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2137212087428126e-05,
+      "loss": 0.7402,
+      "step": 13358
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.213473952184504e-05,
+      "loss": 0.7824,
+      "step": 13359
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.213226688028132e-05,
+      "loss": 0.734,
+      "step": 13360
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.212979416276328e-05,
+      "loss": 0.8195,
+      "step": 13361
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.212732136931726e-05,
+      "loss": 0.7222,
+      "step": 13362
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.21248484999696e-05,
+      "loss": 0.7912,
+      "step": 13363
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.212237555474662e-05,
+      "loss": 0.7967,
+      "step": 13364
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.211990253367467e-05,
+      "loss": 0.7214,
+      "step": 13365
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2117429436780086e-05,
+      "loss": 0.879,
+      "step": 13366
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.211495626408919e-05,
+      "loss": 0.7377,
+      "step": 13367
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2112483015628335e-05,
+      "loss": 0.6534,
+      "step": 13368
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2110009691423854e-05,
+      "loss": 0.7054,
+      "step": 13369
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2107536291502075e-05,
+      "loss": 0.6817,
+      "step": 13370
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.210506281588936e-05,
+      "loss": 0.7817,
+      "step": 13371
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2102589264612046e-05,
+      "loss": 0.7861,
+      "step": 13372
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.210011563769646e-05,
+      "loss": 0.7561,
+      "step": 13373
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2097641935168956e-05,
+      "loss": 0.7231,
+      "step": 13374
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.209516815705586e-05,
+      "loss": 0.7352,
+      "step": 13375
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2092694303383555e-05,
+      "loss": 0.7489,
+      "step": 13376
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.209022037417835e-05,
+      "loss": 0.7885,
+      "step": 13377
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.20877463694666e-05,
+      "loss": 0.7663,
+      "step": 13378
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.208527228927466e-05,
+      "loss": 0.7087,
+      "step": 13379
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2082798133628865e-05,
+      "loss": 0.7319,
+      "step": 13380
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.208032390255557e-05,
+      "loss": 0.7696,
+      "step": 13381
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2077849596081134e-05,
+      "loss": 0.7983,
+      "step": 13382
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.20753752142319e-05,
+      "loss": 0.687,
+      "step": 13383
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2072900757034195e-05,
+      "loss": 0.7031,
+      "step": 13384
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2070426224514406e-05,
+      "loss": 0.7411,
+      "step": 13385
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.206795161669887e-05,
+      "loss": 0.7201,
+      "step": 13386
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.206547693361393e-05,
+      "loss": 0.7287,
+      "step": 13387
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2063002175285954e-05,
+      "loss": 0.8256,
+      "step": 13388
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.20605273417413e-05,
+      "loss": 0.7153,
+      "step": 13389
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2058052433006305e-05,
+      "loss": 0.7673,
+      "step": 13390
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.205557744910735e-05,
+      "loss": 0.7633,
+      "step": 13391
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.205310239007077e-05,
+      "loss": 0.7482,
+      "step": 13392
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2050627255922936e-05,
+      "loss": 0.7528,
+      "step": 13393
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.20481520466902e-05,
+      "loss": 0.7334,
+      "step": 13394
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2045676762398924e-05,
+      "loss": 0.7734,
+      "step": 13395
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.204320140307547e-05,
+      "loss": 0.7658,
+      "step": 13396
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2040725968746196e-05,
+      "loss": 0.8,
+      "step": 13397
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.203825045943746e-05,
+      "loss": 0.7667,
+      "step": 13398
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2035774875175635e-05,
+      "loss": 0.7499,
+      "step": 13399
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.203329921598708e-05,
+      "loss": 0.7743,
+      "step": 13400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.203082348189815e-05,
+      "loss": 0.7358,
+      "step": 13401
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.202834767293523e-05,
+      "loss": 0.7097,
+      "step": 13402
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2025871789124674e-05,
+      "loss": 0.788,
+      "step": 13403
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.202339583049284e-05,
+      "loss": 0.7403,
+      "step": 13404
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.202091979706611e-05,
+      "loss": 0.8012,
+      "step": 13405
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2018443688870845e-05,
+      "loss": 0.6909,
+      "step": 13406
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2015967505933406e-05,
+      "loss": 0.7816,
+      "step": 13407
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.201349124828019e-05,
+      "loss": 0.756,
+      "step": 13408
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.201101491593754e-05,
+      "loss": 0.817,
+      "step": 13409
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.200853850893184e-05,
+      "loss": 0.8116,
+      "step": 13410
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2006062027289465e-05,
+      "loss": 0.7131,
+      "step": 13411
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2003585471036776e-05,
+      "loss": 0.7406,
+      "step": 13412
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.200110884020016e-05,
+      "loss": 0.7628,
+      "step": 13413
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.199863213480598e-05,
+      "loss": 0.7215,
+      "step": 13414
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.199615535488062e-05,
+      "loss": 0.7089,
+      "step": 13415
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.199367850045044e-05,
+      "loss": 0.791,
+      "step": 13416
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1991201571541846e-05,
+      "loss": 0.7071,
+      "step": 13417
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.19887245681812e-05,
+      "loss": 0.77,
+      "step": 13418
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1986247490394865e-05,
+      "loss": 0.7685,
+      "step": 13419
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1983770338209246e-05,
+      "loss": 0.717,
+      "step": 13420
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.198129311165071e-05,
+      "loss": 0.7498,
+      "step": 13421
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.197881581074564e-05,
+      "loss": 0.7665,
+      "step": 13422
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1976338435520426e-05,
+      "loss": 0.6911,
+      "step": 13423
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.197386098600144e-05,
+      "loss": 0.6933,
+      "step": 13424
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.197138346221507e-05,
+      "loss": 0.7509,
+      "step": 13425
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1968905864187694e-05,
+      "loss": 0.8289,
+      "step": 13426
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1966428191945705e-05,
+      "loss": 0.7859,
+      "step": 13427
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1963950445515475e-05,
+      "loss": 0.6675,
+      "step": 13428
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.196147262492342e-05,
+      "loss": 0.7744,
+      "step": 13429
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.195899473019589e-05,
+      "loss": 0.7523,
+      "step": 13430
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.19565167613593e-05,
+      "loss": 0.8184,
+      "step": 13431
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.195403871844003e-05,
+      "loss": 0.7039,
+      "step": 13432
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.195156060146446e-05,
+      "loss": 0.7593,
+      "step": 13433
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1949082410459e-05,
+      "loss": 0.8477,
+      "step": 13434
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.194660414545003e-05,
+      "loss": 0.7077,
+      "step": 13435
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1944125806463945e-05,
+      "loss": 0.6653,
+      "step": 13436
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.194164739352713e-05,
+      "loss": 0.7482,
+      "step": 13437
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.193916890666599e-05,
+      "loss": 0.7926,
+      "step": 13438
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1936690345906905e-05,
+      "loss": 0.7287,
+      "step": 13439
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1934211711276293e-05,
+      "loss": 0.7759,
+      "step": 13440
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.193173300280052e-05,
+      "loss": 0.7953,
+      "step": 13441
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1929254220506014e-05,
+      "loss": 0.8138,
+      "step": 13442
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1926775364419146e-05,
+      "loss": 0.6983,
+      "step": 13443
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1924296434566335e-05,
+      "loss": 0.8394,
+      "step": 13444
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1921817430973965e-05,
+      "loss": 0.6665,
+      "step": 13445
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.191933835366845e-05,
+      "loss": 0.789,
+      "step": 13446
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.191685920267617e-05,
+      "loss": 0.7714,
+      "step": 13447
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1914379978023546e-05,
+      "loss": 0.7457,
+      "step": 13448
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.191190067973697e-05,
+      "loss": 0.6541,
+      "step": 13449
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.190942130784284e-05,
+      "loss": 0.7307,
+      "step": 13450
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.190694186236759e-05,
+      "loss": 0.6934,
+      "step": 13451
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.190446234333759e-05,
+      "loss": 0.8767,
+      "step": 13452
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1901982750779245e-05,
+      "loss": 0.7757,
+      "step": 13453
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1899503084719e-05,
+      "loss": 0.7876,
+      "step": 13454
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1897023345183225e-05,
+      "loss": 0.7018,
+      "step": 13455
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.189454353219833e-05,
+      "loss": 0.8564,
+      "step": 13456
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.189206364579074e-05,
+      "loss": 0.8036,
+      "step": 13457
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.188958368598686e-05,
+      "loss": 0.7625,
+      "step": 13458
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.188710365281308e-05,
+      "loss": 0.7572,
+      "step": 13459
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.188462354629584e-05,
+      "loss": 0.6789,
+      "step": 13460
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.188214336646154e-05,
+      "loss": 0.7826,
+      "step": 13461
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.187966311333659e-05,
+      "loss": 0.7196,
+      "step": 13462
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.18771827869474e-05,
+      "loss": 0.7484,
+      "step": 13463
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.18747023873204e-05,
+      "loss": 0.7612,
+      "step": 13464
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.187222191448198e-05,
+      "loss": 0.7945,
+      "step": 13465
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.186974136845858e-05,
+      "loss": 0.7857,
+      "step": 13466
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.18672607492766e-05,
+      "loss": 0.6923,
+      "step": 13467
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.186478005696246e-05,
+      "loss": 0.7778,
+      "step": 13468
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1862299291542586e-05,
+      "loss": 0.828,
+      "step": 13469
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1859818453043384e-05,
+      "loss": 0.8175,
+      "step": 13470
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1857337541491275e-05,
+      "loss": 0.719,
+      "step": 13471
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1854856556912696e-05,
+      "loss": 0.8256,
+      "step": 13472
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.185237549933405e-05,
+      "loss": 0.7837,
+      "step": 13473
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.184989436878177e-05,
+      "loss": 0.6918,
+      "step": 13474
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.184741316528228e-05,
+      "loss": 0.7208,
+      "step": 13475
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1844931888861984e-05,
+      "loss": 0.7542,
+      "step": 13476
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1842450539547325e-05,
+      "loss": 0.7784,
+      "step": 13477
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.183996911736472e-05,
+      "loss": 0.8151,
+      "step": 13478
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.183748762234059e-05,
+      "loss": 0.7257,
+      "step": 13479
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.183500605450138e-05,
+      "loss": 0.7306,
+      "step": 13480
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.18325244138735e-05,
+      "loss": 0.8124,
+      "step": 13481
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.183004270048338e-05,
+      "loss": 0.7418,
+      "step": 13482
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1827560914357454e-05,
+      "loss": 0.7731,
+      "step": 13483
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.182507905552215e-05,
+      "loss": 0.7686,
+      "step": 13484
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.18225971240039e-05,
+      "loss": 0.7568,
+      "step": 13485
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.182011511982913e-05,
+      "loss": 0.6519,
+      "step": 13486
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.181763304302428e-05,
+      "loss": 0.7758,
+      "step": 13487
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.181515089361577e-05,
+      "loss": 0.8477,
+      "step": 13488
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.181266867163004e-05,
+      "loss": 0.8755,
+      "step": 13489
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.181018637709353e-05,
+      "loss": 0.8487,
+      "step": 13490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1807704010032666e-05,
+      "loss": 0.7866,
+      "step": 13491
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1805221570473894e-05,
+      "loss": 0.7429,
+      "step": 13492
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.180273905844364e-05,
+      "loss": 0.6818,
+      "step": 13493
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1800256473968345e-05,
+      "loss": 0.7126,
+      "step": 13494
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1797773817074444e-05,
+      "loss": 0.7616,
+      "step": 13495
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.179529108778838e-05,
+      "loss": 0.7291,
+      "step": 13496
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1792808286136586e-05,
+      "loss": 0.745,
+      "step": 13497
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.179032541214552e-05,
+      "loss": 0.7683,
+      "step": 13498
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.17878424658416e-05,
+      "loss": 0.8301,
+      "step": 13499
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.178535944725128e-05,
+      "loss": 0.7628,
+      "step": 13500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.178287635640101e-05,
+      "loss": 0.6848,
+      "step": 13501
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.178039319331722e-05,
+      "loss": 0.7229,
+      "step": 13502
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.177790995802635e-05,
+      "loss": 0.8015,
+      "step": 13503
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.177542665055487e-05,
+      "loss": 0.748,
+      "step": 13504
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.17729432709292e-05,
+      "loss": 0.7252,
+      "step": 13505
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.17704598191758e-05,
+      "loss": 0.7156,
+      "step": 13506
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.176797629532112e-05,
+      "loss": 0.6658,
+      "step": 13507
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1765492699391585e-05,
+      "loss": 0.783,
+      "step": 13508
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.176300903141367e-05,
+      "loss": 0.7408,
+      "step": 13509
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1760525291413815e-05,
+      "loss": 0.7339,
+      "step": 13510
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.175804147941847e-05,
+      "loss": 0.7288,
+      "step": 13511
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.175555759545408e-05,
+      "loss": 0.7409,
+      "step": 13512
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.17530736395471e-05,
+      "loss": 0.8197,
+      "step": 13513
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1750589611724e-05,
+      "loss": 0.7147,
+      "step": 13514
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.174810551201122e-05,
+      "loss": 0.7257,
+      "step": 13515
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.174562134043519e-05,
+      "loss": 0.6708,
+      "step": 13516
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1743137097022405e-05,
+      "loss": 0.7717,
+      "step": 13517
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.174065278179931e-05,
+      "loss": 0.7313,
+      "step": 13518
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.173816839479234e-05,
+      "loss": 0.6448,
+      "step": 13519
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.173568393602798e-05,
+      "loss": 0.7642,
+      "step": 13520
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1733199405532663e-05,
+      "loss": 0.7307,
+      "step": 13521
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1730714803332874e-05,
+      "loss": 0.7343,
+      "step": 13522
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.172823012945505e-05,
+      "loss": 0.7962,
+      "step": 13523
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.172574538392566e-05,
+      "loss": 0.8536,
+      "step": 13524
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.172326056677117e-05,
+      "loss": 0.754,
+      "step": 13525
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1720775678018034e-05,
+      "loss": 0.696,
+      "step": 13526
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1718290717692714e-05,
+      "loss": 0.738,
+      "step": 13527
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.171580568582168e-05,
+      "loss": 0.7067,
+      "step": 13528
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1713320582431396e-05,
+      "loss": 0.7892,
+      "step": 13529
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1710835407548314e-05,
+      "loss": 0.6912,
+      "step": 13530
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1708350161198914e-05,
+      "loss": 0.6995,
+      "step": 13531
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.170586484340966e-05,
+      "loss": 0.7327,
+      "step": 13532
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.170337945420701e-05,
+      "loss": 0.781,
+      "step": 13533
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.170089399361744e-05,
+      "loss": 0.7398,
+      "step": 13534
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.169840846166743e-05,
+      "loss": 0.7464,
+      "step": 13535
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.169592285838343e-05,
+      "loss": 0.8133,
+      "step": 13536
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1693437183791904e-05,
+      "loss": 0.7511,
+      "step": 13537
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.169095143791935e-05,
+      "loss": 0.6952,
+      "step": 13538
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.168846562079223e-05,
+      "loss": 0.7572,
+      "step": 13539
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1685979732436996e-05,
+      "loss": 0.7819,
+      "step": 13540
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.168349377288014e-05,
+      "loss": 0.7733,
+      "step": 13541
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.168100774214814e-05,
+      "loss": 0.7268,
+      "step": 13542
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.167852164026746e-05,
+      "loss": 0.7963,
+      "step": 13543
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1676035467264575e-05,
+      "loss": 0.7293,
+      "step": 13544
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1673549223165977e-05,
+      "loss": 0.7712,
+      "step": 13545
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.167106290799812e-05,
+      "loss": 0.8482,
+      "step": 13546
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16685765217875e-05,
+      "loss": 0.8158,
+      "step": 13547
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1666090064560585e-05,
+      "loss": 0.7904,
+      "step": 13548
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.166360353634386e-05,
+      "loss": 0.7801,
+      "step": 13549
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16611169371638e-05,
+      "loss": 0.7693,
+      "step": 13550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1658630267046894e-05,
+      "loss": 0.6791,
+      "step": 13551
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.165614352601961e-05,
+      "loss": 0.6912,
+      "step": 13552
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.165365671410845e-05,
+      "loss": 0.7672,
+      "step": 13553
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1651169831339874e-05,
+      "loss": 0.8458,
+      "step": 13554
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.164868287774039e-05,
+      "loss": 0.6914,
+      "step": 13555
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.164619585333646e-05,
+      "loss": 0.8506,
+      "step": 13556
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.164370875815458e-05,
+      "loss": 0.7293,
+      "step": 13557
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.164122159222124e-05,
+      "loss": 0.7548,
+      "step": 13558
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1638734355562915e-05,
+      "loss": 0.7218,
+      "step": 13559
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16362470482061e-05,
+      "loss": 0.8377,
+      "step": 13560
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.163375967017729e-05,
+      "loss": 0.7769,
+      "step": 13561
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.163127222150296e-05,
+      "loss": 0.7059,
+      "step": 13562
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16287847022096e-05,
+      "loss": 0.7503,
+      "step": 13563
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.162629711232372e-05,
+      "loss": 0.7564,
+      "step": 13564
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16238094518718e-05,
+      "loss": 0.7432,
+      "step": 13565
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1621321720880325e-05,
+      "loss": 0.7546,
+      "step": 13566
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16188339193758e-05,
+      "loss": 0.7259,
+      "step": 13567
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.161634604738471e-05,
+      "loss": 0.7764,
+      "step": 13568
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.161385810493354e-05,
+      "loss": 0.7214,
+      "step": 13569
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.161137009204881e-05,
+      "loss": 0.6935,
+      "step": 13570
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.160888200875699e-05,
+      "loss": 0.7749,
+      "step": 13571
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16063938550846e-05,
+      "loss": 0.7039,
+      "step": 13572
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.160390563105813e-05,
+      "loss": 0.7362,
+      "step": 13573
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1601417336704065e-05,
+      "loss": 0.7508,
+      "step": 13574
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1598928972048915e-05,
+      "loss": 0.7097,
+      "step": 13575
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1596440537119175e-05,
+      "loss": 0.7279,
+      "step": 13576
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.159395203194135e-05,
+      "loss": 0.7529,
+      "step": 13577
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.159146345654194e-05,
+      "loss": 0.7456,
+      "step": 13578
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1588974810947454e-05,
+      "loss": 0.6905,
+      "step": 13579
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.158648609518437e-05,
+      "loss": 0.7097,
+      "step": 13580
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.158399730927922e-05,
+      "loss": 0.7553,
+      "step": 13581
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.158150845325849e-05,
+      "loss": 0.7463,
+      "step": 13582
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.157901952714869e-05,
+      "loss": 0.6982,
+      "step": 13583
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.157653053097633e-05,
+      "loss": 0.7202,
+      "step": 13584
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.157404146476792e-05,
+      "loss": 0.7454,
+      "step": 13585
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.157155232854995e-05,
+      "loss": 0.7323,
+      "step": 13586
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.156906312234894e-05,
+      "loss": 0.7423,
+      "step": 13587
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.156657384619139e-05,
+      "loss": 0.7049,
+      "step": 13588
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1564084500103825e-05,
+      "loss": 0.726,
+      "step": 13589
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1561595084112736e-05,
+      "loss": 0.7701,
+      "step": 13590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1559105598244654e-05,
+      "loss": 0.7262,
+      "step": 13591
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.155661604252607e-05,
+      "loss": 0.7622,
+      "step": 13592
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1554126416983506e-05,
+      "loss": 0.694,
+      "step": 13593
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.155163672164348e-05,
+      "loss": 0.7878,
+      "step": 13594
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.15491469565325e-05,
+      "loss": 0.675,
+      "step": 13595
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.154665712167708e-05,
+      "loss": 0.723,
+      "step": 13596
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.154416721710373e-05,
+      "loss": 0.722,
+      "step": 13597
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.154167724283898e-05,
+      "loss": 0.8604,
+      "step": 13598
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.153918719890934e-05,
+      "loss": 0.7249,
+      "step": 13599
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.153669708534132e-05,
+      "loss": 0.7843,
+      "step": 13600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1534206902161446e-05,
+      "loss": 0.7211,
+      "step": 13601
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.153171664939624e-05,
+      "loss": 0.7118,
+      "step": 13602
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.152922632707221e-05,
+      "loss": 0.783,
+      "step": 13603
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.152673593521589e-05,
+      "loss": 0.8398,
+      "step": 13604
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.15242454738538e-05,
+      "loss": 0.7086,
+      "step": 13605
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.152175494301244e-05,
+      "loss": 0.7263,
+      "step": 13606
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1519264342718356e-05,
+      "loss": 0.784,
+      "step": 13607
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.151677367299807e-05,
+      "loss": 0.7782,
+      "step": 13608
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.15142829338781e-05,
+      "loss": 0.6881,
+      "step": 13609
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.151179212538496e-05,
+      "loss": 0.7784,
+      "step": 13610
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1509301247545205e-05,
+      "loss": 0.7812,
+      "step": 13611
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.150681030038533e-05,
+      "loss": 0.7315,
+      "step": 13612
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.150431928393188e-05,
+      "loss": 0.7699,
+      "step": 13613
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.150182819821138e-05,
+      "loss": 0.7534,
+      "step": 13614
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.149933704325036e-05,
+      "loss": 0.7587,
+      "step": 13615
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.149684581907535e-05,
+      "loss": 0.6935,
+      "step": 13616
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.149435452571286e-05,
+      "loss": 0.7209,
+      "step": 13617
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.149186316318945e-05,
+      "loss": 0.7553,
+      "step": 13618
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1489371731531634e-05,
+      "loss": 0.7031,
+      "step": 13619
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1486880230765944e-05,
+      "loss": 0.7466,
+      "step": 13620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.148438866091893e-05,
+      "loss": 0.8168,
+      "step": 13621
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.14818970220171e-05,
+      "loss": 0.7146,
+      "step": 13622
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1479405314087016e-05,
+      "loss": 0.7488,
+      "step": 13623
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.147691353715519e-05,
+      "loss": 0.7279,
+      "step": 13624
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1474421691248174e-05,
+      "loss": 0.7275,
+      "step": 13625
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.147192977639249e-05,
+      "loss": 0.7346,
+      "step": 13626
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.146943779261469e-05,
+      "loss": 0.7448,
+      "step": 13627
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.14669457399413e-05,
+      "loss": 0.7826,
+      "step": 13628
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.146445361839886e-05,
+      "loss": 0.7204,
+      "step": 13629
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.146196142801392e-05,
+      "loss": 0.7891,
+      "step": 13630
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.145946916881302e-05,
+      "loss": 0.826,
+      "step": 13631
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.145697684082268e-05,
+      "loss": 0.7911,
+      "step": 13632
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.145448444406947e-05,
+      "loss": 0.7682,
+      "step": 13633
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.145199197857991e-05,
+      "loss": 0.6804,
+      "step": 13634
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.144949944438056e-05,
+      "loss": 0.7573,
+      "step": 13635
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1447006841497955e-05,
+      "loss": 0.7092,
+      "step": 13636
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.144451416995863e-05,
+      "loss": 0.7833,
+      "step": 13637
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.144202142978916e-05,
+      "loss": 0.788,
+      "step": 13638
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.143952862101606e-05,
+      "loss": 0.7729,
+      "step": 13639
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.14370357436659e-05,
+      "loss": 0.7935,
+      "step": 13640
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.14345427977652e-05,
+      "loss": 0.6958,
+      "step": 13641
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.143204978334054e-05,
+      "loss": 0.8404,
+      "step": 13642
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.142955670041846e-05,
+      "loss": 0.7749,
+      "step": 13643
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.142706354902549e-05,
+      "loss": 0.8136,
+      "step": 13644
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1424570329188196e-05,
+      "loss": 0.7155,
+      "step": 13645
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.142207704093314e-05,
+      "loss": 0.7016,
+      "step": 13646
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1419583684286856e-05,
+      "loss": 0.7734,
+      "step": 13647
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1417090259275905e-05,
+      "loss": 0.7532,
+      "step": 13648
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.141459676592683e-05,
+      "loss": 0.7661,
+      "step": 13649
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.14121032042662e-05,
+      "loss": 0.7714,
+      "step": 13650
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.140960957432057e-05,
+      "loss": 0.8105,
+      "step": 13651
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1407115876116475e-05,
+      "loss": 0.7927,
+      "step": 13652
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1404622109680495e-05,
+      "loss": 0.7983,
+      "step": 13653
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1402128275039186e-05,
+      "loss": 0.697,
+      "step": 13654
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1399634372219084e-05,
+      "loss": 0.8325,
+      "step": 13655
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.139714040124677e-05,
+      "loss": 0.7157,
+      "step": 13656
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.139464636214879e-05,
+      "loss": 0.8612,
+      "step": 13657
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.139215225495171e-05,
+      "loss": 0.7398,
+      "step": 13658
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.138965807968209e-05,
+      "loss": 0.7238,
+      "step": 13659
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.138716383636649e-05,
+      "loss": 0.8071,
+      "step": 13660
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.138466952503147e-05,
+      "loss": 0.8291,
+      "step": 13661
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1382175145703596e-05,
+      "loss": 0.7428,
+      "step": 13662
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1379680698409435e-05,
+      "loss": 0.7734,
+      "step": 13663
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.137718618317555e-05,
+      "loss": 0.7183,
+      "step": 13664
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1374691600028496e-05,
+      "loss": 0.6846,
+      "step": 13665
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.137219694899485e-05,
+      "loss": 0.7597,
+      "step": 13666
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.136970223010118e-05,
+      "loss": 0.7561,
+      "step": 13667
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.136720744337405e-05,
+      "loss": 0.8028,
+      "step": 13668
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.136471258884001e-05,
+      "loss": 0.7743,
+      "step": 13669
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1362217666525655e-05,
+      "loss": 0.7967,
+      "step": 13670
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.135972267645755e-05,
+      "loss": 0.7682,
+      "step": 13671
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.135722761866225e-05,
+      "loss": 0.7058,
+      "step": 13672
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.135473249316634e-05,
+      "loss": 0.7464,
+      "step": 13673
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.135223729999639e-05,
+      "loss": 0.7198,
+      "step": 13674
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1349742039178964e-05,
+      "loss": 0.7319,
+      "step": 13675
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1347246710740646e-05,
+      "loss": 0.8171,
+      "step": 13676
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1344751314708e-05,
+      "loss": 0.8042,
+      "step": 13677
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1342255851107606e-05,
+      "loss": 0.764,
+      "step": 13678
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.133976031996604e-05,
+      "loss": 0.7757,
+      "step": 13679
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.133726472130987e-05,
+      "loss": 0.7745,
+      "step": 13680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.133476905516568e-05,
+      "loss": 0.8016,
+      "step": 13681
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1332273321560044e-05,
+      "loss": 0.7756,
+      "step": 13682
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1329777520519544e-05,
+      "loss": 0.7685,
+      "step": 13683
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.132728165207075e-05,
+      "loss": 0.7007,
+      "step": 13684
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.132478571624026e-05,
+      "loss": 0.6889,
+      "step": 13685
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1322289713054636e-05,
+      "loss": 0.7266,
+      "step": 13686
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.131979364254047e-05,
+      "loss": 0.7467,
+      "step": 13687
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.131729750472433e-05,
+      "loss": 0.7192,
+      "step": 13688
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1314801299632806e-05,
+      "loss": 0.6592,
+      "step": 13689
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.131230502729249e-05,
+      "loss": 0.736,
+      "step": 13690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1309808687729955e-05,
+      "loss": 0.7784,
+      "step": 13691
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.130731228097179e-05,
+      "loss": 0.8234,
+      "step": 13692
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.130481580704457e-05,
+      "loss": 0.7811,
+      "step": 13693
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.13023192659749e-05,
+      "loss": 0.7456,
+      "step": 13694
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.129982265778935e-05,
+      "loss": 0.7441,
+      "step": 13695
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1297325982514516e-05,
+      "loss": 0.7202,
+      "step": 13696
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.129482924017698e-05,
+      "loss": 0.7483,
+      "step": 13697
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1292332430803335e-05,
+      "loss": 0.6912,
+      "step": 13698
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1289835554420166e-05,
+      "loss": 0.7282,
+      "step": 13699
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.128733861105407e-05,
+      "loss": 0.7456,
+      "step": 13700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.128484160073164e-05,
+      "loss": 0.7144,
+      "step": 13701
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1282344523479464e-05,
+      "loss": 0.7223,
+      "step": 13702
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.127984737932412e-05,
+      "loss": 0.7746,
+      "step": 13703
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.127735016829222e-05,
+      "loss": 0.7168,
+      "step": 13704
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.127485289041036e-05,
+      "loss": 0.7927,
+      "step": 13705
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.127235554570513e-05,
+      "loss": 0.7636,
+      "step": 13706
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.126985813420311e-05,
+      "loss": 0.7015,
+      "step": 13707
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1267360655930905e-05,
+      "loss": 0.7018,
+      "step": 13708
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.126486311091512e-05,
+      "loss": 0.7372,
+      "step": 13709
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.126236549918234e-05,
+      "loss": 0.79,
+      "step": 13710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1259867820759176e-05,
+      "loss": 0.7377,
+      "step": 13711
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1257370075672224e-05,
+      "loss": 0.6721,
+      "step": 13712
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1254872263948065e-05,
+      "loss": 0.8035,
+      "step": 13713
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.125237438561332e-05,
+      "loss": 0.7258,
+      "step": 13714
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1249876440694595e-05,
+      "loss": 0.6775,
+      "step": 13715
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.124737842921847e-05,
+      "loss": 0.9004,
+      "step": 13716
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.124488035121156e-05,
+      "loss": 0.7424,
+      "step": 13717
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.124238220670046e-05,
+      "loss": 0.7533,
+      "step": 13718
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1239883995711786e-05,
+      "loss": 0.7908,
+      "step": 13719
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.123738571827214e-05,
+      "loss": 0.7614,
+      "step": 13720
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.123488737440811e-05,
+      "loss": 0.7084,
+      "step": 13721
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.123238896414633e-05,
+      "loss": 0.7378,
+      "step": 13722
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.122989048751338e-05,
+      "loss": 0.7306,
+      "step": 13723
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.122739194453588e-05,
+      "loss": 0.7153,
+      "step": 13724
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1224893335240435e-05,
+      "loss": 0.7622,
+      "step": 13725
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1222394659653665e-05,
+      "loss": 0.6911,
+      "step": 13726
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.121989591780215e-05,
+      "loss": 0.7729,
+      "step": 13727
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.121739710971254e-05,
+      "loss": 0.7907,
+      "step": 13728
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.121489823541142e-05,
+      "loss": 0.696,
+      "step": 13729
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1212399294925404e-05,
+      "loss": 0.7552,
+      "step": 13730
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1209900288281105e-05,
+      "loss": 0.6701,
+      "step": 13731
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.120740121550514e-05,
+      "loss": 0.6811,
+      "step": 13732
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.120490207662412e-05,
+      "loss": 0.7216,
+      "step": 13733
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.120240287166467e-05,
+      "loss": 0.8122,
+      "step": 13734
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1199903600653385e-05,
+      "loss": 0.7751,
+      "step": 13735
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.119740426361689e-05,
+      "loss": 0.8143,
+      "step": 13736
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.119490486058181e-05,
+      "loss": 0.7858,
+      "step": 13737
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.119240539157474e-05,
+      "loss": 0.7664,
+      "step": 13738
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.118990585662233e-05,
+      "loss": 0.7567,
+      "step": 13739
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1187406255751165e-05,
+      "loss": 0.794,
+      "step": 13740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.118490658898789e-05,
+      "loss": 0.7571,
+      "step": 13741
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.118240685635911e-05,
+      "loss": 0.7219,
+      "step": 13742
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.117990705789145e-05,
+      "loss": 0.7507,
+      "step": 13743
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.117740719361153e-05,
+      "loss": 0.7473,
+      "step": 13744
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1174907263545985e-05,
+      "loss": 0.7049,
+      "step": 13745
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.117240726772142e-05,
+      "loss": 0.7135,
+      "step": 13746
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.116990720616446e-05,
+      "loss": 0.7197,
+      "step": 13747
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.116740707890174e-05,
+      "loss": 0.8381,
+      "step": 13748
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.116490688595988e-05,
+      "loss": 0.8877,
+      "step": 13749
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.116240662736549e-05,
+      "loss": 0.7639,
+      "step": 13750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.115990630314523e-05,
+      "loss": 0.6803,
+      "step": 13751
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.11574059133257e-05,
+      "loss": 0.7615,
+      "step": 13752
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1154905457933525e-05,
+      "loss": 0.6946,
+      "step": 13753
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.115240493699536e-05,
+      "loss": 0.7489,
+      "step": 13754
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.114990435053781e-05,
+      "loss": 0.7052,
+      "step": 13755
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.114740369858751e-05,
+      "loss": 0.7052,
+      "step": 13756
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1144902981171103e-05,
+      "loss": 0.7939,
+      "step": 13757
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.11424021983152e-05,
+      "loss": 0.7968,
+      "step": 13758
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.113990135004645e-05,
+      "loss": 0.7451,
+      "step": 13759
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1137400436391474e-05,
+      "loss": 0.7534,
+      "step": 13760
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.113489945737691e-05,
+      "loss": 0.7107,
+      "step": 13761
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.113239841302939e-05,
+      "loss": 0.7452,
+      "step": 13762
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.112989730337554e-05,
+      "loss": 0.707,
+      "step": 13763
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.112739612844202e-05,
+      "loss": 0.7543,
+      "step": 13764
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1124894888255447e-05,
+      "loss": 0.7774,
+      "step": 13765
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.112239358284246e-05,
+      "loss": 0.8337,
+      "step": 13766
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.111989221222971e-05,
+      "loss": 0.7166,
+      "step": 13767
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.111739077644381e-05,
+      "loss": 0.6896,
+      "step": 13768
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.111488927551142e-05,
+      "loss": 0.7983,
+      "step": 13769
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1112387709459165e-05,
+      "loss": 0.7188,
+      "step": 13770
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.110988607831369e-05,
+      "loss": 0.7432,
+      "step": 13771
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.110738438210165e-05,
+      "loss": 0.7019,
+      "step": 13772
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.110488262084966e-05,
+      "loss": 0.8501,
+      "step": 13773
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1102380794584383e-05,
+      "loss": 0.7802,
+      "step": 13774
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.109987890333246e-05,
+      "loss": 0.7515,
+      "step": 13775
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.109737694712052e-05,
+      "loss": 0.7836,
+      "step": 13776
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1094874925975234e-05,
+      "loss": 0.7388,
+      "step": 13777
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1092372839923214e-05,
+      "loss": 0.7996,
+      "step": 13778
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.108987068899112e-05,
+      "loss": 0.7681,
+      "step": 13779
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.108736847320561e-05,
+      "loss": 0.6848,
+      "step": 13780
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.108486619259333e-05,
+      "loss": 0.734,
+      "step": 13781
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.10823638471809e-05,
+      "loss": 0.8277,
+      "step": 13782
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1079861436995e-05,
+      "loss": 0.7348,
+      "step": 13783
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.107735896206226e-05,
+      "loss": 0.7451,
+      "step": 13784
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1074856422409336e-05,
+      "loss": 0.7442,
+      "step": 13785
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1072353818062885e-05,
+      "loss": 0.6895,
+      "step": 13786
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.106985114904955e-05,
+      "loss": 0.7955,
+      "step": 13787
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.106734841539599e-05,
+      "loss": 0.7344,
+      "step": 13788
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.106484561712884e-05,
+      "loss": 0.8107,
+      "step": 13789
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.106234275427477e-05,
+      "loss": 0.778,
+      "step": 13790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.105983982686044e-05,
+      "loss": 0.803,
+      "step": 13791
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.105733683491249e-05,
+      "loss": 0.7341,
+      "step": 13792
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.105483377845758e-05,
+      "loss": 0.7894,
+      "step": 13793
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1052330657522366e-05,
+      "loss": 0.684,
+      "step": 13794
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.104982747213351e-05,
+      "loss": 0.7212,
+      "step": 13795
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.104732422231765e-05,
+      "loss": 0.7329,
+      "step": 13796
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.104482090810148e-05,
+      "loss": 0.738,
+      "step": 13797
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.104231752951163e-05,
+      "loss": 0.8012,
+      "step": 13798
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1039814086574755e-05,
+      "loss": 0.7738,
+      "step": 13799
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1037310579317535e-05,
+      "loss": 0.7878,
+      "step": 13800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.103480700776662e-05,
+      "loss": 0.7528,
+      "step": 13801
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.103230337194867e-05,
+      "loss": 0.6877,
+      "step": 13802
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.102979967189036e-05,
+      "loss": 0.8128,
+      "step": 13803
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.102729590761835e-05,
+      "loss": 0.7091,
+      "step": 13804
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.102479207915928e-05,
+      "loss": 0.7676,
+      "step": 13805
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.102228818653985e-05,
+      "loss": 0.8049,
+      "step": 13806
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.101978422978669e-05,
+      "loss": 0.7783,
+      "step": 13807
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1017280208926494e-05,
+      "loss": 0.6794,
+      "step": 13808
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.101477612398592e-05,
+      "loss": 0.7148,
+      "step": 13809
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.101227197499162e-05,
+      "loss": 0.6834,
+      "step": 13810
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.100976776197028e-05,
+      "loss": 0.8209,
+      "step": 13811
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1007263484948564e-05,
+      "loss": 0.7384,
+      "step": 13812
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1004759143953134e-05,
+      "loss": 0.7728,
+      "step": 13813
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1002254739010667e-05,
+      "loss": 0.6492,
+      "step": 13814
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0999750270147834e-05,
+      "loss": 0.7939,
+      "step": 13815
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.09972457373913e-05,
+      "loss": 0.708,
+      "step": 13816
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0994741140767744e-05,
+      "loss": 0.7579,
+      "step": 13817
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.099223648030383e-05,
+      "loss": 0.727,
+      "step": 13818
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.098973175602623e-05,
+      "loss": 0.7179,
+      "step": 13819
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.098722696796163e-05,
+      "loss": 0.7882,
+      "step": 13820
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.09847221161367e-05,
+      "loss": 0.6654,
+      "step": 13821
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.098221720057811e-05,
+      "loss": 0.7296,
+      "step": 13822
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0979712221312535e-05,
+      "loss": 0.7966,
+      "step": 13823
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.097720717836666e-05,
+      "loss": 0.762,
+      "step": 13824
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0974702071767156e-05,
+      "loss": 0.761,
+      "step": 13825
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.09721969015407e-05,
+      "loss": 0.8073,
+      "step": 13826
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0969691667713976e-05,
+      "loss": 0.832,
+      "step": 13827
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.096718637031366e-05,
+      "loss": 0.7363,
+      "step": 13828
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0964681009366436e-05,
+      "loss": 0.7864,
+      "step": 13829
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0962175584898976e-05,
+      "loss": 0.8169,
+      "step": 13830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.095967009693796e-05,
+      "loss": 0.7366,
+      "step": 13831
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.095716454551008e-05,
+      "loss": 0.82,
+      "step": 13832
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.095465893064202e-05,
+      "loss": 0.7798,
+      "step": 13833
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.095215325236045e-05,
+      "loss": 0.7739,
+      "step": 13834
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.094964751069206e-05,
+      "loss": 0.6896,
+      "step": 13835
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.094714170566355e-05,
+      "loss": 0.7603,
+      "step": 13836
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.094463583730158e-05,
+      "loss": 0.7553,
+      "step": 13837
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.094212990563285e-05,
+      "loss": 0.7625,
+      "step": 13838
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.093962391068404e-05,
+      "loss": 0.6642,
+      "step": 13839
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.093711785248185e-05,
+      "loss": 0.8189,
+      "step": 13840
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.093461173105295e-05,
+      "loss": 0.7439,
+      "step": 13841
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.093210554642404e-05,
+      "loss": 0.7677,
+      "step": 13842
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.092959929862181e-05,
+      "loss": 0.7255,
+      "step": 13843
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.092709298767295e-05,
+      "loss": 0.6739,
+      "step": 13844
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.092458661360414e-05,
+      "loss": 0.7457,
+      "step": 13845
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0922080176442084e-05,
+      "loss": 0.8521,
+      "step": 13846
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.091957367621348e-05,
+      "loss": 0.7546,
+      "step": 13847
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0917067112945e-05,
+      "loss": 0.7153,
+      "step": 13848
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.091456048666334e-05,
+      "loss": 0.6922,
+      "step": 13849
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.091205379739522e-05,
+      "loss": 0.6855,
+      "step": 13850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0909547045167295e-05,
+      "loss": 0.7941,
+      "step": 13851
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.09070402300063e-05,
+      "loss": 0.7406,
+      "step": 13852
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.09045333519389e-05,
+      "loss": 0.7339,
+      "step": 13853
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.090202641099181e-05,
+      "loss": 0.757,
+      "step": 13854
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0899519407191726e-05,
+      "loss": 0.7079,
+      "step": 13855
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.089701234056534e-05,
+      "loss": 0.712,
+      "step": 13856
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.089450521113935e-05,
+      "loss": 0.6943,
+      "step": 13857
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.089199801894046e-05,
+      "loss": 0.7533,
+      "step": 13858
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.088949076399537e-05,
+      "loss": 0.6684,
+      "step": 13859
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.088698344633077e-05,
+      "loss": 0.7811,
+      "step": 13860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.088447606597338e-05,
+      "loss": 0.7174,
+      "step": 13861
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0881968622949886e-05,
+      "loss": 0.7281,
+      "step": 13862
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0879461117287e-05,
+      "loss": 0.7553,
+      "step": 13863
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0876953549011426e-05,
+      "loss": 0.7691,
+      "step": 13864
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.087444591814986e-05,
+      "loss": 0.7179,
+      "step": 13865
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.087193822472901e-05,
+      "loss": 0.7002,
+      "step": 13866
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.086943046877559e-05,
+      "loss": 0.6736,
+      "step": 13867
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0866922650316295e-05,
+      "loss": 0.7128,
+      "step": 13868
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0864414769377834e-05,
+      "loss": 0.7018,
+      "step": 13869
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0861906825986916e-05,
+      "loss": 0.7817,
+      "step": 13870
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.085939882017025e-05,
+      "loss": 0.7358,
+      "step": 13871
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.085689075195454e-05,
+      "loss": 0.7386,
+      "step": 13872
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.085438262136651e-05,
+      "loss": 0.7314,
+      "step": 13873
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.085187442843285e-05,
+      "loss": 0.7702,
+      "step": 13874
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.084936617318028e-05,
+      "loss": 0.7343,
+      "step": 13875
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.084685785563552e-05,
+      "loss": 0.7204,
+      "step": 13876
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0844349475825264e-05,
+      "loss": 0.703,
+      "step": 13877
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0841841033776244e-05,
+      "loss": 0.7649,
+      "step": 13878
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.083933252951516e-05,
+      "loss": 0.7187,
+      "step": 13879
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.083682396306872e-05,
+      "loss": 0.7596,
+      "step": 13880
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0834315334463656e-05,
+      "loss": 0.7683,
+      "step": 13881
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0831806643726685e-05,
+      "loss": 0.7111,
+      "step": 13882
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.08292978908845e-05,
+      "loss": 0.7001,
+      "step": 13883
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.082678907596384e-05,
+      "loss": 0.7855,
+      "step": 13884
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0824280198991405e-05,
+      "loss": 0.7199,
+      "step": 13885
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.082177125999393e-05,
+      "loss": 0.8732,
+      "step": 13886
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.081926225899812e-05,
+      "loss": 0.847,
+      "step": 13887
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.08167531960307e-05,
+      "loss": 0.7538,
+      "step": 13888
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.08142440711184e-05,
+      "loss": 0.7549,
+      "step": 13889
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0811734884287924e-05,
+      "loss": 0.699,
+      "step": 13890
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.080922563556599e-05,
+      "loss": 0.6876,
+      "step": 13891
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0806716324979343e-05,
+      "loss": 0.7105,
+      "step": 13892
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0804206952554694e-05,
+      "loss": 0.6987,
+      "step": 13893
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.080169751831875e-05,
+      "loss": 0.6615,
+      "step": 13894
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.079918802229826e-05,
+      "loss": 0.7362,
+      "step": 13895
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.079667846451994e-05,
+      "loss": 0.842,
+      "step": 13896
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.079416884501051e-05,
+      "loss": 0.7217,
+      "step": 13897
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0791659163796704e-05,
+      "loss": 0.7891,
+      "step": 13898
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.078914942090524e-05,
+      "loss": 0.7642,
+      "step": 13899
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0786639616362844e-05,
+      "loss": 0.6902,
+      "step": 13900
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.078412975019626e-05,
+      "loss": 0.763,
+      "step": 13901
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.07816198224322e-05,
+      "loss": 0.7581,
+      "step": 13902
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0779109833097395e-05,
+      "loss": 0.7239,
+      "step": 13903
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0776599782218586e-05,
+      "loss": 0.7402,
+      "step": 13904
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.07740896698225e-05,
+      "loss": 0.7384,
+      "step": 13905
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.077157949593585e-05,
+      "loss": 0.7191,
+      "step": 13906
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0769069260585395e-05,
+      "loss": 0.8333,
+      "step": 13907
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0766558963797845e-05,
+      "loss": 0.6899,
+      "step": 13908
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.076404860559995e-05,
+      "loss": 0.7557,
+      "step": 13909
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.076153818601844e-05,
+      "loss": 0.6677,
+      "step": 13910
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.075902770508004e-05,
+      "loss": 0.7254,
+      "step": 13911
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.075651716281149e-05,
+      "loss": 0.7253,
+      "step": 13912
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.075400655923953e-05,
+      "loss": 0.6947,
+      "step": 13913
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0751495894390896e-05,
+      "loss": 0.6942,
+      "step": 13914
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.074898516829232e-05,
+      "loss": 0.7722,
+      "step": 13915
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.074647438097055e-05,
+      "loss": 0.7107,
+      "step": 13916
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0743963532452305e-05,
+      "loss": 0.7719,
+      "step": 13917
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.074145262276434e-05,
+      "loss": 0.7035,
+      "step": 13918
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.073894165193339e-05,
+      "loss": 0.7353,
+      "step": 13919
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0736430619986194e-05,
+      "loss": 0.7855,
+      "step": 13920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0733919526949496e-05,
+      "loss": 0.7884,
+      "step": 13921
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.073140837285004e-05,
+      "loss": 0.6927,
+      "step": 13922
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0728897157714555e-05,
+      "loss": 0.7427,
+      "step": 13923
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0726385881569804e-05,
+      "loss": 0.6847,
+      "step": 13924
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.072387454444251e-05,
+      "loss": 0.7769,
+      "step": 13925
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.072136314635943e-05,
+      "loss": 0.6972,
+      "step": 13926
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.071885168734731e-05,
+      "loss": 0.7129,
+      "step": 13927
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.071634016743289e-05,
+      "loss": 0.7542,
+      "step": 13928
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0713828586642924e-05,
+      "loss": 0.6922,
+      "step": 13929
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.071131694500414e-05,
+      "loss": 0.8171,
+      "step": 13930
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.070880524254331e-05,
+      "loss": 0.7722,
+      "step": 13931
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0706293479287164e-05,
+      "loss": 0.7507,
+      "step": 13932
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0703781655262456e-05,
+      "loss": 0.7495,
+      "step": 13933
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.070126977049594e-05,
+      "loss": 0.7838,
+      "step": 13934
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0698757825014354e-05,
+      "loss": 0.6978,
+      "step": 13935
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.069624581884446e-05,
+      "loss": 0.7341,
+      "step": 13936
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.069373375201301e-05,
+      "loss": 0.7757,
+      "step": 13937
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.069122162454675e-05,
+      "loss": 0.7444,
+      "step": 13938
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.068870943647244e-05,
+      "loss": 0.6983,
+      "step": 13939
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0686197187816826e-05,
+      "loss": 0.752,
+      "step": 13940
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.068368487860666e-05,
+      "loss": 0.7015,
+      "step": 13941
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.06811725088687e-05,
+      "loss": 0.7424,
+      "step": 13942
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.06786600786297e-05,
+      "loss": 0.6359,
+      "step": 13943
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0676147587916425e-05,
+      "loss": 0.7302,
+      "step": 13944
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0673635036755626e-05,
+      "loss": 0.7137,
+      "step": 13945
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.067112242517405e-05,
+      "loss": 0.6689,
+      "step": 13946
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0668609753198466e-05,
+      "loss": 0.7274,
+      "step": 13947
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.066609702085564e-05,
+      "loss": 0.7003,
+      "step": 13948
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0663584228172304e-05,
+      "loss": 0.7301,
+      "step": 13949
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0661071375175244e-05,
+      "loss": 0.7863,
+      "step": 13950
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.065855846189121e-05,
+      "loss": 0.7264,
+      "step": 13951
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.065604548834696e-05,
+      "loss": 0.7512,
+      "step": 13952
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0653532454569265e-05,
+      "loss": 0.7577,
+      "step": 13953
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.065101936058488e-05,
+      "loss": 0.728,
+      "step": 13954
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.064850620642056e-05,
+      "loss": 0.7662,
+      "step": 13955
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.064599299210309e-05,
+      "loss": 0.7362,
+      "step": 13956
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0643479717659223e-05,
+      "loss": 0.7464,
+      "step": 13957
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.064096638311572e-05,
+      "loss": 0.7974,
+      "step": 13958
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.063845298849935e-05,
+      "loss": 0.7073,
+      "step": 13959
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.063593953383687e-05,
+      "loss": 0.7227,
+      "step": 13960
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.063342601915507e-05,
+      "loss": 0.6343,
+      "step": 13961
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.06309124444807e-05,
+      "loss": 0.7786,
+      "step": 13962
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.062839880984052e-05,
+      "loss": 0.7468,
+      "step": 13963
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.062588511526132e-05,
+      "loss": 0.8027,
+      "step": 13964
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.062337136076986e-05,
+      "loss": 0.7469,
+      "step": 13965
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.06208575463929e-05,
+      "loss": 0.7031,
+      "step": 13966
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0618343672157224e-05,
+      "loss": 0.779,
+      "step": 13967
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.06158297380896e-05,
+      "loss": 0.7576,
+      "step": 13968
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.06133157442168e-05,
+      "loss": 0.7097,
+      "step": 13969
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.06108016905656e-05,
+      "loss": 0.7155,
+      "step": 13970
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0608287577162755e-05,
+      "loss": 0.7235,
+      "step": 13971
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.060577340403506e-05,
+      "loss": 0.7007,
+      "step": 13972
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.060325917120928e-05,
+      "loss": 0.7321,
+      "step": 13973
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.06007448787122e-05,
+      "loss": 0.7208,
+      "step": 13974
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.059823052657057e-05,
+      "loss": 0.7331,
+      "step": 13975
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0595716114811196e-05,
+      "loss": 0.6646,
+      "step": 13976
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.059320164346084e-05,
+      "loss": 0.7091,
+      "step": 13977
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.059068711254629e-05,
+      "loss": 0.8036,
+      "step": 13978
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0588172522094313e-05,
+      "loss": 0.695,
+      "step": 13979
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.058565787213168e-05,
+      "loss": 0.7135,
+      "step": 13980
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.058314316268519e-05,
+      "loss": 0.7367,
+      "step": 13981
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.058062839378161e-05,
+      "loss": 0.7448,
+      "step": 13982
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0578113565447733e-05,
+      "loss": 0.7616,
+      "step": 13983
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.057559867771033e-05,
+      "loss": 0.6967,
+      "step": 13984
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.057308373059619e-05,
+      "loss": 0.7059,
+      "step": 13985
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0570568724132086e-05,
+      "loss": 0.7272,
+      "step": 13986
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0568053658344806e-05,
+      "loss": 0.7472,
+      "step": 13987
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.056553853326113e-05,
+      "loss": 0.7047,
+      "step": 13988
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.056302334890786e-05,
+      "loss": 0.7668,
+      "step": 13989
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.056050810531177e-05,
+      "loss": 0.7833,
+      "step": 13990
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0557992802499634e-05,
+      "loss": 0.7804,
+      "step": 13991
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.055547744049825e-05,
+      "loss": 0.7212,
+      "step": 13992
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0552962019334414e-05,
+      "loss": 0.748,
+      "step": 13993
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.05504465390349e-05,
+      "loss": 0.7091,
+      "step": 13994
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.054793099962649e-05,
+      "loss": 0.7017,
+      "step": 13995
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.054541540113599e-05,
+      "loss": 0.6866,
+      "step": 13996
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0542899743590186e-05,
+      "loss": 0.74,
+      "step": 13997
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0540384027015865e-05,
+      "loss": 0.7474,
+      "step": 13998
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0537868251439816e-05,
+      "loss": 0.6898,
+      "step": 13999
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0535352416888835e-05,
+      "loss": 0.7533,
+      "step": 14000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.05328365233897e-05,
+      "loss": 0.7362,
+      "step": 14001
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0530320570969226e-05,
+      "loss": 0.6465,
+      "step": 14002
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.052780455965419e-05,
+      "loss": 0.7545,
+      "step": 14003
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.052528848947139e-05,
+      "loss": 0.8213,
+      "step": 14004
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.052277236044763e-05,
+      "loss": 0.685,
+      "step": 14005
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0520256172609695e-05,
+      "loss": 0.8494,
+      "step": 14006
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0517739925984374e-05,
+      "loss": 0.7594,
+      "step": 14007
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.051522362059848e-05,
+      "loss": 0.7542,
+      "step": 14008
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0512707256478808e-05,
+      "loss": 0.6968,
+      "step": 14009
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0510190833652148e-05,
+      "loss": 0.7082,
+      "step": 14010
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.05076743521453e-05,
+      "loss": 0.7712,
+      "step": 14011
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.050515781198505e-05,
+      "loss": 0.724,
+      "step": 14012
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.050264121319823e-05,
+      "loss": 0.7254,
+      "step": 14013
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0500124555811614e-05,
+      "loss": 0.7804,
+      "step": 14014
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0497607839852012e-05,
+      "loss": 0.7331,
+      "step": 14015
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0495091065346228e-05,
+      "loss": 0.7152,
+      "step": 14016
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.049257423232105e-05,
+      "loss": 0.7604,
+      "step": 14017
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0490057340803297e-05,
+      "loss": 0.7806,
+      "step": 14018
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0487540390819774e-05,
+      "loss": 0.7105,
+      "step": 14019
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.048502338239727e-05,
+      "loss": 0.7542,
+      "step": 14020
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0482506315562597e-05,
+      "loss": 0.8332,
+      "step": 14021
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.047998919034256e-05,
+      "loss": 0.7508,
+      "step": 14022
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.047747200676397e-05,
+      "loss": 0.793,
+      "step": 14023
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.047495476485362e-05,
+      "loss": 0.6992,
+      "step": 14024
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0472437464638332e-05,
+      "loss": 0.772,
+      "step": 14025
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0469920106144906e-05,
+      "loss": 0.7142,
+      "step": 14026
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0467402689400147e-05,
+      "loss": 0.7909,
+      "step": 14027
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0464885214430872e-05,
+      "loss": 0.681,
+      "step": 14028
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.046236768126389e-05,
+      "loss": 0.7981,
+      "step": 14029
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.045985008992601e-05,
+      "loss": 0.7318,
+      "step": 14030
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0457332440444036e-05,
+      "loss": 0.8041,
+      "step": 14031
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.045481473284478e-05,
+      "loss": 0.7805,
+      "step": 14032
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.045229696715507e-05,
+      "loss": 0.7196,
+      "step": 14033
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.044977914340171e-05,
+      "loss": 0.7185,
+      "step": 14034
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.04472612616115e-05,
+      "loss": 0.7382,
+      "step": 14035
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0444743321811264e-05,
+      "loss": 0.7243,
+      "step": 14036
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0442225324027822e-05,
+      "loss": 0.8626,
+      "step": 14037
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0439707268287988e-05,
+      "loss": 0.6962,
+      "step": 14038
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0437189154618574e-05,
+      "loss": 0.8583,
+      "step": 14039
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0434670983046394e-05,
+      "loss": 0.7141,
+      "step": 14040
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.043215275359826e-05,
+      "loss": 0.7637,
+      "step": 14041
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0429634466301006e-05,
+      "loss": 0.7773,
+      "step": 14042
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.042711612118144e-05,
+      "loss": 0.7599,
+      "step": 14043
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0424597718266374e-05,
+      "loss": 0.735,
+      "step": 14044
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.042207925758265e-05,
+      "loss": 0.8031,
+      "step": 14045
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.041956073915706e-05,
+      "loss": 0.7676,
+      "step": 14046
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.041704216301644e-05,
+      "loss": 0.7345,
+      "step": 14047
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0414523529187616e-05,
+      "loss": 0.77,
+      "step": 14048
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.041200483769741e-05,
+      "loss": 0.7151,
+      "step": 14049
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0409486088572625e-05,
+      "loss": 0.8348,
+      "step": 14050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0406967281840105e-05,
+      "loss": 0.718,
+      "step": 14051
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0404448417526664e-05,
+      "loss": 0.6816,
+      "step": 14052
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0401929495659127e-05,
+      "loss": 0.7061,
+      "step": 14053
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.039941051626432e-05,
+      "loss": 0.7443,
+      "step": 14054
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0396891479369073e-05,
+      "loss": 0.7408,
+      "step": 14055
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0394372385000203e-05,
+      "loss": 0.6685,
+      "step": 14056
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0391853233184546e-05,
+      "loss": 0.7187,
+      "step": 14057
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0389334023948922e-05,
+      "loss": 0.7133,
+      "step": 14058
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.038681475732017e-05,
+      "loss": 0.8285,
+      "step": 14059
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.038429543332511e-05,
+      "loss": 0.675,
+      "step": 14060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0381776051990563e-05,
+      "loss": 0.6514,
+      "step": 14061
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0379256613343376e-05,
+      "loss": 0.6427,
+      "step": 14062
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.037673711741037e-05,
+      "loss": 0.7215,
+      "step": 14063
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0374217564218378e-05,
+      "loss": 0.75,
+      "step": 14064
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.037169795379423e-05,
+      "loss": 0.7545,
+      "step": 14065
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0369178286164762e-05,
+      "loss": 0.7833,
+      "step": 14066
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.03666585613568e-05,
+      "loss": 0.7843,
+      "step": 14067
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.036413877939719e-05,
+      "loss": 0.711,
+      "step": 14068
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.036161894031276e-05,
+      "loss": 0.7717,
+      "step": 14069
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.035909904413034e-05,
+      "loss": 0.6366,
+      "step": 14070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0356579090876768e-05,
+      "loss": 0.7371,
+      "step": 14071
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0354059080578884e-05,
+      "loss": 0.7163,
+      "step": 14072
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0351539013263515e-05,
+      "loss": 0.8018,
+      "step": 14073
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.034901888895751e-05,
+      "loss": 0.7794,
+      "step": 14074
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.03464987076877e-05,
+      "loss": 0.8809,
+      "step": 14075
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0343978469480916e-05,
+      "loss": 0.7675,
+      "step": 14076
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.034145817436402e-05,
+      "loss": 0.7301,
+      "step": 14077
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0338937822363827e-05,
+      "loss": 0.7246,
+      "step": 14078
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.033641741350719e-05,
+      "loss": 0.6973,
+      "step": 14079
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0333896947820954e-05,
+      "loss": 0.7302,
+      "step": 14080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0331376425331946e-05,
+      "loss": 0.7771,
+      "step": 14081
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.032885584606702e-05,
+      "loss": 0.7546,
+      "step": 14082
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0326335210053014e-05,
+      "loss": 0.7616,
+      "step": 14083
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.032381451731676e-05,
+      "loss": 0.7485,
+      "step": 14084
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.032129376788513e-05,
+      "loss": 0.678,
+      "step": 14085
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0318772961784948e-05,
+      "loss": 0.7036,
+      "step": 14086
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0316252099043056e-05,
+      "loss": 0.6772,
+      "step": 14087
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0313731179686312e-05,
+      "loss": 0.7376,
+      "step": 14088
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0311210203741557e-05,
+      "loss": 0.713,
+      "step": 14089
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0308689171235645e-05,
+      "loss": 0.7444,
+      "step": 14090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0306168082195407e-05,
+      "loss": 0.7724,
+      "step": 14091
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0303646936647706e-05,
+      "loss": 0.7654,
+      "step": 14092
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0301125734619375e-05,
+      "loss": 0.7686,
+      "step": 14093
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0298604476137283e-05,
+      "loss": 0.7706,
+      "step": 14094
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0296083161228266e-05,
+      "loss": 0.7126,
+      "step": 14095
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0293561789919177e-05,
+      "loss": 0.7237,
+      "step": 14096
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0291040362236873e-05,
+      "loss": 0.7818,
+      "step": 14097
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.02885188782082e-05,
+      "loss": 0.7295,
+      "step": 14098
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0285997337860005e-05,
+      "loss": 0.7205,
+      "step": 14099
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0283475741219163e-05,
+      "loss": 0.7631,
+      "step": 14100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0280954088312496e-05,
+      "loss": 0.7904,
+      "step": 14101
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0278432379166888e-05,
+      "loss": 0.6959,
+      "step": 14102
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.027591061380917e-05,
+      "loss": 0.7407,
+      "step": 14103
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.027338879226621e-05,
+      "loss": 0.785,
+      "step": 14104
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.027086691456486e-05,
+      "loss": 0.7734,
+      "step": 14105
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.026834498073198e-05,
+      "loss": 0.8566,
+      "step": 14106
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0265822990794423e-05,
+      "loss": 0.7307,
+      "step": 14107
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0263300944779044e-05,
+      "loss": 0.7323,
+      "step": 14108
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0260778842712713e-05,
+      "loss": 0.7399,
+      "step": 14109
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0258256684622278e-05,
+      "loss": 0.7406,
+      "step": 14110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0255734470534604e-05,
+      "loss": 0.8604,
+      "step": 14111
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0253212200476544e-05,
+      "loss": 0.8764,
+      "step": 14112
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0250689874474956e-05,
+      "loss": 0.7255,
+      "step": 14113
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0248167492556722e-05,
+      "loss": 0.807,
+      "step": 14114
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.024564505474869e-05,
+      "loss": 0.773,
+      "step": 14115
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0243122561077713e-05,
+      "loss": 0.7637,
+      "step": 14116
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0240600011570673e-05,
+      "loss": 0.7889,
+      "step": 14117
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.023807740625442e-05,
+      "loss": 0.7302,
+      "step": 14118
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0235554745155826e-05,
+      "loss": 0.6795,
+      "step": 14119
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0233032028301757e-05,
+      "loss": 0.7539,
+      "step": 14120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0230509255719063e-05,
+      "loss": 0.7249,
+      "step": 14121
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0227986427434622e-05,
+      "loss": 0.7826,
+      "step": 14122
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.022546354347531e-05,
+      "loss": 0.6866,
+      "step": 14123
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0222940603867977e-05,
+      "loss": 0.7485,
+      "step": 14124
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.022041760863949e-05,
+      "loss": 0.8034,
+      "step": 14125
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0217894557816733e-05,
+      "loss": 0.6906,
+      "step": 14126
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0215371451426568e-05,
+      "loss": 0.7031,
+      "step": 14127
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.021284828949586e-05,
+      "loss": 0.7456,
+      "step": 14128
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.021032507205148e-05,
+      "loss": 0.7229,
+      "step": 14129
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.020780179912031e-05,
+      "loss": 0.717,
+      "step": 14130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.020527847072921e-05,
+      "loss": 0.7685,
+      "step": 14131
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0202755086905054e-05,
+      "loss": 0.7729,
+      "step": 14132
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0200231647674708e-05,
+      "loss": 0.7059,
+      "step": 14133
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0197708153065056e-05,
+      "loss": 0.7243,
+      "step": 14134
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.019518460310297e-05,
+      "loss": 0.7064,
+      "step": 14135
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.019266099781532e-05,
+      "loss": 0.7858,
+      "step": 14136
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0190137337228985e-05,
+      "loss": 0.8164,
+      "step": 14137
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0187613621370842e-05,
+      "loss": 0.776,
+      "step": 14138
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.018508985026775e-05,
+      "loss": 0.7466,
+      "step": 14139
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.018256602394662e-05,
+      "loss": 0.7895,
+      "step": 14140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0180042142434296e-05,
+      "loss": 0.8026,
+      "step": 14141
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.017751820575766e-05,
+      "loss": 0.6975,
+      "step": 14142
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0174994213943607e-05,
+      "loss": 0.7609,
+      "step": 14143
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.017247016701901e-05,
+      "loss": 0.8464,
+      "step": 14144
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.016994606501074e-05,
+      "loss": 0.7477,
+      "step": 14145
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.016742190794569e-05,
+      "loss": 0.7128,
+      "step": 14146
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.016489769585073e-05,
+      "loss": 0.6761,
+      "step": 14147
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0162373428752744e-05,
+      "loss": 0.744,
+      "step": 14148
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.015984910667862e-05,
+      "loss": 0.7722,
+      "step": 14149
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.015732472965523e-05,
+      "loss": 0.741,
+      "step": 14150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0154800297709467e-05,
+      "loss": 0.748,
+      "step": 14151
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.015227581086821e-05,
+      "loss": 0.6903,
+      "step": 14152
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.014975126915834e-05,
+      "loss": 0.6697,
+      "step": 14153
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0147226672606742e-05,
+      "loss": 0.7205,
+      "step": 14154
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0144702021240312e-05,
+      "loss": 0.7097,
+      "step": 14155
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0142177315085927e-05,
+      "loss": 0.7503,
+      "step": 14156
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0139652554170476e-05,
+      "loss": 0.8,
+      "step": 14157
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0137127738520848e-05,
+      "loss": 0.6947,
+      "step": 14158
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0134602868163926e-05,
+      "loss": 0.7854,
+      "step": 14159
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0132077943126602e-05,
+      "loss": 0.8206,
+      "step": 14160
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0129552963435765e-05,
+      "loss": 0.7187,
+      "step": 14161
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0127027929118295e-05,
+      "loss": 0.7826,
+      "step": 14162
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0124502840201096e-05,
+      "loss": 0.7013,
+      "step": 14163
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.012197769671105e-05,
+      "loss": 0.7787,
+      "step": 14164
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0119452498675056e-05,
+      "loss": 0.8508,
+      "step": 14165
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0116927246119995e-05,
+      "loss": 0.7095,
+      "step": 14166
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0114401939072774e-05,
+      "loss": 0.8468,
+      "step": 14167
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0111876577560267e-05,
+      "loss": 0.819,
+      "step": 14168
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0109351161609383e-05,
+      "loss": 0.7871,
+      "step": 14169
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.010682569124702e-05,
+      "loss": 0.7301,
+      "step": 14170
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.010430016650005e-05,
+      "loss": 0.7486,
+      "step": 14171
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0101774587395386e-05,
+      "loss": 0.7834,
+      "step": 14172
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0099248953959923e-05,
+      "loss": 0.6897,
+      "step": 14173
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.009672326622055e-05,
+      "loss": 0.7828,
+      "step": 14174
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0094197524204177e-05,
+      "loss": 0.7896,
+      "step": 14175
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0091671727937688e-05,
+      "loss": 0.8296,
+      "step": 14176
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.008914587744798e-05,
+      "loss": 0.7507,
+      "step": 14177
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0086619972761964e-05,
+      "loss": 0.7597,
+      "step": 14178
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.008409401390654e-05,
+      "loss": 0.7692,
+      "step": 14179
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0081568000908588e-05,
+      "loss": 0.6997,
+      "step": 14180
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0079041933795033e-05,
+      "loss": 0.6834,
+      "step": 14181
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.007651581259276e-05,
+      "loss": 0.7176,
+      "step": 14182
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0073989637328674e-05,
+      "loss": 0.8015,
+      "step": 14183
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0071463408029683e-05,
+      "loss": 0.7417,
+      "step": 14184
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0068937124722678e-05,
+      "loss": 0.7141,
+      "step": 14185
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.006641078743458e-05,
+      "loss": 0.7262,
+      "step": 14186
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0063884396192276e-05,
+      "loss": 0.7306,
+      "step": 14187
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.006135795102268e-05,
+      "loss": 0.7286,
+      "step": 14188
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0058831451952697e-05,
+      "loss": 0.7253,
+      "step": 14189
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.005630489900923e-05,
+      "loss": 0.6999,
+      "step": 14190
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.005377829221918e-05,
+      "loss": 0.6659,
+      "step": 14191
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0051251631609467e-05,
+      "loss": 0.7882,
+      "step": 14192
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.004872491720699e-05,
+      "loss": 0.7242,
+      "step": 14193
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.004619814903865e-05,
+      "loss": 0.6724,
+      "step": 14194
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0043671327131374e-05,
+      "loss": 0.662,
+      "step": 14195
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0041144451512054e-05,
+      "loss": 0.7896,
+      "step": 14196
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0038617522207603e-05,
+      "loss": 0.7593,
+      "step": 14197
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0036090539244936e-05,
+      "loss": 0.7756,
+      "step": 14198
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.003356350265097e-05,
+      "loss": 0.7925,
+      "step": 14199
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0031036412452596e-05,
+      "loss": 0.8084,
+      "step": 14200
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0028509268676754e-05,
+      "loss": 0.6836,
+      "step": 14201
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.002598207135033e-05,
+      "loss": 0.7276,
+      "step": 14202
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0023454820500252e-05,
+      "loss": 0.7507,
+      "step": 14203
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.002092751615343e-05,
+      "loss": 0.7506,
+      "step": 14204
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0018400158336774e-05,
+      "loss": 0.7003,
+      "step": 14205
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0015872747077207e-05,
+      "loss": 0.7668,
+      "step": 14206
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0013345282401638e-05,
+      "loss": 0.7476,
+      "step": 14207
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.001081776433699e-05,
+      "loss": 0.7276,
+      "step": 14208
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0008290192910173e-05,
+      "loss": 0.6789,
+      "step": 14209
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0005762568148104e-05,
+      "loss": 0.7507,
+      "step": 14210
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0003234890077702e-05,
+      "loss": 0.7906,
+      "step": 14211
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0000707158725895e-05,
+      "loss": 0.7614,
+      "step": 14212
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9998179374119585e-05,
+      "loss": 0.7381,
+      "step": 14213
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9995651536285698e-05,
+      "loss": 0.7851,
+      "step": 14214
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9993123645251155e-05,
+      "loss": 0.773,
+      "step": 14215
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.999059570104288e-05,
+      "loss": 0.7361,
+      "step": 14216
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9988067703687784e-05,
+      "loss": 0.7818,
+      "step": 14217
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.99855396532128e-05,
+      "loss": 0.6952,
+      "step": 14218
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.998301154964485e-05,
+      "loss": 0.7244,
+      "step": 14219
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9980483393010845e-05,
+      "loss": 0.7218,
+      "step": 14220
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9977955183337724e-05,
+      "loss": 0.6935,
+      "step": 14221
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.99754269206524e-05,
+      "loss": 0.7677,
+      "step": 14222
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.997289860498179e-05,
+      "loss": 0.6782,
+      "step": 14223
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.997037023635284e-05,
+      "loss": 0.6795,
+      "step": 14224
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9967841814792464e-05,
+      "loss": 0.706,
+      "step": 14225
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9965313340327584e-05,
+      "loss": 0.731,
+      "step": 14226
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.996278481298513e-05,
+      "loss": 0.7416,
+      "step": 14227
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9960256232792035e-05,
+      "loss": 0.6847,
+      "step": 14228
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.995772759977522e-05,
+      "loss": 0.761,
+      "step": 14229
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.995519891396162e-05,
+      "loss": 0.817,
+      "step": 14230
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9952670175378155e-05,
+      "loss": 0.7244,
+      "step": 14231
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.995014138405176e-05,
+      "loss": 0.7156,
+      "step": 14232
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9947612540009373e-05,
+      "loss": 0.7266,
+      "step": 14233
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9945083643277904e-05,
+      "loss": 0.775,
+      "step": 14234
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.99425546938843e-05,
+      "loss": 0.8422,
+      "step": 14235
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9940025691855495e-05,
+      "loss": 0.7379,
+      "step": 14236
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.993749663721841e-05,
+      "loss": 0.8245,
+      "step": 14237
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9934967529999984e-05,
+      "loss": 0.7389,
+      "step": 14238
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.993243837022715e-05,
+      "loss": 0.7104,
+      "step": 14239
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9929909157926837e-05,
+      "loss": 0.7241,
+      "step": 14240
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9927379893126e-05,
+      "loss": 0.7875,
+      "step": 14241
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.992485057585155e-05,
+      "loss": 0.7772,
+      "step": 14242
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.992232120613043e-05,
+      "loss": 0.7786,
+      "step": 14243
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9919791783989576e-05,
+      "loss": 0.6718,
+      "step": 14244
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.991726230945593e-05,
+      "loss": 0.756,
+      "step": 14245
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.991473278255642e-05,
+      "loss": 0.6952,
+      "step": 14246
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9912203203317997e-05,
+      "loss": 0.7675,
+      "step": 14247
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9909673571767587e-05,
+      "loss": 0.7222,
+      "step": 14248
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.990714388793213e-05,
+      "loss": 0.7485,
+      "step": 14249
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9904614151838577e-05,
+      "loss": 0.7266,
+      "step": 14250
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9902084363513865e-05,
+      "loss": 0.7791,
+      "step": 14251
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9899554522984923e-05,
+      "loss": 0.6964,
+      "step": 14252
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9897024630278708e-05,
+      "loss": 0.7369,
+      "step": 14253
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.989449468542214e-05,
+      "loss": 0.7336,
+      "step": 14254
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9891964688442185e-05,
+      "loss": 0.707,
+      "step": 14255
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9889434639365776e-05,
+      "loss": 0.7354,
+      "step": 14256
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.988690453821985e-05,
+      "loss": 0.808,
+      "step": 14257
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9884374385031366e-05,
+      "loss": 0.7402,
+      "step": 14258
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9881844179827256e-05,
+      "loss": 0.7466,
+      "step": 14259
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9879313922634466e-05,
+      "loss": 0.7522,
+      "step": 14260
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9876783613479958e-05,
+      "loss": 0.7413,
+      "step": 14261
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9874253252390656e-05,
+      "loss": 0.7233,
+      "step": 14262
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9871722839393514e-05,
+      "loss": 0.7288,
+      "step": 14263
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9869192374515486e-05,
+      "loss": 0.7084,
+      "step": 14264
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9866661857783513e-05,
+      "loss": 0.8098,
+      "step": 14265
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.986413128922454e-05,
+      "loss": 0.7416,
+      "step": 14266
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.986160066886553e-05,
+      "loss": 0.7319,
+      "step": 14267
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9859069996733426e-05,
+      "loss": 0.789,
+      "step": 14268
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9856539272855165e-05,
+      "loss": 0.7632,
+      "step": 14269
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.985400849725772e-05,
+      "loss": 0.707,
+      "step": 14270
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9851477669968032e-05,
+      "loss": 0.6836,
+      "step": 14271
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.984894679101305e-05,
+      "loss": 0.7641,
+      "step": 14272
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9846415860419724e-05,
+      "loss": 0.6908,
+      "step": 14273
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.984388487821502e-05,
+      "loss": 0.7188,
+      "step": 14274
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9841353844425875e-05,
+      "loss": 0.6847,
+      "step": 14275
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9838822759079248e-05,
+      "loss": 0.7271,
+      "step": 14276
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9836291622202105e-05,
+      "loss": 0.7115,
+      "step": 14277
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9833760433821384e-05,
+      "loss": 0.7368,
+      "step": 14278
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9831229193964056e-05,
+      "loss": 0.7976,
+      "step": 14279
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9828697902657072e-05,
+      "loss": 0.7464,
+      "step": 14280
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9826166559927375e-05,
+      "loss": 0.7317,
+      "step": 14281
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9823635165801955e-05,
+      "loss": 0.6953,
+      "step": 14282
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.982110372030773e-05,
+      "loss": 0.7368,
+      "step": 14283
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9818572223471686e-05,
+      "loss": 0.6966,
+      "step": 14284
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.981604067532077e-05,
+      "loss": 0.7257,
+      "step": 14285
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.981350907588194e-05,
+      "loss": 0.8423,
+      "step": 14286
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9810977425182167e-05,
+      "loss": 0.7279,
+      "step": 14287
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9808445723248407e-05,
+      "loss": 0.7521,
+      "step": 14288
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.980591397010761e-05,
+      "loss": 0.7598,
+      "step": 14289
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9803382165786758e-05,
+      "loss": 0.8556,
+      "step": 14290
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.98008503103128e-05,
+      "loss": 0.6979,
+      "step": 14291
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.97983184037127e-05,
+      "loss": 0.829,
+      "step": 14292
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9795786446013423e-05,
+      "loss": 0.7288,
+      "step": 14293
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.979325443724193e-05,
+      "loss": 0.7293,
+      "step": 14294
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9790722377425185e-05,
+      "loss": 0.7512,
+      "step": 14295
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.978819026659016e-05,
+      "loss": 0.6945,
+      "step": 14296
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9785658104763813e-05,
+      "loss": 0.8035,
+      "step": 14297
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9783125891973108e-05,
+      "loss": 0.6627,
+      "step": 14298
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9780593628245023e-05,
+      "loss": 0.7536,
+      "step": 14299
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9778061313606514e-05,
+      "loss": 0.7666,
+      "step": 14300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9775528948084554e-05,
+      "loss": 0.7856,
+      "step": 14301
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9772996531706115e-05,
+      "loss": 0.7451,
+      "step": 14302
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9770464064498154e-05,
+      "loss": 0.6836,
+      "step": 14303
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9767931546487648e-05,
+      "loss": 0.7227,
+      "step": 14304
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9765398977701568e-05,
+      "loss": 0.7176,
+      "step": 14305
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.976286635816688e-05,
+      "loss": 0.7378,
+      "step": 14306
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9760333687910558e-05,
+      "loss": 0.7399,
+      "step": 14307
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9757800966959577e-05,
+      "loss": 0.6897,
+      "step": 14308
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9755268195340897e-05,
+      "loss": 0.7571,
+      "step": 14309
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9752735373081496e-05,
+      "loss": 0.8161,
+      "step": 14310
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.975020250020836e-05,
+      "loss": 0.7447,
+      "step": 14311
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9747669576748442e-05,
+      "loss": 0.8038,
+      "step": 14312
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.974513660272873e-05,
+      "loss": 0.7716,
+      "step": 14313
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9742603578176193e-05,
+      "loss": 0.7288,
+      "step": 14314
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9740070503117802e-05,
+      "loss": 0.7141,
+      "step": 14315
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9737537377580544e-05,
+      "loss": 0.7904,
+      "step": 14316
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.973500420159139e-05,
+      "loss": 0.778,
+      "step": 14317
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9732470975177307e-05,
+      "loss": 0.7858,
+      "step": 14318
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.972993769836529e-05,
+      "loss": 0.6997,
+      "step": 14319
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9727404371182306e-05,
+      "loss": 0.7422,
+      "step": 14320
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9724870993655328e-05,
+      "loss": 0.7469,
+      "step": 14321
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9722337565811357e-05,
+      "loss": 0.7369,
+      "step": 14322
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.971980408767735e-05,
+      "loss": 0.8313,
+      "step": 14323
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9717270559280296e-05,
+      "loss": 0.6471,
+      "step": 14324
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9714736980647174e-05,
+      "loss": 0.7311,
+      "step": 14325
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.971220335180496e-05,
+      "loss": 0.694,
+      "step": 14326
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9709669672780648e-05,
+      "loss": 0.802,
+      "step": 14327
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9707135943601217e-05,
+      "loss": 0.7059,
+      "step": 14328
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9704602164293637e-05,
+      "loss": 0.7456,
+      "step": 14329
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.97020683348849e-05,
+      "loss": 0.767,
+      "step": 14330
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9699534455401994e-05,
+      "loss": 0.7441,
+      "step": 14331
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9697000525871905e-05,
+      "loss": 0.7621,
+      "step": 14332
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9694466546321603e-05,
+      "loss": 0.7691,
+      "step": 14333
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9691932516778088e-05,
+      "loss": 0.7377,
+      "step": 14334
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9689398437268333e-05,
+      "loss": 0.7912,
+      "step": 14335
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.968686430781934e-05,
+      "loss": 0.6973,
+      "step": 14336
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.968433012845808e-05,
+      "loss": 0.7026,
+      "step": 14337
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.968179589921155e-05,
+      "loss": 0.6635,
+      "step": 14338
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.967926162010674e-05,
+      "loss": 0.6841,
+      "step": 14339
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9676727291170636e-05,
+      "loss": 0.8001,
+      "step": 14340
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9674192912430216e-05,
+      "loss": 0.7289,
+      "step": 14341
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9671658483912495e-05,
+      "loss": 0.7358,
+      "step": 14342
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9669124005644437e-05,
+      "loss": 0.7355,
+      "step": 14343
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.966658947765304e-05,
+      "loss": 0.7463,
+      "step": 14344
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9664054899965304e-05,
+      "loss": 0.7338,
+      "step": 14345
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.966152027260821e-05,
+      "loss": 0.7446,
+      "step": 14346
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9658985595608758e-05,
+      "loss": 0.7404,
+      "step": 14347
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.965645086899394e-05,
+      "loss": 0.7615,
+      "step": 14348
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9653916092790746e-05,
+      "loss": 0.8036,
+      "step": 14349
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9651381267026168e-05,
+      "loss": 0.6993,
+      "step": 14350
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9648846391727204e-05,
+      "loss": 0.7255,
+      "step": 14351
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9646311466920852e-05,
+      "loss": 0.758,
+      "step": 14352
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9643776492634108e-05,
+      "loss": 0.7435,
+      "step": 14353
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9641241468893955e-05,
+      "loss": 0.6841,
+      "step": 14354
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9638706395727394e-05,
+      "loss": 0.7299,
+      "step": 14355
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.963617127316144e-05,
+      "loss": 0.7164,
+      "step": 14356
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9633636101223065e-05,
+      "loss": 0.7863,
+      "step": 14357
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9631100879939284e-05,
+      "loss": 0.7427,
+      "step": 14358
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9628565609337088e-05,
+      "loss": 0.7358,
+      "step": 14359
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.962603028944348e-05,
+      "loss": 0.7865,
+      "step": 14360
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9623494920285456e-05,
+      "loss": 0.7182,
+      "step": 14361
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9620959501890028e-05,
+      "loss": 0.6488,
+      "step": 14362
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9618424034284177e-05,
+      "loss": 0.741,
+      "step": 14363
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.961588851749491e-05,
+      "loss": 0.7507,
+      "step": 14364
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.961335295154924e-05,
+      "loss": 0.7415,
+      "step": 14365
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9610817336474163e-05,
+      "loss": 0.7556,
+      "step": 14366
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.960828167229667e-05,
+      "loss": 0.696,
+      "step": 14367
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9605745959043785e-05,
+      "loss": 0.6851,
+      "step": 14368
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9603210196742503e-05,
+      "loss": 0.7268,
+      "step": 14369
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.960067438541982e-05,
+      "loss": 0.8201,
+      "step": 14370
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9598138525102752e-05,
+      "loss": 0.7328,
+      "step": 14371
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9595602615818303e-05,
+      "loss": 0.7084,
+      "step": 14372
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.959306665759347e-05,
+      "loss": 0.7381,
+      "step": 14373
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.959053065045527e-05,
+      "loss": 0.7843,
+      "step": 14374
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9587994594430703e-05,
+      "loss": 0.6605,
+      "step": 14375
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.958545848954678e-05,
+      "loss": 0.814,
+      "step": 14376
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.958292233583051e-05,
+      "loss": 0.7977,
+      "step": 14377
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9580386133308892e-05,
+      "loss": 0.7785,
+      "step": 14378
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.957784988200895e-05,
+      "loss": 0.762,
+      "step": 14379
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9575313581957686e-05,
+      "loss": 0.6719,
+      "step": 14380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.957277723318211e-05,
+      "loss": 0.7803,
+      "step": 14381
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9570240835709228e-05,
+      "loss": 0.7314,
+      "step": 14382
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9567704389566064e-05,
+      "loss": 0.7581,
+      "step": 14383
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.956516789477961e-05,
+      "loss": 0.7741,
+      "step": 14384
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9562631351376895e-05,
+      "loss": 0.699,
+      "step": 14385
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.956009475938493e-05,
+      "loss": 0.8123,
+      "step": 14386
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9557558118830718e-05,
+      "loss": 0.7192,
+      "step": 14387
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.955502142974128e-05,
+      "loss": 0.7416,
+      "step": 14388
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9552484692143633e-05,
+      "loss": 0.7106,
+      "step": 14389
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9549947906064786e-05,
+      "loss": 0.7773,
+      "step": 14390
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9547411071531754e-05,
+      "loss": 0.777,
+      "step": 14391
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9544874188571558e-05,
+      "loss": 0.8451,
+      "step": 14392
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9542337257211215e-05,
+      "loss": 0.7284,
+      "step": 14393
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9539800277477736e-05,
+      "loss": 0.7022,
+      "step": 14394
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9537263249398133e-05,
+      "loss": 0.7189,
+      "step": 14395
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9534726172999434e-05,
+      "loss": 0.6654,
+      "step": 14396
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9532189048308655e-05,
+      "loss": 0.7884,
+      "step": 14397
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9529651875352822e-05,
+      "loss": 0.7946,
+      "step": 14398
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9527114654158933e-05,
+      "loss": 0.7785,
+      "step": 14399
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9524577384754026e-05,
+      "loss": 0.6865,
+      "step": 14400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9522040067165124e-05,
+      "loss": 0.6818,
+      "step": 14401
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.951950270141923e-05,
+      "loss": 0.7754,
+      "step": 14402
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9516965287543386e-05,
+      "loss": 0.7624,
+      "step": 14403
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9514427825564596e-05,
+      "loss": 0.7144,
+      "step": 14404
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9511890315509894e-05,
+      "loss": 0.7978,
+      "step": 14405
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9509352757406305e-05,
+      "loss": 0.6702,
+      "step": 14406
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9506815151280838e-05,
+      "loss": 0.7504,
+      "step": 14407
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9504277497160525e-05,
+      "loss": 0.686,
+      "step": 14408
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9501739795072398e-05,
+      "loss": 0.7341,
+      "step": 14409
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9499202045043466e-05,
+      "loss": 0.7447,
+      "step": 14410
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9496664247100774e-05,
+      "loss": 0.7773,
+      "step": 14411
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.949412640127134e-05,
+      "loss": 0.6383,
+      "step": 14412
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.949158850758218e-05,
+      "loss": 0.6964,
+      "step": 14413
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9489050566060333e-05,
+      "loss": 0.6724,
+      "step": 14414
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9486512576732824e-05,
+      "loss": 0.7808,
+      "step": 14415
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.948397453962667e-05,
+      "loss": 0.832,
+      "step": 14416
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9481436454768917e-05,
+      "loss": 0.7419,
+      "step": 14417
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9478898322186587e-05,
+      "loss": 0.6921,
+      "step": 14418
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9476360141906704e-05,
+      "loss": 0.7092,
+      "step": 14419
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9473821913956312e-05,
+      "loss": 0.8038,
+      "step": 14420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.947128363836243e-05,
+      "loss": 0.7987,
+      "step": 14421
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9468745315152084e-05,
+      "loss": 0.8054,
+      "step": 14422
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9466206944352326e-05,
+      "loss": 0.7625,
+      "step": 14423
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9463668525990163e-05,
+      "loss": 0.7563,
+      "step": 14424
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.946113006009265e-05,
+      "loss": 0.7663,
+      "step": 14425
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.945859154668681e-05,
+      "loss": 0.7232,
+      "step": 14426
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9456052985799664e-05,
+      "loss": 0.6955,
+      "step": 14427
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9453514377458268e-05,
+      "loss": 0.7084,
+      "step": 14428
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9450975721689657e-05,
+      "loss": 0.7315,
+      "step": 14429
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.944843701852084e-05,
+      "loss": 0.7143,
+      "step": 14430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9445898267978876e-05,
+      "loss": 0.7646,
+      "step": 14431
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9443359470090798e-05,
+      "loss": 0.7559,
+      "step": 14432
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.944082062488364e-05,
+      "loss": 0.7002,
+      "step": 14433
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9438281732384436e-05,
+      "loss": 0.6938,
+      "step": 14434
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.943574279262023e-05,
+      "loss": 0.6984,
+      "step": 14435
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9433203805618047e-05,
+      "loss": 0.6674,
+      "step": 14436
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.943066477140494e-05,
+      "loss": 0.6613,
+      "step": 14437
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9428125690007945e-05,
+      "loss": 0.7029,
+      "step": 14438
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9425586561454095e-05,
+      "loss": 0.6941,
+      "step": 14439
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9423047385770437e-05,
+      "loss": 0.759,
+      "step": 14440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9420508162984012e-05,
+      "loss": 0.7148,
+      "step": 14441
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9417968893121854e-05,
+      "loss": 0.7275,
+      "step": 14442
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9415429576211017e-05,
+      "loss": 0.6829,
+      "step": 14443
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9412890212278525e-05,
+      "loss": 0.7488,
+      "step": 14444
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9410350801351443e-05,
+      "loss": 0.7022,
+      "step": 14445
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9407811343456796e-05,
+      "loss": 0.7686,
+      "step": 14446
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.940527183862164e-05,
+      "loss": 0.706,
+      "step": 14447
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9402732286873003e-05,
+      "loss": 0.7775,
+      "step": 14448
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9400192688237948e-05,
+      "loss": 0.7979,
+      "step": 14449
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.939765304274351e-05,
+      "loss": 0.7646,
+      "step": 14450
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9395113350416735e-05,
+      "loss": 0.715,
+      "step": 14451
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9392573611284674e-05,
+      "loss": 0.7509,
+      "step": 14452
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9390033825374374e-05,
+      "loss": 0.7064,
+      "step": 14453
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9387493992712877e-05,
+      "loss": 0.7666,
+      "step": 14454
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9384954113327233e-05,
+      "loss": 0.7664,
+      "step": 14455
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9382414187244483e-05,
+      "loss": 0.7813,
+      "step": 14456
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.937987421449169e-05,
+      "loss": 0.713,
+      "step": 14457
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.937733419509589e-05,
+      "loss": 0.7884,
+      "step": 14458
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9374794129084137e-05,
+      "loss": 0.7152,
+      "step": 14459
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9372254016483492e-05,
+      "loss": 0.8454,
+      "step": 14460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.936971385732099e-05,
+      "loss": 0.7241,
+      "step": 14461
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9367173651623685e-05,
+      "loss": 0.6914,
+      "step": 14462
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9364633399418644e-05,
+      "loss": 0.6675,
+      "step": 14463
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.93620931007329e-05,
+      "loss": 0.7461,
+      "step": 14464
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9359552755593513e-05,
+      "loss": 0.7743,
+      "step": 14465
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9357012364027537e-05,
+      "loss": 0.6889,
+      "step": 14466
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.935447192606202e-05,
+      "loss": 0.7076,
+      "step": 14467
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9351931441724024e-05,
+      "loss": 0.7114,
+      "step": 14468
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9349390911040598e-05,
+      "loss": 0.7537,
+      "step": 14469
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.93468503340388e-05,
+      "loss": 0.7445,
+      "step": 14470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.934430971074569e-05,
+      "loss": 0.7893,
+      "step": 14471
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.934176904118831e-05,
+      "loss": 0.6805,
+      "step": 14472
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.933922832539373e-05,
+      "loss": 0.7098,
+      "step": 14473
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.933668756338901e-05,
+      "loss": 0.7517,
+      "step": 14474
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9334146755201193e-05,
+      "loss": 0.7147,
+      "step": 14475
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.933160590085734e-05,
+      "loss": 0.7292,
+      "step": 14476
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.932906500038452e-05,
+      "loss": 0.7707,
+      "step": 14477
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.932652405380979e-05,
+      "loss": 0.7058,
+      "step": 14478
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.93239830611602e-05,
+      "loss": 0.6974,
+      "step": 14479
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9321442022462814e-05,
+      "loss": 0.7207,
+      "step": 14480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.93189009377447e-05,
+      "loss": 0.7238,
+      "step": 14481
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.93163598070329e-05,
+      "loss": 0.7825,
+      "step": 14482
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9313818630354506e-05,
+      "loss": 0.729,
+      "step": 14483
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.931127740773656e-05,
+      "loss": 0.7693,
+      "step": 14484
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9308736139206115e-05,
+      "loss": 0.707,
+      "step": 14485
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9306194824790256e-05,
+      "loss": 0.8001,
+      "step": 14486
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9303653464516033e-05,
+      "loss": 0.6969,
+      "step": 14487
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9301112058410507e-05,
+      "loss": 0.7803,
+      "step": 14488
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.929857060650076e-05,
+      "loss": 0.7633,
+      "step": 14489
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9296029108813838e-05,
+      "loss": 0.7422,
+      "step": 14490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9293487565376814e-05,
+      "loss": 0.7284,
+      "step": 14491
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9290945976216762e-05,
+      "loss": 0.7297,
+      "step": 14492
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9288404341360735e-05,
+      "loss": 0.7129,
+      "step": 14493
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.928586266083581e-05,
+      "loss": 0.661,
+      "step": 14494
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9283320934669045e-05,
+      "loss": 0.8009,
+      "step": 14495
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.928077916288751e-05,
+      "loss": 0.7132,
+      "step": 14496
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9278237345518276e-05,
+      "loss": 0.8,
+      "step": 14497
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9275695482588424e-05,
+      "loss": 0.6576,
+      "step": 14498
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9273153574125e-05,
+      "loss": 0.8156,
+      "step": 14499
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.927061162015508e-05,
+      "loss": 0.8096,
+      "step": 14500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.926806962070575e-05,
+      "loss": 0.8691,
+      "step": 14501
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9265527575804063e-05,
+      "loss": 0.8046,
+      "step": 14502
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9262985485477103e-05,
+      "loss": 0.6898,
+      "step": 14503
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.926044334975193e-05,
+      "loss": 0.7613,
+      "step": 14504
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9257901168655623e-05,
+      "loss": 0.8139,
+      "step": 14505
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9255358942215255e-05,
+      "loss": 0.718,
+      "step": 14506
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.92528166704579e-05,
+      "loss": 0.715,
+      "step": 14507
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9250274353410622e-05,
+      "loss": 0.7207,
+      "step": 14508
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9247731991100508e-05,
+      "loss": 0.7115,
+      "step": 14509
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9245189583554627e-05,
+      "loss": 0.7212,
+      "step": 14510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.924264713080005e-05,
+      "loss": 0.7473,
+      "step": 14511
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9240104632863864e-05,
+      "loss": 0.6716,
+      "step": 14512
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9237562089773135e-05,
+      "loss": 0.8657,
+      "step": 14513
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9235019501554943e-05,
+      "loss": 0.7065,
+      "step": 14514
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9232476868236363e-05,
+      "loss": 0.8783,
+      "step": 14515
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9229934189844477e-05,
+      "loss": 0.6411,
+      "step": 14516
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9227391466406352e-05,
+      "loss": 0.7334,
+      "step": 14517
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.922484869794908e-05,
+      "loss": 0.745,
+      "step": 14518
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9222305884499735e-05,
+      "loss": 0.7476,
+      "step": 14519
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.921976302608539e-05,
+      "loss": 0.7289,
+      "step": 14520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.921722012273314e-05,
+      "loss": 0.7121,
+      "step": 14521
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.921467717447005e-05,
+      "loss": 0.7685,
+      "step": 14522
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9212134181323202e-05,
+      "loss": 0.75,
+      "step": 14523
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9209591143319697e-05,
+      "loss": 0.672,
+      "step": 14524
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.920704806048659e-05,
+      "loss": 0.7152,
+      "step": 14525
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.920450493285098e-05,
+      "loss": 0.7512,
+      "step": 14526
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9201961760439945e-05,
+      "loss": 0.7983,
+      "step": 14527
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9199418543280566e-05,
+      "loss": 0.7279,
+      "step": 14528
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9196875281399927e-05,
+      "loss": 0.7674,
+      "step": 14529
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9194331974825124e-05,
+      "loss": 0.8076,
+      "step": 14530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9191788623583217e-05,
+      "loss": 0.7113,
+      "step": 14531
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9189245227701316e-05,
+      "loss": 0.8051,
+      "step": 14532
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9186701787206494e-05,
+      "loss": 0.7395,
+      "step": 14533
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9184158302125847e-05,
+      "loss": 0.755,
+      "step": 14534
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.918161477248645e-05,
+      "loss": 0.6839,
+      "step": 14535
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9179071198315387e-05,
+      "loss": 0.6576,
+      "step": 14536
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9176527579639755e-05,
+      "loss": 0.7374,
+      "step": 14537
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9173983916486646e-05,
+      "loss": 0.7566,
+      "step": 14538
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9171440208883145e-05,
+      "loss": 0.7312,
+      "step": 14539
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9168896456856328e-05,
+      "loss": 0.7744,
+      "step": 14540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9166352660433298e-05,
+      "loss": 0.7482,
+      "step": 14541
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9163808819641147e-05,
+      "loss": 0.7363,
+      "step": 14542
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9161264934506954e-05,
+      "loss": 0.7748,
+      "step": 14543
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.915872100505782e-05,
+      "loss": 0.778,
+      "step": 14544
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.915617703132083e-05,
+      "loss": 0.7407,
+      "step": 14545
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9153633013323078e-05,
+      "loss": 0.7365,
+      "step": 14546
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.915108895109166e-05,
+      "loss": 0.7168,
+      "step": 14547
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9148544844653656e-05,
+      "loss": 0.7395,
+      "step": 14548
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9146000694036173e-05,
+      "loss": 0.7143,
+      "step": 14549
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9143456499266304e-05,
+      "loss": 0.7082,
+      "step": 14550
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9140912260371128e-05,
+      "loss": 0.7454,
+      "step": 14551
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.913836797737776e-05,
+      "loss": 0.6753,
+      "step": 14552
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9135823650313282e-05,
+      "loss": 0.7878,
+      "step": 14553
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9133279279204794e-05,
+      "loss": 0.7255,
+      "step": 14554
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9130734864079396e-05,
+      "loss": 0.7417,
+      "step": 14555
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9128190404964175e-05,
+      "loss": 0.7738,
+      "step": 14556
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.912564590188623e-05,
+      "loss": 0.7543,
+      "step": 14557
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.912310135487266e-05,
+      "loss": 0.7647,
+      "step": 14558
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.912055676395057e-05,
+      "loss": 0.6521,
+      "step": 14559
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9118012129147043e-05,
+      "loss": 0.7577,
+      "step": 14560
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9115467450489197e-05,
+      "loss": 0.7223,
+      "step": 14561
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9112922728004117e-05,
+      "loss": 0.7201,
+      "step": 14562
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9110377961718903e-05,
+      "loss": 0.8598,
+      "step": 14563
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.910783315166067e-05,
+      "loss": 0.7822,
+      "step": 14564
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9105288297856505e-05,
+      "loss": 0.7196,
+      "step": 14565
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.91027434003335e-05,
+      "loss": 0.725,
+      "step": 14566
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9100198459118777e-05,
+      "loss": 0.7329,
+      "step": 14567
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9097653474239428e-05,
+      "loss": 0.7105,
+      "step": 14568
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9095108445722558e-05,
+      "loss": 0.769,
+      "step": 14569
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.909256337359527e-05,
+      "loss": 0.7419,
+      "step": 14570
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9090018257884666e-05,
+      "loss": 0.7869,
+      "step": 14571
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9087473098617846e-05,
+      "loss": 0.7178,
+      "step": 14572
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9084927895821927e-05,
+      "loss": 0.7728,
+      "step": 14573
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9082382649524002e-05,
+      "loss": 0.7678,
+      "step": 14574
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9079837359751183e-05,
+      "loss": 0.7205,
+      "step": 14575
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.907729202653057e-05,
+      "loss": 0.7079,
+      "step": 14576
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9074746649889272e-05,
+      "loss": 0.7137,
+      "step": 14577
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.90722012298544e-05,
+      "loss": 0.7981,
+      "step": 14578
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9069655766453057e-05,
+      "loss": 0.7119,
+      "step": 14579
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9067110259712342e-05,
+      "loss": 0.7061,
+      "step": 14580
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.906456470965938e-05,
+      "loss": 0.6865,
+      "step": 14581
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9062019116321275e-05,
+      "loss": 0.7441,
+      "step": 14582
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9059473479725124e-05,
+      "loss": 0.7274,
+      "step": 14583
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9056927799898054e-05,
+      "loss": 0.7498,
+      "step": 14584
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.905438207686716e-05,
+      "loss": 0.7751,
+      "step": 14585
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.905183631065956e-05,
+      "loss": 0.7832,
+      "step": 14586
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.904929050130236e-05,
+      "loss": 0.7247,
+      "step": 14587
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.904674464882268e-05,
+      "loss": 0.7148,
+      "step": 14588
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9044198753247613e-05,
+      "loss": 0.8018,
+      "step": 14589
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9041652814604296e-05,
+      "loss": 0.7241,
+      "step": 14590
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.903910683291983e-05,
+      "loss": 0.7635,
+      "step": 14591
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9036560808221325e-05,
+      "loss": 0.75,
+      "step": 14592
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9034014740535903e-05,
+      "loss": 0.7349,
+      "step": 14593
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9031468629890675e-05,
+      "loss": 0.74,
+      "step": 14594
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9028922476312748e-05,
+      "loss": 0.7497,
+      "step": 14595
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.902637627982924e-05,
+      "loss": 0.7391,
+      "step": 14596
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.902383004046727e-05,
+      "loss": 0.6939,
+      "step": 14597
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.902128375825396e-05,
+      "loss": 0.744,
+      "step": 14598
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9018737433216415e-05,
+      "loss": 0.7361,
+      "step": 14599
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9016191065381752e-05,
+      "loss": 0.7174,
+      "step": 14600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9013644654777102e-05,
+      "loss": 0.7721,
+      "step": 14601
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9011098201429566e-05,
+      "loss": 0.7185,
+      "step": 14602
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.900855170536627e-05,
+      "loss": 0.6822,
+      "step": 14603
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9006005166614335e-05,
+      "loss": 0.7685,
+      "step": 14604
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9003458585200883e-05,
+      "loss": 0.7839,
+      "step": 14605
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9000911961153014e-05,
+      "loss": 0.6988,
+      "step": 14606
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.899836529449787e-05,
+      "loss": 0.6654,
+      "step": 14607
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.899581858526256e-05,
+      "loss": 0.7937,
+      "step": 14608
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8993271833474206e-05,
+      "loss": 0.7196,
+      "step": 14609
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8990725039159934e-05,
+      "loss": 0.7081,
+      "step": 14610
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8988178202346865e-05,
+      "loss": 0.7923,
+      "step": 14611
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8985631323062114e-05,
+      "loss": 0.7032,
+      "step": 14612
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8983084401332816e-05,
+      "loss": 0.6721,
+      "step": 14613
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8980537437186084e-05,
+      "loss": 0.7529,
+      "step": 14614
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8977990430649042e-05,
+      "loss": 0.8516,
+      "step": 14615
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8975443381748823e-05,
+      "loss": 0.7221,
+      "step": 14616
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8972896290512545e-05,
+      "loss": 0.7051,
+      "step": 14617
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8970349156967325e-05,
+      "loss": 0.8525,
+      "step": 14618
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.896780198114031e-05,
+      "loss": 0.7082,
+      "step": 14619
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8965254763058602e-05,
+      "loss": 0.7434,
+      "step": 14620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8962707502749344e-05,
+      "loss": 0.7004,
+      "step": 14621
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.896016020023965e-05,
+      "loss": 0.662,
+      "step": 14622
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8957612855556658e-05,
+      "loss": 0.7025,
+      "step": 14623
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8955065468727495e-05,
+      "loss": 0.7595,
+      "step": 14624
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8952518039779287e-05,
+      "loss": 0.7504,
+      "step": 14625
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8949970568739155e-05,
+      "loss": 0.6854,
+      "step": 14626
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8947423055634238e-05,
+      "loss": 0.7052,
+      "step": 14627
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8944875500491663e-05,
+      "loss": 0.679,
+      "step": 14628
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8942327903338557e-05,
+      "loss": 0.7191,
+      "step": 14629
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8939780264202054e-05,
+      "loss": 0.7771,
+      "step": 14630
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.893723258310928e-05,
+      "loss": 0.7668,
+      "step": 14631
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.893468486008737e-05,
+      "loss": 0.824,
+      "step": 14632
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.893213709516346e-05,
+      "loss": 0.6973,
+      "step": 14633
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8929589288364674e-05,
+      "loss": 0.7088,
+      "step": 14634
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.892704143971815e-05,
+      "loss": 0.7371,
+      "step": 14635
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8924493549251015e-05,
+      "loss": 0.7672,
+      "step": 14636
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8921945616990408e-05,
+      "loss": 0.8145,
+      "step": 14637
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8919397642963457e-05,
+      "loss": 0.6892,
+      "step": 14638
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8916849627197307e-05,
+      "loss": 0.841,
+      "step": 14639
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.891430156971908e-05,
+      "loss": 0.6822,
+      "step": 14640
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8911753470555925e-05,
+      "loss": 0.7006,
+      "step": 14641
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.890920532973497e-05,
+      "loss": 0.7834,
+      "step": 14642
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8906657147283344e-05,
+      "loss": 0.8134,
+      "step": 14643
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.890410892322819e-05,
+      "loss": 0.7061,
+      "step": 14644
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.890156065759666e-05,
+      "loss": 0.7122,
+      "step": 14645
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8899012350415865e-05,
+      "loss": 0.7117,
+      "step": 14646
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8896464001712963e-05,
+      "loss": 0.7519,
+      "step": 14647
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8893915611515086e-05,
+      "loss": 0.7713,
+      "step": 14648
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.889136717984936e-05,
+      "loss": 0.7792,
+      "step": 14649
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.888881870674295e-05,
+      "loss": 0.7752,
+      "step": 14650
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8886270192222977e-05,
+      "loss": 0.6812,
+      "step": 14651
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.888372163631658e-05,
+      "loss": 0.717,
+      "step": 14652
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8881173039050912e-05,
+      "loss": 0.7455,
+      "step": 14653
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8878624400453102e-05,
+      "loss": 0.7518,
+      "step": 14654
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8876075720550306e-05,
+      "loss": 0.7506,
+      "step": 14655
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.887352699936965e-05,
+      "loss": 0.7515,
+      "step": 14656
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.887097823693829e-05,
+      "loss": 0.7243,
+      "step": 14657
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8868429433283352e-05,
+      "loss": 0.7428,
+      "step": 14658
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8865880588431992e-05,
+      "loss": 0.8096,
+      "step": 14659
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8863331702411355e-05,
+      "loss": 0.7728,
+      "step": 14660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8860782775248573e-05,
+      "loss": 0.7046,
+      "step": 14661
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8858233806970804e-05,
+      "loss": 0.7225,
+      "step": 14662
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.885568479760519e-05,
+      "loss": 0.741,
+      "step": 14663
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8853135747178868e-05,
+      "loss": 0.7666,
+      "step": 14664
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8850586655719e-05,
+      "loss": 0.7347,
+      "step": 14665
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.884803752325271e-05,
+      "loss": 0.7532,
+      "step": 14666
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.884548834980716e-05,
+      "loss": 0.8174,
+      "step": 14667
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8842939135409497e-05,
+      "loss": 0.6595,
+      "step": 14668
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8840389880086855e-05,
+      "loss": 0.9025,
+      "step": 14669
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8837840583866404e-05,
+      "loss": 0.845,
+      "step": 14670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8835291246775277e-05,
+      "loss": 0.7439,
+      "step": 14671
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.883274186884063e-05,
+      "loss": 0.7057,
+      "step": 14672
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.88301924500896e-05,
+      "loss": 0.7,
+      "step": 14673
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8827642990549354e-05,
+      "loss": 0.7472,
+      "step": 14674
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.882509349024703e-05,
+      "loss": 0.7756,
+      "step": 14675
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8822543949209784e-05,
+      "loss": 0.7229,
+      "step": 14676
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8819994367464765e-05,
+      "loss": 0.7622,
+      "step": 14677
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8817444745039125e-05,
+      "loss": 0.7827,
+      "step": 14678
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8814895081960013e-05,
+      "loss": 0.7371,
+      "step": 14679
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.881234537825459e-05,
+      "loss": 0.736,
+      "step": 14680
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8809795633949998e-05,
+      "loss": 0.7284,
+      "step": 14681
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8807245849073406e-05,
+      "loss": 0.7325,
+      "step": 14682
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.880469602365195e-05,
+      "loss": 0.7109,
+      "step": 14683
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8802146157712784e-05,
+      "loss": 0.7229,
+      "step": 14684
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8799596251283085e-05,
+      "loss": 0.6811,
+      "step": 14685
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8797046304389985e-05,
+      "loss": 0.6774,
+      "step": 14686
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8794496317060644e-05,
+      "loss": 0.7446,
+      "step": 14687
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8791946289322223e-05,
+      "loss": 0.7638,
+      "step": 14688
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.878939622120188e-05,
+      "loss": 0.8106,
+      "step": 14689
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8786846112726763e-05,
+      "loss": 0.6911,
+      "step": 14690
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8784295963924034e-05,
+      "loss": 0.7632,
+      "step": 14691
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8781745774820857e-05,
+      "loss": 0.8043,
+      "step": 14692
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.877919554544437e-05,
+      "loss": 0.7277,
+      "step": 14693
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8776645275821756e-05,
+      "loss": 0.8223,
+      "step": 14694
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8774094965980164e-05,
+      "loss": 0.7527,
+      "step": 14695
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.877154461594675e-05,
+      "loss": 0.7257,
+      "step": 14696
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8768994225748675e-05,
+      "loss": 0.7134,
+      "step": 14697
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.876644379541309e-05,
+      "loss": 0.7585,
+      "step": 14698
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8763893324967177e-05,
+      "loss": 0.7441,
+      "step": 14699
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8761342814438086e-05,
+      "loss": 0.8129,
+      "step": 14700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8758792263852967e-05,
+      "loss": 0.7643,
+      "step": 14701
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8756241673238998e-05,
+      "loss": 0.7394,
+      "step": 14702
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8753691042623336e-05,
+      "loss": 0.6779,
+      "step": 14703
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8751140372033143e-05,
+      "loss": 0.8431,
+      "step": 14704
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.874858966149559e-05,
+      "loss": 0.7873,
+      "step": 14705
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8746038911037826e-05,
+      "loss": 0.7814,
+      "step": 14706
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8743488120687017e-05,
+      "loss": 0.7411,
+      "step": 14707
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8740937290470345e-05,
+      "loss": 0.7987,
+      "step": 14708
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8738386420414955e-05,
+      "loss": 0.7434,
+      "step": 14709
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8735835510548014e-05,
+      "loss": 0.6985,
+      "step": 14710
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8733284560896694e-05,
+      "loss": 0.6541,
+      "step": 14711
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8730733571488165e-05,
+      "loss": 0.707,
+      "step": 14712
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8728182542349587e-05,
+      "loss": 0.7084,
+      "step": 14713
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.872563147350813e-05,
+      "loss": 0.7031,
+      "step": 14714
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8723080364990957e-05,
+      "loss": 0.6886,
+      "step": 14715
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8720529216825242e-05,
+      "loss": 0.6857,
+      "step": 14716
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8717978029038144e-05,
+      "loss": 0.7218,
+      "step": 14717
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.871542680165684e-05,
+      "loss": 0.6313,
+      "step": 14718
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8712875534708493e-05,
+      "loss": 0.6532,
+      "step": 14719
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8710324228220286e-05,
+      "loss": 0.7263,
+      "step": 14720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8707772882219363e-05,
+      "loss": 0.7465,
+      "step": 14721
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8705221496732926e-05,
+      "loss": 0.7343,
+      "step": 14722
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.870267007178812e-05,
+      "loss": 0.7564,
+      "step": 14723
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.870011860741213e-05,
+      "loss": 0.7867,
+      "step": 14724
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8697567103632117e-05,
+      "loss": 0.76,
+      "step": 14725
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.869501556047527e-05,
+      "loss": 0.7637,
+      "step": 14726
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8692463977968743e-05,
+      "loss": 0.7382,
+      "step": 14727
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8689912356139713e-05,
+      "loss": 0.7736,
+      "step": 14728
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8687360695015365e-05,
+      "loss": 0.7275,
+      "step": 14729
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8684808994622853e-05,
+      "loss": 0.8114,
+      "step": 14730
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8682257254989376e-05,
+      "loss": 0.7562,
+      "step": 14731
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8679705476142088e-05,
+      "loss": 0.6707,
+      "step": 14732
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8677153658108163e-05,
+      "loss": 0.7056,
+      "step": 14733
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8674601800914796e-05,
+      "loss": 0.7382,
+      "step": 14734
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8672049904589148e-05,
+      "loss": 0.7104,
+      "step": 14735
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8669497969158397e-05,
+      "loss": 0.692,
+      "step": 14736
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8666945994649717e-05,
+      "loss": 0.7039,
+      "step": 14737
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8664393981090298e-05,
+      "loss": 0.714,
+      "step": 14738
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8661841928507294e-05,
+      "loss": 0.6503,
+      "step": 14739
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8659289836927904e-05,
+      "loss": 0.8206,
+      "step": 14740
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.86567377063793e-05,
+      "loss": 0.7475,
+      "step": 14741
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8654185536888655e-05,
+      "loss": 0.6484,
+      "step": 14742
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8651633328483157e-05,
+      "loss": 0.701,
+      "step": 14743
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8649081081189983e-05,
+      "loss": 0.7694,
+      "step": 14744
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8646528795036303e-05,
+      "loss": 0.8538,
+      "step": 14745
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8643976470049316e-05,
+      "loss": 0.8041,
+      "step": 14746
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8641424106256182e-05,
+      "loss": 0.7595,
+      "step": 14747
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8638871703684095e-05,
+      "loss": 0.7644,
+      "step": 14748
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.863631926236024e-05,
+      "loss": 0.7892,
+      "step": 14749
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8633766782311777e-05,
+      "loss": 0.7612,
+      "step": 14750
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.863121426356592e-05,
+      "loss": 0.7518,
+      "step": 14751
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.862866170614983e-05,
+      "loss": 0.6897,
+      "step": 14752
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8626109110090686e-05,
+      "loss": 0.7078,
+      "step": 14753
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8623556475415698e-05,
+      "loss": 0.798,
+      "step": 14754
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8621003802152023e-05,
+      "loss": 0.7584,
+      "step": 14755
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8618451090326858e-05,
+      "loss": 0.689,
+      "step": 14756
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.861589833996739e-05,
+      "loss": 0.6771,
+      "step": 14757
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.86133455511008e-05,
+      "loss": 0.6572,
+      "step": 14758
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8610792723754266e-05,
+      "loss": 0.7161,
+      "step": 14759
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8608239857954988e-05,
+      "loss": 0.7403,
+      "step": 14760
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.860568695373014e-05,
+      "loss": 0.6853,
+      "step": 14761
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.860313401110692e-05,
+      "loss": 0.6781,
+      "step": 14762
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8600581030112512e-05,
+      "loss": 0.6888,
+      "step": 14763
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8598028010774096e-05,
+      "loss": 0.658,
+      "step": 14764
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8595474953118867e-05,
+      "loss": 0.7358,
+      "step": 14765
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.859292185717402e-05,
+      "loss": 0.7371,
+      "step": 14766
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8590368722966733e-05,
+      "loss": 0.766,
+      "step": 14767
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8587815550524194e-05,
+      "loss": 0.7523,
+      "step": 14768
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.85852623398736e-05,
+      "loss": 0.6827,
+      "step": 14769
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8582709091042138e-05,
+      "loss": 0.7383,
+      "step": 14770
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8580155804057e-05,
+      "loss": 0.7394,
+      "step": 14771
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8577602478945382e-05,
+      "loss": 0.7144,
+      "step": 14772
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.857504911573446e-05,
+      "loss": 0.7048,
+      "step": 14773
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.857249571445144e-05,
+      "loss": 0.7146,
+      "step": 14774
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8569942275123512e-05,
+      "loss": 0.7098,
+      "step": 14775
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8567388797777867e-05,
+      "loss": 0.6573,
+      "step": 14776
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8564835282441693e-05,
+      "loss": 0.7131,
+      "step": 14777
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8562281729142194e-05,
+      "loss": 0.717,
+      "step": 14778
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8559728137906542e-05,
+      "loss": 0.7986,
+      "step": 14779
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8557174508761958e-05,
+      "loss": 0.7427,
+      "step": 14780
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8554620841735624e-05,
+      "loss": 0.7354,
+      "step": 14781
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.855206713685473e-05,
+      "loss": 0.7424,
+      "step": 14782
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8549513394146488e-05,
+      "loss": 0.7026,
+      "step": 14783
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.854695961363808e-05,
+      "loss": 0.6913,
+      "step": 14784
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8544405795356698e-05,
+      "loss": 0.7415,
+      "step": 14785
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.854185193932956e-05,
+      "loss": 0.6552,
+      "step": 14786
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.853929804558384e-05,
+      "loss": 0.7012,
+      "step": 14787
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8536744114146747e-05,
+      "loss": 0.7323,
+      "step": 14788
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8534190145045476e-05,
+      "loss": 0.6536,
+      "step": 14789
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8531636138307227e-05,
+      "loss": 0.753,
+      "step": 14790
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8529082093959185e-05,
+      "loss": 0.8196,
+      "step": 14791
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8526528012028575e-05,
+      "loss": 0.7179,
+      "step": 14792
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8523973892542584e-05,
+      "loss": 0.7952,
+      "step": 14793
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8521419735528398e-05,
+      "loss": 0.7655,
+      "step": 14794
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.851886554101324e-05,
+      "loss": 0.7507,
+      "step": 14795
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8516311309024297e-05,
+      "loss": 0.7201,
+      "step": 14796
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8513757039588772e-05,
+      "loss": 0.7933,
+      "step": 14797
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.851120273273388e-05,
+      "loss": 0.7311,
+      "step": 14798
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.850864838848679e-05,
+      "loss": 0.7775,
+      "step": 14799
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8506094006874738e-05,
+      "loss": 0.73,
+      "step": 14800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.850353958792491e-05,
+      "loss": 0.8224,
+      "step": 14801
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8500985131664508e-05,
+      "loss": 0.7857,
+      "step": 14802
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8498430638120748e-05,
+      "loss": 0.7611,
+      "step": 14803
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8495876107320823e-05,
+      "loss": 0.6839,
+      "step": 14804
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8493321539291933e-05,
+      "loss": 0.8858,
+      "step": 14805
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.849076693406131e-05,
+      "loss": 0.7572,
+      "step": 14806
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.848821229165612e-05,
+      "loss": 0.6803,
+      "step": 14807
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.848565761210359e-05,
+      "loss": 0.8031,
+      "step": 14808
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8483102895430923e-05,
+      "loss": 0.7309,
+      "step": 14809
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8480548141665326e-05,
+      "loss": 0.7653,
+      "step": 14810
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8477993350833998e-05,
+      "loss": 0.7122,
+      "step": 14811
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8475438522964158e-05,
+      "loss": 0.7533,
+      "step": 14812
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.847288365808301e-05,
+      "loss": 0.7083,
+      "step": 14813
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8470328756217758e-05,
+      "loss": 0.725,
+      "step": 14814
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8467773817395608e-05,
+      "loss": 0.7713,
+      "step": 14815
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.846521884164378e-05,
+      "loss": 0.7555,
+      "step": 14816
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.846266382898947e-05,
+      "loss": 0.7509,
+      "step": 14817
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8460108779459892e-05,
+      "loss": 0.7716,
+      "step": 14818
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8457553693082258e-05,
+      "loss": 0.766,
+      "step": 14819
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8454998569883773e-05,
+      "loss": 0.717,
+      "step": 14820
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8452443409891654e-05,
+      "loss": 0.7143,
+      "step": 14821
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.84498882131331e-05,
+      "loss": 0.6767,
+      "step": 14822
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8447332979635344e-05,
+      "loss": 0.7854,
+      "step": 14823
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.844477770942558e-05,
+      "loss": 0.7449,
+      "step": 14824
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8442222402531023e-05,
+      "loss": 0.7965,
+      "step": 14825
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8439667058978892e-05,
+      "loss": 0.7767,
+      "step": 14826
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8437111678796398e-05,
+      "loss": 0.7161,
+      "step": 14827
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8434556262010737e-05,
+      "loss": 0.7324,
+      "step": 14828
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8432000808649146e-05,
+      "loss": 0.7657,
+      "step": 14829
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8429445318738836e-05,
+      "loss": 0.6657,
+      "step": 14830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8426889792307e-05,
+      "loss": 0.6687,
+      "step": 14831
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8424334229380877e-05,
+      "loss": 0.7509,
+      "step": 14832
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8421778629987672e-05,
+      "loss": 0.7002,
+      "step": 14833
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8419222994154605e-05,
+      "loss": 0.884,
+      "step": 14834
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8416667321908886e-05,
+      "loss": 0.7076,
+      "step": 14835
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8414111613277732e-05,
+      "loss": 0.7459,
+      "step": 14836
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.841155586828837e-05,
+      "loss": 0.7291,
+      "step": 14837
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.840900008696801e-05,
+      "loss": 0.6987,
+      "step": 14838
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8406444269343856e-05,
+      "loss": 0.7642,
+      "step": 14839
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8403888415443147e-05,
+      "loss": 0.7358,
+      "step": 14840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.840133252529309e-05,
+      "loss": 0.8105,
+      "step": 14841
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.839877659892091e-05,
+      "loss": 0.7047,
+      "step": 14842
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.839622063635382e-05,
+      "loss": 0.6798,
+      "step": 14843
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8393664637619044e-05,
+      "loss": 0.6931,
+      "step": 14844
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.83911086027438e-05,
+      "loss": 0.7836,
+      "step": 14845
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8388552531755304e-05,
+      "loss": 0.782,
+      "step": 14846
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.838599642468079e-05,
+      "loss": 0.7497,
+      "step": 14847
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8383440281547456e-05,
+      "loss": 0.6903,
+      "step": 14848
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.838088410238255e-05,
+      "loss": 0.6905,
+      "step": 14849
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.837832788721328e-05,
+      "loss": 0.728,
+      "step": 14850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.837577163606686e-05,
+      "loss": 0.687,
+      "step": 14851
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.837321534897053e-05,
+      "loss": 0.6431,
+      "step": 14852
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8370659025951498e-05,
+      "loss": 0.7686,
+      "step": 14853
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8368102667036995e-05,
+      "loss": 0.7456,
+      "step": 14854
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8365546272254245e-05,
+      "loss": 0.7563,
+      "step": 14855
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8362989841630472e-05,
+      "loss": 0.8067,
+      "step": 14856
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8360433375192896e-05,
+      "loss": 0.7292,
+      "step": 14857
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8357876872968753e-05,
+      "loss": 0.7085,
+      "step": 14858
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.835532033498525e-05,
+      "loss": 0.8244,
+      "step": 14859
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8352763761269624e-05,
+      "loss": 0.6694,
+      "step": 14860
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.83502071518491e-05,
+      "loss": 0.7542,
+      "step": 14861
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.834765050675091e-05,
+      "loss": 0.7041,
+      "step": 14862
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8345093826002263e-05,
+      "loss": 0.7496,
+      "step": 14863
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8342537109630407e-05,
+      "loss": 0.7713,
+      "step": 14864
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8339980357662554e-05,
+      "loss": 0.6703,
+      "step": 14865
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8337423570125944e-05,
+      "loss": 0.7313,
+      "step": 14866
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.83348667470478e-05,
+      "loss": 0.7268,
+      "step": 14867
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8332309888455343e-05,
+      "loss": 0.7067,
+      "step": 14868
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8329752994375812e-05,
+      "loss": 0.7031,
+      "step": 14869
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8327196064836436e-05,
+      "loss": 0.7109,
+      "step": 14870
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8324639099864432e-05,
+      "loss": 0.7007,
+      "step": 14871
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8322082099487047e-05,
+      "loss": 0.7096,
+      "step": 14872
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8319525063731505e-05,
+      "loss": 0.686,
+      "step": 14873
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8316967992625032e-05,
+      "loss": 0.6948,
+      "step": 14874
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.831441088619487e-05,
+      "loss": 0.7865,
+      "step": 14875
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8311853744468242e-05,
+      "loss": 0.7734,
+      "step": 14876
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.830929656747239e-05,
+      "loss": 0.7106,
+      "step": 14877
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.830673935523453e-05,
+      "loss": 0.7687,
+      "step": 14878
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.83041821077819e-05,
+      "loss": 0.6989,
+      "step": 14879
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8301624825141736e-05,
+      "loss": 0.7247,
+      "step": 14880
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.829906750734128e-05,
+      "loss": 0.6744,
+      "step": 14881
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8296510154407757e-05,
+      "loss": 0.7259,
+      "step": 14882
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8293952766368396e-05,
+      "loss": 0.7295,
+      "step": 14883
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8291395343250444e-05,
+      "loss": 0.7199,
+      "step": 14884
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8288837885081132e-05,
+      "loss": 0.6886,
+      "step": 14885
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8286280391887683e-05,
+      "loss": 0.7495,
+      "step": 14886
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8283722863697353e-05,
+      "loss": 0.6831,
+      "step": 14887
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.828116530053736e-05,
+      "loss": 0.7633,
+      "step": 14888
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8278607702434956e-05,
+      "loss": 0.662,
+      "step": 14889
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.827605006941737e-05,
+      "loss": 0.8197,
+      "step": 14890
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.827349240151183e-05,
+      "loss": 0.6531,
+      "step": 14891
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8270934698745593e-05,
+      "loss": 0.6729,
+      "step": 14892
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8268376961145886e-05,
+      "loss": 0.7149,
+      "step": 14893
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.826581918873994e-05,
+      "loss": 0.7239,
+      "step": 14894
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.826326138155501e-05,
+      "loss": 0.7738,
+      "step": 14895
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.826070353961833e-05,
+      "loss": 0.6795,
+      "step": 14896
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8258145662957136e-05,
+      "loss": 0.802,
+      "step": 14897
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8255587751598665e-05,
+      "loss": 0.6705,
+      "step": 14898
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8253029805570164e-05,
+      "loss": 0.7222,
+      "step": 14899
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8250471824898866e-05,
+      "loss": 0.8017,
+      "step": 14900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.824791380961202e-05,
+      "loss": 0.7304,
+      "step": 14901
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.824535575973687e-05,
+      "loss": 0.7111,
+      "step": 14902
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8242797675300638e-05,
+      "loss": 0.772,
+      "step": 14903
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.824023955633059e-05,
+      "loss": 0.7458,
+      "step": 14904
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8237681402853956e-05,
+      "loss": 0.7575,
+      "step": 14905
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8235123214897974e-05,
+      "loss": 0.745,
+      "step": 14906
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8232564992489903e-05,
+      "loss": 0.7122,
+      "step": 14907
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8230006735656973e-05,
+      "loss": 0.7974,
+      "step": 14908
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.822744844442643e-05,
+      "loss": 0.7756,
+      "step": 14909
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8224890118825526e-05,
+      "loss": 0.7711,
+      "step": 14910
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8222331758881494e-05,
+      "loss": 0.678,
+      "step": 14911
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8219773364621588e-05,
+      "loss": 0.7292,
+      "step": 14912
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.821721493607305e-05,
+      "loss": 0.7155,
+      "step": 14913
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8214656473263128e-05,
+      "loss": 0.7847,
+      "step": 14914
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.821209797621906e-05,
+      "loss": 0.693,
+      "step": 14915
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8209539444968103e-05,
+      "loss": 0.7295,
+      "step": 14916
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8206980879537503e-05,
+      "loss": 0.7526,
+      "step": 14917
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.82044222799545e-05,
+      "loss": 0.7266,
+      "step": 14918
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8201863646246353e-05,
+      "loss": 0.7676,
+      "step": 14919
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8199304978440287e-05,
+      "loss": 0.7802,
+      "step": 14920
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8196746276563575e-05,
+      "loss": 0.7607,
+      "step": 14921
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8194187540643457e-05,
+      "loss": 0.6928,
+      "step": 14922
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8191628770707174e-05,
+      "loss": 0.7882,
+      "step": 14923
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8189069966781982e-05,
+      "loss": 0.7396,
+      "step": 14924
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8186511128895144e-05,
+      "loss": 0.7459,
+      "step": 14925
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.818395225707388e-05,
+      "loss": 0.7414,
+      "step": 14926
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.818139335134547e-05,
+      "loss": 0.6844,
+      "step": 14927
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8178834411737154e-05,
+      "loss": 0.7205,
+      "step": 14928
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8176275438276167e-05,
+      "loss": 0.693,
+      "step": 14929
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.817371643098979e-05,
+      "loss": 0.7773,
+      "step": 14930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8171157389905252e-05,
+      "loss": 0.7306,
+      "step": 14931
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.816859831504981e-05,
+      "loss": 0.7799,
+      "step": 14932
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.816603920645073e-05,
+      "loss": 0.7124,
+      "step": 14933
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8163480064135245e-05,
+      "loss": 0.734,
+      "step": 14934
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.816092088813062e-05,
+      "loss": 0.7675,
+      "step": 14935
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.815836167846411e-05,
+      "loss": 0.7753,
+      "step": 14936
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8155802435162964e-05,
+      "loss": 0.7101,
+      "step": 14937
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.815324315825444e-05,
+      "loss": 0.6994,
+      "step": 14938
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.815068384776579e-05,
+      "loss": 0.7629,
+      "step": 14939
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.814812450372426e-05,
+      "loss": 0.7,
+      "step": 14940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8145565126157126e-05,
+      "loss": 0.7647,
+      "step": 14941
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.814300571509163e-05,
+      "loss": 0.7934,
+      "step": 14942
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.814044627055503e-05,
+      "loss": 0.7756,
+      "step": 14943
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8137886792574586e-05,
+      "loss": 0.8021,
+      "step": 14944
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.813532728117756e-05,
+      "loss": 0.7871,
+      "step": 14945
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8132767736391185e-05,
+      "loss": 0.6931,
+      "step": 14946
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.813020815824275e-05,
+      "loss": 0.7178,
+      "step": 14947
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.81276485467595e-05,
+      "loss": 0.6698,
+      "step": 14948
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8125088901968682e-05,
+      "loss": 0.7389,
+      "step": 14949
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.812252922389757e-05,
+      "loss": 0.7145,
+      "step": 14950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8119969512573413e-05,
+      "loss": 0.6939,
+      "step": 14951
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8117409768023473e-05,
+      "loss": 0.7013,
+      "step": 14952
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8114849990275016e-05,
+      "loss": 0.7445,
+      "step": 14953
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.81122901793553e-05,
+      "loss": 0.755,
+      "step": 14954
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.810973033529158e-05,
+      "loss": 0.711,
+      "step": 14955
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.810717045811112e-05,
+      "loss": 0.7113,
+      "step": 14956
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.810461054784118e-05,
+      "loss": 0.7297,
+      "step": 14957
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8102050604509027e-05,
+      "loss": 0.7197,
+      "step": 14958
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8099490628141917e-05,
+      "loss": 0.7246,
+      "step": 14959
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.809693061876711e-05,
+      "loss": 0.793,
+      "step": 14960
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8094370576411878e-05,
+      "loss": 0.7731,
+      "step": 14961
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.809181050110347e-05,
+      "loss": 0.756,
+      "step": 14962
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8089250392869167e-05,
+      "loss": 0.7038,
+      "step": 14963
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8086690251736213e-05,
+      "loss": 0.785,
+      "step": 14964
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8084130077731885e-05,
+      "loss": 0.7103,
+      "step": 14965
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8081569870883444e-05,
+      "loss": 0.7495,
+      "step": 14966
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8079009631218156e-05,
+      "loss": 0.7177,
+      "step": 14967
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.807644935876329e-05,
+      "loss": 0.6899,
+      "step": 14968
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8073889053546098e-05,
+      "loss": 0.8489,
+      "step": 14969
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8071328715593857e-05,
+      "loss": 0.7685,
+      "step": 14970
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8068768344933833e-05,
+      "loss": 0.7499,
+      "step": 14971
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.806620794159328e-05,
+      "loss": 0.6862,
+      "step": 14972
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8063647505599477e-05,
+      "loss": 0.7155,
+      "step": 14973
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8061087036979692e-05,
+      "loss": 0.7194,
+      "step": 14974
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8058526535761187e-05,
+      "loss": 0.722,
+      "step": 14975
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.805596600197123e-05,
+      "loss": 0.7441,
+      "step": 14976
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8053405435637093e-05,
+      "loss": 0.7549,
+      "step": 14977
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8050844836786044e-05,
+      "loss": 0.7555,
+      "step": 14978
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8048284205445348e-05,
+      "loss": 0.7228,
+      "step": 14979
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8045723541642277e-05,
+      "loss": 0.706,
+      "step": 14980
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.804316284540409e-05,
+      "loss": 0.7067,
+      "step": 14981
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8040602116758073e-05,
+      "loss": 0.7684,
+      "step": 14982
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8038041355731494e-05,
+      "loss": 0.7132,
+      "step": 14983
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8035480562351606e-05,
+      "loss": 0.7021,
+      "step": 14984
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8032919736645702e-05,
+      "loss": 0.8669,
+      "step": 14985
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8030358878641046e-05,
+      "loss": 0.7594,
+      "step": 14986
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8027797988364902e-05,
+      "loss": 0.7585,
+      "step": 14987
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8025237065844555e-05,
+      "loss": 0.6932,
+      "step": 14988
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8022676111107266e-05,
+      "loss": 0.683,
+      "step": 14989
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8020115124180306e-05,
+      "loss": 0.6874,
+      "step": 14990
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8017554105090966e-05,
+      "loss": 0.6847,
+      "step": 14991
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8014993053866494e-05,
+      "loss": 0.7916,
+      "step": 14992
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.801243197053418e-05,
+      "loss": 0.828,
+      "step": 14993
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.80098708551213e-05,
+      "loss": 0.73,
+      "step": 14994
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8007309707655116e-05,
+      "loss": 0.6976,
+      "step": 14995
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.800474852816291e-05,
+      "loss": 0.7064,
+      "step": 14996
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8002187316671963e-05,
+      "loss": 0.7293,
+      "step": 14997
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.799962607320954e-05,
+      "loss": 0.7765,
+      "step": 14998
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.799706479780292e-05,
+      "loss": 0.691,
+      "step": 14999
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7994503490479384e-05,
+      "loss": 0.735,
+      "step": 15000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7991942151266193e-05,
+      "loss": 0.7142,
+      "step": 15001
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7989380780190645e-05,
+      "loss": 0.8554,
+      "step": 15002
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7986819377279997e-05,
+      "loss": 0.7117,
+      "step": 15003
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.798425794256154e-05,
+      "loss": 0.7625,
+      "step": 15004
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.798169647606255e-05,
+      "loss": 0.6991,
+      "step": 15005
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7979134977810305e-05,
+      "loss": 0.7323,
+      "step": 15006
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.797657344783207e-05,
+      "loss": 0.8013,
+      "step": 15007
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7974011886155156e-05,
+      "loss": 0.6724,
+      "step": 15008
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.79714502928068e-05,
+      "loss": 0.7371,
+      "step": 15009
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7968888667814314e-05,
+      "loss": 0.7194,
+      "step": 15010
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7966327011204963e-05,
+      "loss": 0.7067,
+      "step": 15011
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7963765323006025e-05,
+      "loss": 0.6881,
+      "step": 15012
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.796120360324479e-05,
+      "loss": 0.9065,
+      "step": 15013
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7958641851948536e-05,
+      "loss": 0.6744,
+      "step": 15014
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.795608006914454e-05,
+      "loss": 0.7479,
+      "step": 15015
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7953518254860083e-05,
+      "loss": 0.7476,
+      "step": 15016
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.795095640912246e-05,
+      "loss": 0.7013,
+      "step": 15017
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.794839453195894e-05,
+      "loss": 0.7626,
+      "step": 15018
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.79458326233968e-05,
+      "loss": 0.6761,
+      "step": 15019
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.794327068346334e-05,
+      "loss": 0.7236,
+      "step": 15020
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7940708712185827e-05,
+      "loss": 0.6968,
+      "step": 15021
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7938146709591557e-05,
+      "loss": 0.6681,
+      "step": 15022
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7935584675707805e-05,
+      "loss": 0.683,
+      "step": 15023
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7933022610561855e-05,
+      "loss": 0.7253,
+      "step": 15024
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7930460514181e-05,
+      "loss": 0.7818,
+      "step": 15025
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.792789838659252e-05,
+      "loss": 0.8084,
+      "step": 15026
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7925336227823694e-05,
+      "loss": 0.7714,
+      "step": 15027
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7922774037901826e-05,
+      "loss": 0.7279,
+      "step": 15028
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.792021181685418e-05,
+      "loss": 0.7268,
+      "step": 15029
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.791764956470805e-05,
+      "loss": 0.7049,
+      "step": 15030
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7915087281490727e-05,
+      "loss": 0.7587,
+      "step": 15031
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.791252496722949e-05,
+      "loss": 0.717,
+      "step": 15032
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.790996262195163e-05,
+      "loss": 0.6816,
+      "step": 15033
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.790740024568444e-05,
+      "loss": 0.7219,
+      "step": 15034
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.79048378384552e-05,
+      "loss": 0.7203,
+      "step": 15035
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.79022754002912e-05,
+      "loss": 0.7182,
+      "step": 15036
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.789971293121973e-05,
+      "loss": 0.7285,
+      "step": 15037
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7897150431268076e-05,
+      "loss": 0.6847,
+      "step": 15038
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.789458790046353e-05,
+      "loss": 0.6784,
+      "step": 15039
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7892025338833384e-05,
+      "loss": 0.7115,
+      "step": 15040
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7889462746404915e-05,
+      "loss": 0.683,
+      "step": 15041
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7886900123205434e-05,
+      "loss": 0.7217,
+      "step": 15042
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.788433746926221e-05,
+      "loss": 0.7998,
+      "step": 15043
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7881774784602543e-05,
+      "loss": 0.7419,
+      "step": 15044
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.787921206925373e-05,
+      "loss": 0.679,
+      "step": 15045
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7876649323243054e-05,
+      "loss": 0.7679,
+      "step": 15046
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7874086546597804e-05,
+      "loss": 0.6944,
+      "step": 15047
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7871523739345283e-05,
+      "loss": 0.6782,
+      "step": 15048
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7868960901512786e-05,
+      "loss": 0.7628,
+      "step": 15049
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7866398033127583e-05,
+      "loss": 0.7143,
+      "step": 15050
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7863835134216987e-05,
+      "loss": 0.7705,
+      "step": 15051
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7861272204808288e-05,
+      "loss": 0.7114,
+      "step": 15052
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7858709244928765e-05,
+      "loss": 0.7121,
+      "step": 15053
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7856146254605732e-05,
+      "loss": 0.7061,
+      "step": 15054
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7853583233866475e-05,
+      "loss": 0.6645,
+      "step": 15055
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.785102018273829e-05,
+      "loss": 0.793,
+      "step": 15056
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.784845710124847e-05,
+      "loss": 0.7487,
+      "step": 15057
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7845893989424317e-05,
+      "loss": 0.6436,
+      "step": 15058
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7843330847293115e-05,
+      "loss": 0.7346,
+      "step": 15059
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7840767674882167e-05,
+      "loss": 0.7729,
+      "step": 15060
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7838204472218766e-05,
+      "loss": 0.7385,
+      "step": 15061
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7835641239330213e-05,
+      "loss": 0.6197,
+      "step": 15062
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7833077976243798e-05,
+      "loss": 0.7099,
+      "step": 15063
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7830514682986825e-05,
+      "loss": 0.7437,
+      "step": 15064
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7827951359586592e-05,
+      "loss": 0.6154,
+      "step": 15065
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.782538800607039e-05,
+      "loss": 0.7301,
+      "step": 15066
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7822824622465525e-05,
+      "loss": 0.7664,
+      "step": 15067
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7820261208799285e-05,
+      "loss": 0.7424,
+      "step": 15068
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7817697765098988e-05,
+      "loss": 0.7248,
+      "step": 15069
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7815134291391908e-05,
+      "loss": 0.7315,
+      "step": 15070
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.781257078770536e-05,
+      "loss": 0.6968,
+      "step": 15071
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7810007254066644e-05,
+      "loss": 0.6962,
+      "step": 15072
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7807443690503056e-05,
+      "loss": 0.7908,
+      "step": 15073
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7804880097041892e-05,
+      "loss": 0.7943,
+      "step": 15074
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7802316473710467e-05,
+      "loss": 0.7227,
+      "step": 15075
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7799752820536062e-05,
+      "loss": 0.6906,
+      "step": 15076
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7797189137545993e-05,
+      "loss": 0.753,
+      "step": 15077
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7794625424767563e-05,
+      "loss": 0.7834,
+      "step": 15078
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7792061682228065e-05,
+      "loss": 0.641,
+      "step": 15079
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.778949790995481e-05,
+      "loss": 0.712,
+      "step": 15080
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7786934107975094e-05,
+      "loss": 0.7205,
+      "step": 15081
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7784370276316214e-05,
+      "loss": 0.6737,
+      "step": 15082
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7781806415005485e-05,
+      "loss": 0.7218,
+      "step": 15083
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.777924252407021e-05,
+      "loss": 0.7578,
+      "step": 15084
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.777667860353768e-05,
+      "loss": 0.7744,
+      "step": 15085
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.777411465343522e-05,
+      "loss": 0.777,
+      "step": 15086
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.777155067379012e-05,
+      "loss": 0.7155,
+      "step": 15087
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.776898666462968e-05,
+      "loss": 0.7715,
+      "step": 15088
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.776642262598123e-05,
+      "loss": 0.7654,
+      "step": 15089
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7763858557872046e-05,
+      "loss": 0.7997,
+      "step": 15090
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7761294460329445e-05,
+      "loss": 0.7658,
+      "step": 15091
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7758730333380744e-05,
+      "loss": 0.8288,
+      "step": 15092
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7756166177053227e-05,
+      "loss": 0.7794,
+      "step": 15093
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7753601991374223e-05,
+      "loss": 0.719,
+      "step": 15094
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.775103777637103e-05,
+      "loss": 0.7803,
+      "step": 15095
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7748473532070945e-05,
+      "loss": 0.674,
+      "step": 15096
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7745909258501295e-05,
+      "loss": 0.7583,
+      "step": 15097
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7743344955689378e-05,
+      "loss": 0.7349,
+      "step": 15098
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7740780623662504e-05,
+      "loss": 0.7778,
+      "step": 15099
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.773821626244798e-05,
+      "loss": 0.6742,
+      "step": 15100
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.773565187207311e-05,
+      "loss": 0.7853,
+      "step": 15101
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.773308745256521e-05,
+      "loss": 0.8385,
+      "step": 15102
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7730523003951586e-05,
+      "loss": 0.7402,
+      "step": 15103
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7727958526259555e-05,
+      "loss": 0.7443,
+      "step": 15104
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7725394019516416e-05,
+      "loss": 0.6908,
+      "step": 15105
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7722829483749492e-05,
+      "loss": 0.7852,
+      "step": 15106
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7720264918986083e-05,
+      "loss": 0.724,
+      "step": 15107
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7717700325253503e-05,
+      "loss": 0.7938,
+      "step": 15108
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.771513570257907e-05,
+      "loss": 0.765,
+      "step": 15109
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7712571050990083e-05,
+      "loss": 0.6647,
+      "step": 15110
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.771000637051387e-05,
+      "loss": 0.7126,
+      "step": 15111
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.770744166117773e-05,
+      "loss": 0.736,
+      "step": 15112
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.770487692300897e-05,
+      "loss": 0.6873,
+      "step": 15113
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7702312156034928e-05,
+      "loss": 0.7701,
+      "step": 15114
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7699747360282896e-05,
+      "loss": 0.7726,
+      "step": 15115
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.769718253578019e-05,
+      "loss": 0.7066,
+      "step": 15116
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7694617682554132e-05,
+      "loss": 0.7809,
+      "step": 15117
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7692052800632036e-05,
+      "loss": 0.7904,
+      "step": 15118
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7689487890041205e-05,
+      "loss": 0.7131,
+      "step": 15119
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7686922950808962e-05,
+      "loss": 0.7303,
+      "step": 15120
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7684357982962627e-05,
+      "loss": 0.7067,
+      "step": 15121
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7681792986529497e-05,
+      "loss": 0.711,
+      "step": 15122
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.767922796153691e-05,
+      "loss": 0.7236,
+      "step": 15123
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.767666290801217e-05,
+      "loss": 0.7136,
+      "step": 15124
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7674097825982588e-05,
+      "loss": 0.7432,
+      "step": 15125
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7671532715475495e-05,
+      "loss": 0.7908,
+      "step": 15126
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7668967576518202e-05,
+      "loss": 0.7097,
+      "step": 15127
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7666402409138014e-05,
+      "loss": 0.6756,
+      "step": 15128
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7663837213362276e-05,
+      "loss": 0.738,
+      "step": 15129
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7661271989218272e-05,
+      "loss": 0.7165,
+      "step": 15130
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7658706736733347e-05,
+      "loss": 0.7568,
+      "step": 15131
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.765614145593481e-05,
+      "loss": 0.7526,
+      "step": 15132
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.765357614684998e-05,
+      "loss": 0.7291,
+      "step": 15133
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7651010809506166e-05,
+      "loss": 0.7219,
+      "step": 15134
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7648445443930704e-05,
+      "loss": 0.7432,
+      "step": 15135
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.76458800501509e-05,
+      "loss": 0.7021,
+      "step": 15136
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7643314628194083e-05,
+      "loss": 0.7359,
+      "step": 15137
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.764074917808757e-05,
+      "loss": 0.7249,
+      "step": 15138
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.763818369985868e-05,
+      "loss": 0.7041,
+      "step": 15139
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.763561819353474e-05,
+      "loss": 0.7904,
+      "step": 15140
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7633052659143066e-05,
+      "loss": 0.7131,
+      "step": 15141
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7630487096710973e-05,
+      "loss": 0.7236,
+      "step": 15142
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7627921506265796e-05,
+      "loss": 0.7704,
+      "step": 15143
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7625355887834848e-05,
+      "loss": 0.7005,
+      "step": 15144
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.762279024144545e-05,
+      "loss": 0.7382,
+      "step": 15145
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.762022456712493e-05,
+      "loss": 0.7014,
+      "step": 15146
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7617658864900613e-05,
+      "loss": 0.7094,
+      "step": 15147
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.761509313479981e-05,
+      "loss": 0.7115,
+      "step": 15148
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.761252737684987e-05,
+      "loss": 0.7581,
+      "step": 15149
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.760996159107809e-05,
+      "loss": 0.7363,
+      "step": 15150
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7607395777511795e-05,
+      "loss": 0.6789,
+      "step": 15151
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7604829936178326e-05,
+      "loss": 0.7157,
+      "step": 15152
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7602264067105006e-05,
+      "loss": 0.6629,
+      "step": 15153
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7599698170319143e-05,
+      "loss": 0.7195,
+      "step": 15154
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7597132245848072e-05,
+      "loss": 0.7716,
+      "step": 15155
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7594566293719126e-05,
+      "loss": 0.6855,
+      "step": 15156
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7592000313959622e-05,
+      "loss": 0.6975,
+      "step": 15157
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.758943430659689e-05,
+      "loss": 0.7068,
+      "step": 15158
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.758686827165825e-05,
+      "loss": 0.7371,
+      "step": 15159
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7584302209171037e-05,
+      "loss": 0.7702,
+      "step": 15160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7581736119162576e-05,
+      "loss": 0.7269,
+      "step": 15161
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7579170001660188e-05,
+      "loss": 0.7043,
+      "step": 15162
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7576603856691208e-05,
+      "loss": 0.6733,
+      "step": 15163
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.757403768428296e-05,
+      "loss": 0.6842,
+      "step": 15164
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7571471484462774e-05,
+      "loss": 0.7591,
+      "step": 15165
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7568905257257977e-05,
+      "loss": 0.7406,
+      "step": 15166
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.75663390026959e-05,
+      "loss": 0.7497,
+      "step": 15167
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.756377272080387e-05,
+      "loss": 0.6555,
+      "step": 15168
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.756120641160922e-05,
+      "loss": 0.7382,
+      "step": 15169
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.755864007513928e-05,
+      "loss": 0.7897,
+      "step": 15170
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7556073711421365e-05,
+      "loss": 0.7324,
+      "step": 15171
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7553507320482826e-05,
+      "loss": 0.7474,
+      "step": 15172
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.755094090235098e-05,
+      "loss": 0.661,
+      "step": 15173
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7548374457053162e-05,
+      "loss": 0.6982,
+      "step": 15174
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.75458079846167e-05,
+      "loss": 0.7358,
+      "step": 15175
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7543241485068938e-05,
+      "loss": 0.7858,
+      "step": 15176
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7540674958437185e-05,
+      "loss": 0.6982,
+      "step": 15177
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7538108404748797e-05,
+      "loss": 0.6917,
+      "step": 15178
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7535541824031096e-05,
+      "loss": 0.6967,
+      "step": 15179
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.753297521631141e-05,
+      "loss": 0.7514,
+      "step": 15180
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.753040858161708e-05,
+      "loss": 0.721,
+      "step": 15181
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7527841919975423e-05,
+      "loss": 0.7809,
+      "step": 15182
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7525275231413793e-05,
+      "loss": 0.7332,
+      "step": 15183
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7522708515959517e-05,
+      "loss": 0.747,
+      "step": 15184
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7520141773639917e-05,
+      "loss": 0.6651,
+      "step": 15185
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.751757500448235e-05,
+      "loss": 0.7319,
+      "step": 15186
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7515008208514133e-05,
+      "loss": 0.7076,
+      "step": 15187
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7512441385762606e-05,
+      "loss": 0.7311,
+      "step": 15188
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7509874536255096e-05,
+      "loss": 0.7317,
+      "step": 15189
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.750730766001896e-05,
+      "loss": 0.7167,
+      "step": 15190
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7504740757081508e-05,
+      "loss": 0.7149,
+      "step": 15191
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.750217382747009e-05,
+      "loss": 0.7507,
+      "step": 15192
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7499606871212048e-05,
+      "loss": 0.7094,
+      "step": 15193
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.74970398883347e-05,
+      "loss": 0.7744,
+      "step": 15194
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7494472878865397e-05,
+      "loss": 0.7677,
+      "step": 15195
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7491905842831468e-05,
+      "loss": 0.6967,
+      "step": 15196
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7489338780260254e-05,
+      "loss": 0.7929,
+      "step": 15197
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.74867716911791e-05,
+      "loss": 0.7526,
+      "step": 15198
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7484204575615335e-05,
+      "loss": 0.7239,
+      "step": 15199
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7481637433596304e-05,
+      "loss": 0.6673,
+      "step": 15200
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7479070265149336e-05,
+      "loss": 0.7347,
+      "step": 15201
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7476503070301778e-05,
+      "loss": 0.7745,
+      "step": 15202
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7473935849080955e-05,
+      "loss": 0.6848,
+      "step": 15203
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.747136860151423e-05,
+      "loss": 0.8203,
+      "step": 15204
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7468801327628925e-05,
+      "loss": 0.7242,
+      "step": 15205
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.746623402745238e-05,
+      "loss": 0.6975,
+      "step": 15206
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7463666701011948e-05,
+      "loss": 0.705,
+      "step": 15207
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7461099348334955e-05,
+      "loss": 0.7457,
+      "step": 15208
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7458531969448754e-05,
+      "loss": 0.7827,
+      "step": 15209
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7455964564380683e-05,
+      "loss": 0.7955,
+      "step": 15210
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.745339713315807e-05,
+      "loss": 0.6347,
+      "step": 15211
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.745082967580827e-05,
+      "loss": 0.6831,
+      "step": 15212
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7448262192358626e-05,
+      "loss": 0.7111,
+      "step": 15213
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7445694682836472e-05,
+      "loss": 0.7557,
+      "step": 15214
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7443127147269155e-05,
+      "loss": 0.7483,
+      "step": 15215
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7440559585684022e-05,
+      "loss": 0.7581,
+      "step": 15216
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7437991998108405e-05,
+      "loss": 0.7354,
+      "step": 15217
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7435424384569652e-05,
+      "loss": 0.7153,
+      "step": 15218
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7432856745095113e-05,
+      "loss": 0.7105,
+      "step": 15219
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.743028907971213e-05,
+      "loss": 0.6801,
+      "step": 15220
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7427721388448035e-05,
+      "loss": 0.5942,
+      "step": 15221
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.742515367133019e-05,
+      "loss": 0.749,
+      "step": 15222
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7422585928385918e-05,
+      "loss": 0.6781,
+      "step": 15223
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.742001815964259e-05,
+      "loss": 0.7816,
+      "step": 15224
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7417450365127534e-05,
+      "loss": 0.6582,
+      "step": 15225
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7414882544868094e-05,
+      "loss": 0.7284,
+      "step": 15226
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.741231469889162e-05,
+      "loss": 0.7699,
+      "step": 15227
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.740974682722547e-05,
+      "loss": 0.7087,
+      "step": 15228
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.740717892989697e-05,
+      "loss": 0.7074,
+      "step": 15229
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7404611006933485e-05,
+      "loss": 0.7255,
+      "step": 15230
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7402043058362343e-05,
+      "loss": 0.7769,
+      "step": 15231
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.73994750842109e-05,
+      "loss": 0.821,
+      "step": 15232
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7396907084506513e-05,
+      "loss": 0.7467,
+      "step": 15233
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7394339059276514e-05,
+      "loss": 0.7212,
+      "step": 15234
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.739177100854826e-05,
+      "loss": 0.6519,
+      "step": 15235
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7389202932349095e-05,
+      "loss": 0.7387,
+      "step": 15236
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.738663483070637e-05,
+      "loss": 0.722,
+      "step": 15237
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.738406670364743e-05,
+      "loss": 0.7214,
+      "step": 15238
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7381498551199636e-05,
+      "loss": 0.7791,
+      "step": 15239
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7378930373390323e-05,
+      "loss": 0.7498,
+      "step": 15240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.737636217024685e-05,
+      "loss": 0.6569,
+      "step": 15241
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.737379394179656e-05,
+      "loss": 0.7326,
+      "step": 15242
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.73712256880668e-05,
+      "loss": 0.7323,
+      "step": 15243
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.736865740908493e-05,
+      "loss": 0.6311,
+      "step": 15244
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7366089104878297e-05,
+      "loss": 0.7367,
+      "step": 15245
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.736352077547425e-05,
+      "loss": 0.7349,
+      "step": 15246
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7360952420900145e-05,
+      "loss": 0.7463,
+      "step": 15247
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7358384041183326e-05,
+      "loss": 0.675,
+      "step": 15248
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7355815636351145e-05,
+      "loss": 0.6923,
+      "step": 15249
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7353247206430965e-05,
+      "loss": 0.6484,
+      "step": 15250
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7350678751450133e-05,
+      "loss": 0.7013,
+      "step": 15251
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7348110271436e-05,
+      "loss": 0.6506,
+      "step": 15252
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7345541766415915e-05,
+      "loss": 0.7367,
+      "step": 15253
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7342973236417235e-05,
+      "loss": 0.7671,
+      "step": 15254
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.73404046814673e-05,
+      "loss": 0.7822,
+      "step": 15255
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7337836101593494e-05,
+      "loss": 0.7269,
+      "step": 15256
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7335267496823145e-05,
+      "loss": 0.7949,
+      "step": 15257
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7332698867183615e-05,
+      "loss": 0.7021,
+      "step": 15258
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7330130212702258e-05,
+      "loss": 0.6838,
+      "step": 15259
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.732756153340643e-05,
+      "loss": 0.7083,
+      "step": 15260
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7324992829323486e-05,
+      "loss": 0.6781,
+      "step": 15261
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.732242410048078e-05,
+      "loss": 0.7333,
+      "step": 15262
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7319855346905667e-05,
+      "loss": 0.7288,
+      "step": 15263
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.73172865686255e-05,
+      "loss": 0.7469,
+      "step": 15264
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.731471776566764e-05,
+      "loss": 0.7285,
+      "step": 15265
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7312148938059435e-05,
+      "loss": 0.6935,
+      "step": 15266
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.730958008582826e-05,
+      "loss": 0.7125,
+      "step": 15267
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7307011209001448e-05,
+      "loss": 0.6426,
+      "step": 15268
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.730444230760637e-05,
+      "loss": 0.7117,
+      "step": 15269
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7301873381670383e-05,
+      "loss": 0.7573,
+      "step": 15270
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.729930443122085e-05,
+      "loss": 0.7576,
+      "step": 15271
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7296735456285107e-05,
+      "loss": 0.753,
+      "step": 15272
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.729416645689053e-05,
+      "loss": 0.7787,
+      "step": 15273
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7291597433064476e-05,
+      "loss": 0.7176,
+      "step": 15274
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7289028384834297e-05,
+      "loss": 0.746,
+      "step": 15275
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7286459312227357e-05,
+      "loss": 0.6878,
+      "step": 15276
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7283890215271012e-05,
+      "loss": 0.7248,
+      "step": 15277
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7281321093992624e-05,
+      "loss": 0.6401,
+      "step": 15278
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7278751948419552e-05,
+      "loss": 0.7307,
+      "step": 15279
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7276182778579155e-05,
+      "loss": 0.7157,
+      "step": 15280
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7273613584498797e-05,
+      "loss": 0.6626,
+      "step": 15281
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7271044366205832e-05,
+      "loss": 0.7157,
+      "step": 15282
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7268475123727617e-05,
+      "loss": 0.7533,
+      "step": 15283
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7265905857091524e-05,
+      "loss": 0.7131,
+      "step": 15284
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.726333656632491e-05,
+      "loss": 0.6914,
+      "step": 15285
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7260767251455132e-05,
+      "loss": 0.7092,
+      "step": 15286
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7258197912509557e-05,
+      "loss": 0.7331,
+      "step": 15287
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.725562854951555e-05,
+      "loss": 0.7909,
+      "step": 15288
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.725305916250046e-05,
+      "loss": 0.6872,
+      "step": 15289
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7250489751491665e-05,
+      "loss": 0.7067,
+      "step": 15290
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.724792031651653e-05,
+      "loss": 0.7472,
+      "step": 15291
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7245350857602392e-05,
+      "loss": 0.7468,
+      "step": 15292
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7242781374776633e-05,
+      "loss": 0.7339,
+      "step": 15293
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.724021186806662e-05,
+      "loss": 0.7092,
+      "step": 15294
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.72376423374997e-05,
+      "loss": 0.767,
+      "step": 15295
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7235072783103254e-05,
+      "loss": 0.7014,
+      "step": 15296
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7232503204904648e-05,
+      "loss": 0.7407,
+      "step": 15297
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7229933602931223e-05,
+      "loss": 0.7008,
+      "step": 15298
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.722736397721037e-05,
+      "loss": 0.7115,
+      "step": 15299
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.722479432776944e-05,
+      "loss": 0.722,
+      "step": 15300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7222224654635804e-05,
+      "loss": 0.6868,
+      "step": 15301
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.721965495783682e-05,
+      "loss": 0.6671,
+      "step": 15302
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.721708523739985e-05,
+      "loss": 0.6537,
+      "step": 15303
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7214515493352283e-05,
+      "loss": 0.7183,
+      "step": 15304
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7211945725721467e-05,
+      "loss": 0.6968,
+      "step": 15305
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.720937593453477e-05,
+      "loss": 0.6641,
+      "step": 15306
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7206806119819554e-05,
+      "loss": 0.6717,
+      "step": 15307
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7204236281603203e-05,
+      "loss": 0.687,
+      "step": 15308
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7201666419913075e-05,
+      "loss": 0.6885,
+      "step": 15309
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7199096534776526e-05,
+      "loss": 0.7504,
+      "step": 15310
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.719652662622095e-05,
+      "loss": 0.7009,
+      "step": 15311
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.719395669427368e-05,
+      "loss": 0.8064,
+      "step": 15312
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.719138673896211e-05,
+      "loss": 0.8382,
+      "step": 15313
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7188816760313606e-05,
+      "loss": 0.6941,
+      "step": 15314
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7186246758355526e-05,
+      "loss": 0.8,
+      "step": 15315
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.718367673311525e-05,
+      "loss": 0.764,
+      "step": 15316
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7181106684620146e-05,
+      "loss": 0.7148,
+      "step": 15317
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.717853661289757e-05,
+      "loss": 0.7371,
+      "step": 15318
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7175966517974903e-05,
+      "loss": 0.7247,
+      "step": 15319
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7173396399879523e-05,
+      "loss": 0.7529,
+      "step": 15320
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.717082625863879e-05,
+      "loss": 0.7231,
+      "step": 15321
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.716825609428007e-05,
+      "loss": 0.76,
+      "step": 15322
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.716568590683074e-05,
+      "loss": 0.7043,
+      "step": 15323
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7163115696318164e-05,
+      "loss": 0.6556,
+      "step": 15324
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7160545462769727e-05,
+      "loss": 0.7365,
+      "step": 15325
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7157975206212795e-05,
+      "loss": 0.7335,
+      "step": 15326
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7155404926674727e-05,
+      "loss": 0.7252,
+      "step": 15327
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.715283462418291e-05,
+      "loss": 0.7399,
+      "step": 15328
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7150264298764717e-05,
+      "loss": 0.7118,
+      "step": 15329
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7147693950447507e-05,
+      "loss": 0.7384,
+      "step": 15330
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7145123579258668e-05,
+      "loss": 0.7723,
+      "step": 15331
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7142553185225555e-05,
+      "loss": 0.6837,
+      "step": 15332
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.713998276837556e-05,
+      "loss": 0.7826,
+      "step": 15333
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7137412328736046e-05,
+      "loss": 0.6893,
+      "step": 15334
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.713484186633438e-05,
+      "loss": 0.8206,
+      "step": 15335
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.713227138119795e-05,
+      "loss": 0.7491,
+      "step": 15336
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.712970087335413e-05,
+      "loss": 0.7128,
+      "step": 15337
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7127130342830275e-05,
+      "loss": 0.6968,
+      "step": 15338
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7124559789653786e-05,
+      "loss": 0.7828,
+      "step": 15339
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7121989213852023e-05,
+      "loss": 0.6961,
+      "step": 15340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7119418615452363e-05,
+      "loss": 0.706,
+      "step": 15341
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7116847994482182e-05,
+      "loss": 0.7108,
+      "step": 15342
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7114277350968853e-05,
+      "loss": 0.7605,
+      "step": 15343
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.711170668493975e-05,
+      "loss": 0.7782,
+      "step": 15344
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7109135996422258e-05,
+      "loss": 0.7284,
+      "step": 15345
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7106565285443747e-05,
+      "loss": 0.783,
+      "step": 15346
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7103994552031592e-05,
+      "loss": 0.7515,
+      "step": 15347
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7101423796213177e-05,
+      "loss": 0.7355,
+      "step": 15348
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.709885301801587e-05,
+      "loss": 0.7687,
+      "step": 15349
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.709628221746705e-05,
+      "loss": 0.7268,
+      "step": 15350
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.709371139459411e-05,
+      "loss": 0.6505,
+      "step": 15351
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7091140549424403e-05,
+      "loss": 0.6979,
+      "step": 15352
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.708856968198532e-05,
+      "loss": 0.7389,
+      "step": 15353
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7085998792304247e-05,
+      "loss": 0.7546,
+      "step": 15354
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7083427880408542e-05,
+      "loss": 0.694,
+      "step": 15355
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.70808569463256e-05,
+      "loss": 0.728,
+      "step": 15356
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7078285990082796e-05,
+      "loss": 0.7626,
+      "step": 15357
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7075715011707503e-05,
+      "loss": 0.7721,
+      "step": 15358
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7073144011227107e-05,
+      "loss": 0.7595,
+      "step": 15359
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.707057298866899e-05,
+      "loss": 0.7674,
+      "step": 15360
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.706800194406053e-05,
+      "loss": 0.7397,
+      "step": 15361
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.70654308774291e-05,
+      "loss": 0.626,
+      "step": 15362
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.706285978880208e-05,
+      "loss": 0.7584,
+      "step": 15363
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7060288678206856e-05,
+      "loss": 0.6923,
+      "step": 15364
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.705771754567082e-05,
+      "loss": 0.7181,
+      "step": 15365
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.705514639122133e-05,
+      "loss": 0.7339,
+      "step": 15366
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7052575214885785e-05,
+      "loss": 0.7036,
+      "step": 15367
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7050004016691554e-05,
+      "loss": 0.7231,
+      "step": 15368
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7047432796666033e-05,
+      "loss": 0.6415,
+      "step": 15369
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7044861554836588e-05,
+      "loss": 0.7616,
+      "step": 15370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7042290291230615e-05,
+      "loss": 0.7013,
+      "step": 15371
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7039719005875498e-05,
+      "loss": 0.6978,
+      "step": 15372
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7037147698798594e-05,
+      "loss": 0.6526,
+      "step": 15373
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7034576370027313e-05,
+      "loss": 0.7345,
+      "step": 15374
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.703200501958903e-05,
+      "loss": 0.6963,
+      "step": 15375
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.702943364751112e-05,
+      "loss": 0.7701,
+      "step": 15376
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7026862253820978e-05,
+      "loss": 0.6848,
+      "step": 15377
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7024290838545986e-05,
+      "loss": 0.6849,
+      "step": 15378
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7021719401713518e-05,
+      "loss": 0.7428,
+      "step": 15379
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7019147943350977e-05,
+      "loss": 0.7964,
+      "step": 15380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.701657646348573e-05,
+      "loss": 0.7078,
+      "step": 15381
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7014004962145172e-05,
+      "loss": 0.6673,
+      "step": 15382
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.701143343935668e-05,
+      "loss": 0.7452,
+      "step": 15383
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7008861895147642e-05,
+      "loss": 0.7191,
+      "step": 15384
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7006290329545448e-05,
+      "loss": 0.73,
+      "step": 15385
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7003718742577477e-05,
+      "loss": 0.7856,
+      "step": 15386
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7001147134271114e-05,
+      "loss": 0.6925,
+      "step": 15387
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6998575504653757e-05,
+      "loss": 0.727,
+      "step": 15388
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6996003853752783e-05,
+      "loss": 0.6956,
+      "step": 15389
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6993432181595572e-05,
+      "loss": 0.7725,
+      "step": 15390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6990860488209523e-05,
+      "loss": 0.7035,
+      "step": 15391
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.698828877362203e-05,
+      "loss": 0.7959,
+      "step": 15392
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6985717037860453e-05,
+      "loss": 0.6848,
+      "step": 15393
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6983145280952205e-05,
+      "loss": 0.7798,
+      "step": 15394
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.698057350292466e-05,
+      "loss": 0.6807,
+      "step": 15395
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6978001703805206e-05,
+      "loss": 0.7664,
+      "step": 15396
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.697542988362124e-05,
+      "loss": 0.7572,
+      "step": 15397
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.697285804240014e-05,
+      "loss": 0.7608,
+      "step": 15398
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6970286180169302e-05,
+      "loss": 0.7368,
+      "step": 15399
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6967714296956115e-05,
+      "loss": 0.7593,
+      "step": 15400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6965142392787963e-05,
+      "loss": 0.7211,
+      "step": 15401
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.696257046769224e-05,
+      "loss": 0.7319,
+      "step": 15402
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6959998521696327e-05,
+      "loss": 0.7477,
+      "step": 15403
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.695742655482762e-05,
+      "loss": 0.7701,
+      "step": 15404
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6954854567113514e-05,
+      "loss": 0.7359,
+      "step": 15405
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6952282558581393e-05,
+      "loss": 0.6921,
+      "step": 15406
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.694971052925864e-05,
+      "loss": 0.7708,
+      "step": 15407
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6947138479172666e-05,
+      "loss": 0.7186,
+      "step": 15408
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6944566408350847e-05,
+      "loss": 0.7841,
+      "step": 15409
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.694199431682057e-05,
+      "loss": 0.6962,
+      "step": 15410
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6939422204609233e-05,
+      "loss": 0.6742,
+      "step": 15411
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6936850071744236e-05,
+      "loss": 0.7205,
+      "step": 15412
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6934277918252952e-05,
+      "loss": 0.637,
+      "step": 15413
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.693170574416279e-05,
+      "loss": 0.743,
+      "step": 15414
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6929133549501128e-05,
+      "loss": 0.8528,
+      "step": 15415
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6926561334295363e-05,
+      "loss": 0.6912,
+      "step": 15416
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6923989098572895e-05,
+      "loss": 0.7703,
+      "step": 15417
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6921416842361113e-05,
+      "loss": 0.6839,
+      "step": 15418
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.69188445656874e-05,
+      "loss": 0.7154,
+      "step": 15419
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6916272268579164e-05,
+      "loss": 0.9459,
+      "step": 15420
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6913699951063786e-05,
+      "loss": 0.7752,
+      "step": 15421
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6911127613168673e-05,
+      "loss": 0.6618,
+      "step": 15422
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.690855525492121e-05,
+      "loss": 0.6978,
+      "step": 15423
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.690598287634879e-05,
+      "loss": 0.6834,
+      "step": 15424
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6903410477478806e-05,
+      "loss": 0.7965,
+      "step": 15425
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.690083805833866e-05,
+      "loss": 0.7285,
+      "step": 15426
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6898265618955744e-05,
+      "loss": 0.6802,
+      "step": 15427
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6895693159357443e-05,
+      "loss": 0.6936,
+      "step": 15428
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6893120679571172e-05,
+      "loss": 0.8217,
+      "step": 15429
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6890548179624308e-05,
+      "loss": 0.6842,
+      "step": 15430
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6887975659544257e-05,
+      "loss": 0.6745,
+      "step": 15431
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6885403119358414e-05,
+      "loss": 0.7392,
+      "step": 15432
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6882830559094165e-05,
+      "loss": 0.6672,
+      "step": 15433
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6880257978778924e-05,
+      "loss": 0.7678,
+      "step": 15434
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.687768537844007e-05,
+      "loss": 0.8339,
+      "step": 15435
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6875112758105004e-05,
+      "loss": 0.732,
+      "step": 15436
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6872540117801128e-05,
+      "loss": 0.7241,
+      "step": 15437
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.686996745755584e-05,
+      "loss": 0.7329,
+      "step": 15438
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6867394777396525e-05,
+      "loss": 0.7236,
+      "step": 15439
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.68648220773506e-05,
+      "loss": 0.644,
+      "step": 15440
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6862249357445446e-05,
+      "loss": 0.6685,
+      "step": 15441
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6859676617708474e-05,
+      "loss": 0.6581,
+      "step": 15442
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6857103858167066e-05,
+      "loss": 0.655,
+      "step": 15443
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6854531078848637e-05,
+      "loss": 0.7448,
+      "step": 15444
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.685195827978057e-05,
+      "loss": 0.63,
+      "step": 15445
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6849385460990278e-05,
+      "loss": 0.6997,
+      "step": 15446
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.684681262250515e-05,
+      "loss": 0.6993,
+      "step": 15447
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6844239764352587e-05,
+      "loss": 0.7849,
+      "step": 15448
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6841666886559995e-05,
+      "loss": 0.7748,
+      "step": 15449
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.683909398915477e-05,
+      "loss": 0.7645,
+      "step": 15450
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.68365210721643e-05,
+      "loss": 0.776,
+      "step": 15451
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.683394813561601e-05,
+      "loss": 0.7285,
+      "step": 15452
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6831375179537277e-05,
+      "loss": 0.7054,
+      "step": 15453
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.682880220395551e-05,
+      "loss": 0.6744,
+      "step": 15454
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.682622920889812e-05,
+      "loss": 0.7373,
+      "step": 15455
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.682365619439248e-05,
+      "loss": 0.6563,
+      "step": 15456
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.682108316046602e-05,
+      "loss": 0.7498,
+      "step": 15457
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6818510107146132e-05,
+      "loss": 0.7383,
+      "step": 15458
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.681593703446021e-05,
+      "loss": 0.8,
+      "step": 15459
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.681336394243566e-05,
+      "loss": 0.7553,
+      "step": 15460
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6810790831099892e-05,
+      "loss": 0.6893,
+      "step": 15461
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6808217700480298e-05,
+      "loss": 0.7731,
+      "step": 15462
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.680564455060428e-05,
+      "loss": 0.7756,
+      "step": 15463
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.680307138149925e-05,
+      "loss": 0.7181,
+      "step": 15464
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6800498193192592e-05,
+      "loss": 0.7153,
+      "step": 15465
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.679792498571173e-05,
+      "loss": 0.7463,
+      "step": 15466
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6795351759084063e-05,
+      "loss": 0.7327,
+      "step": 15467
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.679277851333698e-05,
+      "loss": 0.7032,
+      "step": 15468
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.67902052484979e-05,
+      "loss": 0.783,
+      "step": 15469
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6787631964594224e-05,
+      "loss": 0.7458,
+      "step": 15470
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.678505866165334e-05,
+      "loss": 0.6913,
+      "step": 15471
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6782485339702677e-05,
+      "loss": 0.6576,
+      "step": 15472
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6779911998769624e-05,
+      "loss": 0.7301,
+      "step": 15473
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6777338638881593e-05,
+      "loss": 0.6906,
+      "step": 15474
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.677476526006598e-05,
+      "loss": 0.7366,
+      "step": 15475
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6772191862350193e-05,
+      "loss": 0.7185,
+      "step": 15476
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.676961844576164e-05,
+      "loss": 0.7383,
+      "step": 15477
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.676704501032773e-05,
+      "loss": 0.7361,
+      "step": 15478
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.676447155607586e-05,
+      "loss": 0.745,
+      "step": 15479
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6761898083033434e-05,
+      "loss": 0.8695,
+      "step": 15480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6759324591227868e-05,
+      "loss": 0.7156,
+      "step": 15481
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.675675108068656e-05,
+      "loss": 0.7392,
+      "step": 15482
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6754177551436927e-05,
+      "loss": 0.688,
+      "step": 15483
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6751604003506365e-05,
+      "loss": 0.682,
+      "step": 15484
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6749030436922274e-05,
+      "loss": 0.6831,
+      "step": 15485
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.674645685171208e-05,
+      "loss": 0.7046,
+      "step": 15486
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6743883247903186e-05,
+      "loss": 0.7218,
+      "step": 15487
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.674130962552298e-05,
+      "loss": 0.7422,
+      "step": 15488
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6738735984598888e-05,
+      "loss": 0.7524,
+      "step": 15489
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.673616232515832e-05,
+      "loss": 0.6587,
+      "step": 15490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6733588647228668e-05,
+      "loss": 0.7205,
+      "step": 15491
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6731014950837357e-05,
+      "loss": 0.7658,
+      "step": 15492
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6728441236011797e-05,
+      "loss": 0.6847,
+      "step": 15493
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.672586750277937e-05,
+      "loss": 0.6721,
+      "step": 15494
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.672329375116751e-05,
+      "loss": 0.7761,
+      "step": 15495
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.672071998120362e-05,
+      "loss": 0.6672,
+      "step": 15496
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6718146192915094e-05,
+      "loss": 0.6994,
+      "step": 15497
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6715572386329368e-05,
+      "loss": 0.7215,
+      "step": 15498
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6712998561473835e-05,
+      "loss": 0.7082,
+      "step": 15499
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.67104247183759e-05,
+      "loss": 0.7404,
+      "step": 15500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.670785085706298e-05,
+      "loss": 0.7428,
+      "step": 15501
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6705276977562494e-05,
+      "loss": 0.7666,
+      "step": 15502
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.670270307990184e-05,
+      "loss": 0.7678,
+      "step": 15503
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6700129164108434e-05,
+      "loss": 0.7008,
+      "step": 15504
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.669755523020968e-05,
+      "loss": 0.7111,
+      "step": 15505
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6694981278232995e-05,
+      "loss": 0.6871,
+      "step": 15506
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.669240730820579e-05,
+      "loss": 0.766,
+      "step": 15507
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6689833320155472e-05,
+      "loss": 0.7157,
+      "step": 15508
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.668725931410946e-05,
+      "loss": 0.6834,
+      "step": 15509
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.668468529009516e-05,
+      "loss": 0.7179,
+      "step": 15510
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6682111248139972e-05,
+      "loss": 0.716,
+      "step": 15511
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.667953718827133e-05,
+      "loss": 0.7247,
+      "step": 15512
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6676963110516645e-05,
+      "loss": 0.7734,
+      "step": 15513
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.667438901490331e-05,
+      "loss": 0.6431,
+      "step": 15514
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6671814901458754e-05,
+      "loss": 0.7392,
+      "step": 15515
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.666924077021038e-05,
+      "loss": 0.6695,
+      "step": 15516
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6666666621185605e-05,
+      "loss": 0.6929,
+      "step": 15517
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6664092454411844e-05,
+      "loss": 0.712,
+      "step": 15518
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6661518269916513e-05,
+      "loss": 0.7903,
+      "step": 15519
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.665894406772701e-05,
+      "loss": 0.6426,
+      "step": 15520
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.665636984787077e-05,
+      "loss": 0.7651,
+      "step": 15521
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6653795610375193e-05,
+      "loss": 0.7421,
+      "step": 15522
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6651221355267703e-05,
+      "loss": 0.815,
+      "step": 15523
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.66486470825757e-05,
+      "loss": 0.7629,
+      "step": 15524
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6646072792326605e-05,
+      "loss": 0.6715,
+      "step": 15525
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.664349848454784e-05,
+      "loss": 0.7501,
+      "step": 15526
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6640924159266812e-05,
+      "loss": 0.7383,
+      "step": 15527
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6638349816510938e-05,
+      "loss": 0.6977,
+      "step": 15528
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.663577545630763e-05,
+      "loss": 0.7336,
+      "step": 15529
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.663320107868431e-05,
+      "loss": 0.6444,
+      "step": 15530
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6630626683668393e-05,
+      "loss": 0.7167,
+      "step": 15531
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6628052271287284e-05,
+      "loss": 0.811,
+      "step": 15532
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.662547784156842e-05,
+      "loss": 0.6764,
+      "step": 15533
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.662290339453919e-05,
+      "loss": 0.757,
+      "step": 15534
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.662032893022703e-05,
+      "loss": 0.7404,
+      "step": 15535
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6617754448659348e-05,
+      "loss": 0.7314,
+      "step": 15536
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.661517994986357e-05,
+      "loss": 0.6852,
+      "step": 15537
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6612605433867098e-05,
+      "loss": 0.6534,
+      "step": 15538
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.661003090069736e-05,
+      "loss": 0.846,
+      "step": 15539
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6607456350381764e-05,
+      "loss": 0.6901,
+      "step": 15540
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6604881782947744e-05,
+      "loss": 0.6738,
+      "step": 15541
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6602307198422706e-05,
+      "loss": 0.6645,
+      "step": 15542
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.659973259683407e-05,
+      "loss": 0.7629,
+      "step": 15543
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6597157978209246e-05,
+      "loss": 0.8035,
+      "step": 15544
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.659458334257567e-05,
+      "loss": 0.7517,
+      "step": 15545
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6592008689960734e-05,
+      "loss": 0.7144,
+      "step": 15546
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6589434020391884e-05,
+      "loss": 0.663,
+      "step": 15547
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6586859333896524e-05,
+      "loss": 0.7156,
+      "step": 15548
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.658428463050207e-05,
+      "loss": 0.7022,
+      "step": 15549
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6581709910235952e-05,
+      "loss": 0.7205,
+      "step": 15550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6579135173125586e-05,
+      "loss": 0.7064,
+      "step": 15551
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.657656041919838e-05,
+      "loss": 0.7168,
+      "step": 15552
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.657398564848178e-05,
+      "loss": 0.7125,
+      "step": 15553
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6571410861003172e-05,
+      "loss": 0.727,
+      "step": 15554
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.656883605679e-05,
+      "loss": 0.744,
+      "step": 15555
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6566261235869677e-05,
+      "loss": 0.7102,
+      "step": 15556
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6563686398269616e-05,
+      "loss": 0.6588,
+      "step": 15557
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6561111544017246e-05,
+      "loss": 0.7327,
+      "step": 15558
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.655853667313999e-05,
+      "loss": 0.7044,
+      "step": 15559
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6555961785665256e-05,
+      "loss": 0.7207,
+      "step": 15560
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6553386881620485e-05,
+      "loss": 0.6891,
+      "step": 15561
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6550811961033086e-05,
+      "loss": 0.6438,
+      "step": 15562
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.654823702393048e-05,
+      "loss": 0.8145,
+      "step": 15563
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6545662070340084e-05,
+      "loss": 0.6921,
+      "step": 15564
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6543087100289326e-05,
+      "loss": 0.8082,
+      "step": 15565
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6540512113805626e-05,
+      "loss": 0.6943,
+      "step": 15566
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6537937110916417e-05,
+      "loss": 0.6742,
+      "step": 15567
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6535362091649103e-05,
+      "loss": 0.7206,
+      "step": 15568
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.653278705603111e-05,
+      "loss": 0.6892,
+      "step": 15569
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6530212004089877e-05,
+      "loss": 0.6978,
+      "step": 15570
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6527636935852814e-05,
+      "loss": 0.7736,
+      "step": 15571
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.652506185134733e-05,
+      "loss": 0.7019,
+      "step": 15572
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6522486750600883e-05,
+      "loss": 0.6886,
+      "step": 15573
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6519911633640872e-05,
+      "loss": 0.7017,
+      "step": 15574
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6517336500494716e-05,
+      "loss": 0.7521,
+      "step": 15575
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6514761351189854e-05,
+      "loss": 0.6912,
+      "step": 15576
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.65121861857537e-05,
+      "loss": 0.6719,
+      "step": 15577
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6509611004213685e-05,
+      "loss": 0.634,
+      "step": 15578
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.650703580659723e-05,
+      "loss": 0.7908,
+      "step": 15579
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.650446059293175e-05,
+      "loss": 0.7266,
+      "step": 15580
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6501885363244682e-05,
+      "loss": 0.7074,
+      "step": 15581
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6499310117563453e-05,
+      "loss": 0.8115,
+      "step": 15582
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.649673485591548e-05,
+      "loss": 0.7009,
+      "step": 15583
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6494159578328186e-05,
+      "loss": 0.6831,
+      "step": 15584
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6491584284829003e-05,
+      "loss": 0.6672,
+      "step": 15585
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6489008975445346e-05,
+      "loss": 0.6718,
+      "step": 15586
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.648643365020465e-05,
+      "loss": 0.6108,
+      "step": 15587
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.648385830913434e-05,
+      "loss": 0.7303,
+      "step": 15588
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6481282952261837e-05,
+      "loss": 0.7346,
+      "step": 15589
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6478707579614577e-05,
+      "loss": 0.7779,
+      "step": 15590
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6476132191219978e-05,
+      "loss": 0.7352,
+      "step": 15591
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6473556787105457e-05,
+      "loss": 0.7665,
+      "step": 15592
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.647098136729846e-05,
+      "loss": 0.6759,
+      "step": 15593
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.64684059318264e-05,
+      "loss": 0.6733,
+      "step": 15594
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6465830480716712e-05,
+      "loss": 0.7595,
+      "step": 15595
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6463255013996818e-05,
+      "loss": 0.7518,
+      "step": 15596
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.646067953169415e-05,
+      "loss": 0.804,
+      "step": 15597
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6458104033836125e-05,
+      "loss": 0.7428,
+      "step": 15598
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.645552852045018e-05,
+      "loss": 0.7056,
+      "step": 15599
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.645295299156374e-05,
+      "loss": 0.7434,
+      "step": 15600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6450377447204228e-05,
+      "loss": 0.7971,
+      "step": 15601
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.644780188739909e-05,
+      "loss": 0.6997,
+      "step": 15602
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.644522631217573e-05,
+      "loss": 0.6905,
+      "step": 15603
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.644265072156159e-05,
+      "loss": 0.6746,
+      "step": 15604
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6440075115584096e-05,
+      "loss": 0.6945,
+      "step": 15605
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6437499494270673e-05,
+      "loss": 0.6708,
+      "step": 15606
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6434923857648756e-05,
+      "loss": 0.7396,
+      "step": 15607
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6432348205745772e-05,
+      "loss": 0.7513,
+      "step": 15608
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.642977253858914e-05,
+      "loss": 0.7244,
+      "step": 15609
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6427196856206317e-05,
+      "loss": 0.6782,
+      "step": 15610
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6424621158624702e-05,
+      "loss": 0.6825,
+      "step": 15611
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6422045445871735e-05,
+      "loss": 0.714,
+      "step": 15612
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.641946971797485e-05,
+      "loss": 0.706,
+      "step": 15613
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.641689397496148e-05,
+      "loss": 0.7108,
+      "step": 15614
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.641431821685904e-05,
+      "loss": 0.6837,
+      "step": 15615
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6411742443694975e-05,
+      "loss": 0.6608,
+      "step": 15616
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6409166655496716e-05,
+      "loss": 0.7559,
+      "step": 15617
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6406590852291675e-05,
+      "loss": 0.6671,
+      "step": 15618
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.64040150341073e-05,
+      "loss": 0.6554,
+      "step": 15619
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.640143920097102e-05,
+      "loss": 0.6891,
+      "step": 15620
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6398863352910258e-05,
+      "loss": 0.762,
+      "step": 15621
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6396287489952454e-05,
+      "loss": 0.7232,
+      "step": 15622
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6393711612125037e-05,
+      "loss": 0.6969,
+      "step": 15623
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.639113571945544e-05,
+      "loss": 0.7915,
+      "step": 15624
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6388559811971087e-05,
+      "loss": 0.7178,
+      "step": 15625
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.638598388969941e-05,
+      "loss": 0.7721,
+      "step": 15626
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.638340795266786e-05,
+      "loss": 0.7191,
+      "step": 15627
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.638083200090385e-05,
+      "loss": 0.6918,
+      "step": 15628
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6378256034434807e-05,
+      "loss": 0.7186,
+      "step": 15629
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.637568005328819e-05,
+      "loss": 0.7215,
+      "step": 15630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6373104057491403e-05,
+      "loss": 0.7028,
+      "step": 15631
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6370528047071895e-05,
+      "loss": 0.7297,
+      "step": 15632
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6367952022057095e-05,
+      "loss": 0.7255,
+      "step": 15633
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.636537598247445e-05,
+      "loss": 0.7181,
+      "step": 15634
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.636279992835136e-05,
+      "loss": 0.6653,
+      "step": 15635
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.636022385971529e-05,
+      "loss": 0.7303,
+      "step": 15636
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.635764777659366e-05,
+      "loss": 0.7176,
+      "step": 15637
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6355071679013898e-05,
+      "loss": 0.6886,
+      "step": 15638
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.635249556700345e-05,
+      "loss": 0.7202,
+      "step": 15639
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.634991944058975e-05,
+      "loss": 0.7561,
+      "step": 15640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6347343299800216e-05,
+      "loss": 0.7108,
+      "step": 15641
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6344767144662303e-05,
+      "loss": 0.6885,
+      "step": 15642
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6342190975203434e-05,
+      "loss": 0.7847,
+      "step": 15643
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.633961479145105e-05,
+      "loss": 0.6778,
+      "step": 15644
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6337038593432576e-05,
+      "loss": 0.7259,
+      "step": 15645
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6334462381175458e-05,
+      "loss": 0.748,
+      "step": 15646
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6331886154707118e-05,
+      "loss": 0.7427,
+      "step": 15647
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6329309914055012e-05,
+      "loss": 0.7573,
+      "step": 15648
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6326733659246556e-05,
+      "loss": 0.7201,
+      "step": 15649
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6324157390309184e-05,
+      "loss": 0.6771,
+      "step": 15650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.632158110727035e-05,
+      "loss": 0.699,
+      "step": 15651
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.631900481015747e-05,
+      "loss": 0.719,
+      "step": 15652
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6316428498997996e-05,
+      "loss": 0.6528,
+      "step": 15653
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6313852173819364e-05,
+      "loss": 0.7537,
+      "step": 15654
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6311275834648992e-05,
+      "loss": 0.7199,
+      "step": 15655
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6308699481514332e-05,
+      "loss": 0.7856,
+      "step": 15656
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.630612311444282e-05,
+      "loss": 0.6931,
+      "step": 15657
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6303546733461888e-05,
+      "loss": 0.757,
+      "step": 15658
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6300970338598973e-05,
+      "loss": 0.6745,
+      "step": 15659
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.629839392988152e-05,
+      "loss": 0.6676,
+      "step": 15660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.629581750733695e-05,
+      "loss": 0.7212,
+      "step": 15661
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6293241070992718e-05,
+      "loss": 0.6832,
+      "step": 15662
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6290664620876255e-05,
+      "loss": 0.6569,
+      "step": 15663
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.628808815701499e-05,
+      "loss": 0.7838,
+      "step": 15664
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6285511679436377e-05,
+      "loss": 0.6835,
+      "step": 15665
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.628293518816784e-05,
+      "loss": 0.7346,
+      "step": 15666
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6280358683236818e-05,
+      "loss": 0.6733,
+      "step": 15667
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6277782164670762e-05,
+      "loss": 0.7598,
+      "step": 15668
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6275205632497095e-05,
+      "loss": 0.7214,
+      "step": 15669
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.627262908674326e-05,
+      "loss": 0.742,
+      "step": 15670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6270052527436706e-05,
+      "loss": 0.6844,
+      "step": 15671
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6267475954604858e-05,
+      "loss": 0.7293,
+      "step": 15672
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6264899368275164e-05,
+      "loss": 0.7437,
+      "step": 15673
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.626232276847506e-05,
+      "loss": 0.708,
+      "step": 15674
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6259746155231986e-05,
+      "loss": 0.8143,
+      "step": 15675
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.625716952857338e-05,
+      "loss": 0.78,
+      "step": 15676
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6254592888526675e-05,
+      "loss": 0.6893,
+      "step": 15677
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.625201623511932e-05,
+      "loss": 0.6983,
+      "step": 15678
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6249439568378752e-05,
+      "loss": 0.7469,
+      "step": 15679
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6246862888332414e-05,
+      "loss": 0.6614,
+      "step": 15680
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.624428619500774e-05,
+      "loss": 0.6857,
+      "step": 15681
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6241709488432175e-05,
+      "loss": 0.7083,
+      "step": 15682
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6239132768633162e-05,
+      "loss": 0.6954,
+      "step": 15683
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6236556035638127e-05,
+      "loss": 0.7488,
+      "step": 15684
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6233979289474535e-05,
+      "loss": 0.6927,
+      "step": 15685
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6231402530169807e-05,
+      "loss": 0.7378,
+      "step": 15686
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6228825757751386e-05,
+      "loss": 0.6678,
+      "step": 15687
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.622624897224672e-05,
+      "loss": 0.7945,
+      "step": 15688
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6223672173683246e-05,
+      "loss": 0.644,
+      "step": 15689
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6221095362088406e-05,
+      "loss": 0.71,
+      "step": 15690
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6218518537489638e-05,
+      "loss": 0.6658,
+      "step": 15691
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6215941699914393e-05,
+      "loss": 0.7109,
+      "step": 15692
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6213364849390105e-05,
+      "loss": 0.7211,
+      "step": 15693
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6210787985944217e-05,
+      "loss": 0.6624,
+      "step": 15694
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6208211109604174e-05,
+      "loss": 0.7858,
+      "step": 15695
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.620563422039742e-05,
+      "loss": 0.7287,
+      "step": 15696
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6203057318351386e-05,
+      "loss": 0.6536,
+      "step": 15697
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6200480403493522e-05,
+      "loss": 0.7043,
+      "step": 15698
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6197903475851275e-05,
+      "loss": 0.7274,
+      "step": 15699
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6195326535452085e-05,
+      "loss": 0.6834,
+      "step": 15700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6192749582323382e-05,
+      "loss": 0.7502,
+      "step": 15701
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.619017261649263e-05,
+      "loss": 0.7285,
+      "step": 15702
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6187595637987265e-05,
+      "loss": 0.7676,
+      "step": 15703
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6185018646834725e-05,
+      "loss": 0.7479,
+      "step": 15704
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6182441643062454e-05,
+      "loss": 0.6626,
+      "step": 15705
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6179864626697897e-05,
+      "loss": 0.6919,
+      "step": 15706
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6177287597768497e-05,
+      "loss": 0.731,
+      "step": 15707
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6174710556301695e-05,
+      "loss": 0.6762,
+      "step": 15708
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6172133502324948e-05,
+      "loss": 0.7452,
+      "step": 15709
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6169556435865683e-05,
+      "loss": 0.6755,
+      "step": 15710
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6166979356951354e-05,
+      "loss": 0.6327,
+      "step": 15711
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6164402265609405e-05,
+      "loss": 0.6907,
+      "step": 15712
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6161825161867277e-05,
+      "loss": 0.6804,
+      "step": 15713
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.615924804575242e-05,
+      "loss": 0.6895,
+      "step": 15714
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.615667091729228e-05,
+      "loss": 0.7271,
+      "step": 15715
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6154093776514288e-05,
+      "loss": 0.7355,
+      "step": 15716
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6151516623445898e-05,
+      "loss": 0.7054,
+      "step": 15717
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6148939458114557e-05,
+      "loss": 0.7372,
+      "step": 15718
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6146362280547704e-05,
+      "loss": 0.7611,
+      "step": 15719
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.61437850907728e-05,
+      "loss": 0.7852,
+      "step": 15720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.614120788881727e-05,
+      "loss": 0.7394,
+      "step": 15721
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6138630674708563e-05,
+      "loss": 0.7778,
+      "step": 15722
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6136053448474147e-05,
+      "loss": 0.8023,
+      "step": 15723
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6133476210141438e-05,
+      "loss": 0.7306,
+      "step": 15724
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6130898959737903e-05,
+      "loss": 0.6784,
+      "step": 15725
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.612832169729098e-05,
+      "loss": 0.7773,
+      "step": 15726
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6125744422828114e-05,
+      "loss": 0.6958,
+      "step": 15727
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6123167136376754e-05,
+      "loss": 0.7308,
+      "step": 15728
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6120589837964343e-05,
+      "loss": 0.6717,
+      "step": 15729
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.611801252761833e-05,
+      "loss": 0.7005,
+      "step": 15730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6115435205366167e-05,
+      "loss": 0.6603,
+      "step": 15731
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6112857871235295e-05,
+      "loss": 0.6922,
+      "step": 15732
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6110280525253162e-05,
+      "loss": 0.7833,
+      "step": 15733
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.610770316744721e-05,
+      "loss": 0.705,
+      "step": 15734
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.610512579784491e-05,
+      "loss": 0.6921,
+      "step": 15735
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.610254841647367e-05,
+      "loss": 0.7075,
+      "step": 15736
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.609997102336097e-05,
+      "loss": 0.7219,
+      "step": 15737
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6097393618534245e-05,
+      "loss": 0.8022,
+      "step": 15738
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.609481620202094e-05,
+      "loss": 0.6943,
+      "step": 15739
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.609223877384851e-05,
+      "loss": 0.768,
+      "step": 15740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6089661334044402e-05,
+      "loss": 0.6485,
+      "step": 15741
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6087083882636053e-05,
+      "loss": 0.7379,
+      "step": 15742
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6084506419650934e-05,
+      "loss": 0.7767,
+      "step": 15743
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6081928945116474e-05,
+      "loss": 0.7273,
+      "step": 15744
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6079351459060127e-05,
+      "loss": 0.6214,
+      "step": 15745
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6076773961509342e-05,
+      "loss": 0.7574,
+      "step": 15746
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6074196452491568e-05,
+      "loss": 0.6137,
+      "step": 15747
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6071618932034254e-05,
+      "loss": 0.6692,
+      "step": 15748
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6069041400164852e-05,
+      "loss": 0.6726,
+      "step": 15749
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.60664638569108e-05,
+      "loss": 0.7153,
+      "step": 15750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.606388630229957e-05,
+      "loss": 0.7077,
+      "step": 15751
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6061308736358585e-05,
+      "loss": 0.6944,
+      "step": 15752
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.605873115911531e-05,
+      "loss": 0.6868,
+      "step": 15753
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6056153570597187e-05,
+      "loss": 0.7376,
+      "step": 15754
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6053575970831677e-05,
+      "loss": 0.7681,
+      "step": 15755
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.605099835984622e-05,
+      "loss": 0.709,
+      "step": 15756
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6048420737668265e-05,
+      "loss": 0.7664,
+      "step": 15757
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.604584310432527e-05,
+      "loss": 0.6642,
+      "step": 15758
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6043265459844673e-05,
+      "loss": 0.7827,
+      "step": 15759
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.604068780425394e-05,
+      "loss": 0.6733,
+      "step": 15760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6038110137580514e-05,
+      "loss": 0.7543,
+      "step": 15761
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6035532459851835e-05,
+      "loss": 0.8049,
+      "step": 15762
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6032954771095373e-05,
+      "loss": 0.7032,
+      "step": 15763
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6030377071338573e-05,
+      "loss": 0.7416,
+      "step": 15764
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6027799360608877e-05,
+      "loss": 0.6936,
+      "step": 15765
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6025221638933744e-05,
+      "loss": 0.7025,
+      "step": 15766
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6022643906340627e-05,
+      "loss": 0.7437,
+      "step": 15767
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6020066162856967e-05,
+      "loss": 0.7485,
+      "step": 15768
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.601748840851022e-05,
+      "loss": 0.7178,
+      "step": 15769
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6014910643327844e-05,
+      "loss": 0.7311,
+      "step": 15770
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.601233286733728e-05,
+      "loss": 0.7147,
+      "step": 15771
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.600975508056599e-05,
+      "loss": 0.8354,
+      "step": 15772
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6007177283041424e-05,
+      "loss": 0.6937,
+      "step": 15773
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6004599474791026e-05,
+      "loss": 0.7131,
+      "step": 15774
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6002021655842263e-05,
+      "loss": 0.6856,
+      "step": 15775
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5999443826222562e-05,
+      "loss": 0.7207,
+      "step": 15776
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.59968659859594e-05,
+      "loss": 0.6326,
+      "step": 15777
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5994288135080223e-05,
+      "loss": 0.7317,
+      "step": 15778
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5991710273612473e-05,
+      "loss": 0.6758,
+      "step": 15779
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5989132401583622e-05,
+      "loss": 0.7603,
+      "step": 15780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5986554519021106e-05,
+      "loss": 0.6939,
+      "step": 15781
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.598397662595238e-05,
+      "loss": 0.7083,
+      "step": 15782
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5981398722404903e-05,
+      "loss": 0.7043,
+      "step": 15783
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5978820808406125e-05,
+      "loss": 0.7195,
+      "step": 15784
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5976242883983505e-05,
+      "loss": 0.7106,
+      "step": 15785
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5973664949164483e-05,
+      "loss": 0.6478,
+      "step": 15786
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5971087003976525e-05,
+      "loss": 0.6598,
+      "step": 15787
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5968509048447074e-05,
+      "loss": 0.6952,
+      "step": 15788
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5965931082603595e-05,
+      "loss": 0.6849,
+      "step": 15789
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5963353106473537e-05,
+      "loss": 0.6608,
+      "step": 15790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5960775120084348e-05,
+      "loss": 0.8133,
+      "step": 15791
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.595819712346349e-05,
+      "loss": 0.7202,
+      "step": 15792
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.595561911663842e-05,
+      "loss": 0.6759,
+      "step": 15793
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5953041099636576e-05,
+      "loss": 0.6929,
+      "step": 15794
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.595046307248543e-05,
+      "loss": 0.7161,
+      "step": 15795
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5947885035212432e-05,
+      "loss": 0.7194,
+      "step": 15796
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.594530698784503e-05,
+      "loss": 0.627,
+      "step": 15797
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.594272893041068e-05,
+      "loss": 0.7533,
+      "step": 15798
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.594015086293684e-05,
+      "loss": 0.7362,
+      "step": 15799
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.593757278545097e-05,
+      "loss": 0.7189,
+      "step": 15800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5934994697980514e-05,
+      "loss": 0.7404,
+      "step": 15801
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.593241660055293e-05,
+      "loss": 0.7096,
+      "step": 15802
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5929838493195678e-05,
+      "loss": 0.6612,
+      "step": 15803
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5927260375936213e-05,
+      "loss": 0.6886,
+      "step": 15804
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.592468224880198e-05,
+      "loss": 0.703,
+      "step": 15805
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5922104111820456e-05,
+      "loss": 0.6925,
+      "step": 15806
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5919525965019077e-05,
+      "loss": 0.6863,
+      "step": 15807
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5916947808425295e-05,
+      "loss": 0.7559,
+      "step": 15808
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.591436964206659e-05,
+      "loss": 0.811,
+      "step": 15809
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.59117914659704e-05,
+      "loss": 0.6716,
+      "step": 15810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5909213280164175e-05,
+      "loss": 0.6729,
+      "step": 15811
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5906635084675386e-05,
+      "loss": 0.6988,
+      "step": 15812
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5904056879531485e-05,
+      "loss": 0.829,
+      "step": 15813
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5901478664759926e-05,
+      "loss": 0.7087,
+      "step": 15814
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5898900440388167e-05,
+      "loss": 0.7373,
+      "step": 15815
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.589632220644367e-05,
+      "loss": 0.7216,
+      "step": 15816
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5893743962953875e-05,
+      "loss": 0.7405,
+      "step": 15817
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5891165709946256e-05,
+      "loss": 0.6866,
+      "step": 15818
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.588858744744826e-05,
+      "loss": 0.6698,
+      "step": 15819
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.588600917548735e-05,
+      "loss": 0.7646,
+      "step": 15820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5883430894090983e-05,
+      "loss": 0.75,
+      "step": 15821
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.588085260328661e-05,
+      "loss": 0.734,
+      "step": 15822
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5878274303101685e-05,
+      "loss": 0.7227,
+      "step": 15823
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.587569599356368e-05,
+      "loss": 0.7086,
+      "step": 15824
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5873117674700043e-05,
+      "loss": 0.6318,
+      "step": 15825
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.587053934653823e-05,
+      "loss": 0.7623,
+      "step": 15826
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5867961009105706e-05,
+      "loss": 0.8196,
+      "step": 15827
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.586538266242992e-05,
+      "loss": 0.6018,
+      "step": 15828
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5862804306538334e-05,
+      "loss": 0.7666,
+      "step": 15829
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.586022594145841e-05,
+      "loss": 0.7378,
+      "step": 15830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5857647567217595e-05,
+      "loss": 0.7045,
+      "step": 15831
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5855069183843362e-05,
+      "loss": 0.8075,
+      "step": 15832
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.585249079136316e-05,
+      "loss": 0.7104,
+      "step": 15833
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.584991238980444e-05,
+      "loss": 0.7107,
+      "step": 15834
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5847333979194677e-05,
+      "loss": 0.6728,
+      "step": 15835
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.584475555956133e-05,
+      "loss": 0.6328,
+      "step": 15836
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5842177130931834e-05,
+      "loss": 0.7142,
+      "step": 15837
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5839598693333673e-05,
+      "loss": 0.7186,
+      "step": 15838
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5837020246794293e-05,
+      "loss": 0.6845,
+      "step": 15839
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5834441791341148e-05,
+      "loss": 0.67,
+      "step": 15840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.583186332700171e-05,
+      "loss": 0.7457,
+      "step": 15841
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.582928485380343e-05,
+      "loss": 0.7182,
+      "step": 15842
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5826706371773766e-05,
+      "loss": 0.7223,
+      "step": 15843
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.582412788094019e-05,
+      "loss": 0.6331,
+      "step": 15844
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5821549381330158e-05,
+      "loss": 0.687,
+      "step": 15845
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5818970872971115e-05,
+      "loss": 0.7417,
+      "step": 15846
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.581639235589053e-05,
+      "loss": 0.7542,
+      "step": 15847
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.581381383011586e-05,
+      "loss": 0.7071,
+      "step": 15848
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5811235295674567e-05,
+      "loss": 0.727,
+      "step": 15849
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5808656752594112e-05,
+      "loss": 0.7253,
+      "step": 15850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.580607820090195e-05,
+      "loss": 0.7742,
+      "step": 15851
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.580349964062555e-05,
+      "loss": 0.7864,
+      "step": 15852
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5800921071792368e-05,
+      "loss": 0.6568,
+      "step": 15853
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5798342494429857e-05,
+      "loss": 0.7179,
+      "step": 15854
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.579576390856549e-05,
+      "loss": 0.7486,
+      "step": 15855
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.579318531422672e-05,
+      "loss": 0.7384,
+      "step": 15856
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5790606711441e-05,
+      "loss": 0.6816,
+      "step": 15857
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5788028100235806e-05,
+      "loss": 0.697,
+      "step": 15858
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.578544948063859e-05,
+      "loss": 0.7197,
+      "step": 15859
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.578287085267681e-05,
+      "loss": 0.7463,
+      "step": 15860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5780292216377933e-05,
+      "loss": 0.7323,
+      "step": 15861
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.577771357176942e-05,
+      "loss": 0.7798,
+      "step": 15862
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.577513491887872e-05,
+      "loss": 0.7098,
+      "step": 15863
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5772556257733316e-05,
+      "loss": 0.7646,
+      "step": 15864
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5769977588360648e-05,
+      "loss": 0.6771,
+      "step": 15865
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5767398910788193e-05,
+      "loss": 0.6628,
+      "step": 15866
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5764820225043406e-05,
+      "loss": 0.7045,
+      "step": 15867
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5762241531153742e-05,
+      "loss": 0.7422,
+      "step": 15868
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.575966282914667e-05,
+      "loss": 0.6999,
+      "step": 15869
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.575708411904965e-05,
+      "loss": 0.6723,
+      "step": 15870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.575450540089014e-05,
+      "loss": 0.7478,
+      "step": 15871
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5751926674695615e-05,
+      "loss": 0.7503,
+      "step": 15872
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5749347940493522e-05,
+      "loss": 0.7684,
+      "step": 15873
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5746769198311332e-05,
+      "loss": 0.7808,
+      "step": 15874
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5744190448176492e-05,
+      "loss": 0.6958,
+      "step": 15875
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5741611690116497e-05,
+      "loss": 0.7202,
+      "step": 15876
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.573903292415877e-05,
+      "loss": 0.6792,
+      "step": 15877
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5736454150330795e-05,
+      "loss": 0.689,
+      "step": 15878
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.573387536866003e-05,
+      "loss": 0.7391,
+      "step": 15879
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5731296579173937e-05,
+      "loss": 0.739,
+      "step": 15880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.572871778189998e-05,
+      "loss": 0.6595,
+      "step": 15881
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5726138976865627e-05,
+      "loss": 0.6886,
+      "step": 15882
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5723560164098325e-05,
+      "loss": 0.656,
+      "step": 15883
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5720981343625554e-05,
+      "loss": 0.7113,
+      "step": 15884
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5718402515474765e-05,
+      "loss": 0.7227,
+      "step": 15885
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5715823679673428e-05,
+      "loss": 0.7692,
+      "step": 15886
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5713244836249005e-05,
+      "loss": 0.6979,
+      "step": 15887
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.571066598522896e-05,
+      "loss": 0.7443,
+      "step": 15888
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5708087126640746e-05,
+      "loss": 0.7371,
+      "step": 15889
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5705508260511835e-05,
+      "loss": 0.7043,
+      "step": 15890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5702929386869695e-05,
+      "loss": 0.724,
+      "step": 15891
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5700350505741776e-05,
+      "loss": 0.6888,
+      "step": 15892
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.569777161715555e-05,
+      "loss": 0.7851,
+      "step": 15893
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.569519272113849e-05,
+      "loss": 0.7781,
+      "step": 15894
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5692613817718036e-05,
+      "loss": 0.6295,
+      "step": 15895
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.569003490692168e-05,
+      "loss": 0.7372,
+      "step": 15896
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5687455988776858e-05,
+      "loss": 0.7161,
+      "step": 15897
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5684877063311057e-05,
+      "loss": 0.6569,
+      "step": 15898
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.568229813055173e-05,
+      "loss": 0.7588,
+      "step": 15899
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.567971919052633e-05,
+      "loss": 0.7873,
+      "step": 15900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5677140243262348e-05,
+      "loss": 0.739,
+      "step": 15901
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5674561288787223e-05,
+      "loss": 0.6317,
+      "step": 15902
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5671982327128436e-05,
+      "loss": 0.7365,
+      "step": 15903
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5669403358313443e-05,
+      "loss": 0.712,
+      "step": 15904
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5666824382369716e-05,
+      "loss": 0.7476,
+      "step": 15905
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5664245399324705e-05,
+      "loss": 0.7059,
+      "step": 15906
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.566166640920589e-05,
+      "loss": 0.749,
+      "step": 15907
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5659087412040733e-05,
+      "loss": 0.7423,
+      "step": 15908
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5656508407856684e-05,
+      "loss": 0.7664,
+      "step": 15909
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5653929396681224e-05,
+      "loss": 0.7084,
+      "step": 15910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5651350378541816e-05,
+      "loss": 0.7736,
+      "step": 15911
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5648771353465915e-05,
+      "loss": 0.7148,
+      "step": 15912
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5646192321481e-05,
+      "loss": 0.6714,
+      "step": 15913
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.564361328261453e-05,
+      "loss": 0.7149,
+      "step": 15914
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5641034236893964e-05,
+      "loss": 0.7356,
+      "step": 15915
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5638455184346772e-05,
+      "loss": 0.7053,
+      "step": 15916
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5635876125000424e-05,
+      "loss": 0.7197,
+      "step": 15917
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5633297058882382e-05,
+      "loss": 0.8071,
+      "step": 15918
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.563071798602011e-05,
+      "loss": 0.665,
+      "step": 15919
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5628138906441067e-05,
+      "loss": 0.654,
+      "step": 15920
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5625559820172735e-05,
+      "loss": 0.726,
+      "step": 15921
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5622980727242567e-05,
+      "loss": 0.6387,
+      "step": 15922
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.562040162767803e-05,
+      "loss": 0.6658,
+      "step": 15923
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5617822521506602e-05,
+      "loss": 0.7095,
+      "step": 15924
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5615243408755728e-05,
+      "loss": 0.782,
+      "step": 15925
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.561266428945289e-05,
+      "loss": 0.7899,
+      "step": 15926
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.561008516362556e-05,
+      "loss": 0.7983,
+      "step": 15927
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5607506031301182e-05,
+      "loss": 0.7308,
+      "step": 15928
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5604926892507226e-05,
+      "loss": 0.7207,
+      "step": 15929
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5602347747271178e-05,
+      "loss": 0.7538,
+      "step": 15930
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5599768595620488e-05,
+      "loss": 0.6906,
+      "step": 15931
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5597189437582624e-05,
+      "loss": 0.7248,
+      "step": 15932
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.559461027318506e-05,
+      "loss": 0.7356,
+      "step": 15933
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5592031102455255e-05,
+      "loss": 0.776,
+      "step": 15934
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5589451925420675e-05,
+      "loss": 0.7558,
+      "step": 15935
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5586872742108798e-05,
+      "loss": 0.7712,
+      "step": 15936
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.558429355254708e-05,
+      "loss": 0.7223,
+      "step": 15937
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5581714356762987e-05,
+      "loss": 0.7056,
+      "step": 15938
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5579135154783988e-05,
+      "loss": 0.6614,
+      "step": 15939
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.557655594663756e-05,
+      "loss": 0.7701,
+      "step": 15940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.557397673235115e-05,
+      "loss": 0.6807,
+      "step": 15941
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5571397511952245e-05,
+      "loss": 0.7135,
+      "step": 15942
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5568818285468298e-05,
+      "loss": 0.6718,
+      "step": 15943
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5566239052926782e-05,
+      "loss": 0.7247,
+      "step": 15944
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5563659814355167e-05,
+      "loss": 0.7048,
+      "step": 15945
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5561080569780914e-05,
+      "loss": 0.7302,
+      "step": 15946
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5558501319231497e-05,
+      "loss": 0.6967,
+      "step": 15947
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5555922062734385e-05,
+      "loss": 0.6914,
+      "step": 15948
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5553342800317027e-05,
+      "loss": 0.6629,
+      "step": 15949
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.555076353200691e-05,
+      "loss": 0.7211,
+      "step": 15950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5548184257831496e-05,
+      "loss": 0.7178,
+      "step": 15951
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.554560497781825e-05,
+      "loss": 0.6388,
+      "step": 15952
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.554302569199465e-05,
+      "loss": 0.7189,
+      "step": 15953
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.554044640038815e-05,
+      "loss": 0.6982,
+      "step": 15954
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5537867103026218e-05,
+      "loss": 0.7103,
+      "step": 15955
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.553528779993634e-05,
+      "loss": 0.6589,
+      "step": 15956
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5532708491145975e-05,
+      "loss": 0.68,
+      "step": 15957
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5530129176682576e-05,
+      "loss": 0.7155,
+      "step": 15958
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5527549856573628e-05,
+      "loss": 0.7084,
+      "step": 15959
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.552497053084659e-05,
+      "loss": 0.6829,
+      "step": 15960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5522391199528938e-05,
+      "loss": 0.7067,
+      "step": 15961
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.551981186264814e-05,
+      "loss": 0.6899,
+      "step": 15962
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.551723252023166e-05,
+      "loss": 0.6477,
+      "step": 15963
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5514653172306962e-05,
+      "loss": 0.715,
+      "step": 15964
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.551207381890152e-05,
+      "loss": 0.6791,
+      "step": 15965
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.550949446004281e-05,
+      "loss": 0.6651,
+      "step": 15966
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5506915095758293e-05,
+      "loss": 0.6812,
+      "step": 15967
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5504335726075434e-05,
+      "loss": 0.7103,
+      "step": 15968
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5501756351021705e-05,
+      "loss": 0.7812,
+      "step": 15969
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5499176970624577e-05,
+      "loss": 0.7043,
+      "step": 15970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5496597584911518e-05,
+      "loss": 0.7026,
+      "step": 15971
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5494018193909995e-05,
+      "loss": 0.7279,
+      "step": 15972
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5491438797647483e-05,
+      "loss": 0.6852,
+      "step": 15973
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5488859396151438e-05,
+      "loss": 0.683,
+      "step": 15974
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.548627998944934e-05,
+      "loss": 0.6892,
+      "step": 15975
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5483700577568658e-05,
+      "loss": 0.7083,
+      "step": 15976
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5481121160536863e-05,
+      "loss": 0.7467,
+      "step": 15977
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.547854173838141e-05,
+      "loss": 0.6707,
+      "step": 15978
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5475962311129785e-05,
+      "loss": 0.6967,
+      "step": 15979
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.547338287880945e-05,
+      "loss": 0.7147,
+      "step": 15980
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5470803441447865e-05,
+      "loss": 0.8009,
+      "step": 15981
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5468223999072516e-05,
+      "loss": 0.7439,
+      "step": 15982
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.546564455171087e-05,
+      "loss": 0.6771,
+      "step": 15983
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5463065099390383e-05,
+      "loss": 0.6515,
+      "step": 15984
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5460485642138544e-05,
+      "loss": 0.6911,
+      "step": 15985
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5457906179982804e-05,
+      "loss": 0.6769,
+      "step": 15986
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5455326712950645e-05,
+      "loss": 0.7736,
+      "step": 15987
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5452747241069537e-05,
+      "loss": 0.6473,
+      "step": 15988
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5450167764366935e-05,
+      "loss": 0.7054,
+      "step": 15989
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5447588282870326e-05,
+      "loss": 0.7971,
+      "step": 15990
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5445008796607178e-05,
+      "loss": 0.6915,
+      "step": 15991
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5442429305604952e-05,
+      "loss": 0.6749,
+      "step": 15992
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.543984980989112e-05,
+      "loss": 0.7636,
+      "step": 15993
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5437270309493156e-05,
+      "loss": 0.7027,
+      "step": 15994
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.543469080443853e-05,
+      "loss": 0.7405,
+      "step": 15995
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5432111294754706e-05,
+      "loss": 0.8566,
+      "step": 15996
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5429531780469174e-05,
+      "loss": 0.7574,
+      "step": 15997
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5426952261609373e-05,
+      "loss": 0.6817,
+      "step": 15998
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5424372738202797e-05,
+      "loss": 0.7262,
+      "step": 15999
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.542179321027691e-05,
+      "loss": 0.8113,
+      "step": 16000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.541921367785917e-05,
+      "loss": 0.6584,
+      "step": 16001
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5416634140977073e-05,
+      "loss": 0.6967,
+      "step": 16002
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.541405459965807e-05,
+      "loss": 0.7357,
+      "step": 16003
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5411475053929633e-05,
+      "loss": 0.7065,
+      "step": 16004
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5408895503819242e-05,
+      "loss": 0.7061,
+      "step": 16005
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5406315949354363e-05,
+      "loss": 0.7621,
+      "step": 16006
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.540373639056246e-05,
+      "loss": 0.7089,
+      "step": 16007
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.540115682747102e-05,
+      "loss": 0.7207,
+      "step": 16008
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.53985772601075e-05,
+      "loss": 0.8112,
+      "step": 16009
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.539599768849936e-05,
+      "loss": 0.6913,
+      "step": 16010
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.53934181126741e-05,
+      "loss": 0.7437,
+      "step": 16011
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.539083853265917e-05,
+      "loss": 0.7758,
+      "step": 16012
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5388258948482047e-05,
+      "loss": 0.6991,
+      "step": 16013
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5385679360170206e-05,
+      "loss": 0.786,
+      "step": 16014
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5383099767751112e-05,
+      "loss": 0.742,
+      "step": 16015
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.538052017125223e-05,
+      "loss": 0.7489,
+      "step": 16016
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5377940570701048e-05,
+      "loss": 0.6949,
+      "step": 16017
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.537536096612503e-05,
+      "loss": 0.7121,
+      "step": 16018
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5372781357551645e-05,
+      "loss": 0.6751,
+      "step": 16019
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5370201745008367e-05,
+      "loss": 0.689,
+      "step": 16020
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.536762212852265e-05,
+      "loss": 0.732,
+      "step": 16021
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5365042508121994e-05,
+      "loss": 0.7667,
+      "step": 16022
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5362462883833855e-05,
+      "loss": 0.6858,
+      "step": 16023
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5359883255685702e-05,
+      "loss": 0.7856,
+      "step": 16024
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5357303623705013e-05,
+      "loss": 0.6978,
+      "step": 16025
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.535472398791926e-05,
+      "loss": 0.755,
+      "step": 16026
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.535214434835591e-05,
+      "loss": 0.7243,
+      "step": 16027
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.534956470504244e-05,
+      "loss": 0.6897,
+      "step": 16028
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5346985058006316e-05,
+      "loss": 0.6936,
+      "step": 16029
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5344405407275008e-05,
+      "loss": 0.7303,
+      "step": 16030
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5341825752875996e-05,
+      "loss": 0.7503,
+      "step": 16031
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.533924609483675e-05,
+      "loss": 0.7197,
+      "step": 16032
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5336666433184726e-05,
+      "loss": 0.7031,
+      "step": 16033
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.533408676794742e-05,
+      "loss": 0.7186,
+      "step": 16034
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.533150709915229e-05,
+      "loss": 0.7405,
+      "step": 16035
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.532892742682681e-05,
+      "loss": 0.7447,
+      "step": 16036
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5326347750998453e-05,
+      "loss": 0.6471,
+      "step": 16037
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5323768071694694e-05,
+      "loss": 0.7766,
+      "step": 16038
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5321188388942996e-05,
+      "loss": 0.7083,
+      "step": 16039
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.531860870277084e-05,
+      "loss": 0.6951,
+      "step": 16040
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5316029013205693e-05,
+      "loss": 0.7738,
+      "step": 16041
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5313449320275027e-05,
+      "loss": 0.7153,
+      "step": 16042
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5310869624006323e-05,
+      "loss": 0.7935,
+      "step": 16043
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5308289924427035e-05,
+      "loss": 0.699,
+      "step": 16044
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.530571022156466e-05,
+      "loss": 0.7008,
+      "step": 16045
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.530313051544665e-05,
+      "loss": 0.7716,
+      "step": 16046
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5300550806100486e-05,
+      "loss": 0.6715,
+      "step": 16047
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5297971093553642e-05,
+      "loss": 0.6924,
+      "step": 16048
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5295391377833578e-05,
+      "loss": 0.7218,
+      "step": 16049
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5292811658967773e-05,
+      "loss": 0.7232,
+      "step": 16050
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.529023193698371e-05,
+      "loss": 0.8094,
+      "step": 16051
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5287652211908853e-05,
+      "loss": 0.656,
+      "step": 16052
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5285072483770667e-05,
+      "loss": 0.702,
+      "step": 16053
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5282492752596636e-05,
+      "loss": 0.6533,
+      "step": 16054
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5279913018414236e-05,
+      "loss": 0.7428,
+      "step": 16055
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.527733328125092e-05,
+      "loss": 0.7026,
+      "step": 16056
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5274753541134183e-05,
+      "loss": 0.6228,
+      "step": 16057
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5272173798091486e-05,
+      "loss": 0.6885,
+      "step": 16058
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5269594052150297e-05,
+      "loss": 0.7052,
+      "step": 16059
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5267014303338103e-05,
+      "loss": 0.6035,
+      "step": 16060
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5264434551682365e-05,
+      "loss": 0.6833,
+      "step": 16061
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5261854797210554e-05,
+      "loss": 0.7075,
+      "step": 16062
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.525927503995016e-05,
+      "loss": 0.7357,
+      "step": 16063
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5256695279928632e-05,
+      "loss": 0.6408,
+      "step": 16064
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.525411551717346e-05,
+      "loss": 0.7675,
+      "step": 16065
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5251535751712114e-05,
+      "loss": 0.7551,
+      "step": 16066
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5248955983572065e-05,
+      "loss": 0.671,
+      "step": 16067
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5246376212780786e-05,
+      "loss": 0.682,
+      "step": 16068
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5243796439365754e-05,
+      "loss": 0.7173,
+      "step": 16069
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5241216663354427e-05,
+      "loss": 0.7515,
+      "step": 16070
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5238636884774297e-05,
+      "loss": 0.6499,
+      "step": 16071
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5236057103652827e-05,
+      "loss": 0.6775,
+      "step": 16072
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5233477320017486e-05,
+      "loss": 0.7446,
+      "step": 16073
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5230897533895757e-05,
+      "loss": 0.6842,
+      "step": 16074
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5228317745315116e-05,
+      "loss": 0.7685,
+      "step": 16075
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.522573795430302e-05,
+      "loss": 0.6973,
+      "step": 16076
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.522315816088696e-05,
+      "loss": 0.6751,
+      "step": 16077
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5220578365094406e-05,
+      "loss": 0.6944,
+      "step": 16078
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5217998566952815e-05,
+      "loss": 0.702,
+      "step": 16079
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5215418766489674e-05,
+      "loss": 0.7878,
+      "step": 16080
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5212838963732462e-05,
+      "loss": 0.7445,
+      "step": 16081
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5210259158708627e-05,
+      "loss": 0.7143,
+      "step": 16082
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5207679351445674e-05,
+      "loss": 0.6718,
+      "step": 16083
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5205099541971054e-05,
+      "loss": 0.7491,
+      "step": 16084
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.520251973031225e-05,
+      "loss": 0.7222,
+      "step": 16085
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.519993991649674e-05,
+      "loss": 0.7113,
+      "step": 16086
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5197360100551987e-05,
+      "loss": 0.7504,
+      "step": 16087
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.519478028250547e-05,
+      "loss": 0.6693,
+      "step": 16088
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5192200462384663e-05,
+      "loss": 0.6727,
+      "step": 16089
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.518962064021703e-05,
+      "loss": 0.7557,
+      "step": 16090
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.518704081603006e-05,
+      "loss": 0.7008,
+      "step": 16091
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5184460989851216e-05,
+      "loss": 0.7048,
+      "step": 16092
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.518188116170797e-05,
+      "loss": 0.6687,
+      "step": 16093
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.517930133162781e-05,
+      "loss": 0.7339,
+      "step": 16094
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5176721499638196e-05,
+      "loss": 0.6428,
+      "step": 16095
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5174141665766598e-05,
+      "loss": 0.6632,
+      "step": 16096
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5171561830040502e-05,
+      "loss": 0.6902,
+      "step": 16097
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.516898199248739e-05,
+      "loss": 0.6539,
+      "step": 16098
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5166402153134706e-05,
+      "loss": 0.6777,
+      "step": 16099
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5163822312009944e-05,
+      "loss": 0.7096,
+      "step": 16100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5161242469140572e-05,
+      "loss": 0.7024,
+      "step": 16101
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5158662624554065e-05,
+      "loss": 0.7095,
+      "step": 16102
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.51560827782779e-05,
+      "loss": 0.6976,
+      "step": 16103
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.515350293033955e-05,
+      "loss": 0.6676,
+      "step": 16104
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5150923080766482e-05,
+      "loss": 0.624,
+      "step": 16105
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.514834322958618e-05,
+      "loss": 0.6559,
+      "step": 16106
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5145763376826108e-05,
+      "loss": 0.6375,
+      "step": 16107
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5143183522513748e-05,
+      "loss": 0.7,
+      "step": 16108
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5140603666676573e-05,
+      "loss": 0.7599,
+      "step": 16109
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5138023809342043e-05,
+      "loss": 0.6686,
+      "step": 16110
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5135443950537657e-05,
+      "loss": 0.7137,
+      "step": 16111
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.513286409029087e-05,
+      "loss": 0.6536,
+      "step": 16112
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.513028422862916e-05,
+      "loss": 0.6729,
+      "step": 16113
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5127704365579997e-05,
+      "loss": 0.6735,
+      "step": 16114
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5125124501170865e-05,
+      "loss": 0.7665,
+      "step": 16115
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5122544635429235e-05,
+      "loss": 0.698,
+      "step": 16116
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5119964768382575e-05,
+      "loss": 0.6978,
+      "step": 16117
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5117384900058366e-05,
+      "loss": 0.7396,
+      "step": 16118
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5114805030484078e-05,
+      "loss": 0.7759,
+      "step": 16119
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5112225159687193e-05,
+      "loss": 0.6528,
+      "step": 16120
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.510964528769517e-05,
+      "loss": 0.6485,
+      "step": 16121
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.510706541453549e-05,
+      "loss": 0.6943,
+      "step": 16122
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5104485540235634e-05,
+      "loss": 0.6497,
+      "step": 16123
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5101905664823067e-05,
+      "loss": 0.7301,
+      "step": 16124
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5099325788325264e-05,
+      "loss": 0.6746,
+      "step": 16125
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5096745910769703e-05,
+      "loss": 0.7437,
+      "step": 16126
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5094166032183862e-05,
+      "loss": 0.7125,
+      "step": 16127
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5091586152595202e-05,
+      "loss": 0.7973,
+      "step": 16128
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5089006272031217e-05,
+      "loss": 0.6878,
+      "step": 16129
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5086426390519362e-05,
+      "loss": 0.6311,
+      "step": 16130
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5083846508087117e-05,
+      "loss": 0.7041,
+      "step": 16131
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5081266624761958e-05,
+      "loss": 0.6583,
+      "step": 16132
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.507868674057136e-05,
+      "loss": 0.6801,
+      "step": 16133
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.507610685554279e-05,
+      "loss": 0.7476,
+      "step": 16134
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5073526969703736e-05,
+      "loss": 0.6415,
+      "step": 16135
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5070947083081663e-05,
+      "loss": 0.7399,
+      "step": 16136
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5068367195704047e-05,
+      "loss": 0.7468,
+      "step": 16137
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.506578730759836e-05,
+      "loss": 0.6541,
+      "step": 16138
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.506320741879208e-05,
+      "loss": 0.7153,
+      "step": 16139
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.506062752931268e-05,
+      "loss": 0.7208,
+      "step": 16140
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.505804763918763e-05,
+      "loss": 0.7269,
+      "step": 16141
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.505546774844441e-05,
+      "loss": 0.6689,
+      "step": 16142
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.505288785711049e-05,
+      "loss": 0.7601,
+      "step": 16143
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5050307965213355e-05,
+      "loss": 0.7437,
+      "step": 16144
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5047728072780456e-05,
+      "loss": 0.7392,
+      "step": 16145
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5045148179839294e-05,
+      "loss": 0.6708,
+      "step": 16146
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.504256828641733e-05,
+      "loss": 0.6533,
+      "step": 16147
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.503998839254203e-05,
+      "loss": 0.6859,
+      "step": 16148
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5037408498240894e-05,
+      "loss": 0.7262,
+      "step": 16149
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5034828603541373e-05,
+      "loss": 0.7457,
+      "step": 16150
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.503224870847094e-05,
+      "loss": 0.7092,
+      "step": 16151
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5029668813057085e-05,
+      "loss": 0.688,
+      "step": 16152
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.502708891732728e-05,
+      "loss": 0.7158,
+      "step": 16153
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5024509021308983e-05,
+      "loss": 0.7109,
+      "step": 16154
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5021929125029687e-05,
+      "loss": 0.7677,
+      "step": 16155
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.501934922851686e-05,
+      "loss": 0.774,
+      "step": 16156
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5016769331797967e-05,
+      "loss": 0.7342,
+      "step": 16157
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5014189434900497e-05,
+      "loss": 0.6986,
+      "step": 16158
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5011609537851923e-05,
+      "loss": 0.7504,
+      "step": 16159
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.500902964067971e-05,
+      "loss": 0.7555,
+      "step": 16160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5006449743411337e-05,
+      "loss": 0.6629,
+      "step": 16161
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5003869846074273e-05,
+      "loss": 0.6751,
+      "step": 16162
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5001289948696004e-05,
+      "loss": 0.7008,
+      "step": 16163
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4998710051304e-05,
+      "loss": 0.6934,
+      "step": 16164
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4996130153925732e-05,
+      "loss": 0.7141,
+      "step": 16165
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4993550256588666e-05,
+      "loss": 0.6823,
+      "step": 16166
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4990970359320297e-05,
+      "loss": 0.7206,
+      "step": 16167
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4988390462148083e-05,
+      "loss": 0.6957,
+      "step": 16168
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4985810565099502e-05,
+      "loss": 0.7679,
+      "step": 16169
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4983230668202032e-05,
+      "loss": 0.7052,
+      "step": 16170
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4980650771483147e-05,
+      "loss": 0.7144,
+      "step": 16171
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4978070874970315e-05,
+      "loss": 0.7499,
+      "step": 16172
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.497549097869102e-05,
+      "loss": 0.7281,
+      "step": 16173
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4972911082672733e-05,
+      "loss": 0.7573,
+      "step": 16174
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4970331186942925e-05,
+      "loss": 0.7166,
+      "step": 16175
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4967751291529067e-05,
+      "loss": 0.7035,
+      "step": 16176
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.496517139645864e-05,
+      "loss": 0.7533,
+      "step": 16177
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.496259150175912e-05,
+      "loss": 0.7355,
+      "step": 16178
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4960011607457974e-05,
+      "loss": 0.7166,
+      "step": 16179
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4957431713582677e-05,
+      "loss": 0.7527,
+      "step": 16180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.495485182016071e-05,
+      "loss": 0.7483,
+      "step": 16181
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4952271927219546e-05,
+      "loss": 0.6051,
+      "step": 16182
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4949692034786655e-05,
+      "loss": 0.7219,
+      "step": 16183
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.494711214288951e-05,
+      "loss": 0.7061,
+      "step": 16184
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4944532251555598e-05,
+      "loss": 0.701,
+      "step": 16185
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.494195236081237e-05,
+      "loss": 0.7429,
+      "step": 16186
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4939372470687326e-05,
+      "loss": 0.7371,
+      "step": 16187
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4936792581207923e-05,
+      "loss": 0.6676,
+      "step": 16188
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.493421269240164e-05,
+      "loss": 0.7304,
+      "step": 16189
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4931632804295955e-05,
+      "loss": 0.6909,
+      "step": 16190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.492905291691834e-05,
+      "loss": 0.7719,
+      "step": 16191
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.492647303029626e-05,
+      "loss": 0.6989,
+      "step": 16192
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4923893144457205e-05,
+      "loss": 0.682,
+      "step": 16193
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4921313259428648e-05,
+      "loss": 0.7272,
+      "step": 16194
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.491873337523805e-05,
+      "loss": 0.6978,
+      "step": 16195
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.49161534919129e-05,
+      "loss": 0.7766,
+      "step": 16196
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4913573609480647e-05,
+      "loss": 0.671,
+      "step": 16197
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4910993727968792e-05,
+      "loss": 0.6967,
+      "step": 16198
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4908413847404804e-05,
+      "loss": 0.7319,
+      "step": 16199
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4905833967816144e-05,
+      "loss": 0.6835,
+      "step": 16200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4903254089230303e-05,
+      "loss": 0.7492,
+      "step": 16201
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4900674211674742e-05,
+      "loss": 0.7249,
+      "step": 16202
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.489809433517694e-05,
+      "loss": 0.6991,
+      "step": 16203
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4895514459764375e-05,
+      "loss": 0.6774,
+      "step": 16204
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4892934585464518e-05,
+      "loss": 0.6852,
+      "step": 16205
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4890354712304838e-05,
+      "loss": 0.6815,
+      "step": 16206
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4887774840312816e-05,
+      "loss": 0.6312,
+      "step": 16207
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4885194969515928e-05,
+      "loss": 0.7581,
+      "step": 16208
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4882615099941633e-05,
+      "loss": 0.691,
+      "step": 16209
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4880035231617427e-05,
+      "loss": 0.7687,
+      "step": 16210
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.487745536457077e-05,
+      "loss": 0.7677,
+      "step": 16211
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4874875498829134e-05,
+      "loss": 0.6929,
+      "step": 16212
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4872295634420002e-05,
+      "loss": 0.7175,
+      "step": 16213
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4869715771370853e-05,
+      "loss": 0.74,
+      "step": 16214
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4867135909709144e-05,
+      "loss": 0.7765,
+      "step": 16215
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4864556049462356e-05,
+      "loss": 0.7406,
+      "step": 16216
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.486197619065796e-05,
+      "loss": 0.7738,
+      "step": 16217
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.485939633332344e-05,
+      "loss": 0.7486,
+      "step": 16218
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.485681647748626e-05,
+      "loss": 0.7116,
+      "step": 16219
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4854236623173895e-05,
+      "loss": 0.7416,
+      "step": 16220
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4851656770413827e-05,
+      "loss": 0.6952,
+      "step": 16221
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4849076919233524e-05,
+      "loss": 0.7263,
+      "step": 16222
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4846497069660456e-05,
+      "loss": 0.7511,
+      "step": 16223
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4843917221722103e-05,
+      "loss": 0.7025,
+      "step": 16224
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.484133737544594e-05,
+      "loss": 0.7612,
+      "step": 16225
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4838757530859433e-05,
+      "loss": 0.684,
+      "step": 16226
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4836177687990065e-05,
+      "loss": 0.7022,
+      "step": 16227
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4833597846865303e-05,
+      "loss": 0.7886,
+      "step": 16228
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.483101800751262e-05,
+      "loss": 0.7144,
+      "step": 16229
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4828438169959494e-05,
+      "loss": 0.6822,
+      "step": 16230
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.48258583342334e-05,
+      "loss": 0.7611,
+      "step": 16231
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4823278500361806e-05,
+      "loss": 0.7839,
+      "step": 16232
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4820698668372193e-05,
+      "loss": 0.7339,
+      "step": 16233
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4818118838292026e-05,
+      "loss": 0.714,
+      "step": 16234
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4815539010148793e-05,
+      "loss": 0.8703,
+      "step": 16235
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.481295918396995e-05,
+      "loss": 0.7257,
+      "step": 16236
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4810379359782972e-05,
+      "loss": 0.663,
+      "step": 16237
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.480779953761535e-05,
+      "loss": 0.7351,
+      "step": 16238
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4805219717494538e-05,
+      "loss": 0.7529,
+      "step": 16239
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4802639899448016e-05,
+      "loss": 0.7812,
+      "step": 16240
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4800060083503266e-05,
+      "loss": 0.7367,
+      "step": 16241
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4797480269687755e-05,
+      "loss": 0.6947,
+      "step": 16242
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.479490045802895e-05,
+      "loss": 0.7402,
+      "step": 16243
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.479232064855433e-05,
+      "loss": 0.7157,
+      "step": 16244
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4789740841291375e-05,
+      "loss": 0.7759,
+      "step": 16245
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4787161036267547e-05,
+      "loss": 0.6855,
+      "step": 16246
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.478458123351033e-05,
+      "loss": 0.6335,
+      "step": 16247
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.478200143304719e-05,
+      "loss": 0.7049,
+      "step": 16248
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.47794216349056e-05,
+      "loss": 0.699,
+      "step": 16249
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4776841839113042e-05,
+      "loss": 0.7473,
+      "step": 16250
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4774262045696977e-05,
+      "loss": 0.7629,
+      "step": 16251
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4771682254684883e-05,
+      "loss": 0.7126,
+      "step": 16252
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.476910246610424e-05,
+      "loss": 0.6597,
+      "step": 16253
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4766522679982513e-05,
+      "loss": 0.7659,
+      "step": 16254
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4763942896347185e-05,
+      "loss": 0.6707,
+      "step": 16255
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4761363115225716e-05,
+      "loss": 0.7405,
+      "step": 16256
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.475878333664558e-05,
+      "loss": 0.7098,
+      "step": 16257
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4756203560634262e-05,
+      "loss": 0.7648,
+      "step": 16258
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4753623787219224e-05,
+      "loss": 0.6917,
+      "step": 16259
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.475104401642794e-05,
+      "loss": 0.6685,
+      "step": 16260
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.474846424828789e-05,
+      "loss": 0.6771,
+      "step": 16261
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4745884482826547e-05,
+      "loss": 0.741,
+      "step": 16262
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.474330472007137e-05,
+      "loss": 0.7157,
+      "step": 16263
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.474072496004985e-05,
+      "loss": 0.7238,
+      "step": 16264
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.473814520278945e-05,
+      "loss": 0.7201,
+      "step": 16265
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.473556544831764e-05,
+      "loss": 0.6979,
+      "step": 16266
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4732985696661903e-05,
+      "loss": 0.7188,
+      "step": 16267
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.473040594784971e-05,
+      "loss": 0.6763,
+      "step": 16268
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4727826201908516e-05,
+      "loss": 0.6541,
+      "step": 16269
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.472524645886582e-05,
+      "loss": 0.7901,
+      "step": 16270
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.472266671874908e-05,
+      "loss": 0.6986,
+      "step": 16271
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4720086981585773e-05,
+      "loss": 0.6747,
+      "step": 16272
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.471750724740336e-05,
+      "loss": 0.685,
+      "step": 16273
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4714927516229332e-05,
+      "loss": 0.6925,
+      "step": 16274
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.471234778809116e-05,
+      "loss": 0.6932,
+      "step": 16275
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.47097680630163e-05,
+      "loss": 0.7185,
+      "step": 16276
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4707188341032232e-05,
+      "loss": 0.6913,
+      "step": 16277
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.470460862216643e-05,
+      "loss": 0.7304,
+      "step": 16278
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4702028906446374e-05,
+      "loss": 0.671,
+      "step": 16279
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.469944919389952e-05,
+      "loss": 0.6895,
+      "step": 16280
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4696869484553355e-05,
+      "loss": 0.7268,
+      "step": 16281
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.469428977843535e-05,
+      "loss": 0.7455,
+      "step": 16282
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4691710075572967e-05,
+      "loss": 0.697,
+      "step": 16283
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4689130375993683e-05,
+      "loss": 0.7112,
+      "step": 16284
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4686550679724975e-05,
+      "loss": 0.7131,
+      "step": 16285
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4683970986794313e-05,
+      "loss": 0.6758,
+      "step": 16286
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4681391297229163e-05,
+      "loss": 0.7033,
+      "step": 16287
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4678811611057007e-05,
+      "loss": 0.6927,
+      "step": 16288
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4676231928305312e-05,
+      "loss": 0.6956,
+      "step": 16289
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4673652249001546e-05,
+      "loss": 0.7931,
+      "step": 16290
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4671072573173194e-05,
+      "loss": 0.6665,
+      "step": 16291
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4668492900847716e-05,
+      "loss": 0.7136,
+      "step": 16292
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.466591323205258e-05,
+      "loss": 0.7157,
+      "step": 16293
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4663333566815273e-05,
+      "loss": 0.6678,
+      "step": 16294
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4660753905163264e-05,
+      "loss": 0.6714,
+      "step": 16295
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4658174247124017e-05,
+      "loss": 0.6954,
+      "step": 16296
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4655594592725005e-05,
+      "loss": 0.6642,
+      "step": 16297
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4653014941993693e-05,
+      "loss": 0.7839,
+      "step": 16298
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.465043529495757e-05,
+      "loss": 0.6742,
+      "step": 16299
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.46478556516441e-05,
+      "loss": 0.6474,
+      "step": 16300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4645276012080743e-05,
+      "loss": 0.8088,
+      "step": 16301
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.464269637629499e-05,
+      "loss": 0.7172,
+      "step": 16302
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4640116744314304e-05,
+      "loss": 0.7083,
+      "step": 16303
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.463753711616615e-05,
+      "loss": 0.7393,
+      "step": 16304
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4634957491878012e-05,
+      "loss": 0.7144,
+      "step": 16305
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.463237787147735e-05,
+      "loss": 0.6619,
+      "step": 16306
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4629798254991643e-05,
+      "loss": 0.6311,
+      "step": 16307
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.462721864244836e-05,
+      "loss": 0.6963,
+      "step": 16308
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4624639033874973e-05,
+      "loss": 0.6643,
+      "step": 16309
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4622059429298948e-05,
+      "loss": 0.731,
+      "step": 16310
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.461947982874777e-05,
+      "loss": 0.7242,
+      "step": 16311
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4616900232248894e-05,
+      "loss": 0.8155,
+      "step": 16312
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4614320639829797e-05,
+      "loss": 0.7174,
+      "step": 16313
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4611741051517952e-05,
+      "loss": 0.6938,
+      "step": 16314
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.460916146734084e-05,
+      "loss": 0.667,
+      "step": 16315
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4606581887325912e-05,
+      "loss": 0.7542,
+      "step": 16316
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4604002311500648e-05,
+      "loss": 0.7078,
+      "step": 16317
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4601422739892514e-05,
+      "loss": 0.9176,
+      "step": 16318
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4598843172528993e-05,
+      "loss": 0.6905,
+      "step": 16319
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.459626360943755e-05,
+      "loss": 0.729,
+      "step": 16320
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4593684050645642e-05,
+      "loss": 0.7119,
+      "step": 16321
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.459110449618076e-05,
+      "loss": 0.6935,
+      "step": 16322
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.458852494607037e-05,
+      "loss": 0.615,
+      "step": 16323
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4585945400341934e-05,
+      "loss": 0.6685,
+      "step": 16324
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4583365859022932e-05,
+      "loss": 0.6532,
+      "step": 16325
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.458078632214083e-05,
+      "loss": 0.6856,
+      "step": 16326
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4578206789723095e-05,
+      "loss": 0.7664,
+      "step": 16327
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4575627261797206e-05,
+      "loss": 0.7209,
+      "step": 16328
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.457304773839063e-05,
+      "loss": 0.693,
+      "step": 16329
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4570468219530832e-05,
+      "loss": 0.6906,
+      "step": 16330
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4567888705245293e-05,
+      "loss": 0.6545,
+      "step": 16331
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4565309195561474e-05,
+      "loss": 0.715,
+      "step": 16332
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4562729690506843e-05,
+      "loss": 0.6839,
+      "step": 16333
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.456015019010888e-05,
+      "loss": 0.6438,
+      "step": 16334
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.455757069439505e-05,
+      "loss": 0.6794,
+      "step": 16335
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.455499120339283e-05,
+      "loss": 0.8149,
+      "step": 16336
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4552411717129683e-05,
+      "loss": 0.779,
+      "step": 16337
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4549832235633067e-05,
+      "loss": 0.7448,
+      "step": 16338
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4547252758930475e-05,
+      "loss": 0.7444,
+      "step": 16339
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4544673287049364e-05,
+      "loss": 0.7215,
+      "step": 16340
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.45420938200172e-05,
+      "loss": 0.7633,
+      "step": 16341
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4539514357861465e-05,
+      "loss": 0.7228,
+      "step": 16342
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4536934900609623e-05,
+      "loss": 0.6959,
+      "step": 16343
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.453435544828914e-05,
+      "loss": 0.7876,
+      "step": 16344
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4531776000927486e-05,
+      "loss": 0.7473,
+      "step": 16345
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4529196558552138e-05,
+      "loss": 0.7205,
+      "step": 16346
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.452661712119056e-05,
+      "loss": 0.701,
+      "step": 16347
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.452403768887022e-05,
+      "loss": 0.6305,
+      "step": 16348
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4521458261618595e-05,
+      "loss": 0.6904,
+      "step": 16349
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4518878839463142e-05,
+      "loss": 0.6676,
+      "step": 16350
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4516299422431345e-05,
+      "loss": 0.6903,
+      "step": 16351
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4513720010550663e-05,
+      "loss": 0.7402,
+      "step": 16352
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.451114060384856e-05,
+      "loss": 0.6808,
+      "step": 16353
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.450856120235252e-05,
+      "loss": 0.6816,
+      "step": 16354
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4505981806090007e-05,
+      "loss": 0.7008,
+      "step": 16355
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4503402415088488e-05,
+      "loss": 0.7356,
+      "step": 16356
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4500823029375432e-05,
+      "loss": 0.6463,
+      "step": 16357
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.44982436489783e-05,
+      "loss": 0.7104,
+      "step": 16358
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4495664273924575e-05,
+      "loss": 0.7082,
+      "step": 16359
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4493084904241716e-05,
+      "loss": 0.7622,
+      "step": 16360
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4490505539957194e-05,
+      "loss": 0.6839,
+      "step": 16361
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.448792618109848e-05,
+      "loss": 0.6972,
+      "step": 16362
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4485346827693044e-05,
+      "loss": 0.6836,
+      "step": 16363
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4482767479768347e-05,
+      "loss": 0.6766,
+      "step": 16364
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4480188137351868e-05,
+      "loss": 0.711,
+      "step": 16365
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4477608800471068e-05,
+      "loss": 0.7128,
+      "step": 16366
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.447502946915341e-05,
+      "loss": 0.6903,
+      "step": 16367
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4472450143426378e-05,
+      "loss": 0.6623,
+      "step": 16368
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.446987082331743e-05,
+      "loss": 0.6321,
+      "step": 16369
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.446729150885403e-05,
+      "loss": 0.6964,
+      "step": 16370
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4464712200063662e-05,
+      "loss": 0.6658,
+      "step": 16371
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4462132896973778e-05,
+      "loss": 0.74,
+      "step": 16372
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.445955359961185e-05,
+      "loss": 0.6745,
+      "step": 16373
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4456974308005353e-05,
+      "loss": 0.6922,
+      "step": 16374
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4454395022181747e-05,
+      "loss": 0.6936,
+      "step": 16375
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4451815742168513e-05,
+      "loss": 0.7019,
+      "step": 16376
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.44492364679931e-05,
+      "loss": 0.7608,
+      "step": 16377
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.444665719968298e-05,
+      "loss": 0.6791,
+      "step": 16378
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.444407793726563e-05,
+      "loss": 0.7508,
+      "step": 16379
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4441498680768512e-05,
+      "loss": 0.7742,
+      "step": 16380
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.443891943021909e-05,
+      "loss": 0.6996,
+      "step": 16381
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.443634018564484e-05,
+      "loss": 0.6878,
+      "step": 16382
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4433760947073224e-05,
+      "loss": 0.7879,
+      "step": 16383
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4431181714531707e-05,
+      "loss": 0.7134,
+      "step": 16384
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.442860248804776e-05,
+      "loss": 0.6622,
+      "step": 16385
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4426023267648855e-05,
+      "loss": 0.6464,
+      "step": 16386
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4423444053362447e-05,
+      "loss": 0.6294,
+      "step": 16387
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4420864845216014e-05,
+      "loss": 0.7483,
+      "step": 16388
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.441828564323702e-05,
+      "loss": 0.709,
+      "step": 16389
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.441570644745293e-05,
+      "loss": 0.686,
+      "step": 16390
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4413127257891204e-05,
+      "loss": 0.7081,
+      "step": 16391
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4410548074579327e-05,
+      "loss": 0.7181,
+      "step": 16392
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.440796889754475e-05,
+      "loss": 0.6919,
+      "step": 16393
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4405389726814942e-05,
+      "loss": 0.6805,
+      "step": 16394
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.440281056241738e-05,
+      "loss": 0.7364,
+      "step": 16395
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.440023140437952e-05,
+      "loss": 0.7474,
+      "step": 16396
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4397652252728835e-05,
+      "loss": 0.7332,
+      "step": 16397
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4395073107492776e-05,
+      "loss": 0.7082,
+      "step": 16398
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4392493968698833e-05,
+      "loss": 0.7905,
+      "step": 16399
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4389914836374457e-05,
+      "loss": 0.6974,
+      "step": 16400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4387335710547112e-05,
+      "loss": 0.734,
+      "step": 16401
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4384756591244275e-05,
+      "loss": 0.6966,
+      "step": 16402
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4382177478493407e-05,
+      "loss": 0.6821,
+      "step": 16403
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4379598372321972e-05,
+      "loss": 0.7167,
+      "step": 16404
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4377019272757436e-05,
+      "loss": 0.6779,
+      "step": 16405
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4374440179827268e-05,
+      "loss": 0.6947,
+      "step": 16406
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4371861093558936e-05,
+      "loss": 0.6558,
+      "step": 16407
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4369282013979894e-05,
+      "loss": 0.7831,
+      "step": 16408
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4366702941117624e-05,
+      "loss": 0.7305,
+      "step": 16409
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4364123874999582e-05,
+      "loss": 0.7332,
+      "step": 16410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4361544815653227e-05,
+      "loss": 0.7369,
+      "step": 16411
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.435896576310604e-05,
+      "loss": 0.715,
+      "step": 16412
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4356386717385476e-05,
+      "loss": 0.7014,
+      "step": 16413
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4353807678518998e-05,
+      "loss": 0.6575,
+      "step": 16414
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4351228646534084e-05,
+      "loss": 0.7463,
+      "step": 16415
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4348649621458194e-05,
+      "loss": 0.69,
+      "step": 16416
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4346070603318785e-05,
+      "loss": 0.7265,
+      "step": 16417
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.434349159214333e-05,
+      "loss": 0.7334,
+      "step": 16418
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.434091258795928e-05,
+      "loss": 0.7381,
+      "step": 16419
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4338333590794118e-05,
+      "loss": 0.7103,
+      "step": 16420
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4335754600675304e-05,
+      "loss": 0.6794,
+      "step": 16421
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.433317561763029e-05,
+      "loss": 0.6782,
+      "step": 16422
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4330596641686563e-05,
+      "loss": 0.6869,
+      "step": 16423
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.432801767287157e-05,
+      "loss": 0.7,
+      "step": 16424
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.432543871121278e-05,
+      "loss": 0.6158,
+      "step": 16425
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4322859756737658e-05,
+      "loss": 0.7036,
+      "step": 16426
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4320280809473674e-05,
+      "loss": 0.7106,
+      "step": 16427
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4317701869448277e-05,
+      "loss": 0.7151,
+      "step": 16428
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.431512293668895e-05,
+      "loss": 0.6948,
+      "step": 16429
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4312544011223144e-05,
+      "loss": 0.6857,
+      "step": 16430
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4309965093078323e-05,
+      "loss": 0.7498,
+      "step": 16431
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.430738618228196e-05,
+      "loss": 0.7244,
+      "step": 16432
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4304807278861516e-05,
+      "loss": 0.6282,
+      "step": 16433
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4302228382844444e-05,
+      "loss": 0.7576,
+      "step": 16434
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4299649494258227e-05,
+      "loss": 0.6888,
+      "step": 16435
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.429707061313031e-05,
+      "loss": 0.6956,
+      "step": 16436
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4294491739488174e-05,
+      "loss": 0.7967,
+      "step": 16437
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.429191287335927e-05,
+      "loss": 0.7156,
+      "step": 16438
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.428933401477105e-05,
+      "loss": 0.7381,
+      "step": 16439
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4286755163751004e-05,
+      "loss": 0.7,
+      "step": 16440
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4284176320326578e-05,
+      "loss": 0.6699,
+      "step": 16441
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4281597484525238e-05,
+      "loss": 0.7388,
+      "step": 16442
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4279018656374452e-05,
+      "loss": 0.7259,
+      "step": 16443
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.427643983590168e-05,
+      "loss": 0.737,
+      "step": 16444
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4273861023134382e-05,
+      "loss": 0.6976,
+      "step": 16445
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4271282218100023e-05,
+      "loss": 0.8297,
+      "step": 16446
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.426870342082607e-05,
+      "loss": 0.7385,
+      "step": 16447
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.426612463133997e-05,
+      "loss": 0.7286,
+      "step": 16448
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.426354584966921e-05,
+      "loss": 0.7301,
+      "step": 16449
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.426096707584124e-05,
+      "loss": 0.7232,
+      "step": 16450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4258388309883512e-05,
+      "loss": 0.7784,
+      "step": 16451
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4255809551823504e-05,
+      "loss": 0.711,
+      "step": 16452
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4253230801688677e-05,
+      "loss": 0.8023,
+      "step": 16453
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.425065205950648e-05,
+      "loss": 0.7161,
+      "step": 16454
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4248073325304388e-05,
+      "loss": 0.6724,
+      "step": 16455
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4245494599109857e-05,
+      "loss": 0.6445,
+      "step": 16456
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4242915880950358e-05,
+      "loss": 0.6746,
+      "step": 16457
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4240337170853342e-05,
+      "loss": 0.6802,
+      "step": 16458
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4237758468846267e-05,
+      "loss": 0.7432,
+      "step": 16459
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4235179774956607e-05,
+      "loss": 0.6896,
+      "step": 16460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.423260108921182e-05,
+      "loss": 0.6795,
+      "step": 16461
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4230022411639354e-05,
+      "loss": 0.7452,
+      "step": 16462
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4227443742266693e-05,
+      "loss": 0.752,
+      "step": 16463
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4224865081121284e-05,
+      "loss": 0.6777,
+      "step": 16464
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4222286428230587e-05,
+      "loss": 0.7668,
+      "step": 16465
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4219707783622072e-05,
+      "loss": 0.7029,
+      "step": 16466
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.42171291473232e-05,
+      "loss": 0.6903,
+      "step": 16467
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4214550519361416e-05,
+      "loss": 0.8103,
+      "step": 16468
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.42119718997642e-05,
+      "loss": 0.7043,
+      "step": 16469
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4209393288559005e-05,
+      "loss": 0.6869,
+      "step": 16470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4206814685773285e-05,
+      "loss": 0.7334,
+      "step": 16471
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4204236091434513e-05,
+      "loss": 0.7095,
+      "step": 16472
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4201657505570142e-05,
+      "loss": 0.7098,
+      "step": 16473
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.419907892820763e-05,
+      "loss": 0.6743,
+      "step": 16474
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4196500359374447e-05,
+      "loss": 0.7063,
+      "step": 16475
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4193921799098045e-05,
+      "loss": 0.6909,
+      "step": 16476
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4191343247405894e-05,
+      "loss": 0.7643,
+      "step": 16477
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4188764704325442e-05,
+      "loss": 0.7575,
+      "step": 16478
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4186186169884147e-05,
+      "loss": 0.7476,
+      "step": 16479
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4183607644109483e-05,
+      "loss": 0.7047,
+      "step": 16480
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4181029127028898e-05,
+      "loss": 0.6579,
+      "step": 16481
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.417845061866985e-05,
+      "loss": 0.6921,
+      "step": 16482
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4175872119059812e-05,
+      "loss": 0.7597,
+      "step": 16483
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4173293628226236e-05,
+      "loss": 0.7235,
+      "step": 16484
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4170715146196572e-05,
+      "loss": 0.7471,
+      "step": 16485
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4168136672998297e-05,
+      "loss": 0.7275,
+      "step": 16486
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.416555820865886e-05,
+      "loss": 0.7253,
+      "step": 16487
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4162979753205717e-05,
+      "loss": 0.6906,
+      "step": 16488
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4160401306666336e-05,
+      "loss": 0.7208,
+      "step": 16489
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4157822869068172e-05,
+      "loss": 0.7027,
+      "step": 16490
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4155244440438677e-05,
+      "loss": 0.7362,
+      "step": 16491
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.415266602080532e-05,
+      "loss": 0.7235,
+      "step": 16492
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.415008761019556e-05,
+      "loss": 0.6752,
+      "step": 16493
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.414750920863684e-05,
+      "loss": 0.6865,
+      "step": 16494
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.414493081615664e-05,
+      "loss": 0.6994,
+      "step": 16495
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4142352432782404e-05,
+      "loss": 0.7311,
+      "step": 16496
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.41397740585416e-05,
+      "loss": 0.7314,
+      "step": 16497
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4137195693461676e-05,
+      "loss": 0.6773,
+      "step": 16498
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4134617337570088e-05,
+      "loss": 0.6961,
+      "step": 16499
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4132038990894303e-05,
+      "loss": 0.6528,
+      "step": 16500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.412946065346178e-05,
+      "loss": 0.7115,
+      "step": 16501
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4126882325299963e-05,
+      "loss": 0.7311,
+      "step": 16502
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4124304006436325e-05,
+      "loss": 0.6524,
+      "step": 16503
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.412172569689832e-05,
+      "loss": 0.6367,
+      "step": 16504
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.41191473967134e-05,
+      "loss": 0.7194,
+      "step": 16505
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4116569105909026e-05,
+      "loss": 0.6576,
+      "step": 16506
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4113990824512654e-05,
+      "loss": 0.7275,
+      "step": 16507
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4111412552551745e-05,
+      "loss": 0.6843,
+      "step": 16508
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4108834290053746e-05,
+      "loss": 0.8408,
+      "step": 16509
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4106256037046127e-05,
+      "loss": 0.7485,
+      "step": 16510
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.410367779355634e-05,
+      "loss": 0.7128,
+      "step": 16511
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.410109955961183e-05,
+      "loss": 0.6995,
+      "step": 16512
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4098521335240076e-05,
+      "loss": 0.6555,
+      "step": 16513
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4095943120468517e-05,
+      "loss": 0.6847,
+      "step": 16514
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4093364915324613e-05,
+      "loss": 0.6816,
+      "step": 16515
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4090786719835827e-05,
+      "loss": 0.6602,
+      "step": 16516
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4088208534029614e-05,
+      "loss": 0.6414,
+      "step": 16517
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4085630357933424e-05,
+      "loss": 0.6507,
+      "step": 16518
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4083052191574707e-05,
+      "loss": 0.7051,
+      "step": 16519
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4080474034980936e-05,
+      "loss": 0.6971,
+      "step": 16520
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4077895888179556e-05,
+      "loss": 0.6697,
+      "step": 16521
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4075317751198023e-05,
+      "loss": 0.7273,
+      "step": 16522
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4072739624063793e-05,
+      "loss": 0.6199,
+      "step": 16523
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4070161506804324e-05,
+      "loss": 0.6994,
+      "step": 16524
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4067583399447077e-05,
+      "loss": 0.722,
+      "step": 16525
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4065005302019492e-05,
+      "loss": 0.7071,
+      "step": 16526
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4062427214549037e-05,
+      "loss": 0.7026,
+      "step": 16527
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4059849137063165e-05,
+      "loss": 0.715,
+      "step": 16528
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.405727106958932e-05,
+      "loss": 0.6747,
+      "step": 16529
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4054693012154973e-05,
+      "loss": 0.7002,
+      "step": 16530
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4052114964787574e-05,
+      "loss": 0.6533,
+      "step": 16531
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4049536927514568e-05,
+      "loss": 0.7248,
+      "step": 16532
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4046958900363423e-05,
+      "loss": 0.6633,
+      "step": 16533
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4044380883361586e-05,
+      "loss": 0.6948,
+      "step": 16534
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4041802876536507e-05,
+      "loss": 0.702,
+      "step": 16535
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.403922487991565e-05,
+      "loss": 0.7546,
+      "step": 16536
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4036646893526473e-05,
+      "loss": 0.7313,
+      "step": 16537
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4034068917396415e-05,
+      "loss": 0.6313,
+      "step": 16538
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4031490951552938e-05,
+      "loss": 0.6933,
+      "step": 16539
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4028912996023487e-05,
+      "loss": 0.6486,
+      "step": 16540
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4026335050835526e-05,
+      "loss": 0.6904,
+      "step": 16541
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.402375711601651e-05,
+      "loss": 0.731,
+      "step": 16542
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.402117919159388e-05,
+      "loss": 0.748,
+      "step": 16543
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4018601277595103e-05,
+      "loss": 0.6991,
+      "step": 16544
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4016023374047627e-05,
+      "loss": 0.7841,
+      "step": 16545
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4013445480978903e-05,
+      "loss": 0.7228,
+      "step": 16546
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4010867598416387e-05,
+      "loss": 0.7633,
+      "step": 16547
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.400828972638753e-05,
+      "loss": 0.6478,
+      "step": 16548
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.400571186491978e-05,
+      "loss": 0.663,
+      "step": 16549
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.40031340140406e-05,
+      "loss": 0.6589,
+      "step": 16550
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.400055617377744e-05,
+      "loss": 0.62,
+      "step": 16551
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3997978344157746e-05,
+      "loss": 0.7266,
+      "step": 16552
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3995400525208977e-05,
+      "loss": 0.7361,
+      "step": 16553
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3992822716958582e-05,
+      "loss": 0.7026,
+      "step": 16554
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.399024491943401e-05,
+      "loss": 0.7189,
+      "step": 16555
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.398766713266272e-05,
+      "loss": 0.6772,
+      "step": 16556
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3985089356672162e-05,
+      "loss": 0.6313,
+      "step": 16557
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3982511591489788e-05,
+      "loss": 0.6688,
+      "step": 16558
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.397993383714305e-05,
+      "loss": 0.717,
+      "step": 16559
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3977356093659382e-05,
+      "loss": 0.7539,
+      "step": 16560
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3974778361066262e-05,
+      "loss": 0.6591,
+      "step": 16561
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3972200639391132e-05,
+      "loss": 0.639,
+      "step": 16562
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3969622928661433e-05,
+      "loss": 0.7456,
+      "step": 16563
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.396704522890463e-05,
+      "loss": 0.681,
+      "step": 16564
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.396446754014817e-05,
+      "loss": 0.7252,
+      "step": 16565
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3961889862419492e-05,
+      "loss": 0.6927,
+      "step": 16566
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3959312195746067e-05,
+      "loss": 0.6466,
+      "step": 16567
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3956734540155333e-05,
+      "loss": 0.6875,
+      "step": 16568
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3954156895674737e-05,
+      "loss": 0.6894,
+      "step": 16569
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.395157926233174e-05,
+      "loss": 0.7241,
+      "step": 16570
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.394900164015379e-05,
+      "loss": 0.6694,
+      "step": 16571
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3946424029168326e-05,
+      "loss": 0.6832,
+      "step": 16572
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3943846429402812e-05,
+      "loss": 0.6886,
+      "step": 16573
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3941268840884694e-05,
+      "loss": 0.7185,
+      "step": 16574
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3938691263641418e-05,
+      "loss": 0.736,
+      "step": 16575
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3936113697700434e-05,
+      "loss": 0.7522,
+      "step": 16576
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3933536143089197e-05,
+      "loss": 0.6979,
+      "step": 16577
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3930958599835157e-05,
+      "loss": 0.7157,
+      "step": 16578
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3928381067965755e-05,
+      "loss": 0.6956,
+      "step": 16579
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3925803547508438e-05,
+      "loss": 0.7632,
+      "step": 16580
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3923226038490664e-05,
+      "loss": 0.6587,
+      "step": 16581
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3920648540939883e-05,
+      "loss": 0.6843,
+      "step": 16582
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3918071054883535e-05,
+      "loss": 0.7764,
+      "step": 16583
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3915493580349075e-05,
+      "loss": 0.8368,
+      "step": 16584
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.391291611736395e-05,
+      "loss": 0.7199,
+      "step": 16585
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3910338665955603e-05,
+      "loss": 0.7151,
+      "step": 16586
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3907761226151497e-05,
+      "loss": 0.7415,
+      "step": 16587
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3905183797979067e-05,
+      "loss": 0.685,
+      "step": 16588
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.390260638146576e-05,
+      "loss": 0.8183,
+      "step": 16589
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3900028976639035e-05,
+      "loss": 0.7015,
+      "step": 16590
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3897451583526332e-05,
+      "loss": 0.7077,
+      "step": 16591
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.38948742021551e-05,
+      "loss": 0.7448,
+      "step": 16592
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3892296832552784e-05,
+      "loss": 0.6638,
+      "step": 16593
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.388971947474684e-05,
+      "loss": 0.6622,
+      "step": 16594
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3887142128764704e-05,
+      "loss": 0.6611,
+      "step": 16595
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3884564794633832e-05,
+      "loss": 0.701,
+      "step": 16596
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.388198747238167e-05,
+      "loss": 0.8386,
+      "step": 16597
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3879410162035666e-05,
+      "loss": 0.7506,
+      "step": 16598
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.387683286362326e-05,
+      "loss": 0.7813,
+      "step": 16599
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3874255577171896e-05,
+      "loss": 0.7068,
+      "step": 16600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.387167830270903e-05,
+      "loss": 0.7098,
+      "step": 16601
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3869101040262106e-05,
+      "loss": 0.6823,
+      "step": 16602
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3866523789858564e-05,
+      "loss": 0.7707,
+      "step": 16603
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3863946551525862e-05,
+      "loss": 0.7645,
+      "step": 16604
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.386136932529144e-05,
+      "loss": 0.7677,
+      "step": 16605
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3858792111182735e-05,
+      "loss": 0.7433,
+      "step": 16606
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.385621490922721e-05,
+      "loss": 0.7371,
+      "step": 16607
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.38536377194523e-05,
+      "loss": 0.6992,
+      "step": 16608
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.385106054188545e-05,
+      "loss": 0.7031,
+      "step": 16609
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3848483376554108e-05,
+      "loss": 0.6907,
+      "step": 16610
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.384590622348572e-05,
+      "loss": 0.6813,
+      "step": 16611
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3843329082707728e-05,
+      "loss": 0.6678,
+      "step": 16612
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3840751954247582e-05,
+      "loss": 0.6838,
+      "step": 16613
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3838174838132722e-05,
+      "loss": 0.6634,
+      "step": 16614
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3835597734390597e-05,
+      "loss": 0.7299,
+      "step": 16615
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.383302064304864e-05,
+      "loss": 0.6389,
+      "step": 16616
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3830443564134316e-05,
+      "loss": 0.6849,
+      "step": 16617
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3827866497675065e-05,
+      "loss": 0.7895,
+      "step": 16618
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.382528944369831e-05,
+      "loss": 0.6709,
+      "step": 16619
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3822712402231512e-05,
+      "loss": 0.7681,
+      "step": 16620
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3820135373302112e-05,
+      "loss": 0.7452,
+      "step": 16621
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.381755835693756e-05,
+      "loss": 0.7243,
+      "step": 16622
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3814981353165284e-05,
+      "loss": 0.7384,
+      "step": 16623
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.381240436201274e-05,
+      "loss": 0.734,
+      "step": 16624
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.380982738350737e-05,
+      "loss": 0.6816,
+      "step": 16625
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.380725041767662e-05,
+      "loss": 0.7362,
+      "step": 16626
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.380467346454792e-05,
+      "loss": 0.7149,
+      "step": 16627
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3802096524148727e-05,
+      "loss": 0.7395,
+      "step": 16628
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.379951959650648e-05,
+      "loss": 0.7798,
+      "step": 16629
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3796942681648616e-05,
+      "loss": 0.7046,
+      "step": 16630
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3794365779602587e-05,
+      "loss": 0.6965,
+      "step": 16631
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.379178889039583e-05,
+      "loss": 0.7353,
+      "step": 16632
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3789212014055782e-05,
+      "loss": 0.7179,
+      "step": 16633
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3786635150609894e-05,
+      "loss": 0.7587,
+      "step": 16634
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.378405830008561e-05,
+      "loss": 0.7478,
+      "step": 16635
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.378148146251036e-05,
+      "loss": 0.7381,
+      "step": 16636
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3778904637911596e-05,
+      "loss": 0.6712,
+      "step": 16637
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3776327826316763e-05,
+      "loss": 0.6942,
+      "step": 16638
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.377375102775329e-05,
+      "loss": 0.73,
+      "step": 16639
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3771174242248627e-05,
+      "loss": 0.6863,
+      "step": 16640
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3768597469830202e-05,
+      "loss": 0.6556,
+      "step": 16641
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3766020710525474e-05,
+      "loss": 0.7301,
+      "step": 16642
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3763443964361875e-05,
+      "loss": 0.734,
+      "step": 16643
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3760867231366844e-05,
+      "loss": 0.6542,
+      "step": 16644
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.375829051156783e-05,
+      "loss": 0.7451,
+      "step": 16645
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3755713804992265e-05,
+      "loss": 0.6521,
+      "step": 16646
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3753137111667588e-05,
+      "loss": 0.6508,
+      "step": 16647
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.375056043162125e-05,
+      "loss": 0.7704,
+      "step": 16648
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.374798376488069e-05,
+      "loss": 0.749,
+      "step": 16649
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.374540711147333e-05,
+      "loss": 0.7025,
+      "step": 16650
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.374283047142663e-05,
+      "loss": 0.7359,
+      "step": 16651
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3740253844768024e-05,
+      "loss": 0.7073,
+      "step": 16652
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3737677231524945e-05,
+      "loss": 0.6529,
+      "step": 16653
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.373510063172484e-05,
+      "loss": 0.68,
+      "step": 16654
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3732524045395144e-05,
+      "loss": 0.71,
+      "step": 16655
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3729947472563293e-05,
+      "loss": 0.647,
+      "step": 16656
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3727370913256738e-05,
+      "loss": 0.6829,
+      "step": 16657
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.372479436750291e-05,
+      "loss": 0.6742,
+      "step": 16658
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3722217835329254e-05,
+      "loss": 0.7708,
+      "step": 16659
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.371964131676319e-05,
+      "loss": 0.6677,
+      "step": 16660
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.371706481183217e-05,
+      "loss": 0.7135,
+      "step": 16661
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3714488320563636e-05,
+      "loss": 0.7131,
+      "step": 16662
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.371191184298502e-05,
+      "loss": 0.6717,
+      "step": 16663
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3709335379123754e-05,
+      "loss": 0.6995,
+      "step": 16664
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3706758929007288e-05,
+      "loss": 0.7572,
+      "step": 16665
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3704182492663056e-05,
+      "loss": 0.6282,
+      "step": 16666
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.370160607011849e-05,
+      "loss": 0.6925,
+      "step": 16667
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3699029661401032e-05,
+      "loss": 0.7181,
+      "step": 16668
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3696453266538118e-05,
+      "loss": 0.7275,
+      "step": 16669
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3693876885557183e-05,
+      "loss": 0.672,
+      "step": 16670
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.369130051848567e-05,
+      "loss": 0.68,
+      "step": 16671
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3688724165351014e-05,
+      "loss": 0.7567,
+      "step": 16672
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3686147826180645e-05,
+      "loss": 0.725,
+      "step": 16673
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3683571501002006e-05,
+      "loss": 0.6931,
+      "step": 16674
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3680995189842532e-05,
+      "loss": 0.6846,
+      "step": 16675
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3678418892729652e-05,
+      "loss": 0.7804,
+      "step": 16676
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.367584260969082e-05,
+      "loss": 0.7105,
+      "step": 16677
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3673266340753453e-05,
+      "loss": 0.7319,
+      "step": 16678
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3670690085945e-05,
+      "loss": 0.7266,
+      "step": 16679
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3668113845292888e-05,
+      "loss": 0.804,
+      "step": 16680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.366553761882455e-05,
+      "loss": 0.7098,
+      "step": 16681
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.366296140656743e-05,
+      "loss": 0.7494,
+      "step": 16682
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.366038520854896e-05,
+      "loss": 0.6477,
+      "step": 16683
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3657809024796568e-05,
+      "loss": 0.7113,
+      "step": 16684
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3655232855337703e-05,
+      "loss": 0.6611,
+      "step": 16685
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3652656700199787e-05,
+      "loss": 0.6824,
+      "step": 16686
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3650080559410256e-05,
+      "loss": 0.7434,
+      "step": 16687
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3647504432996556e-05,
+      "loss": 0.6698,
+      "step": 16688
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3644928320986108e-05,
+      "loss": 0.7016,
+      "step": 16689
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3642352223406347e-05,
+      "loss": 0.6812,
+      "step": 16690
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3639776140284717e-05,
+      "loss": 0.7894,
+      "step": 16691
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3637200071648645e-05,
+      "loss": 0.7536,
+      "step": 16692
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3634624017525557e-05,
+      "loss": 0.6991,
+      "step": 16693
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3632047977942904e-05,
+      "loss": 0.7163,
+      "step": 16694
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3629471952928107e-05,
+      "loss": 0.7258,
+      "step": 16695
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3626895942508596e-05,
+      "loss": 0.6582,
+      "step": 16696
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3624319946711814e-05,
+      "loss": 0.6525,
+      "step": 16697
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.362174396556519e-05,
+      "loss": 0.6851,
+      "step": 16698
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3619167999096163e-05,
+      "loss": 0.6885,
+      "step": 16699
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3616592047332152e-05,
+      "loss": 0.7298,
+      "step": 16700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.361401611030059e-05,
+      "loss": 0.7161,
+      "step": 16701
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3611440188028922e-05,
+      "loss": 0.7324,
+      "step": 16702
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3608864280544575e-05,
+      "loss": 0.7559,
+      "step": 16703
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.360628838787497e-05,
+      "loss": 0.6009,
+      "step": 16704
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3603712510047552e-05,
+      "loss": 0.7163,
+      "step": 16705
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.360113664708975e-05,
+      "loss": 0.6435,
+      "step": 16706
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3598560799028987e-05,
+      "loss": 0.6602,
+      "step": 16707
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3595984965892705e-05,
+      "loss": 0.6877,
+      "step": 16708
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3593409147708334e-05,
+      "loss": 0.7219,
+      "step": 16709
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3590833344503294e-05,
+      "loss": 0.7471,
+      "step": 16710
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.358825755630503e-05,
+      "loss": 0.7053,
+      "step": 16711
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3585681783140966e-05,
+      "loss": 0.7139,
+      "step": 16712
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3583106025038522e-05,
+      "loss": 0.6414,
+      "step": 16713
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.358053028202515e-05,
+      "loss": 0.6518,
+      "step": 16714
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3577954554128267e-05,
+      "loss": 0.6784,
+      "step": 16715
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3575378841375297e-05,
+      "loss": 0.7109,
+      "step": 16716
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3572803143793686e-05,
+      "loss": 0.7448,
+      "step": 16717
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3570227461410855e-05,
+      "loss": 0.7208,
+      "step": 16718
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.356765179425424e-05,
+      "loss": 0.6694,
+      "step": 16719
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3565076142351256e-05,
+      "loss": 0.7267,
+      "step": 16720
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3562500505729332e-05,
+      "loss": 0.682,
+      "step": 16721
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3559924884415913e-05,
+      "loss": 0.6855,
+      "step": 16722
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3557349278438422e-05,
+      "loss": 0.6965,
+      "step": 16723
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.355477368782428e-05,
+      "loss": 0.6965,
+      "step": 16724
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.355219811260092e-05,
+      "loss": 0.6597,
+      "step": 16725
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3549622552795774e-05,
+      "loss": 0.7866,
+      "step": 16726
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3547047008436264e-05,
+      "loss": 0.8093,
+      "step": 16727
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3544471479549826e-05,
+      "loss": 0.7341,
+      "step": 16728
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.354189596616388e-05,
+      "loss": 0.7418,
+      "step": 16729
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3539320468305856e-05,
+      "loss": 0.6899,
+      "step": 16730
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3536744986003185e-05,
+      "loss": 0.6833,
+      "step": 16731
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.353416951928329e-05,
+      "loss": 0.6384,
+      "step": 16732
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3531594068173603e-05,
+      "loss": 0.6811,
+      "step": 16733
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.352901863270154e-05,
+      "loss": 0.7043,
+      "step": 16734
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3526443212894545e-05,
+      "loss": 0.6833,
+      "step": 16735
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.352386780878003e-05,
+      "loss": 0.71,
+      "step": 16736
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3521292420385422e-05,
+      "loss": 0.7852,
+      "step": 16737
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3518717047738162e-05,
+      "loss": 0.6935,
+      "step": 16738
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3516141690865668e-05,
+      "loss": 0.7086,
+      "step": 16739
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.351356634979536e-05,
+      "loss": 0.7187,
+      "step": 16740
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.351099102455466e-05,
+      "loss": 0.7896,
+      "step": 16741
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.350841571517101e-05,
+      "loss": 0.6938,
+      "step": 16742
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3505840421671823e-05,
+      "loss": 0.741,
+      "step": 16743
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3503265144084526e-05,
+      "loss": 0.694,
+      "step": 16744
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3500689882436553e-05,
+      "loss": 0.6583,
+      "step": 16745
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.349811463675532e-05,
+      "loss": 0.6696,
+      "step": 16746
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3495539407068258e-05,
+      "loss": 0.7048,
+      "step": 16747
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3492964193402777e-05,
+      "loss": 0.6844,
+      "step": 16748
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.349038899578632e-05,
+      "loss": 0.7878,
+      "step": 16749
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3487813814246305e-05,
+      "loss": 0.6844,
+      "step": 16750
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.348523864881015e-05,
+      "loss": 0.7261,
+      "step": 16751
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3482663499505287e-05,
+      "loss": 0.6956,
+      "step": 16752
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3480088366359137e-05,
+      "loss": 0.7343,
+      "step": 16753
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.347751324939912e-05,
+      "loss": 0.6317,
+      "step": 16754
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3474938148652665e-05,
+      "loss": 0.6535,
+      "step": 16755
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3472363064147192e-05,
+      "loss": 0.7146,
+      "step": 16756
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3469787995910122e-05,
+      "loss": 0.6906,
+      "step": 16757
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3467212943968886e-05,
+      "loss": 0.7214,
+      "step": 16758
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3464637908350903e-05,
+      "loss": 0.6802,
+      "step": 16759
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3462062889083596e-05,
+      "loss": 0.7611,
+      "step": 16760
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3459487886194383e-05,
+      "loss": 0.7619,
+      "step": 16761
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.345691289971068e-05,
+      "loss": 0.6907,
+      "step": 16762
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.345433792965993e-05,
+      "loss": 0.7096,
+      "step": 16763
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3451762976069537e-05,
+      "loss": 0.7196,
+      "step": 16764
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3449188038966924e-05,
+      "loss": 0.7389,
+      "step": 16765
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.344661311837952e-05,
+      "loss": 0.7152,
+      "step": 16766
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3444038214334747e-05,
+      "loss": 0.7107,
+      "step": 16767
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3441463326860014e-05,
+      "loss": 0.688,
+      "step": 16768
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.343888845598276e-05,
+      "loss": 0.7142,
+      "step": 16769
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3436313601730393e-05,
+      "loss": 0.6868,
+      "step": 16770
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3433738764130332e-05,
+      "loss": 0.7587,
+      "step": 16771
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3431163943210004e-05,
+      "loss": 0.6703,
+      "step": 16772
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.342858913899683e-05,
+      "loss": 0.7815,
+      "step": 16773
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3426014351518227e-05,
+      "loss": 0.6357,
+      "step": 16774
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3423439580801615e-05,
+      "loss": 0.7284,
+      "step": 16775
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3420864826874416e-05,
+      "loss": 0.7271,
+      "step": 16776
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3418290089764044e-05,
+      "loss": 0.712,
+      "step": 16777
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3415715369497928e-05,
+      "loss": 0.6966,
+      "step": 16778
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.341314066610348e-05,
+      "loss": 0.7467,
+      "step": 16779
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3410565979608125e-05,
+      "loss": 0.7195,
+      "step": 16780
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3407991310039272e-05,
+      "loss": 0.7628,
+      "step": 16781
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3405416657424342e-05,
+      "loss": 0.673,
+      "step": 16782
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.340284202179076e-05,
+      "loss": 0.7049,
+      "step": 16783
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3400267403165945e-05,
+      "loss": 0.704,
+      "step": 16784
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.33976928015773e-05,
+      "loss": 0.6199,
+      "step": 16785
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3395118217052262e-05,
+      "loss": 0.6782,
+      "step": 16786
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.339254364961824e-05,
+      "loss": 0.6615,
+      "step": 16787
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3389969099302643e-05,
+      "loss": 0.7036,
+      "step": 16788
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3387394566132908e-05,
+      "loss": 0.6648,
+      "step": 16789
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.338482005013644e-05,
+      "loss": 0.679,
+      "step": 16790
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3382245551340655e-05,
+      "loss": 0.7193,
+      "step": 16791
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3379671069772973e-05,
+      "loss": 0.5983,
+      "step": 16792
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3377096605460817e-05,
+      "loss": 0.7546,
+      "step": 16793
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3374522158431585e-05,
+      "loss": 0.6651,
+      "step": 16794
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3371947728712715e-05,
+      "loss": 0.633,
+      "step": 16795
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.336937331633161e-05,
+      "loss": 0.6842,
+      "step": 16796
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3366798921315686e-05,
+      "loss": 0.6735,
+      "step": 16797
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3364224543692368e-05,
+      "loss": 0.6347,
+      "step": 16798
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3361650183489065e-05,
+      "loss": 0.6988,
+      "step": 16799
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3359075840733197e-05,
+      "loss": 0.6892,
+      "step": 16800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.335650151545217e-05,
+      "loss": 0.734,
+      "step": 16801
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3353927207673397e-05,
+      "loss": 0.6229,
+      "step": 16802
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.335135291742431e-05,
+      "loss": 0.7061,
+      "step": 16803
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.334877864473231e-05,
+      "loss": 0.6983,
+      "step": 16804
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.334620438962481e-05,
+      "loss": 0.776,
+      "step": 16805
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3343630152129238e-05,
+      "loss": 0.7158,
+      "step": 16806
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3341055932272993e-05,
+      "loss": 0.6681,
+      "step": 16807
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3338481730083496e-05,
+      "loss": 0.6251,
+      "step": 16808
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.333590754558816e-05,
+      "loss": 0.7077,
+      "step": 16809
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.33333333788144e-05,
+      "loss": 0.6776,
+      "step": 16810
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.333075922978962e-05,
+      "loss": 0.727,
+      "step": 16811
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3328185098541252e-05,
+      "loss": 0.7849,
+      "step": 16812
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3325610985096695e-05,
+      "loss": 0.685,
+      "step": 16813
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.332303688948336e-05,
+      "loss": 0.7008,
+      "step": 16814
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3320462811728666e-05,
+      "loss": 0.6865,
+      "step": 16815
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.331788875186003e-05,
+      "loss": 0.7612,
+      "step": 16816
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3315314709904844e-05,
+      "loss": 0.7076,
+      "step": 16817
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3312740685890544e-05,
+      "loss": 0.7434,
+      "step": 16818
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.331016667984453e-05,
+      "loss": 0.679,
+      "step": 16819
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.330759269179422e-05,
+      "loss": 0.6475,
+      "step": 16820
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3305018721767017e-05,
+      "loss": 0.6681,
+      "step": 16821
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3302444769790328e-05,
+      "loss": 0.7513,
+      "step": 16822
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3299870835891576e-05,
+      "loss": 0.7143,
+      "step": 16823
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.329729692009817e-05,
+      "loss": 0.6937,
+      "step": 16824
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3294723022437512e-05,
+      "loss": 0.7363,
+      "step": 16825
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3292149142937024e-05,
+      "loss": 0.663,
+      "step": 16826
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.328957528162411e-05,
+      "loss": 0.7134,
+      "step": 16827
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3287001438526174e-05,
+      "loss": 0.6246,
+      "step": 16828
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.328442761367064e-05,
+      "loss": 0.6734,
+      "step": 16829
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.328185380708491e-05,
+      "loss": 0.6589,
+      "step": 16830
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.327928001879639e-05,
+      "loss": 0.7014,
+      "step": 16831
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3276706248832493e-05,
+      "loss": 0.6479,
+      "step": 16832
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3274132497220634e-05,
+      "loss": 0.8442,
+      "step": 16833
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3271558763988212e-05,
+      "loss": 0.7244,
+      "step": 16834
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.326898504916264e-05,
+      "loss": 0.6982,
+      "step": 16835
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3266411352771328e-05,
+      "loss": 0.8305,
+      "step": 16836
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.326383767484168e-05,
+      "loss": 0.6764,
+      "step": 16837
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3261264015401107e-05,
+      "loss": 0.6822,
+      "step": 16838
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.325869037447702e-05,
+      "loss": 0.7156,
+      "step": 16839
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.325611675209683e-05,
+      "loss": 0.7299,
+      "step": 16840
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3253543148287928e-05,
+      "loss": 0.6665,
+      "step": 16841
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.325096956307773e-05,
+      "loss": 0.6345,
+      "step": 16842
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3248395996493648e-05,
+      "loss": 0.6747,
+      "step": 16843
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3245822448563086e-05,
+      "loss": 0.6767,
+      "step": 16844
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3243248919313443e-05,
+      "loss": 0.6856,
+      "step": 16845
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3240675408772138e-05,
+      "loss": 0.7464,
+      "step": 16846
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3238101916966572e-05,
+      "loss": 0.7262,
+      "step": 16847
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3235528443924148e-05,
+      "loss": 0.7042,
+      "step": 16848
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.323295498967228e-05,
+      "loss": 0.9294,
+      "step": 16849
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3230381554238365e-05,
+      "loss": 0.6941,
+      "step": 16850
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3227808137649813e-05,
+      "loss": 0.6742,
+      "step": 16851
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3225234739934022e-05,
+      "loss": 0.6787,
+      "step": 16852
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3222661361118413e-05,
+      "loss": 0.705,
+      "step": 16853
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.322008800123038e-05,
+      "loss": 0.6833,
+      "step": 16854
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3217514660297325e-05,
+      "loss": 0.6393,
+      "step": 16855
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3214941338346656e-05,
+      "loss": 0.71,
+      "step": 16856
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3212368035405785e-05,
+      "loss": 0.7067,
+      "step": 16857
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.32097947515021e-05,
+      "loss": 0.6914,
+      "step": 16858
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3207221486663018e-05,
+      "loss": 0.68,
+      "step": 16859
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.320464824091595e-05,
+      "loss": 0.7022,
+      "step": 16860
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3202075014288278e-05,
+      "loss": 0.7437,
+      "step": 16861
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.319950180680741e-05,
+      "loss": 0.7197,
+      "step": 16862
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3196928618500763e-05,
+      "loss": 0.6739,
+      "step": 16863
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.319435544939573e-05,
+      "loss": 0.666,
+      "step": 16864
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3191782299519715e-05,
+      "loss": 0.7133,
+      "step": 16865
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3189209168900117e-05,
+      "loss": 0.6322,
+      "step": 16866
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3186636057564347e-05,
+      "loss": 0.725,
+      "step": 16867
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.31840629655398e-05,
+      "loss": 0.7269,
+      "step": 16868
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3181489892853874e-05,
+      "loss": 0.7069,
+      "step": 16869
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3178916839533985e-05,
+      "loss": 0.686,
+      "step": 16870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.317634380560752e-05,
+      "loss": 0.6886,
+      "step": 16871
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.317377079110189e-05,
+      "loss": 0.7857,
+      "step": 16872
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3171197796044492e-05,
+      "loss": 0.6533,
+      "step": 16873
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.316862482046273e-05,
+      "loss": 0.7153,
+      "step": 16874
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3166051864383995e-05,
+      "loss": 0.6485,
+      "step": 16875
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.31634789278357e-05,
+      "loss": 0.7564,
+      "step": 16876
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3160906010845237e-05,
+      "loss": 0.6787,
+      "step": 16877
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3158333113440007e-05,
+      "loss": 0.741,
+      "step": 16878
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3155760235647415e-05,
+      "loss": 0.673,
+      "step": 16879
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3153187377494854e-05,
+      "loss": 0.7323,
+      "step": 16880
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3150614539009734e-05,
+      "loss": 0.6806,
+      "step": 16881
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3148041720219445e-05,
+      "loss": 0.7416,
+      "step": 16882
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3145468921151376e-05,
+      "loss": 0.6817,
+      "step": 16883
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3142896141832944e-05,
+      "loss": 0.7172,
+      "step": 16884
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.314032338229154e-05,
+      "loss": 0.7264,
+      "step": 16885
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.313775064255456e-05,
+      "loss": 0.7021,
+      "step": 16886
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.313517792264941e-05,
+      "loss": 0.6693,
+      "step": 16887
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.313260522260348e-05,
+      "loss": 0.6609,
+      "step": 16888
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3130032542444166e-05,
+      "loss": 0.7323,
+      "step": 16889
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3127459882198878e-05,
+      "loss": 0.6676,
+      "step": 16890
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3124887241895006e-05,
+      "loss": 0.6416,
+      "step": 16891
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3122314621559938e-05,
+      "loss": 0.6029,
+      "step": 16892
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3119742021221086e-05,
+      "loss": 0.6182,
+      "step": 16893
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3117169440905838e-05,
+      "loss": 0.7376,
+      "step": 16894
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.311459688064159e-05,
+      "loss": 0.6669,
+      "step": 16895
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3112024340455746e-05,
+      "loss": 0.7468,
+      "step": 16896
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3109451820375694e-05,
+      "loss": 0.7133,
+      "step": 16897
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.310687932042883e-05,
+      "loss": 0.6738,
+      "step": 16898
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3104306840642553e-05,
+      "loss": 0.6612,
+      "step": 16899
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.310173438104426e-05,
+      "loss": 0.6346,
+      "step": 16900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.309916194166135e-05,
+      "loss": 0.6541,
+      "step": 16901
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3096589522521206e-05,
+      "loss": 0.7463,
+      "step": 16902
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.309401712365122e-05,
+      "loss": 0.6531,
+      "step": 16903
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.30914447450788e-05,
+      "loss": 0.8429,
+      "step": 16904
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3088872386831336e-05,
+      "loss": 0.6378,
+      "step": 16905
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3086300048936217e-05,
+      "loss": 0.71,
+      "step": 16906
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3083727731420842e-05,
+      "loss": 0.7222,
+      "step": 16907
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3081155434312605e-05,
+      "loss": 0.6727,
+      "step": 16908
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3078583157638896e-05,
+      "loss": 0.7172,
+      "step": 16909
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.307601090142711e-05,
+      "loss": 0.6272,
+      "step": 16910
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3073438665704642e-05,
+      "loss": 0.7139,
+      "step": 16911
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3070866450498878e-05,
+      "loss": 0.7834,
+      "step": 16912
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.306829425583722e-05,
+      "loss": 0.6969,
+      "step": 16913
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3065722081747053e-05,
+      "loss": 0.737,
+      "step": 16914
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3063149928255767e-05,
+      "loss": 0.6587,
+      "step": 16915
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3060577795390766e-05,
+      "loss": 0.689,
+      "step": 16916
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3058005683179433e-05,
+      "loss": 0.688,
+      "step": 16917
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3055433591649156e-05,
+      "loss": 0.7104,
+      "step": 16918
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3052861520827336e-05,
+      "loss": 0.6317,
+      "step": 16919
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3050289470741357e-05,
+      "loss": 0.7439,
+      "step": 16920
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3047717441418616e-05,
+      "loss": 0.6435,
+      "step": 16921
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3045145432886495e-05,
+      "loss": 0.7026,
+      "step": 16922
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3042573445172383e-05,
+      "loss": 0.6944,
+      "step": 16923
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.304000147830368e-05,
+      "loss": 0.6635,
+      "step": 16924
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3037429532307776e-05,
+      "loss": 0.6754,
+      "step": 16925
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3034857607212042e-05,
+      "loss": 0.7048,
+      "step": 16926
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3032285703043894e-05,
+      "loss": 0.7347,
+      "step": 16927
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3029713819830707e-05,
+      "loss": 0.6814,
+      "step": 16928
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3027141957599862e-05,
+      "loss": 0.6938,
+      "step": 16929
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3024570116378767e-05,
+      "loss": 0.7358,
+      "step": 16930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3021998296194803e-05,
+      "loss": 0.7725,
+      "step": 16931
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3019426497075345e-05,
+      "loss": 0.6945,
+      "step": 16932
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3016854719047804e-05,
+      "loss": 0.7638,
+      "step": 16933
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.301428296213955e-05,
+      "loss": 0.6874,
+      "step": 16934
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3011711226377973e-05,
+      "loss": 0.7285,
+      "step": 16935
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3009139511790473e-05,
+      "loss": 0.6523,
+      "step": 16936
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.300656781840443e-05,
+      "loss": 0.7423,
+      "step": 16937
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3003996146247216e-05,
+      "loss": 0.8097,
+      "step": 16938
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3001424495346245e-05,
+      "loss": 0.7609,
+      "step": 16939
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2998852865728885e-05,
+      "loss": 0.821,
+      "step": 16940
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2996281257422532e-05,
+      "loss": 0.6777,
+      "step": 16941
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2993709670454565e-05,
+      "loss": 0.7346,
+      "step": 16942
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2991138104852363e-05,
+      "loss": 0.7312,
+      "step": 16943
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2988566560643328e-05,
+      "loss": 0.6628,
+      "step": 16944
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.298599503785484e-05,
+      "loss": 0.6843,
+      "step": 16945
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2983423536514274e-05,
+      "loss": 0.6724,
+      "step": 16946
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.298085205664903e-05,
+      "loss": 0.6543,
+      "step": 16947
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2978280598286484e-05,
+      "loss": 0.7528,
+      "step": 16948
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2975709161454016e-05,
+      "loss": 0.7993,
+      "step": 16949
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2973137746179024e-05,
+      "loss": 0.7493,
+      "step": 16950
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2970566352488885e-05,
+      "loss": 0.6905,
+      "step": 16951
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2967994980410975e-05,
+      "loss": 0.7007,
+      "step": 16952
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2965423629972692e-05,
+      "loss": 0.8051,
+      "step": 16953
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2962852301201408e-05,
+      "loss": 0.6753,
+      "step": 16954
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.296028099412451e-05,
+      "loss": 0.6502,
+      "step": 16955
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2957709708769384e-05,
+      "loss": 0.6693,
+      "step": 16956
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.295513844516341e-05,
+      "loss": 0.6285,
+      "step": 16957
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2952567203333973e-05,
+      "loss": 0.6739,
+      "step": 16958
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.294999598330844e-05,
+      "loss": 0.7131,
+      "step": 16959
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2947424785114218e-05,
+      "loss": 0.6825,
+      "step": 16960
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2944853608778675e-05,
+      "loss": 0.6769,
+      "step": 16961
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.294228245432919e-05,
+      "loss": 0.7418,
+      "step": 16962
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2939711321793146e-05,
+      "loss": 0.6429,
+      "step": 16963
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.293714021119793e-05,
+      "loss": 0.7083,
+      "step": 16964
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2934569122570913e-05,
+      "loss": 0.721,
+      "step": 16965
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.293199805593948e-05,
+      "loss": 0.7036,
+      "step": 16966
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.292942701133102e-05,
+      "loss": 0.6416,
+      "step": 16967
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.29268559887729e-05,
+      "loss": 0.6446,
+      "step": 16968
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.29242849882925e-05,
+      "loss": 0.6362,
+      "step": 16969
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.292171400991721e-05,
+      "loss": 0.7045,
+      "step": 16970
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2919143053674408e-05,
+      "loss": 0.6836,
+      "step": 16971
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2916572119591464e-05,
+      "loss": 0.6632,
+      "step": 16972
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2914001207695762e-05,
+      "loss": 0.7083,
+      "step": 16973
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2911430318014682e-05,
+      "loss": 0.6831,
+      "step": 16974
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2908859450575603e-05,
+      "loss": 0.6997,
+      "step": 16975
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2906288605405894e-05,
+      "loss": 0.7306,
+      "step": 16976
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.290371778253295e-05,
+      "loss": 0.8001,
+      "step": 16977
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2901146981984133e-05,
+      "loss": 0.6646,
+      "step": 16978
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2898576203786826e-05,
+      "loss": 0.7402,
+      "step": 16979
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.289600544796841e-05,
+      "loss": 0.6505,
+      "step": 16980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.289343471455626e-05,
+      "loss": 0.7242,
+      "step": 16981
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.289086400357775e-05,
+      "loss": 0.7076,
+      "step": 16982
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.288829331506026e-05,
+      "loss": 0.7016,
+      "step": 16983
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.288572264903116e-05,
+      "loss": 0.6752,
+      "step": 16984
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.288315200551783e-05,
+      "loss": 0.6313,
+      "step": 16985
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2880581384547646e-05,
+      "loss": 0.7197,
+      "step": 16986
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2878010786147983e-05,
+      "loss": 0.6781,
+      "step": 16987
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.287544021034622e-05,
+      "loss": 0.7081,
+      "step": 16988
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2872869657169727e-05,
+      "loss": 0.6589,
+      "step": 16989
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2870299126645877e-05,
+      "loss": 0.6613,
+      "step": 16990
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2867728618802052e-05,
+      "loss": 0.7366,
+      "step": 16991
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2865158133665625e-05,
+      "loss": 0.6928,
+      "step": 16992
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2862587671263963e-05,
+      "loss": 0.7031,
+      "step": 16993
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2860017231624447e-05,
+      "loss": 0.6548,
+      "step": 16994
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.285744681477445e-05,
+      "loss": 0.6143,
+      "step": 16995
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2854876420741338e-05,
+      "loss": 0.6801,
+      "step": 16996
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2852306049552496e-05,
+      "loss": 0.6823,
+      "step": 16997
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.284973570123529e-05,
+      "loss": 0.7339,
+      "step": 16998
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2847165375817085e-05,
+      "loss": 0.6396,
+      "step": 16999
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.284459507332527e-05,
+      "loss": 0.7833,
+      "step": 17000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.284202479378721e-05,
+      "loss": 0.6707,
+      "step": 17001
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2839454537230282e-05,
+      "loss": 0.6864,
+      "step": 17002
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2836884303681845e-05,
+      "loss": 0.671,
+      "step": 17003
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.283431409316927e-05,
+      "loss": 0.7135,
+      "step": 17004
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.283174390571994e-05,
+      "loss": 0.6584,
+      "step": 17005
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2829173741361224e-05,
+      "loss": 0.7332,
+      "step": 17006
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2826603600120482e-05,
+      "loss": 0.835,
+      "step": 17007
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.28240334820251e-05,
+      "loss": 0.7458,
+      "step": 17008
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2821463387102436e-05,
+      "loss": 0.6943,
+      "step": 17009
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2818893315379864e-05,
+      "loss": 0.712,
+      "step": 17010
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2816323266884754e-05,
+      "loss": 0.713,
+      "step": 17011
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.281375324164448e-05,
+      "loss": 0.7983,
+      "step": 17012
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.28111832396864e-05,
+      "loss": 0.6585,
+      "step": 17013
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.280861326103789e-05,
+      "loss": 0.7002,
+      "step": 17014
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2806043305726326e-05,
+      "loss": 0.6962,
+      "step": 17015
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.280347337377906e-05,
+      "loss": 0.6554,
+      "step": 17016
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2800903465223473e-05,
+      "loss": 0.6442,
+      "step": 17017
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2798333580086935e-05,
+      "loss": 0.6337,
+      "step": 17018
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2795763718396796e-05,
+      "loss": 0.707,
+      "step": 17019
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.279319388018044e-05,
+      "loss": 0.7282,
+      "step": 17020
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2790624065465236e-05,
+      "loss": 0.6879,
+      "step": 17021
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2788054274278545e-05,
+      "loss": 0.671,
+      "step": 17022
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.278548450664773e-05,
+      "loss": 0.7552,
+      "step": 17023
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.278291476260015e-05,
+      "loss": 0.6638,
+      "step": 17024
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2780345042163192e-05,
+      "loss": 0.7251,
+      "step": 17025
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2777775345364212e-05,
+      "loss": 0.6545,
+      "step": 17026
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2775205672230564e-05,
+      "loss": 0.6643,
+      "step": 17027
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2772636022789636e-05,
+      "loss": 0.7764,
+      "step": 17028
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2770066397068782e-05,
+      "loss": 0.646,
+      "step": 17029
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.276749679509536e-05,
+      "loss": 0.7175,
+      "step": 17030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.276492721689675e-05,
+      "loss": 0.6709,
+      "step": 17031
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2762357662500304e-05,
+      "loss": 0.6628,
+      "step": 17032
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2759788131933387e-05,
+      "loss": 0.6717,
+      "step": 17033
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.275721862522337e-05,
+      "loss": 0.6681,
+      "step": 17034
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2754649142397617e-05,
+      "loss": 0.7694,
+      "step": 17035
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.275207968348348e-05,
+      "loss": 0.6627,
+      "step": 17036
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2749510248508334e-05,
+      "loss": 0.6574,
+      "step": 17037
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2746940837499537e-05,
+      "loss": 0.7325,
+      "step": 17038
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2744371450484447e-05,
+      "loss": 0.7298,
+      "step": 17039
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2741802087490438e-05,
+      "loss": 0.725,
+      "step": 17040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2739232748544867e-05,
+      "loss": 0.8031,
+      "step": 17041
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.27366634336751e-05,
+      "loss": 0.7341,
+      "step": 17042
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2734094142908485e-05,
+      "loss": 0.7322,
+      "step": 17043
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.273152487627239e-05,
+      "loss": 0.6618,
+      "step": 17044
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.272895563379418e-05,
+      "loss": 0.7708,
+      "step": 17045
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2726386415501216e-05,
+      "loss": 0.7171,
+      "step": 17046
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2723817221420847e-05,
+      "loss": 0.7098,
+      "step": 17047
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2721248051580454e-05,
+      "loss": 0.6744,
+      "step": 17048
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2718678906007382e-05,
+      "loss": 0.733,
+      "step": 17049
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.271610978472899e-05,
+      "loss": 0.6973,
+      "step": 17050
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.271354068777265e-05,
+      "loss": 0.636,
+      "step": 17051
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2710971615165712e-05,
+      "loss": 0.7593,
+      "step": 17052
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2708402566935526e-05,
+      "loss": 0.6652,
+      "step": 17053
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.270583354310947e-05,
+      "loss": 0.8223,
+      "step": 17054
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2703264543714896e-05,
+      "loss": 0.7272,
+      "step": 17055
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2700695568779155e-05,
+      "loss": 0.6479,
+      "step": 17056
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2698126618329616e-05,
+      "loss": 0.7671,
+      "step": 17057
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2695557692393628e-05,
+      "loss": 0.7896,
+      "step": 17058
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2692988790998548e-05,
+      "loss": 0.7386,
+      "step": 17059
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2690419914171744e-05,
+      "loss": 0.6592,
+      "step": 17060
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2687851061940564e-05,
+      "loss": 0.6798,
+      "step": 17061
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.268528223433237e-05,
+      "loss": 0.6855,
+      "step": 17062
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.268271343137451e-05,
+      "loss": 0.6466,
+      "step": 17063
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2680144653094343e-05,
+      "loss": 0.6981,
+      "step": 17064
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2677575899519228e-05,
+      "loss": 0.6773,
+      "step": 17065
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2675007170676524e-05,
+      "loss": 0.7467,
+      "step": 17066
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2672438466593575e-05,
+      "loss": 0.7292,
+      "step": 17067
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2669869787297748e-05,
+      "loss": 0.7336,
+      "step": 17068
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2667301132816394e-05,
+      "loss": 0.7175,
+      "step": 17069
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2664732503176857e-05,
+      "loss": 0.6856,
+      "step": 17070
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2662163898406512e-05,
+      "loss": 0.6886,
+      "step": 17071
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.26595953185327e-05,
+      "loss": 0.7563,
+      "step": 17072
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.265702676358277e-05,
+      "loss": 0.7079,
+      "step": 17073
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.265445823358409e-05,
+      "loss": 0.634,
+      "step": 17074
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2651889728564008e-05,
+      "loss": 0.693,
+      "step": 17075
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.264932124854987e-05,
+      "loss": 0.7305,
+      "step": 17076
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.264675279356903e-05,
+      "loss": 0.6585,
+      "step": 17077
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.264418436364885e-05,
+      "loss": 0.7228,
+      "step": 17078
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.264161595881668e-05,
+      "loss": 0.7251,
+      "step": 17079
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2639047579099858e-05,
+      "loss": 0.6997,
+      "step": 17080
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2636479224525753e-05,
+      "loss": 0.7696,
+      "step": 17081
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2633910895121706e-05,
+      "loss": 0.6819,
+      "step": 17082
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.263134259091508e-05,
+      "loss": 0.6685,
+      "step": 17083
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2628774311933206e-05,
+      "loss": 0.7426,
+      "step": 17084
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2626206058203452e-05,
+      "loss": 0.7607,
+      "step": 17085
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2623637829753163e-05,
+      "loss": 0.6583,
+      "step": 17086
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.262106962660968e-05,
+      "loss": 0.7133,
+      "step": 17087
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.261850144880037e-05,
+      "loss": 0.7102,
+      "step": 17088
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2615933296352574e-05,
+      "loss": 0.7523,
+      "step": 17089
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2613365169293636e-05,
+      "loss": 0.6684,
+      "step": 17090
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.261079706765091e-05,
+      "loss": 0.7195,
+      "step": 17091
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2608228991451748e-05,
+      "loss": 0.6897,
+      "step": 17092
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2605660940723495e-05,
+      "loss": 0.6636,
+      "step": 17093
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2603092915493493e-05,
+      "loss": 0.7565,
+      "step": 17094
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2600524915789102e-05,
+      "loss": 0.7308,
+      "step": 17095
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2597956941637667e-05,
+      "loss": 0.6716,
+      "step": 17096
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.259538899306652e-05,
+      "loss": 0.6678,
+      "step": 17097
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.259282107010303e-05,
+      "loss": 0.6943,
+      "step": 17098
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2590253172774536e-05,
+      "loss": 0.6523,
+      "step": 17099
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2587685301108374e-05,
+      "loss": 0.7148,
+      "step": 17100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.258511745513191e-05,
+      "loss": 0.6586,
+      "step": 17101
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2582549634872475e-05,
+      "loss": 0.6279,
+      "step": 17102
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.257998184035742e-05,
+      "loss": 0.7611,
+      "step": 17103
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2577414071614088e-05,
+      "loss": 0.6657,
+      "step": 17104
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2574846328669822e-05,
+      "loss": 0.7114,
+      "step": 17105
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.257227861155197e-05,
+      "loss": 0.6355,
+      "step": 17106
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2569710920287883e-05,
+      "loss": 0.6647,
+      "step": 17107
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.256714325490489e-05,
+      "loss": 0.663,
+      "step": 17108
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2564575615430354e-05,
+      "loss": 0.6396,
+      "step": 17109
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2562008001891605e-05,
+      "loss": 0.752,
+      "step": 17110
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2559440414315984e-05,
+      "loss": 0.6629,
+      "step": 17111
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.255687285273085e-05,
+      "loss": 0.679,
+      "step": 17112
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2554305317163534e-05,
+      "loss": 0.7451,
+      "step": 17113
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.255173780764138e-05,
+      "loss": 0.6501,
+      "step": 17114
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2549170324191736e-05,
+      "loss": 0.6512,
+      "step": 17115
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.254660286684194e-05,
+      "loss": 0.7788,
+      "step": 17116
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2544035435619323e-05,
+      "loss": 0.6794,
+      "step": 17117
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.254146803055125e-05,
+      "loss": 0.71,
+      "step": 17118
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2538900651665047e-05,
+      "loss": 0.665,
+      "step": 17119
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.253633329898805e-05,
+      "loss": 0.706,
+      "step": 17120
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.253376597254762e-05,
+      "loss": 0.7625,
+      "step": 17121
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2531198672371077e-05,
+      "loss": 0.6558,
+      "step": 17122
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.252863139848578e-05,
+      "loss": 0.7905,
+      "step": 17123
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.252606415091905e-05,
+      "loss": 0.6674,
+      "step": 17124
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2523496929698234e-05,
+      "loss": 0.7534,
+      "step": 17125
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2520929734850673e-05,
+      "loss": 0.6934,
+      "step": 17126
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.251836256640371e-05,
+      "loss": 0.7705,
+      "step": 17127
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2515795424384667e-05,
+      "loss": 0.6217,
+      "step": 17128
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2513228308820904e-05,
+      "loss": 0.7421,
+      "step": 17129
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.251066121973975e-05,
+      "loss": 0.7273,
+      "step": 17130
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2508094157168535e-05,
+      "loss": 0.6607,
+      "step": 17131
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2505527121134613e-05,
+      "loss": 0.6764,
+      "step": 17132
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.250296011166531e-05,
+      "loss": 0.7926,
+      "step": 17133
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.250039312878796e-05,
+      "loss": 0.7704,
+      "step": 17134
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2497826172529916e-05,
+      "loss": 0.7063,
+      "step": 17135
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2495259242918498e-05,
+      "loss": 0.7146,
+      "step": 17136
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2492692339981045e-05,
+      "loss": 0.6645,
+      "step": 17137
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2490125463744906e-05,
+      "loss": 0.7547,
+      "step": 17138
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.24875586142374e-05,
+      "loss": 0.6451,
+      "step": 17139
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.248499179148587e-05,
+      "loss": 0.7376,
+      "step": 17140
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.248242499551765e-05,
+      "loss": 0.6587,
+      "step": 17141
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.247985822636008e-05,
+      "loss": 0.688,
+      "step": 17142
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2477291484040496e-05,
+      "loss": 0.6782,
+      "step": 17143
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2474724768586216e-05,
+      "loss": 0.7177,
+      "step": 17144
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.247215808002458e-05,
+      "loss": 0.7097,
+      "step": 17145
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2469591418382933e-05,
+      "loss": 0.6829,
+      "step": 17146
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.24670247836886e-05,
+      "loss": 0.732,
+      "step": 17147
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.246445817596891e-05,
+      "loss": 0.7455,
+      "step": 17148
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.246189159525121e-05,
+      "loss": 0.6932,
+      "step": 17149
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2459325041562818e-05,
+      "loss": 0.6051,
+      "step": 17150
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.245675851493107e-05,
+      "loss": 0.6795,
+      "step": 17151
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2454192015383305e-05,
+      "loss": 0.6708,
+      "step": 17152
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2451625542946847e-05,
+      "loss": 0.6221,
+      "step": 17153
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2449059097649024e-05,
+      "loss": 0.7459,
+      "step": 17154
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.244649267951718e-05,
+      "loss": 0.6241,
+      "step": 17155
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.244392628857864e-05,
+      "loss": 0.6681,
+      "step": 17156
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2441359924860726e-05,
+      "loss": 0.6409,
+      "step": 17157
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2438793588390783e-05,
+      "loss": 0.7137,
+      "step": 17158
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.243622727919613e-05,
+      "loss": 0.6663,
+      "step": 17159
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.24336609973041e-05,
+      "loss": 0.6877,
+      "step": 17160
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2431094742742022e-05,
+      "loss": 0.717,
+      "step": 17161
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.242852851553723e-05,
+      "loss": 0.6454,
+      "step": 17162
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2425962315717048e-05,
+      "loss": 0.6369,
+      "step": 17163
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.24233961433088e-05,
+      "loss": 0.6672,
+      "step": 17164
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2420829998339818e-05,
+      "loss": 0.7149,
+      "step": 17165
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2418263880837433e-05,
+      "loss": 0.6889,
+      "step": 17166
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2415697790828972e-05,
+      "loss": 0.6529,
+      "step": 17167
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2413131728341756e-05,
+      "loss": 0.6994,
+      "step": 17168
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.241056569340312e-05,
+      "loss": 0.6904,
+      "step": 17169
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2407999686040383e-05,
+      "loss": 0.7057,
+      "step": 17170
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2405433706280877e-05,
+      "loss": 0.7467,
+      "step": 17171
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.240286775415193e-05,
+      "loss": 0.7832,
+      "step": 17172
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2400301829680863e-05,
+      "loss": 0.673,
+      "step": 17173
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2397735932895e-05,
+      "loss": 0.7513,
+      "step": 17174
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2395170063821676e-05,
+      "loss": 0.7308,
+      "step": 17175
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2392604222488207e-05,
+      "loss": 0.6965,
+      "step": 17176
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2390038408921914e-05,
+      "loss": 0.7048,
+      "step": 17177
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2387472623150134e-05,
+      "loss": 0.7421,
+      "step": 17178
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2384906865200184e-05,
+      "loss": 0.7017,
+      "step": 17179
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2382341135099383e-05,
+      "loss": 0.6808,
+      "step": 17180
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2379775432875067e-05,
+      "loss": 0.7803,
+      "step": 17181
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2377209758554553e-05,
+      "loss": 0.7195,
+      "step": 17182
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2374644112165165e-05,
+      "loss": 0.7248,
+      "step": 17183
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2372078493734216e-05,
+      "loss": 0.6458,
+      "step": 17184
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2369512903289033e-05,
+      "loss": 0.6785,
+      "step": 17185
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2366947340856944e-05,
+      "loss": 0.7347,
+      "step": 17186
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.236438180646527e-05,
+      "loss": 0.7129,
+      "step": 17187
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.236181630014132e-05,
+      "loss": 0.7273,
+      "step": 17188
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2359250821912438e-05,
+      "loss": 0.7169,
+      "step": 17189
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2356685371805923e-05,
+      "loss": 0.7732,
+      "step": 17190
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.23541199498491e-05,
+      "loss": 0.7519,
+      "step": 17191
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2351554556069305e-05,
+      "loss": 0.6223,
+      "step": 17192
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.234898919049384e-05,
+      "loss": 0.6637,
+      "step": 17193
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.234642385315003e-05,
+      "loss": 0.6619,
+      "step": 17194
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2343858544065193e-05,
+      "loss": 0.721,
+      "step": 17195
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2341293263266656e-05,
+      "loss": 0.6871,
+      "step": 17196
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.233872801078173e-05,
+      "loss": 0.7669,
+      "step": 17197
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.233616278663773e-05,
+      "loss": 0.7669,
+      "step": 17198
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2333597590861982e-05,
+      "loss": 0.7349,
+      "step": 17199
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2331032423481804e-05,
+      "loss": 0.6799,
+      "step": 17200
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2328467284524504e-05,
+      "loss": 0.6158,
+      "step": 17201
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.232590217401741e-05,
+      "loss": 0.7324,
+      "step": 17202
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2323337091987837e-05,
+      "loss": 0.6549,
+      "step": 17203
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.23207720384631e-05,
+      "loss": 0.7221,
+      "step": 17204
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.231820701347051e-05,
+      "loss": 0.7758,
+      "step": 17205
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2315642017037385e-05,
+      "loss": 0.6168,
+      "step": 17206
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2313077049191047e-05,
+      "loss": 0.7329,
+      "step": 17207
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2310512109958807e-05,
+      "loss": 0.6579,
+      "step": 17208
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2307947199367973e-05,
+      "loss": 0.6523,
+      "step": 17209
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.230538231744587e-05,
+      "loss": 0.8002,
+      "step": 17210
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2302817464219818e-05,
+      "loss": 0.6593,
+      "step": 17211
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.230025263971711e-05,
+      "loss": 0.733,
+      "step": 17212
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2297687843965078e-05,
+      "loss": 0.6985,
+      "step": 17213
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.229512307699103e-05,
+      "loss": 0.7141,
+      "step": 17214
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.229255833882228e-05,
+      "loss": 0.6705,
+      "step": 17215
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2289993629486137e-05,
+      "loss": 0.6672,
+      "step": 17216
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.228742894900992e-05,
+      "loss": 0.6769,
+      "step": 17217
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.228486429742093e-05,
+      "loss": 0.727,
+      "step": 17218
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.22822996747465e-05,
+      "loss": 0.6331,
+      "step": 17219
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2279735081013923e-05,
+      "loss": 0.7035,
+      "step": 17220
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.227717051625051e-05,
+      "loss": 0.7095,
+      "step": 17221
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2274605980483583e-05,
+      "loss": 0.6867,
+      "step": 17222
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.227204147374045e-05,
+      "loss": 0.6364,
+      "step": 17223
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2269476996048423e-05,
+      "loss": 0.6481,
+      "step": 17224
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2266912547434803e-05,
+      "loss": 0.7083,
+      "step": 17225
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.22643481279269e-05,
+      "loss": 0.7036,
+      "step": 17226
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2261783737552032e-05,
+      "loss": 0.7236,
+      "step": 17227
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2259219376337508e-05,
+      "loss": 0.6816,
+      "step": 17228
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2256655044310625e-05,
+      "loss": 0.6774,
+      "step": 17229
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.225409074149871e-05,
+      "loss": 0.7293,
+      "step": 17230
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2251526467929057e-05,
+      "loss": 0.7143,
+      "step": 17231
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2248962223628976e-05,
+      "loss": 0.6237,
+      "step": 17232
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2246398008625783e-05,
+      "loss": 0.6458,
+      "step": 17233
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.224383382294678e-05,
+      "loss": 0.6479,
+      "step": 17234
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2241269666619265e-05,
+      "loss": 0.7165,
+      "step": 17235
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2238705539670557e-05,
+      "loss": 0.6655,
+      "step": 17236
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.223614144212796e-05,
+      "loss": 0.6916,
+      "step": 17237
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2233577374018776e-05,
+      "loss": 0.6194,
+      "step": 17238
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.223101333537032e-05,
+      "loss": 0.7306,
+      "step": 17239
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2228449326209884e-05,
+      "loss": 0.625,
+      "step": 17240
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.222588534656478e-05,
+      "loss": 0.6845,
+      "step": 17241
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2223321396462318e-05,
+      "loss": 0.7205,
+      "step": 17242
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2220757475929796e-05,
+      "loss": 0.7144,
+      "step": 17243
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2218193584994527e-05,
+      "loss": 0.6405,
+      "step": 17244
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.22156297236838e-05,
+      "loss": 0.6793,
+      "step": 17245
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2213065892024922e-05,
+      "loss": 0.6726,
+      "step": 17246
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2210502090045202e-05,
+      "loss": 0.5855,
+      "step": 17247
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2207938317771944e-05,
+      "loss": 0.7253,
+      "step": 17248
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2205374575232443e-05,
+      "loss": 0.7173,
+      "step": 17249
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.220281086245401e-05,
+      "loss": 0.7462,
+      "step": 17250
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2200247179463947e-05,
+      "loss": 0.6872,
+      "step": 17251
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2197683526289542e-05,
+      "loss": 0.6313,
+      "step": 17252
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.219511990295811e-05,
+      "loss": 0.6786,
+      "step": 17253
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2192556309496953e-05,
+      "loss": 0.7126,
+      "step": 17254
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2189992745933362e-05,
+      "loss": 0.5937,
+      "step": 17255
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2187429212294642e-05,
+      "loss": 0.7021,
+      "step": 17256
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2184865708608095e-05,
+      "loss": 0.7462,
+      "step": 17257
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2182302234901018e-05,
+      "loss": 0.6692,
+      "step": 17258
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2179738791200714e-05,
+      "loss": 0.7476,
+      "step": 17259
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.217717537753448e-05,
+      "loss": 0.7282,
+      "step": 17260
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.217461199392961e-05,
+      "loss": 0.7426,
+      "step": 17261
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.217204864041341e-05,
+      "loss": 0.6814,
+      "step": 17262
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2169485317013174e-05,
+      "loss": 0.7063,
+      "step": 17263
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2166922023756208e-05,
+      "loss": 0.6211,
+      "step": 17264
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2164358760669803e-05,
+      "loss": 0.6696,
+      "step": 17265
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2161795527781244e-05,
+      "loss": 0.6516,
+      "step": 17266
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2159232325117842e-05,
+      "loss": 0.7112,
+      "step": 17267
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2156669152706897e-05,
+      "loss": 0.7168,
+      "step": 17268
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.215410601057569e-05,
+      "loss": 0.6589,
+      "step": 17269
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2151542898751533e-05,
+      "loss": 0.8322,
+      "step": 17270
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2148979817261718e-05,
+      "loss": 0.6854,
+      "step": 17271
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2146416766133527e-05,
+      "loss": 0.6878,
+      "step": 17272
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.214385374539427e-05,
+      "loss": 0.7577,
+      "step": 17273
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.214129075507124e-05,
+      "loss": 0.665,
+      "step": 17274
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.213872779519172e-05,
+      "loss": 0.7762,
+      "step": 17275
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2136164865783023e-05,
+      "loss": 0.6938,
+      "step": 17276
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2133601966872426e-05,
+      "loss": 0.7047,
+      "step": 17277
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2131039098487223e-05,
+      "loss": 0.7605,
+      "step": 17278
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.212847626065472e-05,
+      "loss": 0.7214,
+      "step": 17279
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2125913453402195e-05,
+      "loss": 0.8167,
+      "step": 17280
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2123350676756945e-05,
+      "loss": 0.6508,
+      "step": 17281
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.212078793074627e-05,
+      "loss": 0.699,
+      "step": 17282
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2118225215397456e-05,
+      "loss": 0.7309,
+      "step": 17283
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.21156625307378e-05,
+      "loss": 0.6582,
+      "step": 17284
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.211309987679458e-05,
+      "loss": 0.7091,
+      "step": 17285
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2110537253595087e-05,
+      "loss": 0.6987,
+      "step": 17286
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2107974661166625e-05,
+      "loss": 0.8291,
+      "step": 17287
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2105412099536478e-05,
+      "loss": 0.6761,
+      "step": 17288
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2102849568731926e-05,
+      "loss": 0.6817,
+      "step": 17289
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2100287068780275e-05,
+      "loss": 0.6693,
+      "step": 17290
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.209772459970881e-05,
+      "loss": 0.6951,
+      "step": 17291
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2095162161544803e-05,
+      "loss": 0.6326,
+      "step": 17292
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2092599754315565e-05,
+      "loss": 0.6729,
+      "step": 17293
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2090037378048373e-05,
+      "loss": 0.6817,
+      "step": 17294
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2087475032770512e-05,
+      "loss": 0.6343,
+      "step": 17295
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.208491271850928e-05,
+      "loss": 0.6835,
+      "step": 17296
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2082350435291957e-05,
+      "loss": 0.6807,
+      "step": 17297
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2079788183145825e-05,
+      "loss": 0.6469,
+      "step": 17298
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.207722596209818e-05,
+      "loss": 0.6289,
+      "step": 17299
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2074663772176305e-05,
+      "loss": 0.7255,
+      "step": 17300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2072101613407485e-05,
+      "loss": 0.6893,
+      "step": 17301
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2069539485819e-05,
+      "loss": 0.6649,
+      "step": 17302
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2066977389438144e-05,
+      "loss": 0.6436,
+      "step": 17303
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.20644153242922e-05,
+      "loss": 0.7235,
+      "step": 17304
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.206185329040846e-05,
+      "loss": 0.6349,
+      "step": 17305
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.205929128781418e-05,
+      "loss": 0.6717,
+      "step": 17306
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2056729316536674e-05,
+      "loss": 0.6606,
+      "step": 17307
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2054167376603208e-05,
+      "loss": 0.6556,
+      "step": 17308
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.205160546804107e-05,
+      "loss": 0.6725,
+      "step": 17309
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2049043590877546e-05,
+      "loss": 0.6318,
+      "step": 17310
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.204648174513992e-05,
+      "loss": 0.6438,
+      "step": 17311
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2043919930855463e-05,
+      "loss": 0.7122,
+      "step": 17312
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.204135814805147e-05,
+      "loss": 0.6579,
+      "step": 17313
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2038796396755215e-05,
+      "loss": 0.7075,
+      "step": 17314
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2036234676993984e-05,
+      "loss": 0.6602,
+      "step": 17315
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2033672988795047e-05,
+      "loss": 0.6528,
+      "step": 17316
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2031111332185695e-05,
+      "loss": 0.6554,
+      "step": 17317
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2028549707193207e-05,
+      "loss": 0.6672,
+      "step": 17318
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2025988113844853e-05,
+      "loss": 0.6505,
+      "step": 17319
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2023426552167928e-05,
+      "loss": 0.6678,
+      "step": 17320
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.20208650221897e-05,
+      "loss": 0.7077,
+      "step": 17321
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.201830352393745e-05,
+      "loss": 0.6477,
+      "step": 17322
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.201574205743846e-05,
+      "loss": 0.6846,
+      "step": 17323
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2013180622720005e-05,
+      "loss": 0.6736,
+      "step": 17324
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.201061921980937e-05,
+      "loss": 0.7205,
+      "step": 17325
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2008057848733812e-05,
+      "loss": 0.7376,
+      "step": 17326
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.200549650952063e-05,
+      "loss": 0.6443,
+      "step": 17327
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.200293520219709e-05,
+      "loss": 0.6656,
+      "step": 17328
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.200037392679047e-05,
+      "loss": 0.7729,
+      "step": 17329
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1997812683328042e-05,
+      "loss": 0.6191,
+      "step": 17330
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1995251471837093e-05,
+      "loss": 0.727,
+      "step": 17331
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1992690292344893e-05,
+      "loss": 0.682,
+      "step": 17332
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1990129144878708e-05,
+      "loss": 0.691,
+      "step": 17333
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1987568029465825e-05,
+      "loss": 0.7275,
+      "step": 17334
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1985006946133512e-05,
+      "loss": 0.7414,
+      "step": 17335
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1982445894909043e-05,
+      "loss": 0.7477,
+      "step": 17336
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1979884875819696e-05,
+      "loss": 0.7379,
+      "step": 17337
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1977323888892743e-05,
+      "loss": 0.7187,
+      "step": 17338
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.197476293415545e-05,
+      "loss": 0.7071,
+      "step": 17339
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1972202011635097e-05,
+      "loss": 0.7009,
+      "step": 17340
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1969641121358957e-05,
+      "loss": 0.6355,
+      "step": 17341
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1967080263354294e-05,
+      "loss": 0.7982,
+      "step": 17342
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1964519437648393e-05,
+      "loss": 0.8083,
+      "step": 17343
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1961958644268516e-05,
+      "loss": 0.6469,
+      "step": 17344
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1959397883241936e-05,
+      "loss": 0.8425,
+      "step": 17345
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.195683715459592e-05,
+      "loss": 0.6749,
+      "step": 17346
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1954276458357736e-05,
+      "loss": 0.6384,
+      "step": 17347
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1951715794554665e-05,
+      "loss": 0.7021,
+      "step": 17348
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.194915516321397e-05,
+      "loss": 0.6869,
+      "step": 17349
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.194659456436291e-05,
+      "loss": 0.7724,
+      "step": 17350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1944033998028773e-05,
+      "loss": 0.7878,
+      "step": 17351
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1941473464238822e-05,
+      "loss": 0.6237,
+      "step": 17352
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.193891296302031e-05,
+      "loss": 0.7399,
+      "step": 17353
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1936352494400525e-05,
+      "loss": 0.7687,
+      "step": 17354
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1933792058406728e-05,
+      "loss": 0.7572,
+      "step": 17355
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1931231655066176e-05,
+      "loss": 0.6758,
+      "step": 17356
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1928671284406148e-05,
+      "loss": 0.669,
+      "step": 17357
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1926110946453908e-05,
+      "loss": 0.6812,
+      "step": 17358
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1923550641236716e-05,
+      "loss": 0.7839,
+      "step": 17359
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1920990368781847e-05,
+      "loss": 0.682,
+      "step": 17360
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1918430129116558e-05,
+      "loss": 0.6699,
+      "step": 17361
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1915869922268114e-05,
+      "loss": 0.7419,
+      "step": 17362
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.191330974826379e-05,
+      "loss": 0.6692,
+      "step": 17363
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.191074960713084e-05,
+      "loss": 0.6733,
+      "step": 17364
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.190818949889654e-05,
+      "loss": 0.7021,
+      "step": 17365
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1905629423588138e-05,
+      "loss": 0.625,
+      "step": 17366
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1903069381232896e-05,
+      "loss": 0.6676,
+      "step": 17367
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1900509371858092e-05,
+      "loss": 0.6672,
+      "step": 17368
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1897949395490982e-05,
+      "loss": 0.6517,
+      "step": 17369
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1895389452158822e-05,
+      "loss": 0.6979,
+      "step": 17370
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1892829541888885e-05,
+      "loss": 0.6925,
+      "step": 17371
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1890269664708426e-05,
+      "loss": 0.6772,
+      "step": 17372
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.18877098206447e-05,
+      "loss": 0.618,
+      "step": 17373
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1885150009724987e-05,
+      "loss": 0.7712,
+      "step": 17374
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.188259023197653e-05,
+      "loss": 0.7273,
+      "step": 17375
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1880030487426593e-05,
+      "loss": 0.6222,
+      "step": 17376
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.187747077610244e-05,
+      "loss": 0.7095,
+      "step": 17377
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1874911098031324e-05,
+      "loss": 0.754,
+      "step": 17378
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1872351453240507e-05,
+      "loss": 0.7066,
+      "step": 17379
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.186979184175725e-05,
+      "loss": 0.6394,
+      "step": 17380
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.186723226360881e-05,
+      "loss": 0.6888,
+      "step": 17381
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1864672718822443e-05,
+      "loss": 0.6875,
+      "step": 17382
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.186211320742541e-05,
+      "loss": 0.7515,
+      "step": 17383
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.185955372944497e-05,
+      "loss": 0.645,
+      "step": 17384
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1856994284908376e-05,
+      "loss": 0.6992,
+      "step": 17385
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1854434873842883e-05,
+      "loss": 0.6695,
+      "step": 17386
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.185187549627574e-05,
+      "loss": 0.6684,
+      "step": 17387
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.184931615223422e-05,
+      "loss": 0.7012,
+      "step": 17388
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1846756841745573e-05,
+      "loss": 0.6082,
+      "step": 17389
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1844197564837045e-05,
+      "loss": 0.7424,
+      "step": 17390
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1841638321535897e-05,
+      "loss": 0.7496,
+      "step": 17391
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1839079111869388e-05,
+      "loss": 0.7147,
+      "step": 17392
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1836519935864757e-05,
+      "loss": 0.6192,
+      "step": 17393
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.183396079354928e-05,
+      "loss": 0.7248,
+      "step": 17394
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1831401684950196e-05,
+      "loss": 0.6799,
+      "step": 17395
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.182884261009475e-05,
+      "loss": 0.6978,
+      "step": 17396
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1826283569010217e-05,
+      "loss": 0.6931,
+      "step": 17397
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1823724561723836e-05,
+      "loss": 0.6912,
+      "step": 17398
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1821165588262855e-05,
+      "loss": 0.6904,
+      "step": 17399
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1818606648654532e-05,
+      "loss": 0.7111,
+      "step": 17400
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1816047742926118e-05,
+      "loss": 0.6761,
+      "step": 17401
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.181348887110486e-05,
+      "loss": 0.6396,
+      "step": 17402
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1810930033218013e-05,
+      "loss": 0.6672,
+      "step": 17403
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.180837122929283e-05,
+      "loss": 0.6998,
+      "step": 17404
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1805812459356545e-05,
+      "loss": 0.682,
+      "step": 17405
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1803253723436437e-05,
+      "loss": 0.6762,
+      "step": 17406
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.180069502155972e-05,
+      "loss": 0.692,
+      "step": 17407
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1798136353753663e-05,
+      "loss": 0.674,
+      "step": 17408
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.179557772004551e-05,
+      "loss": 0.7773,
+      "step": 17409
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1793019120462503e-05,
+      "loss": 0.6828,
+      "step": 17410
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.17904605550319e-05,
+      "loss": 0.7632,
+      "step": 17411
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1787902023780945e-05,
+      "loss": 0.763,
+      "step": 17412
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1785343526736878e-05,
+      "loss": 0.7438,
+      "step": 17413
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1782785063926955e-05,
+      "loss": 0.7875,
+      "step": 17414
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.178022663537842e-05,
+      "loss": 0.7435,
+      "step": 17415
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.177766824111851e-05,
+      "loss": 0.6784,
+      "step": 17416
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1775109881174483e-05,
+      "loss": 0.7795,
+      "step": 17417
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1772551555573574e-05,
+      "loss": 0.6902,
+      "step": 17418
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.176999326434303e-05,
+      "loss": 0.7155,
+      "step": 17419
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1767435007510103e-05,
+      "loss": 0.714,
+      "step": 17420
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1764876785102028e-05,
+      "loss": 0.7174,
+      "step": 17421
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1762318597146053e-05,
+      "loss": 0.6579,
+      "step": 17422
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1759760443669412e-05,
+      "loss": 0.691,
+      "step": 17423
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1757202324699365e-05,
+      "loss": 0.6847,
+      "step": 17424
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.175464424026314e-05,
+      "loss": 0.6617,
+      "step": 17425
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1752086190387988e-05,
+      "loss": 0.7379,
+      "step": 17426
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.174952817510114e-05,
+      "loss": 0.7378,
+      "step": 17427
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1746970194429845e-05,
+      "loss": 0.5961,
+      "step": 17428
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1744412248401344e-05,
+      "loss": 0.778,
+      "step": 17429
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.174185433704287e-05,
+      "loss": 0.6606,
+      "step": 17430
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1739296460381675e-05,
+      "loss": 0.6593,
+      "step": 17431
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1736738618444995e-05,
+      "loss": 0.7186,
+      "step": 17432
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1734180811260064e-05,
+      "loss": 0.6724,
+      "step": 17433
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.173162303885412e-05,
+      "loss": 0.7389,
+      "step": 17434
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1729065301254416e-05,
+      "loss": 0.6711,
+      "step": 17435
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1726507598488173e-05,
+      "loss": 0.7145,
+      "step": 17436
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.172394993058264e-05,
+      "loss": 0.7144,
+      "step": 17437
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.172139229756505e-05,
+      "loss": 0.7111,
+      "step": 17438
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1718834699462643e-05,
+      "loss": 0.7072,
+      "step": 17439
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1716277136302652e-05,
+      "loss": 0.6868,
+      "step": 17440
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.171371960811232e-05,
+      "loss": 0.6925,
+      "step": 17441
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1711162114918877e-05,
+      "loss": 0.6738,
+      "step": 17442
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1708604656749555e-05,
+      "loss": 0.7338,
+      "step": 17443
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1706047233631603e-05,
+      "loss": 0.7274,
+      "step": 17444
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.170348984559225e-05,
+      "loss": 0.6937,
+      "step": 17445
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1700932492658728e-05,
+      "loss": 0.6571,
+      "step": 17446
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.169837517485827e-05,
+      "loss": 0.6892,
+      "step": 17447
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1695817892218106e-05,
+      "loss": 0.71,
+      "step": 17448
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1693260644765483e-05,
+      "loss": 0.6852,
+      "step": 17449
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1690703432527627e-05,
+      "loss": 0.6582,
+      "step": 17450
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.168814625553176e-05,
+      "loss": 0.6706,
+      "step": 17451
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1685589113805132e-05,
+      "loss": 0.6358,
+      "step": 17452
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.168303200737497e-05,
+      "loss": 0.6609,
+      "step": 17453
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1680474936268497e-05,
+      "loss": 0.6465,
+      "step": 17454
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1677917900512955e-05,
+      "loss": 0.6305,
+      "step": 17455
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.167536090013557e-05,
+      "loss": 0.6812,
+      "step": 17456
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1672803935163573e-05,
+      "loss": 0.6422,
+      "step": 17457
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1670247005624194e-05,
+      "loss": 0.6857,
+      "step": 17458
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1667690111544663e-05,
+      "loss": 0.613,
+      "step": 17459
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1665133252952204e-05,
+      "loss": 0.6732,
+      "step": 17460
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1662576429874058e-05,
+      "loss": 0.7567,
+      "step": 17461
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.166001964233745e-05,
+      "loss": 0.704,
+      "step": 17462
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1657462890369596e-05,
+      "loss": 0.7072,
+      "step": 17463
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1654906173997736e-05,
+      "loss": 0.7145,
+      "step": 17464
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1652349493249098e-05,
+      "loss": 0.6601,
+      "step": 17465
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1649792848150908e-05,
+      "loss": 0.66,
+      "step": 17466
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.164723623873039e-05,
+      "loss": 0.6337,
+      "step": 17467
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1644679665014758e-05,
+      "loss": 0.756,
+      "step": 17468
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.164212312703126e-05,
+      "loss": 0.7205,
+      "step": 17469
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.163956662480711e-05,
+      "loss": 0.6841,
+      "step": 17470
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1637010158369534e-05,
+      "loss": 0.7039,
+      "step": 17471
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.163445372774576e-05,
+      "loss": 0.7019,
+      "step": 17472
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.163189733296301e-05,
+      "loss": 0.767,
+      "step": 17473
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1629340974048505e-05,
+      "loss": 0.7045,
+      "step": 17474
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.162678465102948e-05,
+      "loss": 0.6616,
+      "step": 17475
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1624228363933146e-05,
+      "loss": 0.6391,
+      "step": 17476
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.162167211278673e-05,
+      "loss": 0.6672,
+      "step": 17477
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1619115897617452e-05,
+      "loss": 0.6621,
+      "step": 17478
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1616559718452546e-05,
+      "loss": 0.6616,
+      "step": 17479
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1614003575319213e-05,
+      "loss": 0.6545,
+      "step": 17480
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.16114474682447e-05,
+      "loss": 0.6829,
+      "step": 17481
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1608891397256204e-05,
+      "loss": 0.7157,
+      "step": 17482
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1606335362380955e-05,
+      "loss": 0.6844,
+      "step": 17483
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1603779363646183e-05,
+      "loss": 0.6449,
+      "step": 17484
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1601223401079095e-05,
+      "loss": 0.6785,
+      "step": 17485
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.159866747470692e-05,
+      "loss": 0.6496,
+      "step": 17486
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1596111584556866e-05,
+      "loss": 0.7973,
+      "step": 17487
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.159355573065615e-05,
+      "loss": 0.7266,
+      "step": 17488
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1590999913032007e-05,
+      "loss": 0.7029,
+      "step": 17489
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.158844413171164e-05,
+      "loss": 0.6895,
+      "step": 17490
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.158588838672227e-05,
+      "loss": 0.7139,
+      "step": 17491
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.158333267809112e-05,
+      "loss": 0.7438,
+      "step": 17492
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1580777005845404e-05,
+      "loss": 0.7351,
+      "step": 17493
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.157822137001233e-05,
+      "loss": 0.7555,
+      "step": 17494
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.157566577061913e-05,
+      "loss": 0.6446,
+      "step": 17495
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1573110207693006e-05,
+      "loss": 0.6637,
+      "step": 17496
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1570554681261173e-05,
+      "loss": 0.6549,
+      "step": 17497
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1567999191350856e-05,
+      "loss": 0.7876,
+      "step": 17498
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1565443737989266e-05,
+      "loss": 0.6848,
+      "step": 17499
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.156288832120361e-05,
+      "loss": 0.7006,
+      "step": 17500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.156033294102111e-05,
+      "loss": 0.7305,
+      "step": 17501
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.155777759746898e-05,
+      "loss": 0.6509,
+      "step": 17502
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.155522229057442e-05,
+      "loss": 0.7033,
+      "step": 17503
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1552667020364655e-05,
+      "loss": 0.6931,
+      "step": 17504
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1550111786866895e-05,
+      "loss": 0.7604,
+      "step": 17505
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1547556590108355e-05,
+      "loss": 0.6624,
+      "step": 17506
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1545001430116237e-05,
+      "loss": 0.731,
+      "step": 17507
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.154244630691775e-05,
+      "loss": 0.7087,
+      "step": 17508
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1539891220540114e-05,
+      "loss": 0.7227,
+      "step": 17509
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.153733617101054e-05,
+      "loss": 0.699,
+      "step": 17510
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1534781158356225e-05,
+      "loss": 0.5912,
+      "step": 17511
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1532226182604394e-05,
+      "loss": 0.7114,
+      "step": 17512
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.152967124378225e-05,
+      "loss": 0.6532,
+      "step": 17513
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1527116341916992e-05,
+      "loss": 0.6405,
+      "step": 17514
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1524561477035844e-05,
+      "loss": 0.6313,
+      "step": 17515
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1522006649166005e-05,
+      "loss": 0.6316,
+      "step": 17516
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.151945185833468e-05,
+      "loss": 0.6204,
+      "step": 17517
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1516897104569083e-05,
+      "loss": 0.6706,
+      "step": 17518
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.151434238789642e-05,
+      "loss": 0.6496,
+      "step": 17519
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1511787708343883e-05,
+      "loss": 0.8258,
+      "step": 17520
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.15092330659387e-05,
+      "loss": 0.6694,
+      "step": 17521
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1506678460708062e-05,
+      "loss": 0.6316,
+      "step": 17522
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1504123892679176e-05,
+      "loss": 0.6933,
+      "step": 17523
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1501569361879254e-05,
+      "loss": 0.6578,
+      "step": 17524
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.149901486833549e-05,
+      "loss": 0.7105,
+      "step": 17525
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1496460412075095e-05,
+      "loss": 0.7419,
+      "step": 17526
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1493905993125275e-05,
+      "loss": 0.7447,
+      "step": 17527
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1491351611513215e-05,
+      "loss": 0.6668,
+      "step": 17528
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1488797267266137e-05,
+      "loss": 0.6246,
+      "step": 17529
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1486242960411234e-05,
+      "loss": 0.7111,
+      "step": 17530
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1483688690975705e-05,
+      "loss": 0.6483,
+      "step": 17531
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1481134458986767e-05,
+      "loss": 0.6772,
+      "step": 17532
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1478580264471608e-05,
+      "loss": 0.6626,
+      "step": 17533
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1476026107457426e-05,
+      "loss": 0.6928,
+      "step": 17534
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.147347198797143e-05,
+      "loss": 0.6895,
+      "step": 17535
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1470917906040818e-05,
+      "loss": 0.65,
+      "step": 17536
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1468363861692782e-05,
+      "loss": 0.699,
+      "step": 17537
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1465809854954533e-05,
+      "loss": 0.7432,
+      "step": 17538
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1463255885853262e-05,
+      "loss": 0.7569,
+      "step": 17539
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1460701954416168e-05,
+      "loss": 0.7022,
+      "step": 17540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1458148060670444e-05,
+      "loss": 0.7001,
+      "step": 17541
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.14555942046433e-05,
+      "loss": 0.671,
+      "step": 17542
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1453040386361924e-05,
+      "loss": 0.7195,
+      "step": 17543
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.145048660585351e-05,
+      "loss": 0.7997,
+      "step": 17544
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1447932863145265e-05,
+      "loss": 0.6845,
+      "step": 17545
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1445379158264378e-05,
+      "loss": 0.7183,
+      "step": 17546
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.144282549123805e-05,
+      "loss": 0.6944,
+      "step": 17547
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.144027186209346e-05,
+      "loss": 0.7346,
+      "step": 17548
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1437718270857822e-05,
+      "loss": 0.7372,
+      "step": 17549
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.143516471755832e-05,
+      "loss": 0.6981,
+      "step": 17550
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1432611202222146e-05,
+      "loss": 0.6576,
+      "step": 17551
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1430057724876494e-05,
+      "loss": 0.6998,
+      "step": 17552
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1427504285548564e-05,
+      "loss": 0.7056,
+      "step": 17553
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1424950884265547e-05,
+      "loss": 0.7077,
+      "step": 17554
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1422397521054623e-05,
+      "loss": 0.7581,
+      "step": 17555
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1419844195943005e-05,
+      "loss": 0.686,
+      "step": 17556
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1417290908957868e-05,
+      "loss": 0.709,
+      "step": 17557
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.14147376601264e-05,
+      "loss": 0.6532,
+      "step": 17558
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1412184449475812e-05,
+      "loss": 0.6965,
+      "step": 17559
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1409631277033276e-05,
+      "loss": 0.651,
+      "step": 17560
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1407078142825984e-05,
+      "loss": 0.6558,
+      "step": 17561
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1404525046881132e-05,
+      "loss": 0.6557,
+      "step": 17562
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1401971989225906e-05,
+      "loss": 0.7472,
+      "step": 17563
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.139941896988749e-05,
+      "loss": 0.776,
+      "step": 17564
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.139686598889308e-05,
+      "loss": 0.6895,
+      "step": 17565
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.139431304626986e-05,
+      "loss": 0.6813,
+      "step": 17566
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1391760142045024e-05,
+      "loss": 0.6783,
+      "step": 17567
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1389207276245746e-05,
+      "loss": 0.7197,
+      "step": 17568
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1386654448899212e-05,
+      "loss": 0.6757,
+      "step": 17569
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.138410166003262e-05,
+      "loss": 0.7151,
+      "step": 17570
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1381548909673148e-05,
+      "loss": 0.7534,
+      "step": 17571
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.137899619784798e-05,
+      "loss": 0.665,
+      "step": 17572
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.137644352458431e-05,
+      "loss": 0.6643,
+      "step": 17573
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1373890889909316e-05,
+      "loss": 0.6713,
+      "step": 17574
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1371338293850177e-05,
+      "loss": 0.6912,
+      "step": 17575
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1368785736434087e-05,
+      "loss": 0.6472,
+      "step": 17576
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1366233217688225e-05,
+      "loss": 0.6859,
+      "step": 17577
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.136368073763977e-05,
+      "loss": 0.729,
+      "step": 17578
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1361128296315907e-05,
+      "loss": 0.6611,
+      "step": 17579
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1358575893743823e-05,
+      "loss": 0.7419,
+      "step": 17580
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.135602352995069e-05,
+      "loss": 0.699,
+      "step": 17581
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.13534712049637e-05,
+      "loss": 0.6615,
+      "step": 17582
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1350918918810023e-05,
+      "loss": 0.7616,
+      "step": 17583
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1348366671516842e-05,
+      "loss": 0.7745,
+      "step": 17584
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1345814463111344e-05,
+      "loss": 0.6537,
+      "step": 17585
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1343262293620704e-05,
+      "loss": 0.707,
+      "step": 17586
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1340710163072102e-05,
+      "loss": 0.7301,
+      "step": 17587
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1338158071492715e-05,
+      "loss": 0.6316,
+      "step": 17588
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1335606018909715e-05,
+      "loss": 0.7142,
+      "step": 17589
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.133305400535029e-05,
+      "loss": 0.6366,
+      "step": 17590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1330502030841615e-05,
+      "loss": 0.6172,
+      "step": 17591
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1327950095410858e-05,
+      "loss": 0.7023,
+      "step": 17592
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.132539819908521e-05,
+      "loss": 0.6811,
+      "step": 17593
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.132284634189184e-05,
+      "loss": 0.7708,
+      "step": 17594
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.132029452385792e-05,
+      "loss": 0.6929,
+      "step": 17595
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1317742745010633e-05,
+      "loss": 0.6251,
+      "step": 17596
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.131519100537715e-05,
+      "loss": 0.6743,
+      "step": 17597
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.131263930498464e-05,
+      "loss": 0.6557,
+      "step": 17598
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.131008764386029e-05,
+      "loss": 0.7264,
+      "step": 17599
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1307536022031267e-05,
+      "loss": 0.6795,
+      "step": 17600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1304984439524735e-05,
+      "loss": 0.6306,
+      "step": 17601
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1302432896367882e-05,
+      "loss": 0.6816,
+      "step": 17602
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1299881392587876e-05,
+      "loss": 0.7047,
+      "step": 17603
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.129732992821188e-05,
+      "loss": 0.6694,
+      "step": 17604
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1294778503267077e-05,
+      "loss": 0.676,
+      "step": 17605
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1292227117780632e-05,
+      "loss": 0.6598,
+      "step": 17606
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1289675771779727e-05,
+      "loss": 0.6488,
+      "step": 17607
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1287124465291512e-05,
+      "loss": 0.633,
+      "step": 17608
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1284573198343165e-05,
+      "loss": 0.801,
+      "step": 17609
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.128202197096186e-05,
+      "loss": 0.6333,
+      "step": 17610
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.127947078317477e-05,
+      "loss": 0.7685,
+      "step": 17611
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.127691963500905e-05,
+      "loss": 0.6862,
+      "step": 17612
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.127436852649188e-05,
+      "loss": 0.6442,
+      "step": 17613
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.127181745765042e-05,
+      "loss": 0.724,
+      "step": 17614
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1269266428511837e-05,
+      "loss": 0.712,
+      "step": 17615
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1266715439103308e-05,
+      "loss": 0.6582,
+      "step": 17616
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1264164489451992e-05,
+      "loss": 0.7137,
+      "step": 17617
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1261613579585054e-05,
+      "loss": 0.6735,
+      "step": 17618
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1259062709529665e-05,
+      "loss": 0.6891,
+      "step": 17619
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1256511879312986e-05,
+      "loss": 0.7126,
+      "step": 17620
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1253961088962176e-05,
+      "loss": 0.7053,
+      "step": 17621
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1251410338504417e-05,
+      "loss": 0.7753,
+      "step": 17622
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1248859627966856e-05,
+      "loss": 0.6973,
+      "step": 17623
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.124630895737666e-05,
+      "loss": 0.7213,
+      "step": 17624
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1243758326760998e-05,
+      "loss": 0.7465,
+      "step": 17625
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1241207736147036e-05,
+      "loss": 0.6444,
+      "step": 17626
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1238657185561916e-05,
+      "loss": 0.7585,
+      "step": 17627
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1236106675032836e-05,
+      "loss": 0.7349,
+      "step": 17628
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1233556204586913e-05,
+      "loss": 0.7435,
+      "step": 17629
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.123100577425134e-05,
+      "loss": 0.6186,
+      "step": 17630
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1228455384053265e-05,
+      "loss": 0.6878,
+      "step": 17631
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1225905034019845e-05,
+      "loss": 0.6866,
+      "step": 17632
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.122335472417825e-05,
+      "loss": 0.6874,
+      "step": 17633
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1220804454555633e-05,
+      "loss": 0.6903,
+      "step": 17634
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1218254225179153e-05,
+      "loss": 0.744,
+      "step": 17635
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.121570403607597e-05,
+      "loss": 0.6439,
+      "step": 17636
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1213153887273247e-05,
+      "loss": 0.6866,
+      "step": 17637
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1210603778798126e-05,
+      "loss": 0.7166,
+      "step": 17638
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.120805371067778e-05,
+      "loss": 0.6379,
+      "step": 17639
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1205503682939362e-05,
+      "loss": 0.6814,
+      "step": 17640
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.120295369561002e-05,
+      "loss": 0.6834,
+      "step": 17641
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.120040374871692e-05,
+      "loss": 0.733,
+      "step": 17642
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1197853842287212e-05,
+      "loss": 0.6798,
+      "step": 17643
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1195303976348053e-05,
+      "loss": 0.642,
+      "step": 17644
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1192754150926596e-05,
+      "loss": 0.7008,
+      "step": 17645
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1190204366049998e-05,
+      "loss": 0.6326,
+      "step": 17646
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1187654621745414e-05,
+      "loss": 0.7176,
+      "step": 17647
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1185104918039993e-05,
+      "loss": 0.6715,
+      "step": 17648
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.118255525496088e-05,
+      "loss": 0.71,
+      "step": 17649
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1180005632535244e-05,
+      "loss": 0.6346,
+      "step": 17650
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1177456050790225e-05,
+      "loss": 0.7086,
+      "step": 17651
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1174906509752975e-05,
+      "loss": 0.6357,
+      "step": 17652
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1172357009450655e-05,
+      "loss": 0.6364,
+      "step": 17653
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.116980754991041e-05,
+      "loss": 0.734,
+      "step": 17654
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1167258131159377e-05,
+      "loss": 0.7035,
+      "step": 17655
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.116470875322473e-05,
+      "loss": 0.7379,
+      "step": 17656
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1162159416133602e-05,
+      "loss": 0.7121,
+      "step": 17657
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.115961011991315e-05,
+      "loss": 0.6626,
+      "step": 17658
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1157060864590512e-05,
+      "loss": 0.7218,
+      "step": 17659
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.115451165019285e-05,
+      "loss": 0.6775,
+      "step": 17660
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.11519624767473e-05,
+      "loss": 0.7677,
+      "step": 17661
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.114941334428101e-05,
+      "loss": 0.7323,
+      "step": 17662
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1146864252821134e-05,
+      "loss": 0.6222,
+      "step": 17663
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1144315202394814e-05,
+      "loss": 0.6592,
+      "step": 17664
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1141766193029195e-05,
+      "loss": 0.6995,
+      "step": 17665
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1139217224751426e-05,
+      "loss": 0.7282,
+      "step": 17666
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.113666829758865e-05,
+      "loss": 0.6188,
+      "step": 17667
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1134119411568017e-05,
+      "loss": 0.7046,
+      "step": 17668
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1131570566716654e-05,
+      "loss": 0.6598,
+      "step": 17669
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1129021763061724e-05,
+      "loss": 0.6844,
+      "step": 17670
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.112647300063036e-05,
+      "loss": 0.6133,
+      "step": 17671
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1123924279449707e-05,
+      "loss": 0.642,
+      "step": 17672
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.11213755995469e-05,
+      "loss": 0.6626,
+      "step": 17673
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1118826960949097e-05,
+      "loss": 0.7685,
+      "step": 17674
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1116278363683428e-05,
+      "loss": 0.7505,
+      "step": 17675
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.111372980777703e-05,
+      "loss": 0.6921,
+      "step": 17676
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1111181293257057e-05,
+      "loss": 0.7214,
+      "step": 17677
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1108632820150643e-05,
+      "loss": 0.6884,
+      "step": 17678
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.110608438848492e-05,
+      "loss": 0.6766,
+      "step": 17679
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1103535998287043e-05,
+      "loss": 0.7647,
+      "step": 17680
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.110098764958414e-05,
+      "loss": 0.6133,
+      "step": 17681
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1098439342403343e-05,
+      "loss": 0.6873,
+      "step": 17682
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1095891076771808e-05,
+      "loss": 0.7329,
+      "step": 17683
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1093342852716662e-05,
+      "loss": 0.7819,
+      "step": 17684
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1090794670265033e-05,
+      "loss": 0.7059,
+      "step": 17685
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1088246529444078e-05,
+      "loss": 0.6978,
+      "step": 17686
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.108569843028092e-05,
+      "loss": 0.7099,
+      "step": 17687
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1083150372802705e-05,
+      "loss": 0.7533,
+      "step": 17688
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1080602357036556e-05,
+      "loss": 0.6727,
+      "step": 17689
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.10780543830096e-05,
+      "loss": 0.6583,
+      "step": 17690
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1075506450748998e-05,
+      "loss": 0.7039,
+      "step": 17691
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1072958560281864e-05,
+      "loss": 0.7049,
+      "step": 17692
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1070410711635332e-05,
+      "loss": 0.67,
+      "step": 17693
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.106786290483655e-05,
+      "loss": 0.7887,
+      "step": 17694
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1065315139912635e-05,
+      "loss": 0.8372,
+      "step": 17695
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1062767416890722e-05,
+      "loss": 0.7055,
+      "step": 17696
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1060219735797952e-05,
+      "loss": 0.6842,
+      "step": 17697
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.105767209666145e-05,
+      "loss": 0.6878,
+      "step": 17698
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1055124499508343e-05,
+      "loss": 0.6799,
+      "step": 17699
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1052576944365764e-05,
+      "loss": 0.7161,
+      "step": 17700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1050029431260848e-05,
+      "loss": 0.6317,
+      "step": 17701
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1047481960220716e-05,
+      "loss": 0.7529,
+      "step": 17702
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1044934531272507e-05,
+      "loss": 0.6732,
+      "step": 17703
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1042387144443344e-05,
+      "loss": 0.6385,
+      "step": 17704
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1039839799760348e-05,
+      "loss": 0.6819,
+      "step": 17705
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.103729249725066e-05,
+      "loss": 0.73,
+      "step": 17706
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.10347452369414e-05,
+      "loss": 0.6491,
+      "step": 17707
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1032198018859706e-05,
+      "loss": 0.6535,
+      "step": 17708
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1029650843032685e-05,
+      "loss": 0.7175,
+      "step": 17709
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1027103709487467e-05,
+      "loss": 0.6987,
+      "step": 17710
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1024556618251186e-05,
+      "loss": 0.6711,
+      "step": 17711
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1022009569350964e-05,
+      "loss": 0.7363,
+      "step": 17712
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1019462562813925e-05,
+      "loss": 0.7242,
+      "step": 17713
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1016915598667193e-05,
+      "loss": 0.6593,
+      "step": 17714
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1014368676937892e-05,
+      "loss": 0.6613,
+      "step": 17715
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.101182179765314e-05,
+      "loss": 0.7775,
+      "step": 17716
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1009274960840068e-05,
+      "loss": 0.6038,
+      "step": 17717
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1006728166525803e-05,
+      "loss": 0.7103,
+      "step": 17718
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1004181414737444e-05,
+      "loss": 0.6767,
+      "step": 17719
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1001634705502136e-05,
+      "loss": 0.7135,
+      "step": 17720
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0999088038846992e-05,
+      "loss": 0.6689,
+      "step": 17721
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0996541414799126e-05,
+      "loss": 0.6531,
+      "step": 17722
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0993994833385667e-05,
+      "loss": 0.7333,
+      "step": 17723
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.099144829463373e-05,
+      "loss": 0.6194,
+      "step": 17724
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0988901798570433e-05,
+      "loss": 0.7514,
+      "step": 17725
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.09863553452229e-05,
+      "loss": 0.6271,
+      "step": 17726
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0983808934618247e-05,
+      "loss": 0.6742,
+      "step": 17727
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0981262566783584e-05,
+      "loss": 0.6648,
+      "step": 17728
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.097871624174605e-05,
+      "loss": 0.6645,
+      "step": 17729
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0976169959532732e-05,
+      "loss": 0.6753,
+      "step": 17730
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0973623720170764e-05,
+      "loss": 0.6982,
+      "step": 17731
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0971077523687264e-05,
+      "loss": 0.7152,
+      "step": 17732
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0968531370109334e-05,
+      "loss": 0.7392,
+      "step": 17733
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0965985259464103e-05,
+      "loss": 0.6654,
+      "step": 17734
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0963439191778678e-05,
+      "loss": 0.6925,
+      "step": 17735
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.096089316708017e-05,
+      "loss": 0.6488,
+      "step": 17736
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0958347185395706e-05,
+      "loss": 0.6874,
+      "step": 17737
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.095580124675239e-05,
+      "loss": 0.645,
+      "step": 17738
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0953255351177327e-05,
+      "loss": 0.6687,
+      "step": 17739
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0950709498697645e-05,
+      "loss": 0.6874,
+      "step": 17740
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.094816368934045e-05,
+      "loss": 0.6955,
+      "step": 17741
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.094561792313284e-05,
+      "loss": 0.677,
+      "step": 17742
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.094307220010195e-05,
+      "loss": 0.7883,
+      "step": 17743
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.094052652027488e-05,
+      "loss": 0.646,
+      "step": 17744
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0937980883678727e-05,
+      "loss": 0.7397,
+      "step": 17745
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0935435290340616e-05,
+      "loss": 0.6387,
+      "step": 17746
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0932889740287657e-05,
+      "loss": 0.6801,
+      "step": 17747
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0930344233546942e-05,
+      "loss": 0.7296,
+      "step": 17748
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.092779877014561e-05,
+      "loss": 0.6206,
+      "step": 17749
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.092525335011073e-05,
+      "loss": 0.659,
+      "step": 17750
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0922707973469435e-05,
+      "loss": 0.6724,
+      "step": 17751
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0920162640248826e-05,
+      "loss": 0.7692,
+      "step": 17752
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0917617350476e-05,
+      "loss": 0.681,
+      "step": 17753
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.091507210417808e-05,
+      "loss": 0.6795,
+      "step": 17754
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0912526901382157e-05,
+      "loss": 0.7545,
+      "step": 17755
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.090998174211534e-05,
+      "loss": 0.7782,
+      "step": 17756
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0907436626404735e-05,
+      "loss": 0.6808,
+      "step": 17757
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.090489155427745e-05,
+      "loss": 0.6624,
+      "step": 17758
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0902346525760574e-05,
+      "loss": 0.7201,
+      "step": 17759
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.089980154088123e-05,
+      "loss": 0.7074,
+      "step": 17760
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0897256599666506e-05,
+      "loss": 0.6785,
+      "step": 17761
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0894711702143505e-05,
+      "loss": 0.7031,
+      "step": 17762
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0892166848339336e-05,
+      "loss": 0.7408,
+      "step": 17763
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.08896220382811e-05,
+      "loss": 0.6829,
+      "step": 17764
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.088707727199589e-05,
+      "loss": 0.6727,
+      "step": 17765
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0884532549510806e-05,
+      "loss": 0.6694,
+      "step": 17766
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0881987870852956e-05,
+      "loss": 0.6954,
+      "step": 17767
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0879443236049434e-05,
+      "loss": 0.6658,
+      "step": 17768
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0876898645127348e-05,
+      "loss": 0.6927,
+      "step": 17769
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0874354098113777e-05,
+      "loss": 0.7039,
+      "step": 17770
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0871809595035835e-05,
+      "loss": 0.6539,
+      "step": 17771
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0869265135920617e-05,
+      "loss": 0.7553,
+      "step": 17772
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.086672072079521e-05,
+      "loss": 0.6969,
+      "step": 17773
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.086417634968672e-05,
+      "loss": 0.7313,
+      "step": 17774
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0861632022622245e-05,
+      "loss": 0.6418,
+      "step": 17775
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0859087739628874e-05,
+      "loss": 0.6395,
+      "step": 17776
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.08565435007337e-05,
+      "loss": 0.6532,
+      "step": 17777
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.085399930596383e-05,
+      "loss": 0.7098,
+      "step": 17778
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.085145515534635e-05,
+      "loss": 0.6985,
+      "step": 17779
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0848911048908346e-05,
+      "loss": 0.6823,
+      "step": 17780
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0846366986676928e-05,
+      "loss": 0.6997,
+      "step": 17781
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0843822968679176e-05,
+      "loss": 0.7401,
+      "step": 17782
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0841278994942182e-05,
+      "loss": 0.6857,
+      "step": 17783
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0838735065493048e-05,
+      "loss": 0.6265,
+      "step": 17784
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.083619118035886e-05,
+      "loss": 0.6525,
+      "step": 17785
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0833647339566698e-05,
+      "loss": 0.6862,
+      "step": 17786
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.083110354314367e-05,
+      "loss": 0.7412,
+      "step": 17787
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.082855979111686e-05,
+      "loss": 0.6733,
+      "step": 17788
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.082601608351336e-05,
+      "loss": 0.7529,
+      "step": 17789
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.082347242036025e-05,
+      "loss": 0.6374,
+      "step": 17790
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.082092880168462e-05,
+      "loss": 0.7026,
+      "step": 17791
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.081838522751356e-05,
+      "loss": 0.7221,
+      "step": 17792
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0815841697874166e-05,
+      "loss": 0.6353,
+      "step": 17793
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0813298212793508e-05,
+      "loss": 0.6574,
+      "step": 17794
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0810754772298686e-05,
+      "loss": 0.7403,
+      "step": 17795
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0808211376416785e-05,
+      "loss": 0.6883,
+      "step": 17796
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0805668025174885e-05,
+      "loss": 0.7826,
+      "step": 17797
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0803124718600075e-05,
+      "loss": 0.7359,
+      "step": 17798
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0800581456719444e-05,
+      "loss": 0.71,
+      "step": 17799
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0798038239560057e-05,
+      "loss": 0.7827,
+      "step": 17800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0795495067149022e-05,
+      "loss": 0.724,
+      "step": 17801
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0792951939513412e-05,
+      "loss": 0.6838,
+      "step": 17802
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.079040885668031e-05,
+      "loss": 0.7132,
+      "step": 17803
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0787865818676794e-05,
+      "loss": 0.7108,
+      "step": 17804
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0785322825529955e-05,
+      "loss": 0.6264,
+      "step": 17805
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0782779877266863e-05,
+      "loss": 0.7101,
+      "step": 17806
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0780236973914607e-05,
+      "loss": 0.7539,
+      "step": 17807
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0777694115500267e-05,
+      "loss": 0.6879,
+      "step": 17808
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.077515130205093e-05,
+      "loss": 0.7861,
+      "step": 17809
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0772608533593657e-05,
+      "loss": 0.682,
+      "step": 17810
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0770065810155535e-05,
+      "loss": 0.7078,
+      "step": 17811
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0767523131763646e-05,
+      "loss": 0.6995,
+      "step": 17812
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.076498049844507e-05,
+      "loss": 0.6603,
+      "step": 17813
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.076243791022687e-05,
+      "loss": 0.6781,
+      "step": 17814
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0759895367136142e-05,
+      "loss": 0.7349,
+      "step": 17815
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0757352869199953e-05,
+      "loss": 0.6218,
+      "step": 17816
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0754810416445375e-05,
+      "loss": 0.6583,
+      "step": 17817
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0752268008899494e-05,
+      "loss": 0.6345,
+      "step": 17818
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0749725646589384e-05,
+      "loss": 0.6615,
+      "step": 17819
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0747183329542105e-05,
+      "loss": 0.6358,
+      "step": 17820
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0744641057784747e-05,
+      "loss": 0.7059,
+      "step": 17821
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.074209883134438e-05,
+      "loss": 0.6589,
+      "step": 17822
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.073955665024807e-05,
+      "loss": 0.6235,
+      "step": 17823
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.07370145145229e-05,
+      "loss": 0.631,
+      "step": 17824
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.073447242419594e-05,
+      "loss": 0.6778,
+      "step": 17825
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.073193037929425e-05,
+      "loss": 0.7098,
+      "step": 17826
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0729388379844917e-05,
+      "loss": 0.6619,
+      "step": 17827
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0726846425875005e-05,
+      "loss": 0.7046,
+      "step": 17828
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0724304517411592e-05,
+      "loss": 0.7035,
+      "step": 17829
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.072176265448173e-05,
+      "loss": 0.638,
+      "step": 17830
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0719220837112495e-05,
+      "loss": 0.6358,
+      "step": 17831
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0716679065330967e-05,
+      "loss": 0.6641,
+      "step": 17832
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0714137339164203e-05,
+      "loss": 0.6552,
+      "step": 17833
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0711595658639267e-05,
+      "loss": 0.6191,
+      "step": 17834
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0709054023783244e-05,
+      "loss": 0.666,
+      "step": 17835
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0706512434623188e-05,
+      "loss": 0.688,
+      "step": 17836
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0703970891186164e-05,
+      "loss": 0.6377,
+      "step": 17837
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0701429393499246e-05,
+      "loss": 0.7321,
+      "step": 17838
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0698887941589495e-05,
+      "loss": 0.679,
+      "step": 17839
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0696346535483973e-05,
+      "loss": 0.73,
+      "step": 17840
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0693805175209753e-05,
+      "loss": 0.7041,
+      "step": 17841
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.069126386079389e-05,
+      "loss": 0.7334,
+      "step": 17842
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.068872259226345e-05,
+      "loss": 0.7508,
+      "step": 17843
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0686181369645497e-05,
+      "loss": 0.715,
+      "step": 17844
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0683640192967098e-05,
+      "loss": 0.6765,
+      "step": 17845
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.06810990622553e-05,
+      "loss": 0.6394,
+      "step": 17846
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0678557977537185e-05,
+      "loss": 0.652,
+      "step": 17847
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0676016938839803e-05,
+      "loss": 0.6755,
+      "step": 17848
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.067347594619021e-05,
+      "loss": 0.7201,
+      "step": 17849
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.067093499961549e-05,
+      "loss": 0.6166,
+      "step": 17850
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0668394099142662e-05,
+      "loss": 0.6637,
+      "step": 17851
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0665853244798816e-05,
+      "loss": 0.6861,
+      "step": 17852
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0663312436611002e-05,
+      "loss": 0.6787,
+      "step": 17853
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0660771674606273e-05,
+      "loss": 0.6477,
+      "step": 17854
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0658230958811694e-05,
+      "loss": 0.7172,
+      "step": 17855
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.065569028925432e-05,
+      "loss": 0.6896,
+      "step": 17856
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.06531496659612e-05,
+      "loss": 0.7457,
+      "step": 17857
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0650609088959408e-05,
+      "loss": 0.691,
+      "step": 17858
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0648068558275986e-05,
+      "loss": 0.6802,
+      "step": 17859
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0645528073937984e-05,
+      "loss": 0.7091,
+      "step": 17860
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0642987635972472e-05,
+      "loss": 0.6106,
+      "step": 17861
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0640447244406493e-05,
+      "loss": 0.7457,
+      "step": 17862
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0637906899267105e-05,
+      "loss": 0.6433,
+      "step": 17863
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0635366600581362e-05,
+      "loss": 0.7284,
+      "step": 17864
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0632826348376314e-05,
+      "loss": 0.7289,
+      "step": 17865
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.063028614267901e-05,
+      "loss": 0.6305,
+      "step": 17866
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0627745983516507e-05,
+      "loss": 0.694,
+      "step": 17867
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.062520587091586e-05,
+      "loss": 0.7003,
+      "step": 17868
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0622665804904106e-05,
+      "loss": 0.6627,
+      "step": 17869
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0620125785508324e-05,
+      "loss": 0.6112,
+      "step": 17870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0617585812755526e-05,
+      "loss": 0.7069,
+      "step": 17871
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.061504588667278e-05,
+      "loss": 0.7279,
+      "step": 17872
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0612506007287136e-05,
+      "loss": 0.6993,
+      "step": 17873
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0609966174625635e-05,
+      "loss": 0.6544,
+      "step": 17874
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.060742638871533e-05,
+      "loss": 0.7246,
+      "step": 17875
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.060488664958327e-05,
+      "loss": 0.6743,
+      "step": 17876
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0602346957256494e-05,
+      "loss": 0.724,
+      "step": 17877
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0599807311762058e-05,
+      "loss": 0.6754,
+      "step": 17878
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0597267713127e-05,
+      "loss": 0.6947,
+      "step": 17879
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0594728161378365e-05,
+      "loss": 0.6849,
+      "step": 17880
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0592188656543206e-05,
+      "loss": 0.6511,
+      "step": 17881
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0589649198648563e-05,
+      "loss": 0.7156,
+      "step": 17882
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0587109787721477e-05,
+      "loss": 0.7262,
+      "step": 17883
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.058457042378899e-05,
+      "loss": 0.7324,
+      "step": 17884
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0582031106878148e-05,
+      "loss": 0.6628,
+      "step": 17885
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0579491837015994e-05,
+      "loss": 0.6889,
+      "step": 17886
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0576952614229562e-05,
+      "loss": 0.7788,
+      "step": 17887
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0574413438545907e-05,
+      "loss": 0.7272,
+      "step": 17888
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.057187430999206e-05,
+      "loss": 0.7983,
+      "step": 17889
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0569335228595072e-05,
+      "loss": 0.6436,
+      "step": 17890
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.056679619438196e-05,
+      "loss": 0.6871,
+      "step": 17891
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0564257207379783e-05,
+      "loss": 0.7115,
+      "step": 17892
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0561718267615573e-05,
+      "loss": 0.6836,
+      "step": 17893
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0559179375116368e-05,
+      "loss": 0.7075,
+      "step": 17894
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0556640529909204e-05,
+      "loss": 0.7303,
+      "step": 17895
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0554101732021126e-05,
+      "loss": 0.6683,
+      "step": 17896
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0551562981479168e-05,
+      "loss": 0.6565,
+      "step": 17897
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0549024278310352e-05,
+      "loss": 0.7287,
+      "step": 17898
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0546485622541735e-05,
+      "loss": 0.6884,
+      "step": 17899
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0543947014200338e-05,
+      "loss": 0.6737,
+      "step": 17900
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.05414084533132e-05,
+      "loss": 0.6632,
+      "step": 17901
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0538869939907358e-05,
+      "loss": 0.7329,
+      "step": 17902
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.053633147400984e-05,
+      "loss": 0.6456,
+      "step": 17903
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.053379305564768e-05,
+      "loss": 0.6249,
+      "step": 17904
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.053125468484792e-05,
+      "loss": 0.6878,
+      "step": 17905
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.052871636163758e-05,
+      "loss": 0.8402,
+      "step": 17906
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.052617808604369e-05,
+      "loss": 0.8256,
+      "step": 17907
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0523639858093295e-05,
+      "loss": 0.6037,
+      "step": 17908
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.052110167781342e-05,
+      "loss": 0.7213,
+      "step": 17909
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0518563545231092e-05,
+      "loss": 0.638,
+      "step": 17910
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.051602546037334e-05,
+      "loss": 0.7347,
+      "step": 17911
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.051348742326719e-05,
+      "loss": 0.6648,
+      "step": 17912
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.051094943393968e-05,
+      "loss": 0.6655,
+      "step": 17913
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0508411492417832e-05,
+      "loss": 0.6951,
+      "step": 17914
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.050587359872867e-05,
+      "loss": 0.7625,
+      "step": 17915
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.050333575289923e-05,
+      "loss": 0.6967,
+      "step": 17916
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0500797954956536e-05,
+      "loss": 0.7076,
+      "step": 17917
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0498260204927608e-05,
+      "loss": 0.7122,
+      "step": 17918
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.049572250283948e-05,
+      "loss": 0.6565,
+      "step": 17919
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.049318484871917e-05,
+      "loss": 0.6683,
+      "step": 17920
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0490647242593704e-05,
+      "loss": 0.6578,
+      "step": 17921
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.048810968449011e-05,
+      "loss": 0.6594,
+      "step": 17922
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0485572174435406e-05,
+      "loss": 0.6867,
+      "step": 17923
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0483034712456617e-05,
+      "loss": 0.7388,
+      "step": 17924
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.048049729858077e-05,
+      "loss": 0.685,
+      "step": 17925
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0477959932834885e-05,
+      "loss": 0.7119,
+      "step": 17926
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.047542261524597e-05,
+      "loss": 0.6183,
+      "step": 17927
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.047288534584107e-05,
+      "loss": 0.717,
+      "step": 17928
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0470348124647187e-05,
+      "loss": 0.7087,
+      "step": 17929
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0467810951691354e-05,
+      "loss": 0.706,
+      "step": 17930
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.046527382700058e-05,
+      "loss": 0.7048,
+      "step": 17931
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0462736750601873e-05,
+      "loss": 0.712,
+      "step": 17932
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0460199722522277e-05,
+      "loss": 0.72,
+      "step": 17933
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0457662742788797e-05,
+      "loss": 0.7063,
+      "step": 17934
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0455125811428444e-05,
+      "loss": 0.6665,
+      "step": 17935
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.045258892846825e-05,
+      "loss": 0.652,
+      "step": 17936
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.045005209393522e-05,
+      "loss": 0.6431,
+      "step": 17937
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.044751530785637e-05,
+      "loss": 0.7166,
+      "step": 17938
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0444978570258723e-05,
+      "loss": 0.7104,
+      "step": 17939
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0442441881169288e-05,
+      "loss": 0.6895,
+      "step": 17940
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0439905240615075e-05,
+      "loss": 0.712,
+      "step": 17941
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0437368648623107e-05,
+      "loss": 0.7249,
+      "step": 17942
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0434832105220395e-05,
+      "loss": 0.6484,
+      "step": 17943
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0432295610433945e-05,
+      "loss": 0.6467,
+      "step": 17944
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0429759164290775e-05,
+      "loss": 0.6921,
+      "step": 17945
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0427222766817897e-05,
+      "loss": 0.6277,
+      "step": 17946
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0424686418042316e-05,
+      "loss": 0.6632,
+      "step": 17947
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.042215011799105e-05,
+      "loss": 0.7323,
+      "step": 17948
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0419613866691107e-05,
+      "loss": 0.703,
+      "step": 17949
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.041707766416949e-05,
+      "loss": 0.6123,
+      "step": 17950
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.041454151045323e-05,
+      "loss": 0.6243,
+      "step": 17951
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0412005405569303e-05,
+      "loss": 0.7668,
+      "step": 17952
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0409469349544738e-05,
+      "loss": 0.6443,
+      "step": 17953
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.040693334240654e-05,
+      "loss": 0.6868,
+      "step": 17954
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0404397384181706e-05,
+      "loss": 0.6856,
+      "step": 17955
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0401861474897254e-05,
+      "loss": 0.7214,
+      "step": 17956
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0399325614580186e-05,
+      "loss": 0.6929,
+      "step": 17957
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0396789803257503e-05,
+      "loss": 0.7499,
+      "step": 17958
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0394254040956218e-05,
+      "loss": 0.6369,
+      "step": 17959
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.039171832770333e-05,
+      "loss": 0.6824,
+      "step": 17960
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0389182663525843e-05,
+      "loss": 0.7146,
+      "step": 17961
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0386647048450763e-05,
+      "loss": 0.7884,
+      "step": 17962
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0384111482505095e-05,
+      "loss": 0.6853,
+      "step": 17963
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0381575965715826e-05,
+      "loss": 0.7198,
+      "step": 17964
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.037904049810998e-05,
+      "loss": 0.737,
+      "step": 17965
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0376505079714546e-05,
+      "loss": 0.7281,
+      "step": 17966
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.037396971055652e-05,
+      "loss": 0.6974,
+      "step": 17967
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.037143439066291e-05,
+      "loss": 0.6536,
+      "step": 17968
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0368899120060718e-05,
+      "loss": 0.6762,
+      "step": 17969
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0366363898776934e-05,
+      "loss": 0.7005,
+      "step": 17970
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0363828726838575e-05,
+      "loss": 0.639,
+      "step": 17971
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.036129360427261e-05,
+      "loss": 0.7199,
+      "step": 17972
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0358758531106057e-05,
+      "loss": 0.6283,
+      "step": 17973
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0356223507365908e-05,
+      "loss": 0.7017,
+      "step": 17974
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.035368853307915e-05,
+      "loss": 0.6921,
+      "step": 17975
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.03511536082728e-05,
+      "loss": 0.7194,
+      "step": 17976
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.034861873297384e-05,
+      "loss": 0.6248,
+      "step": 17977
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.034608390720926e-05,
+      "loss": 0.6942,
+      "step": 17978
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0343549131006066e-05,
+      "loss": 0.7056,
+      "step": 17979
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0341014404391248e-05,
+      "loss": 0.6597,
+      "step": 17980
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.033847972739179e-05,
+      "loss": 0.6616,
+      "step": 17981
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0335945100034702e-05,
+      "loss": 0.6661,
+      "step": 17982
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0333410522346965e-05,
+      "loss": 0.6659,
+      "step": 17983
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.033087599435557e-05,
+      "loss": 0.676,
+      "step": 17984
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0328341516087514e-05,
+      "loss": 0.6779,
+      "step": 17985
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.032580708756978e-05,
+      "loss": 0.6523,
+      "step": 17986
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0323272708829367e-05,
+      "loss": 0.6833,
+      "step": 17987
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0320738379893258e-05,
+      "loss": 0.6296,
+      "step": 17988
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.031820410078845e-05,
+      "loss": 0.7007,
+      "step": 17989
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.031566987154192e-05,
+      "loss": 0.6148,
+      "step": 17990
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0313135692180674e-05,
+      "loss": 0.6163,
+      "step": 17991
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0310601562731672e-05,
+      "loss": 0.7697,
+      "step": 17992
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.030806748322192e-05,
+      "loss": 0.677,
+      "step": 17993
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0305533453678406e-05,
+      "loss": 0.6376,
+      "step": 17994
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0302999474128104e-05,
+      "loss": 0.7586,
+      "step": 17995
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0300465544598008e-05,
+      "loss": 0.6548,
+      "step": 17996
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0297931665115104e-05,
+      "loss": 0.7087,
+      "step": 17997
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0295397835706366e-05,
+      "loss": 0.7062,
+      "step": 17998
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0292864056398793e-05,
+      "loss": 0.667,
+      "step": 17999
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0290330327219358e-05,
+      "loss": 0.6744,
+      "step": 18000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0287796648195044e-05,
+      "loss": 0.6302,
+      "step": 18001
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0285263019352832e-05,
+      "loss": 0.7839,
+      "step": 18002
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.028272944071971e-05,
+      "loss": 0.7345,
+      "step": 18003
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.028019591232266e-05,
+      "loss": 0.6938,
+      "step": 18004
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0277662434188645e-05,
+      "loss": 0.6982,
+      "step": 18005
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.027512900634467e-05,
+      "loss": 0.6877,
+      "step": 18006
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.02725956288177e-05,
+      "loss": 0.6861,
+      "step": 18007
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.027006230163471e-05,
+      "loss": 0.7574,
+      "step": 18008
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.026752902482269e-05,
+      "loss": 0.7078,
+      "step": 18009
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0264995798408617e-05,
+      "loss": 0.6798,
+      "step": 18010
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0262462622419468e-05,
+      "loss": 0.6754,
+      "step": 18011
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.02599294968822e-05,
+      "loss": 0.6982,
+      "step": 18012
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0257396421823816e-05,
+      "loss": 0.7745,
+      "step": 18013
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.025486339727128e-05,
+      "loss": 0.635,
+      "step": 18014
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0252330423251567e-05,
+      "loss": 0.7097,
+      "step": 18015
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0249797499791647e-05,
+      "loss": 0.6538,
+      "step": 18016
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0247264626918507e-05,
+      "loss": 0.7356,
+      "step": 18017
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0244731804659112e-05,
+      "loss": 0.6625,
+      "step": 18018
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.024219903304043e-05,
+      "loss": 0.6566,
+      "step": 18019
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0239666312089444e-05,
+      "loss": 0.6676,
+      "step": 18020
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0237133641833124e-05,
+      "loss": 0.7917,
+      "step": 18021
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0234601022298434e-05,
+      "loss": 0.7707,
+      "step": 18022
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0232068453512355e-05,
+      "loss": 0.6739,
+      "step": 18023
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.022953593550185e-05,
+      "loss": 0.7511,
+      "step": 18024
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0227003468293887e-05,
+      "loss": 0.5906,
+      "step": 18025
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.022447105191545e-05,
+      "loss": 0.6779,
+      "step": 18026
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.022193868639349e-05,
+      "loss": 0.6681,
+      "step": 18027
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.021940637175498e-05,
+      "loss": 0.69,
+      "step": 18028
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0216874108026894e-05,
+      "loss": 0.6947,
+      "step": 18029
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0214341895236192e-05,
+      "loss": 0.6751,
+      "step": 18030
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0211809733409854e-05,
+      "loss": 0.7285,
+      "step": 18031
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0209277622574827e-05,
+      "loss": 0.7664,
+      "step": 18032
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.020674556275808e-05,
+      "loss": 0.6656,
+      "step": 18033
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0204213553986586e-05,
+      "loss": 0.8072,
+      "step": 18034
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.020168159628731e-05,
+      "loss": 0.6323,
+      "step": 18035
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0199149689687204e-05,
+      "loss": 0.7122,
+      "step": 18036
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0196617834213248e-05,
+      "loss": 0.6462,
+      "step": 18037
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.019408602989239e-05,
+      "loss": 0.6436,
+      "step": 18038
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0191554276751596e-05,
+      "loss": 0.6899,
+      "step": 18039
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0189022574817835e-05,
+      "loss": 0.7059,
+      "step": 18040
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.018649092411806e-05,
+      "loss": 0.6765,
+      "step": 18041
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0183959324679234e-05,
+      "loss": 0.7569,
+      "step": 18042
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0181427776528324e-05,
+      "loss": 0.6453,
+      "step": 18043
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0178896279692278e-05,
+      "loss": 0.7028,
+      "step": 18044
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0176364834198054e-05,
+      "loss": 0.6942,
+      "step": 18045
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.017383344007262e-05,
+      "loss": 0.6234,
+      "step": 18046
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0171302097342934e-05,
+      "loss": 0.6578,
+      "step": 18047
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0168770806035943e-05,
+      "loss": 0.6568,
+      "step": 18048
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0166239566178615e-05,
+      "loss": 0.6521,
+      "step": 18049
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.01637083777979e-05,
+      "loss": 0.7065,
+      "step": 18050
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0161177240920747e-05,
+      "loss": 0.6361,
+      "step": 18051
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.015864615557414e-05,
+      "loss": 0.6972,
+      "step": 18052
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.015611512178499e-05,
+      "loss": 0.7263,
+      "step": 18053
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0153584139580285e-05,
+      "loss": 0.6785,
+      "step": 18054
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0151053208986964e-05,
+      "loss": 0.6666,
+      "step": 18055
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0148522330031977e-05,
+      "loss": 0.6737,
+      "step": 18056
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0145991502742283e-05,
+      "loss": 0.6772,
+      "step": 18057
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0143460727144837e-05,
+      "loss": 0.6761,
+      "step": 18058
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.014093000326658e-05,
+      "loss": 0.7397,
+      "step": 18059
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0138399331134474e-05,
+      "loss": 0.6108,
+      "step": 18060
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0135868710775464e-05,
+      "loss": 0.7136,
+      "step": 18061
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0133338142216493e-05,
+      "loss": 0.7179,
+      "step": 18062
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.013080762548452e-05,
+      "loss": 0.7077,
+      "step": 18063
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0128277160606492e-05,
+      "loss": 0.6863,
+      "step": 18064
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0125746747609347e-05,
+      "loss": 0.6737,
+      "step": 18065
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.012321638652005e-05,
+      "loss": 0.7542,
+      "step": 18066
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0120686077365533e-05,
+      "loss": 0.7046,
+      "step": 18067
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.011815582017274e-05,
+      "loss": 0.7354,
+      "step": 18068
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0115625614968633e-05,
+      "loss": 0.6956,
+      "step": 18069
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0113095461780145e-05,
+      "loss": 0.6634,
+      "step": 18070
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0110565360634223e-05,
+      "loss": 0.6517,
+      "step": 18071
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0108035311557824e-05,
+      "loss": 0.693,
+      "step": 18072
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.010550531457786e-05,
+      "loss": 0.6685,
+      "step": 18073
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0102975369721305e-05,
+      "loss": 0.6995,
+      "step": 18074
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0100445477015083e-05,
+      "loss": 0.6871,
+      "step": 18075
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.009791563648614e-05,
+      "loss": 0.6891,
+      "step": 18076
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0095385848161425e-05,
+      "loss": 0.6519,
+      "step": 18077
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0092856112067875e-05,
+      "loss": 0.6653,
+      "step": 18078
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.009032642823242e-05,
+      "loss": 0.753,
+      "step": 18079
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.008779679668201e-05,
+      "loss": 0.6625,
+      "step": 18080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.008526721744359e-05,
+      "loss": 0.6116,
+      "step": 18081
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0082737690544076e-05,
+      "loss": 0.5915,
+      "step": 18082
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.008020821601043e-05,
+      "loss": 0.668,
+      "step": 18083
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0077678793869577e-05,
+      "loss": 0.7451,
+      "step": 18084
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0075149424148453e-05,
+      "loss": 0.6578,
+      "step": 18085
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0072620106874006e-05,
+      "loss": 0.6065,
+      "step": 18086
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0070090842073158e-05,
+      "loss": 0.7039,
+      "step": 18087
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.006756162977285e-05,
+      "loss": 0.698,
+      "step": 18088
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0065032470000018e-05,
+      "loss": 0.7959,
+      "step": 18089
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0062503362781594e-05,
+      "loss": 0.7354,
+      "step": 18090
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0059974308144504e-05,
+      "loss": 0.6258,
+      "step": 18091
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0057445306115707e-05,
+      "loss": 0.7416,
+      "step": 18092
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0054916356722102e-05,
+      "loss": 0.767,
+      "step": 18093
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.005238745999064e-05,
+      "loss": 0.6905,
+      "step": 18094
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.004985861594825e-05,
+      "loss": 0.6647,
+      "step": 18095
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.004732982462185e-05,
+      "loss": 0.7203,
+      "step": 18096
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.004480108603839e-05,
+      "loss": 0.7035,
+      "step": 18097
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.004227240022479e-05,
+      "loss": 0.6557,
+      "step": 18098
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0039743767207968e-05,
+      "loss": 0.6737,
+      "step": 18099
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0037215187014873e-05,
+      "loss": 0.69,
+      "step": 18100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0034686659672425e-05,
+      "loss": 0.6453,
+      "step": 18101
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.003215818520754e-05,
+      "loss": 0.714,
+      "step": 18102
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0029629763647166e-05,
+      "loss": 0.6822,
+      "step": 18103
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0027101395018212e-05,
+      "loss": 0.6538,
+      "step": 18104
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0024573079347604e-05,
+      "loss": 0.6241,
+      "step": 18105
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.002204481666228e-05,
+      "loss": 0.6678,
+      "step": 18106
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0019516606989154e-05,
+      "loss": 0.6303,
+      "step": 18107
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0016988450355155e-05,
+      "loss": 0.7487,
+      "step": 18108
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0014460346787198e-05,
+      "loss": 0.6418,
+      "step": 18109
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0011932296312215e-05,
+      "loss": 0.6767,
+      "step": 18110
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0009404298957124e-05,
+      "loss": 0.6949,
+      "step": 18111
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0006876354748854e-05,
+      "loss": 0.6557,
+      "step": 18112
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.000434846371431e-05,
+      "loss": 0.6553,
+      "step": 18113
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0001820625880427e-05,
+      "loss": 0.6882,
+      "step": 18114
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.999929284127412e-05,
+      "loss": 0.7063,
+      "step": 18115
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.99967651099223e-05,
+      "loss": 0.6833,
+      "step": 18116
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.99942374318519e-05,
+      "loss": 0.7176,
+      "step": 18117
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9991709807089836e-05,
+      "loss": 0.6785,
+      "step": 18118
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9989182235663013e-05,
+      "loss": 0.6887,
+      "step": 18119
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9986654717598365e-05,
+      "loss": 0.6547,
+      "step": 18120
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.99841272529228e-05,
+      "loss": 0.6679,
+      "step": 18121
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9981599841663235e-05,
+      "loss": 0.6248,
+      "step": 18122
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9979072483846578e-05,
+      "loss": 0.6681,
+      "step": 18123
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9976545179499757e-05,
+      "loss": 0.7365,
+      "step": 18124
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9974017928649676e-05,
+      "loss": 0.5602,
+      "step": 18125
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9971490731323252e-05,
+      "loss": 0.6453,
+      "step": 18126
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9968963587547403e-05,
+      "loss": 0.6597,
+      "step": 18127
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.996643649734904e-05,
+      "loss": 0.7389,
+      "step": 18128
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9963909460755063e-05,
+      "loss": 0.6917,
+      "step": 18129
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.99613824777924e-05,
+      "loss": 0.6763,
+      "step": 18130
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9958855548487955e-05,
+      "loss": 0.7075,
+      "step": 18131
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9956328672868642e-05,
+      "loss": 0.7176,
+      "step": 18132
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.995380185096136e-05,
+      "loss": 0.7312,
+      "step": 18133
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.995127508279302e-05,
+      "loss": 0.6521,
+      "step": 18134
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9948748368390542e-05,
+      "loss": 0.7781,
+      "step": 18135
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9946221707780825e-05,
+      "loss": 0.7475,
+      "step": 18136
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.994369510099078e-05,
+      "loss": 0.6873,
+      "step": 18137
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.994116854804731e-05,
+      "loss": 0.6602,
+      "step": 18138
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9938642048977323e-05,
+      "loss": 0.6613,
+      "step": 18139
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9936115603807726e-05,
+      "loss": 0.6505,
+      "step": 18140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9933589212565427e-05,
+      "loss": 0.6744,
+      "step": 18141
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9931062875277328e-05,
+      "loss": 0.6453,
+      "step": 18142
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.992853659197032e-05,
+      "loss": 0.6432,
+      "step": 18143
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9926010362671328e-05,
+      "loss": 0.6877,
+      "step": 18144
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.992348418740725e-05,
+      "loss": 0.644,
+      "step": 18145
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.992095806620497e-05,
+      "loss": 0.7395,
+      "step": 18146
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9918431999091415e-05,
+      "loss": 0.6557,
+      "step": 18147
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.991590598609347e-05,
+      "loss": 0.7026,
+      "step": 18148
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9913380027238035e-05,
+      "loss": 0.7222,
+      "step": 18149
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.991085412255202e-05,
+      "loss": 0.6471,
+      "step": 18150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9908328272062318e-05,
+      "loss": 0.6983,
+      "step": 18151
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9905802475795836e-05,
+      "loss": 0.8328,
+      "step": 18152
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.990327673377946e-05,
+      "loss": 0.7161,
+      "step": 18153
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9900751046040083e-05,
+      "loss": 0.6242,
+      "step": 18154
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.989822541260462e-05,
+      "loss": 0.6414,
+      "step": 18155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.989569983349996e-05,
+      "loss": 0.6835,
+      "step": 18156
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.989317430875299e-05,
+      "loss": 0.6383,
+      "step": 18157
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9890648838390623e-05,
+      "loss": 0.6863,
+      "step": 18158
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.988812342243974e-05,
+      "loss": 0.6658,
+      "step": 18159
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9885598060927232e-05,
+      "loss": 0.7312,
+      "step": 18160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9883072753880007e-05,
+      "loss": 0.6616,
+      "step": 18161
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9880547501324953e-05,
+      "loss": 0.6625,
+      "step": 18162
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.987802230328895e-05,
+      "loss": 0.6129,
+      "step": 18163
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.987549715979891e-05,
+      "loss": 0.7152,
+      "step": 18164
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9872972070881714e-05,
+      "loss": 0.6391,
+      "step": 18165
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9870447036564245e-05,
+      "loss": 0.6824,
+      "step": 18166
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9867922056873407e-05,
+      "loss": 0.7082,
+      "step": 18167
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.986539713183608e-05,
+      "loss": 0.7497,
+      "step": 18168
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9862872261479155e-05,
+      "loss": 0.7791,
+      "step": 18169
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9860347445829526e-05,
+      "loss": 0.6956,
+      "step": 18170
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9857822684914075e-05,
+      "loss": 0.6304,
+      "step": 18171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9855297978759684e-05,
+      "loss": 0.6265,
+      "step": 18172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9852773327393264e-05,
+      "loss": 0.666,
+      "step": 18173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.985024873084167e-05,
+      "loss": 0.8177,
+      "step": 18174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9847724189131804e-05,
+      "loss": 0.6345,
+      "step": 18175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9845199702290546e-05,
+      "loss": 0.6799,
+      "step": 18176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9842675270344776e-05,
+      "loss": 0.6952,
+      "step": 18177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.984015089332139e-05,
+      "loss": 0.6698,
+      "step": 18178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9837626571247265e-05,
+      "loss": 0.7079,
+      "step": 18179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9835102304149275e-05,
+      "loss": 0.7255,
+      "step": 18180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9832578092054315e-05,
+      "loss": 0.7361,
+      "step": 18181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9830053934989262e-05,
+      "loss": 0.6738,
+      "step": 18182
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.982752983298099e-05,
+      "loss": 0.7097,
+      "step": 18183
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9825005786056395e-05,
+      "loss": 0.6981,
+      "step": 18184
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9822481794242344e-05,
+      "loss": 0.6182,
+      "step": 18185
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.981995785756571e-05,
+      "loss": 0.6937,
+      "step": 18186
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9817433976053387e-05,
+      "loss": 0.6515,
+      "step": 18187
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9814910149732247e-05,
+      "loss": 0.675,
+      "step": 18188
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.981238637862916e-05,
+      "loss": 0.6659,
+      "step": 18189
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9809862662771014e-05,
+      "loss": 0.6518,
+      "step": 18190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.980733900218468e-05,
+      "loss": 0.6928,
+      "step": 18191
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9804815396897025e-05,
+      "loss": 0.7077,
+      "step": 18192
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.980229184693495e-05,
+      "loss": 0.665,
+      "step": 18193
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9799768352325298e-05,
+      "loss": 0.6895,
+      "step": 18194
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.979724491309496e-05,
+      "loss": 0.5961,
+      "step": 18195
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9794721529270803e-05,
+      "loss": 0.6319,
+      "step": 18196
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9792198200879696e-05,
+      "loss": 0.7026,
+      "step": 18197
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.978967492794852e-05,
+      "loss": 0.6358,
+      "step": 18198
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9787151710504147e-05,
+      "loss": 0.7572,
+      "step": 18199
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9784628548573435e-05,
+      "loss": 0.7129,
+      "step": 18200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9782105442183272e-05,
+      "loss": 0.6248,
+      "step": 18201
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.977958239136051e-05,
+      "loss": 0.6606,
+      "step": 18202
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.977705939613203e-05,
+      "loss": 0.6947,
+      "step": 18203
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9774536456524697e-05,
+      "loss": 0.6444,
+      "step": 18204
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.977201357256538e-05,
+      "loss": 0.7091,
+      "step": 18205
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.976949074428094e-05,
+      "loss": 0.6718,
+      "step": 18206
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9766967971698252e-05,
+      "loss": 0.6175,
+      "step": 18207
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9764445254844177e-05,
+      "loss": 0.6911,
+      "step": 18208
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9761922593745578e-05,
+      "loss": 0.7061,
+      "step": 18209
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.975939998842933e-05,
+      "loss": 0.6209,
+      "step": 18210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9756877438922286e-05,
+      "loss": 0.7367,
+      "step": 18211
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.975435494525131e-05,
+      "loss": 0.6526,
+      "step": 18212
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9751832507443287e-05,
+      "loss": 0.6716,
+      "step": 18213
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9749310125525046e-05,
+      "loss": 0.6853,
+      "step": 18214
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9746787799523465e-05,
+      "loss": 0.689,
+      "step": 18215
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9744265529465412e-05,
+      "loss": 0.7218,
+      "step": 18216
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.974174331537773e-05,
+      "loss": 0.6905,
+      "step": 18217
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9739221157287296e-05,
+      "loss": 0.6517,
+      "step": 18218
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9736699055220962e-05,
+      "loss": 0.6924,
+      "step": 18219
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9734177009205583e-05,
+      "loss": 0.6641,
+      "step": 18220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9731655019268024e-05,
+      "loss": 0.6701,
+      "step": 18221
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.972913308543515e-05,
+      "loss": 0.7186,
+      "step": 18222
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9726611207733792e-05,
+      "loss": 0.7351,
+      "step": 18223
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9724089386190835e-05,
+      "loss": 0.7552,
+      "step": 18224
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.972156762083312e-05,
+      "loss": 0.6946,
+      "step": 18225
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9719045911687506e-05,
+      "loss": 0.7192,
+      "step": 18226
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9716524258780843e-05,
+      "loss": 0.7145,
+      "step": 18227
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9714002662139994e-05,
+      "loss": 0.6543,
+      "step": 18228
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9711481121791805e-05,
+      "loss": 0.6501,
+      "step": 18229
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.970895963776313e-05,
+      "loss": 0.6178,
+      "step": 18230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9706438210080826e-05,
+      "loss": 0.6882,
+      "step": 18231
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9703916838771737e-05,
+      "loss": 0.707,
+      "step": 18232
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.970139552386273e-05,
+      "loss": 0.6481,
+      "step": 18233
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9698874265380628e-05,
+      "loss": 0.6877,
+      "step": 18234
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9696353063352306e-05,
+      "loss": 0.7637,
+      "step": 18235
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9693831917804602e-05,
+      "loss": 0.6804,
+      "step": 18236
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9691310828764364e-05,
+      "loss": 0.6934,
+      "step": 18237
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9688789796258445e-05,
+      "loss": 0.6494,
+      "step": 18238
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.968626882031369e-05,
+      "loss": 0.6816,
+      "step": 18239
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.968374790095695e-05,
+      "loss": 0.6313,
+      "step": 18240
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9681227038215058e-05,
+      "loss": 0.6356,
+      "step": 18241
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9678706232114873e-05,
+      "loss": 0.6649,
+      "step": 18242
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.967618548268324e-05,
+      "loss": 0.6756,
+      "step": 18243
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9673664789946995e-05,
+      "loss": 0.6578,
+      "step": 18244
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9671144153932984e-05,
+      "loss": 0.7479,
+      "step": 18245
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.966862357466806e-05,
+      "loss": 0.6652,
+      "step": 18246
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9666103052179052e-05,
+      "loss": 0.6322,
+      "step": 18247
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9663582586492812e-05,
+      "loss": 0.669,
+      "step": 18248
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.966106217763618e-05,
+      "loss": 0.6757,
+      "step": 18249
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9658541825635984e-05,
+      "loss": 0.722,
+      "step": 18250
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.965602153051908e-05,
+      "loss": 0.6708,
+      "step": 18251
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9653501292312307e-05,
+      "loss": 0.6733,
+      "step": 18252
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9650981111042506e-05,
+      "loss": 0.7093,
+      "step": 18253
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9648460986736497e-05,
+      "loss": 0.6783,
+      "step": 18254
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9645940919421125e-05,
+      "loss": 0.6458,
+      "step": 18255
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.964342090912324e-05,
+      "loss": 0.7165,
+      "step": 18256
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.964090095586967e-05,
+      "loss": 0.6618,
+      "step": 18257
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9638381059687247e-05,
+      "loss": 0.6924,
+      "step": 18258
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9635861220602812e-05,
+      "loss": 0.6211,
+      "step": 18259
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.96333414386432e-05,
+      "loss": 0.6687,
+      "step": 18260
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.963082171383524e-05,
+      "loss": 0.731,
+      "step": 18261
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9628302046205778e-05,
+      "loss": 0.6383,
+      "step": 18262
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.962578243578163e-05,
+      "loss": 0.6823,
+      "step": 18263
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9623262882589634e-05,
+      "loss": 0.7532,
+      "step": 18264
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.962074338665663e-05,
+      "loss": 0.6759,
+      "step": 18265
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9618223948009443e-05,
+      "loss": 0.6801,
+      "step": 18266
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9615704566674898e-05,
+      "loss": 0.7151,
+      "step": 18267
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9613185242679835e-05,
+      "loss": 0.6833,
+      "step": 18268
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.961066597605108e-05,
+      "loss": 0.691,
+      "step": 18269
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9608146766815456e-05,
+      "loss": 0.6944,
+      "step": 18270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9605627614999796e-05,
+      "loss": 0.7046,
+      "step": 18271
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.960310852063093e-05,
+      "loss": 0.6259,
+      "step": 18272
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9600589483735678e-05,
+      "loss": 0.6806,
+      "step": 18273
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9598070504340885e-05,
+      "loss": 0.6816,
+      "step": 18274
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9595551582473345e-05,
+      "loss": 0.6793,
+      "step": 18275
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9593032718159904e-05,
+      "loss": 0.6688,
+      "step": 18276
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9590513911427384e-05,
+      "loss": 0.6698,
+      "step": 18277
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.95879951623026e-05,
+      "loss": 0.6613,
+      "step": 18278
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9585476470812386e-05,
+      "loss": 0.6886,
+      "step": 18279
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9582957836983563e-05,
+      "loss": 0.6689,
+      "step": 18280
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9580439260842942e-05,
+      "loss": 0.7111,
+      "step": 18281
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.957792074241736e-05,
+      "loss": 0.6675,
+      "step": 18282
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.957540228173363e-05,
+      "loss": 0.709,
+      "step": 18283
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9572883878818564e-05,
+      "loss": 0.6629,
+      "step": 18284
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9570365533699e-05,
+      "loss": 0.5901,
+      "step": 18285
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9567847246401747e-05,
+      "loss": 0.6862,
+      "step": 18286
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9565329016953615e-05,
+      "loss": 0.694,
+      "step": 18287
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9562810845381435e-05,
+      "loss": 0.6837,
+      "step": 18288
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9560292731712015e-05,
+      "loss": 0.7597,
+      "step": 18289
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9557774675972173e-05,
+      "loss": 0.6937,
+      "step": 18290
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9555256678188732e-05,
+      "loss": 0.6831,
+      "step": 18291
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9552738738388504e-05,
+      "loss": 0.7183,
+      "step": 18292
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9550220856598293e-05,
+      "loss": 0.6886,
+      "step": 18293
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.954770303284494e-05,
+      "loss": 0.7559,
+      "step": 18294
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.954518526715522e-05,
+      "loss": 0.6039,
+      "step": 18295
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.954266755955597e-05,
+      "loss": 0.6755,
+      "step": 18296
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9540149910074e-05,
+      "loss": 0.7083,
+      "step": 18297
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9537632318736114e-05,
+      "loss": 0.7498,
+      "step": 18298
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.953511478556913e-05,
+      "loss": 0.6547,
+      "step": 18299
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.953259731059986e-05,
+      "loss": 0.6596,
+      "step": 18300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9530079893855103e-05,
+      "loss": 0.6233,
+      "step": 18301
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9527562535361677e-05,
+      "loss": 0.6754,
+      "step": 18302
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.952504523514639e-05,
+      "loss": 0.6807,
+      "step": 18303
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9522527993236037e-05,
+      "loss": 0.7853,
+      "step": 18304
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9520010809657442e-05,
+      "loss": 0.6052,
+      "step": 18305
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.951749368443741e-05,
+      "loss": 0.6851,
+      "step": 18306
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9514976617602733e-05,
+      "loss": 0.6742,
+      "step": 18307
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9512459609180232e-05,
+      "loss": 0.7127,
+      "step": 18308
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9509942659196702e-05,
+      "loss": 0.6978,
+      "step": 18309
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9507425767678947e-05,
+      "loss": 0.6956,
+      "step": 18310
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9504908934653775e-05,
+      "loss": 0.7482,
+      "step": 18311
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.950239216014799e-05,
+      "loss": 0.7106,
+      "step": 18312
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9499875444188385e-05,
+      "loss": 0.6734,
+      "step": 18313
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9497358786801782e-05,
+      "loss": 0.7035,
+      "step": 18314
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.949484218801495e-05,
+      "loss": 0.6628,
+      "step": 18315
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9492325647854713e-05,
+      "loss": 0.7212,
+      "step": 18316
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9489809166347868e-05,
+      "loss": 0.679,
+      "step": 18317
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9487292743521198e-05,
+      "loss": 0.6383,
+      "step": 18318
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9484776379401522e-05,
+      "loss": 0.7509,
+      "step": 18319
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9482260074015628e-05,
+      "loss": 0.7062,
+      "step": 18320
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.947974382739031e-05,
+      "loss": 0.6675,
+      "step": 18321
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9477227639552377e-05,
+      "loss": 0.7164,
+      "step": 18322
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9474711510528616e-05,
+      "loss": 0.6628,
+      "step": 18323
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9472195440345815e-05,
+      "loss": 0.6375,
+      "step": 18324
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.946967942903078e-05,
+      "loss": 0.6662,
+      "step": 18325
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9467163476610302e-05,
+      "loss": 0.7043,
+      "step": 18326
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9464647583111174e-05,
+      "loss": 0.6842,
+      "step": 18327
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.946213174856019e-05,
+      "loss": 0.6695,
+      "step": 18328
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.945961597298414e-05,
+      "loss": 0.6416,
+      "step": 18329
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9457100256409813e-05,
+      "loss": 0.6947,
+      "step": 18330
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9454584598864008e-05,
+      "loss": 0.6821,
+      "step": 18331
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.945206900037351e-05,
+      "loss": 0.5958,
+      "step": 18332
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9449553460965107e-05,
+      "loss": 0.6803,
+      "step": 18333
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9447037980665598e-05,
+      "loss": 0.7109,
+      "step": 18334
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9444522559501752e-05,
+      "loss": 0.6662,
+      "step": 18335
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9442007197500372e-05,
+      "loss": 0.601,
+      "step": 18336
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9439491894688245e-05,
+      "loss": 0.6898,
+      "step": 18337
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9436976651092144e-05,
+      "loss": 0.7009,
+      "step": 18338
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.943446146673887e-05,
+      "loss": 0.6374,
+      "step": 18339
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9431946341655203e-05,
+      "loss": 0.6625,
+      "step": 18340
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.942943127586792e-05,
+      "loss": 0.6616,
+      "step": 18341
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.942691626940382e-05,
+      "loss": 0.6505,
+      "step": 18342
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9424401322289676e-05,
+      "loss": 0.7978,
+      "step": 18343
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.942188643455227e-05,
+      "loss": 0.677,
+      "step": 18344
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.941937160621839e-05,
+      "loss": 0.62,
+      "step": 18345
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9416856837314816e-05,
+      "loss": 0.6155,
+      "step": 18346
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9414342127868327e-05,
+      "loss": 0.7779,
+      "step": 18347
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.94118274779057e-05,
+      "loss": 0.7245,
+      "step": 18348
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.940931288745372e-05,
+      "loss": 0.6802,
+      "step": 18349
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9406798356539163e-05,
+      "loss": 0.701,
+      "step": 18350
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.94042838851888e-05,
+      "loss": 0.6679,
+      "step": 18351
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9401769473429428e-05,
+      "loss": 0.6719,
+      "step": 18352
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9399255121287806e-05,
+      "loss": 0.6662,
+      "step": 18353
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9396740828790727e-05,
+      "loss": 0.7674,
+      "step": 18354
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9394226595964944e-05,
+      "loss": 0.6401,
+      "step": 18355
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.939171242283725e-05,
+      "loss": 0.6874,
+      "step": 18356
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9389198309434416e-05,
+      "loss": 0.6908,
+      "step": 18357
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.938668425578321e-05,
+      "loss": 0.6615,
+      "step": 18358
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9384170261910405e-05,
+      "loss": 0.6554,
+      "step": 18359
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.938165632784278e-05,
+      "loss": 0.7136,
+      "step": 18360
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9379142453607106e-05,
+      "loss": 0.6431,
+      "step": 18361
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9376628639230148e-05,
+      "loss": 0.6534,
+      "step": 18362
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9374114884738686e-05,
+      "loss": 0.7008,
+      "step": 18363
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9371601190159485e-05,
+      "loss": 0.6769,
+      "step": 18364
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.936908755551931e-05,
+      "loss": 0.6772,
+      "step": 18365
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9366573980844937e-05,
+      "loss": 0.6676,
+      "step": 18366
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.936406046616313e-05,
+      "loss": 0.7188,
+      "step": 18367
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9361547011500654e-05,
+      "loss": 0.7106,
+      "step": 18368
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9359033616884286e-05,
+      "loss": 0.8027,
+      "step": 18369
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9356520282340786e-05,
+      "loss": 0.6968,
+      "step": 18370
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9354007007896906e-05,
+      "loss": 0.6667,
+      "step": 18371
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9351493793579436e-05,
+      "loss": 0.7254,
+      "step": 18372
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9348980639415125e-05,
+      "loss": 0.6676,
+      "step": 18373
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9346467545430734e-05,
+      "loss": 0.6892,
+      "step": 18374
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.934395451165305e-05,
+      "loss": 0.7092,
+      "step": 18375
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.93414415381088e-05,
+      "loss": 0.6705,
+      "step": 18376
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9338928624824765e-05,
+      "loss": 0.6709,
+      "step": 18377
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9336415771827705e-05,
+      "loss": 0.7585,
+      "step": 18378
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.933390297914437e-05,
+      "loss": 0.6884,
+      "step": 18379
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.933139024680154e-05,
+      "loss": 0.5838,
+      "step": 18380
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9328877574825958e-05,
+      "loss": 0.6649,
+      "step": 18381
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.932636496324438e-05,
+      "loss": 0.7078,
+      "step": 18382
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.932385241208358e-05,
+      "loss": 0.6529,
+      "step": 18383
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.93213399213703e-05,
+      "loss": 0.6648,
+      "step": 18384
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.93188274911313e-05,
+      "loss": 0.6235,
+      "step": 18385
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9316315121393348e-05,
+      "loss": 0.6449,
+      "step": 18386
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9313802812183186e-05,
+      "loss": 0.6728,
+      "step": 18387
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9311290563527566e-05,
+      "loss": 0.671,
+      "step": 18388
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9308778375453253e-05,
+      "loss": 0.7861,
+      "step": 18389
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9306266247986993e-05,
+      "loss": 0.6324,
+      "step": 18390
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.930375418115554e-05,
+      "loss": 0.6978,
+      "step": 18391
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.930124217498565e-05,
+      "loss": 0.6612,
+      "step": 18392
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929873022950407e-05,
+      "loss": 0.7398,
+      "step": 18393
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9296218344737547e-05,
+      "loss": 0.6493,
+      "step": 18394
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.929370652071285e-05,
+      "loss": 0.6322,
+      "step": 18395
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9291194757456702e-05,
+      "loss": 0.6386,
+      "step": 18396
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9288683054995864e-05,
+      "loss": 0.6168,
+      "step": 18397
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9286171413357092e-05,
+      "loss": 0.6683,
+      "step": 18398
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9283659832567113e-05,
+      "loss": 0.7041,
+      "step": 18399
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9281148312652694e-05,
+      "loss": 0.7566,
+      "step": 18400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9278636853640574e-05,
+      "loss": 0.6496,
+      "step": 18401
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9276125455557492e-05,
+      "loss": 0.6102,
+      "step": 18402
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9273614118430205e-05,
+      "loss": 0.6412,
+      "step": 18403
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9271102842285448e-05,
+      "loss": 0.6162,
+      "step": 18404
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9268591627149966e-05,
+      "loss": 0.6823,
+      "step": 18405
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9266080473050507e-05,
+      "loss": 0.7025,
+      "step": 18406
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9263569380013812e-05,
+      "loss": 0.6808,
+      "step": 18407
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9261058348066613e-05,
+      "loss": 0.6086,
+      "step": 18408
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9258547377235665e-05,
+      "loss": 0.6754,
+      "step": 18409
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.92560364675477e-05,
+      "loss": 0.691,
+      "step": 18410
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9253525619029455e-05,
+      "loss": 0.6417,
+      "step": 18411
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9251014831707683e-05,
+      "loss": 0.6931,
+      "step": 18412
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9248504105609106e-05,
+      "loss": 0.6908,
+      "step": 18413
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9245993440760464e-05,
+      "loss": 0.675,
+      "step": 18414
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9243482837188517e-05,
+      "loss": 0.7229,
+      "step": 18415
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9240972294919966e-05,
+      "loss": 0.6796,
+      "step": 18416
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.923846181398157e-05,
+      "loss": 0.7439,
+      "step": 18417
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9235951394400058e-05,
+      "loss": 0.8292,
+      "step": 18418
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9233441036202154e-05,
+      "loss": 0.6541,
+      "step": 18419
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9230930739414614e-05,
+      "loss": 0.5642,
+      "step": 18420
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9228420504064155e-05,
+      "loss": 0.6936,
+      "step": 18421
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.922591033017751e-05,
+      "loss": 0.6326,
+      "step": 18422
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9223400217781417e-05,
+      "loss": 0.6994,
+      "step": 18423
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9220890166902608e-05,
+      "loss": 0.6225,
+      "step": 18424
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9218380177567804e-05,
+      "loss": 0.7131,
+      "step": 18425
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9215870249803746e-05,
+      "loss": 0.7887,
+      "step": 18426
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9213360383637158e-05,
+      "loss": 0.6686,
+      "step": 18427
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9210850579094765e-05,
+      "loss": 0.6886,
+      "step": 18428
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9208340836203302e-05,
+      "loss": 0.7181,
+      "step": 18429
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9205831154989496e-05,
+      "loss": 0.685,
+      "step": 18430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9203321535480064e-05,
+      "loss": 0.7236,
+      "step": 18431
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.920081197770174e-05,
+      "loss": 0.6667,
+      "step": 18432
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9198302481681248e-05,
+      "loss": 0.7573,
+      "step": 18433
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9195793047445308e-05,
+      "loss": 0.6608,
+      "step": 18434
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.919328367502067e-05,
+      "loss": 0.6697,
+      "step": 18435
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.919077436443401e-05,
+      "loss": 0.6257,
+      "step": 18436
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9188265115712088e-05,
+      "loss": 0.655,
+      "step": 18437
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9185755928881614e-05,
+      "loss": 0.6448,
+      "step": 18438
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9183246803969303e-05,
+      "loss": 0.7695,
+      "step": 18439
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9180737741001887e-05,
+      "loss": 0.7161,
+      "step": 18440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.917822874000608e-05,
+      "loss": 0.7643,
+      "step": 18441
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9175719801008594e-05,
+      "loss": 0.8054,
+      "step": 18442
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.917321092403617e-05,
+      "loss": 0.632,
+      "step": 18443
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9170702109115507e-05,
+      "loss": 0.7436,
+      "step": 18444
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9168193356273325e-05,
+      "loss": 0.61,
+      "step": 18445
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9165684665536343e-05,
+      "loss": 0.6373,
+      "step": 18446
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.916317603693128e-05,
+      "loss": 0.7351,
+      "step": 18447
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9160667470484844e-05,
+      "loss": 0.7253,
+      "step": 18448
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9158158966223762e-05,
+      "loss": 0.6937,
+      "step": 18449
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.915565052417474e-05,
+      "loss": 0.7634,
+      "step": 18450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9153142144364487e-05,
+      "loss": 0.7406,
+      "step": 18451
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9150633826819715e-05,
+      "loss": 0.6377,
+      "step": 18452
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9148125571567152e-05,
+      "loss": 0.7479,
+      "step": 18453
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9145617378633497e-05,
+      "loss": 0.6938,
+      "step": 18454
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9143109248045465e-05,
+      "loss": 0.6754,
+      "step": 18455
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9140601179829757e-05,
+      "loss": 0.77,
+      "step": 18456
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9138093174013093e-05,
+      "loss": 0.7156,
+      "step": 18457
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9135585230622178e-05,
+      "loss": 0.6679,
+      "step": 18458
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9133077349683714e-05,
+      "loss": 0.8011,
+      "step": 18459
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.913056953122442e-05,
+      "loss": 0.6663,
+      "step": 18460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9128061775271e-05,
+      "loss": 0.7695,
+      "step": 18461
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9125554081850145e-05,
+      "loss": 0.6817,
+      "step": 18462
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9123046450988583e-05,
+      "loss": 0.6619,
+      "step": 18463
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9120538882713006e-05,
+      "loss": 0.6812,
+      "step": 18464
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.911803137705012e-05,
+      "loss": 0.654,
+      "step": 18465
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.911552393402663e-05,
+      "loss": 0.6654,
+      "step": 18466
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9113016553669234e-05,
+      "loss": 0.7042,
+      "step": 18467
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9110509236004644e-05,
+      "loss": 0.7229,
+      "step": 18468
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9108001981059547e-05,
+      "loss": 0.7052,
+      "step": 18469
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9105494788860655e-05,
+      "loss": 0.6742,
+      "step": 18470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.910298765943467e-05,
+      "loss": 0.656,
+      "step": 18471
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9100480592808276e-05,
+      "loss": 0.6906,
+      "step": 18472
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.909797358900819e-05,
+      "loss": 0.6882,
+      "step": 18473
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.90954666480611e-05,
+      "loss": 0.7232,
+      "step": 18474
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.909295976999371e-05,
+      "loss": 0.5894,
+      "step": 18475
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.909045295483271e-05,
+      "loss": 0.7007,
+      "step": 18476
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9087946202604795e-05,
+      "loss": 0.6858,
+      "step": 18477
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9085439513336666e-05,
+      "loss": 0.6913,
+      "step": 18478
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9082932887055014e-05,
+      "loss": 0.5928,
+      "step": 18479
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9080426323786532e-05,
+      "loss": 0.6345,
+      "step": 18480
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9077919823557918e-05,
+      "loss": 0.8598,
+      "step": 18481
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9075413386395864e-05,
+      "loss": 0.6992,
+      "step": 18482
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9072907012327058e-05,
+      "loss": 0.6896,
+      "step": 18483
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9070400701378195e-05,
+      "loss": 0.6317,
+      "step": 18484
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9067894453575966e-05,
+      "loss": 0.7441,
+      "step": 18485
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9065388268947055e-05,
+      "loss": 0.7144,
+      "step": 18486
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9062882147518158e-05,
+      "loss": 0.6729,
+      "step": 18487
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9060376089315964e-05,
+      "loss": 0.6846,
+      "step": 18488
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9057870094367155e-05,
+      "loss": 0.6812,
+      "step": 18489
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9055364162698426e-05,
+      "loss": 0.7591,
+      "step": 18490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9052858294336458e-05,
+      "loss": 0.6574,
+      "step": 18491
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9050352489307933e-05,
+      "loss": 0.6848,
+      "step": 18492
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9047846747639553e-05,
+      "loss": 0.6743,
+      "step": 18493
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9045341069357986e-05,
+      "loss": 0.6982,
+      "step": 18494
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9042835454489916e-05,
+      "loss": 0.6547,
+      "step": 18495
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9040329903062053e-05,
+      "loss": 0.6716,
+      "step": 18496
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9037824415101037e-05,
+      "loss": 0.7066,
+      "step": 18497
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9035318990633577e-05,
+      "loss": 0.6621,
+      "step": 18498
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.903281362968635e-05,
+      "loss": 0.6734,
+      "step": 18499
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.903030833228603e-05,
+      "loss": 0.7246,
+      "step": 18500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9027803098459305e-05,
+      "loss": 0.669,
+      "step": 18501
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9025297928232853e-05,
+      "loss": 0.6278,
+      "step": 18502
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9022792821633345e-05,
+      "loss": 0.6347,
+      "step": 18503
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.902028777868747e-05,
+      "loss": 0.6955,
+      "step": 18504
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9017782799421903e-05,
+      "loss": 0.6146,
+      "step": 18505
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.901527788386331e-05,
+      "loss": 0.6519,
+      "step": 18506
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9012773032038376e-05,
+      "loss": 0.6012,
+      "step": 18507
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9010268243973775e-05,
+      "loss": 0.6237,
+      "step": 18508
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9007763519696177e-05,
+      "loss": 0.686,
+      "step": 18509
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9005258859232266e-05,
+      "loss": 0.6125,
+      "step": 18510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9002754262608706e-05,
+      "loss": 0.6971,
+      "step": 18511
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9000249729852172e-05,
+      "loss": 0.6493,
+      "step": 18512
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8997745260989336e-05,
+      "loss": 0.7011,
+      "step": 18513
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.899524085604687e-05,
+      "loss": 0.7165,
+      "step": 18514
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.899273651505144e-05,
+      "loss": 0.7032,
+      "step": 18515
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.899023223802973e-05,
+      "loss": 0.6945,
+      "step": 18516
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8987728025008388e-05,
+      "loss": 0.6808,
+      "step": 18517
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8985223876014094e-05,
+      "loss": 0.6521,
+      "step": 18518
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8982719791073518e-05,
+      "loss": 0.706,
+      "step": 18519
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8980215770213313e-05,
+      "loss": 0.6945,
+      "step": 18520
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8977711813460162e-05,
+      "loss": 0.7026,
+      "step": 18521
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8975207920840725e-05,
+      "loss": 0.6479,
+      "step": 18522
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.897270409238166e-05,
+      "loss": 0.6759,
+      "step": 18523
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8970200328109644e-05,
+      "loss": 0.6659,
+      "step": 18524
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8967696628051332e-05,
+      "loss": 0.7422,
+      "step": 18525
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8965192992233382e-05,
+      "loss": 0.6529,
+      "step": 18526
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.896268942068247e-05,
+      "loss": 0.6761,
+      "step": 18527
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.896018591342525e-05,
+      "loss": 0.7211,
+      "step": 18528
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.895768247048838e-05,
+      "loss": 0.7134,
+      "step": 18529
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8955179091898525e-05,
+      "loss": 0.6656,
+      "step": 18530
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8952675777682342e-05,
+      "loss": 0.6576,
+      "step": 18531
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.895017252786649e-05,
+      "loss": 0.6568,
+      "step": 18532
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.894766934247763e-05,
+      "loss": 0.6803,
+      "step": 18533
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.894516622154242e-05,
+      "loss": 0.7398,
+      "step": 18534
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8942663165087504e-05,
+      "loss": 0.7417,
+      "step": 18535
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8940160173139568e-05,
+      "loss": 0.6636,
+      "step": 18536
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8937657245725226e-05,
+      "loss": 0.7192,
+      "step": 18537
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8935154382871166e-05,
+      "loss": 0.6276,
+      "step": 18538
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8932651584604026e-05,
+      "loss": 0.6765,
+      "step": 18539
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8930148850950456e-05,
+      "loss": 0.6704,
+      "step": 18540
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.892764618193712e-05,
+      "loss": 0.6489,
+      "step": 18541
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.892514357759067e-05,
+      "loss": 0.6812,
+      "step": 18542
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8922641037937748e-05,
+      "loss": 0.649,
+      "step": 18543
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.892013856300501e-05,
+      "loss": 0.6501,
+      "step": 18544
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8917636152819108e-05,
+      "loss": 0.6423,
+      "step": 18545
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.891513380740668e-05,
+      "loss": 0.6426,
+      "step": 18546
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8912631526794392e-05,
+      "loss": 0.6489,
+      "step": 18547
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8910129311008882e-05,
+      "loss": 0.6307,
+      "step": 18548
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.890762716007679e-05,
+      "loss": 0.7059,
+      "step": 18549
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8905125074024775e-05,
+      "loss": 0.6606,
+      "step": 18550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.890262305287948e-05,
+      "loss": 0.6976,
+      "step": 18551
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8900121096667543e-05,
+      "loss": 0.6923,
+      "step": 18552
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8897619205415616e-05,
+      "loss": 0.6816,
+      "step": 18553
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.889511737915034e-05,
+      "loss": 0.597,
+      "step": 18554
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.889261561789835e-05,
+      "loss": 0.6715,
+      "step": 18555
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8890113921686318e-05,
+      "loss": 0.6128,
+      "step": 18556
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8887612290540844e-05,
+      "loss": 0.6602,
+      "step": 18557
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.888511072448859e-05,
+      "loss": 0.6693,
+      "step": 18558
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.88826092235562e-05,
+      "loss": 0.6694,
+      "step": 18559
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.88801077877703e-05,
+      "loss": 0.7472,
+      "step": 18560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8877606417157543e-05,
+      "loss": 0.6124,
+      "step": 18561
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8875105111744556e-05,
+      "loss": 0.6335,
+      "step": 18562
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.887260387155798e-05,
+      "loss": 0.762,
+      "step": 18563
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.887010269662446e-05,
+      "loss": 0.7119,
+      "step": 18564
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8867601586970622e-05,
+      "loss": 0.6747,
+      "step": 18565
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.88651005426231e-05,
+      "loss": 0.7065,
+      "step": 18566
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8862599563608535e-05,
+      "loss": 0.6935,
+      "step": 18567
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8860098649953558e-05,
+      "loss": 0.6578,
+      "step": 18568
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.88575978016848e-05,
+      "loss": 0.6689,
+      "step": 18569
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8855097018828906e-05,
+      "loss": 0.7367,
+      "step": 18570
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.885259630141249e-05,
+      "loss": 0.6832,
+      "step": 18571
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8850095649462194e-05,
+      "loss": 0.6325,
+      "step": 18572
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.884759506300464e-05,
+      "loss": 0.7389,
+      "step": 18573
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.884509454206647e-05,
+      "loss": 0.6084,
+      "step": 18574
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8842594086674308e-05,
+      "loss": 0.6676,
+      "step": 18575
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8840093696854782e-05,
+      "loss": 0.7425,
+      "step": 18576
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.883759337263451e-05,
+      "loss": 0.6405,
+      "step": 18577
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8835093114040135e-05,
+      "loss": 0.6517,
+      "step": 18578
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8832592921098272e-05,
+      "loss": 0.7458,
+      "step": 18579
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8830092793835547e-05,
+      "loss": 0.6956,
+      "step": 18580
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.882759273227859e-05,
+      "loss": 0.6472,
+      "step": 18581
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8825092736454024e-05,
+      "loss": 0.6892,
+      "step": 18582
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8822592806388474e-05,
+      "loss": 0.6536,
+      "step": 18583
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.882009294210855e-05,
+      "loss": 0.7266,
+      "step": 18584
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8817593143640896e-05,
+      "loss": 0.7275,
+      "step": 18585
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.881509341101212e-05,
+      "loss": 0.7184,
+      "step": 18586
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8812593744248834e-05,
+      "loss": 0.6601,
+      "step": 18587
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.881009414337768e-05,
+      "loss": 0.6735,
+      "step": 18588
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.880759460842526e-05,
+      "loss": 0.6692,
+      "step": 18589
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8805095139418198e-05,
+      "loss": 0.6347,
+      "step": 18590
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.880259573638311e-05,
+      "loss": 0.7183,
+      "step": 18591
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.880009639934662e-05,
+      "loss": 0.6479,
+      "step": 18592
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8797597128335334e-05,
+      "loss": 0.6663,
+      "step": 18593
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8795097923375875e-05,
+      "loss": 0.6675,
+      "step": 18594
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.879259878449486e-05,
+      "loss": 0.6439,
+      "step": 18595
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.879009971171889e-05,
+      "loss": 0.6821,
+      "step": 18596
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.878760070507461e-05,
+      "loss": 0.6738,
+      "step": 18597
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8785101764588588e-05,
+      "loss": 0.7467,
+      "step": 18598
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8782602890287468e-05,
+      "loss": 0.6241,
+      "step": 18599
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8780104082197854e-05,
+      "loss": 0.6828,
+      "step": 18600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8777605340346347e-05,
+      "loss": 0.719,
+      "step": 18601
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.877510666475957e-05,
+      "loss": 0.6291,
+      "step": 18602
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.877260805546413e-05,
+      "loss": 0.6277,
+      "step": 18603
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.877010951248663e-05,
+      "loss": 0.6654,
+      "step": 18604
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.876761103585368e-05,
+      "loss": 0.7297,
+      "step": 18605
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8765112625591898e-05,
+      "loss": 0.6462,
+      "step": 18606
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8762614281727868e-05,
+      "loss": 0.7288,
+      "step": 18607
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8760116004288217e-05,
+      "loss": 0.7694,
+      "step": 18608
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8757617793299543e-05,
+      "loss": 0.7032,
+      "step": 18609
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8755119648788445e-05,
+      "loss": 0.6331,
+      "step": 18610
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8752621570781535e-05,
+      "loss": 0.6322,
+      "step": 18611
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8750123559305415e-05,
+      "loss": 0.6245,
+      "step": 18612
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8747625614386676e-05,
+      "loss": 0.6857,
+      "step": 18613
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8745127736051934e-05,
+      "loss": 0.7147,
+      "step": 18614
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8742629924327785e-05,
+      "loss": 0.6388,
+      "step": 18615
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8740132179240823e-05,
+      "loss": 0.629,
+      "step": 18616
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8737634500817667e-05,
+      "loss": 0.6691,
+      "step": 18617
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8735136889084892e-05,
+      "loss": 0.7845,
+      "step": 18618
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8732639344069104e-05,
+      "loss": 0.6815,
+      "step": 18619
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8730141865796907e-05,
+      "loss": 0.7142,
+      "step": 18620
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8727644454294883e-05,
+      "loss": 0.6668,
+      "step": 18621
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8725147109589648e-05,
+      "loss": 0.7484,
+      "step": 18622
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.872264983170778e-05,
+      "loss": 0.64,
+      "step": 18623
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.872015262067588e-05,
+      "loss": 0.6714,
+      "step": 18624
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8717655476520545e-05,
+      "loss": 0.679,
+      "step": 18625
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.871515839926837e-05,
+      "loss": 0.7579,
+      "step": 18626
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.871266138894593e-05,
+      "loss": 0.6717,
+      "step": 18627
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8710164445579836e-05,
+      "loss": 0.7104,
+      "step": 18628
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8707667569196674e-05,
+      "loss": 0.6166,
+      "step": 18629
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8705170759823028e-05,
+      "loss": 0.663,
+      "step": 18630
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8702674017485493e-05,
+      "loss": 0.6429,
+      "step": 18631
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.870017734221066e-05,
+      "loss": 0.6853,
+      "step": 18632
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8697680734025106e-05,
+      "loss": 0.7499,
+      "step": 18633
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.869518419295543e-05,
+      "loss": 0.7232,
+      "step": 18634
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.869268771902822e-05,
+      "loss": 0.6334,
+      "step": 18635
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8690191312270048e-05,
+      "loss": 0.6781,
+      "step": 18636
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.868769497270752e-05,
+      "loss": 0.6752,
+      "step": 18637
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.86851987003672e-05,
+      "loss": 0.7849,
+      "step": 18638
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.868270249527568e-05,
+      "loss": 0.6497,
+      "step": 18639
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8680206357459546e-05,
+      "loss": 0.7725,
+      "step": 18640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.867771028694537e-05,
+      "loss": 0.6472,
+      "step": 18641
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8675214283759746e-05,
+      "loss": 0.7068,
+      "step": 18642
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8672718347929253e-05,
+      "loss": 0.6345,
+      "step": 18643
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.867022247948046e-05,
+      "loss": 0.6715,
+      "step": 18644
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8667726678439962e-05,
+      "loss": 0.6549,
+      "step": 18645
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8665230944834327e-05,
+      "loss": 0.6172,
+      "step": 18646
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8662735278690136e-05,
+      "loss": 0.7713,
+      "step": 18647
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.866023968003397e-05,
+      "loss": 0.6765,
+      "step": 18648
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.86577441488924e-05,
+      "loss": 0.6859,
+      "step": 18649
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8655248685292e-05,
+      "loss": 0.6304,
+      "step": 18650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8652753289259356e-05,
+      "loss": 0.7306,
+      "step": 18651
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8650257960821038e-05,
+      "loss": 0.689,
+      "step": 18652
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.864776270000361e-05,
+      "loss": 0.6065,
+      "step": 18653
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.864526750683366e-05,
+      "loss": 0.7219,
+      "step": 18654
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.864277238133775e-05,
+      "loss": 0.6326,
+      "step": 18655
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8640277323542448e-05,
+      "loss": 0.6331,
+      "step": 18656
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.863778233347435e-05,
+      "loss": 0.7134,
+      "step": 18657
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8635287411159996e-05,
+      "loss": 0.6545,
+      "step": 18658
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8632792556625967e-05,
+      "loss": 0.6744,
+      "step": 18659
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.863029776989883e-05,
+      "loss": 0.7027,
+      "step": 18660
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8627803051005156e-05,
+      "loss": 0.7642,
+      "step": 18661
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8625308399971507e-05,
+      "loss": 0.6296,
+      "step": 18662
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8622813816824462e-05,
+      "loss": 0.6759,
+      "step": 18663
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.862031930159057e-05,
+      "loss": 0.6254,
+      "step": 18664
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8617824854296406e-05,
+      "loss": 0.7738,
+      "step": 18665
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8615330474968538e-05,
+      "loss": 0.6185,
+      "step": 18666
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8612836163633515e-05,
+      "loss": 0.7129,
+      "step": 18667
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8610341920317915e-05,
+      "loss": 0.7048,
+      "step": 18668
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8607847745048297e-05,
+      "loss": 0.696,
+      "step": 18669
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.860535363785121e-05,
+      "loss": 0.6611,
+      "step": 18670
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8602859598753234e-05,
+      "loss": 0.6109,
+      "step": 18671
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.860036562778092e-05,
+      "loss": 0.6819,
+      "step": 18672
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8597871724960816e-05,
+      "loss": 0.6449,
+      "step": 18673
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8595377890319497e-05,
+      "loss": 0.7105,
+      "step": 18674
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.859288412388352e-05,
+      "loss": 0.6143,
+      "step": 18675
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8590390425679437e-05,
+      "loss": 0.6744,
+      "step": 18676
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8587896795733807e-05,
+      "loss": 0.6928,
+      "step": 18677
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8585403234073172e-05,
+      "loss": 0.7628,
+      "step": 18678
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8582909740724108e-05,
+      "loss": 0.7088,
+      "step": 18679
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8580416315713157e-05,
+      "loss": 0.752,
+      "step": 18680
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.857792295906687e-05,
+      "loss": 0.6966,
+      "step": 18681
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8575429670811806e-05,
+      "loss": 0.6043,
+      "step": 18682
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8572936450974522e-05,
+      "loss": 0.6437,
+      "step": 18683
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8570443299581552e-05,
+      "loss": 0.7187,
+      "step": 18684
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8567950216659463e-05,
+      "loss": 0.6158,
+      "step": 18685
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8565457202234804e-05,
+      "loss": 0.709,
+      "step": 18686
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8562964256334108e-05,
+      "loss": 0.6403,
+      "step": 18687
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8560471378983944e-05,
+      "loss": 0.7438,
+      "step": 18688
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8557978570210848e-05,
+      "loss": 0.726,
+      "step": 18689
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.855548583004137e-05,
+      "loss": 0.6658,
+      "step": 18690
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.855299315850205e-05,
+      "loss": 0.6675,
+      "step": 18691
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8550500555619442e-05,
+      "loss": 0.7117,
+      "step": 18692
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8548008021420092e-05,
+      "loss": 0.6741,
+      "step": 18693
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.854551555593053e-05,
+      "loss": 0.6419,
+      "step": 18694
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.854302315917732e-05,
+      "loss": 0.6718,
+      "step": 18695
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.854053083118699e-05,
+      "loss": 0.6449,
+      "step": 18696
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8538038571986077e-05,
+      "loss": 0.6744,
+      "step": 18697
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8535546381601145e-05,
+      "loss": 0.6921,
+      "step": 18698
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.853305426005871e-05,
+      "loss": 0.6777,
+      "step": 18699
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8530562207385322e-05,
+      "loss": 0.6931,
+      "step": 18700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8528070223607517e-05,
+      "loss": 0.6829,
+      "step": 18701
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8525578308751835e-05,
+      "loss": 0.7113,
+      "step": 18702
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8523086462844817e-05,
+      "loss": 0.6957,
+      "step": 18703
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8520594685912997e-05,
+      "loss": 0.7369,
+      "step": 18704
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.85181029779829e-05,
+      "loss": 0.6552,
+      "step": 18705
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8515611339081078e-05,
+      "loss": 0.6766,
+      "step": 18706
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8513119769234058e-05,
+      "loss": 0.6132,
+      "step": 18707
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8510628268468372e-05,
+      "loss": 0.6664,
+      "step": 18708
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.850813683681056e-05,
+      "loss": 0.6746,
+      "step": 18709
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.850564547428714e-05,
+      "loss": 0.703,
+      "step": 18710
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.850315418092466e-05,
+      "loss": 0.6286,
+      "step": 18711
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8500662956749643e-05,
+      "loss": 0.7291,
+      "step": 18712
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.849817180178862e-05,
+      "loss": 0.6255,
+      "step": 18713
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8495680716068116e-05,
+      "loss": 0.7052,
+      "step": 18714
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.849318969961467e-05,
+      "loss": 0.6467,
+      "step": 18715
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.84906987524548e-05,
+      "loss": 0.645,
+      "step": 18716
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8488207874615033e-05,
+      "loss": 0.6461,
+      "step": 18717
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8485717066121917e-05,
+      "loss": 0.64,
+      "step": 18718
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.848322632700194e-05,
+      "loss": 0.6822,
+      "step": 18719
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.848073565728165e-05,
+      "loss": 0.7673,
+      "step": 18720
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.847824505698757e-05,
+      "loss": 0.703,
+      "step": 18721
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8475754526146215e-05,
+      "loss": 0.6567,
+      "step": 18722
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.847326406478412e-05,
+      "loss": 0.7017,
+      "step": 18723
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.84707736729278e-05,
+      "loss": 0.6963,
+      "step": 18724
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.846828335060377e-05,
+      "loss": 0.751,
+      "step": 18725
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.846579309783856e-05,
+      "loss": 0.6742,
+      "step": 18726
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.846330291465869e-05,
+      "loss": 0.6747,
+      "step": 18727
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8460812801090667e-05,
+      "loss": 0.6268,
+      "step": 18728
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8458322757161026e-05,
+      "loss": 0.6641,
+      "step": 18729
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8455832782896274e-05,
+      "loss": 0.6317,
+      "step": 18730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8453342878322927e-05,
+      "loss": 0.6834,
+      "step": 18731
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8450853043467506e-05,
+      "loss": 0.6432,
+      "step": 18732
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8448363278356524e-05,
+      "loss": 0.6535,
+      "step": 18733
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8445873583016493e-05,
+      "loss": 0.6428,
+      "step": 18734
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.844338395747393e-05,
+      "loss": 0.6447,
+      "step": 18735
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8440894401755355e-05,
+      "loss": 0.7046,
+      "step": 18736
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.843840491588726e-05,
+      "loss": 0.6844,
+      "step": 18737
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8435915499896187e-05,
+      "loss": 0.6756,
+      "step": 18738
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8433426153808613e-05,
+      "loss": 0.6268,
+      "step": 18739
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.843093687765107e-05,
+      "loss": 0.6568,
+      "step": 18740
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.842844767145006e-05,
+      "loss": 0.7257,
+      "step": 18741
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8425958535232087e-05,
+      "loss": 0.6608,
+      "step": 18742
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.842346946902367e-05,
+      "loss": 0.6535,
+      "step": 18743
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8420980472851312e-05,
+      "loss": 0.7507,
+      "step": 18744
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.841849154674151e-05,
+      "loss": 0.6149,
+      "step": 18745
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8416002690720786e-05,
+      "loss": 0.6609,
+      "step": 18746
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8413513904815632e-05,
+      "loss": 0.5785,
+      "step": 18747
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8411025189052555e-05,
+      "loss": 0.609,
+      "step": 18748
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.840853654345806e-05,
+      "loss": 0.6366,
+      "step": 18749
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8406047968058653e-05,
+      "loss": 0.6627,
+      "step": 18750
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8403559462880824e-05,
+      "loss": 0.6903,
+      "step": 18751
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8401071027951087e-05,
+      "loss": 0.6235,
+      "step": 18752
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8398582663295944e-05,
+      "loss": 0.6587,
+      "step": 18753
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8396094368941875e-05,
+      "loss": 0.6277,
+      "step": 18754
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8393606144915397e-05,
+      "loss": 0.689,
+      "step": 18755
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8391117991243008e-05,
+      "loss": 0.622,
+      "step": 18756
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8388629907951192e-05,
+      "loss": 0.6693,
+      "step": 18757
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8386141895066467e-05,
+      "loss": 0.7121,
+      "step": 18758
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8383653952615306e-05,
+      "loss": 0.6925,
+      "step": 18759
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.838116608062421e-05,
+      "loss": 0.6436,
+      "step": 18760
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.837867827911968e-05,
+      "loss": 0.6727,
+      "step": 18761
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8376190548128205e-05,
+      "loss": 0.7052,
+      "step": 18762
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.837370288767628e-05,
+      "loss": 0.6816,
+      "step": 18763
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.83712152977904e-05,
+      "loss": 0.7443,
+      "step": 18764
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8368727778497048e-05,
+      "loss": 0.6697,
+      "step": 18765
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.836624032982272e-05,
+      "loss": 0.7201,
+      "step": 18766
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8363752951793908e-05,
+      "loss": 0.6373,
+      "step": 18767
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8361265644437094e-05,
+      "loss": 0.6174,
+      "step": 18768
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8358778407778772e-05,
+      "loss": 0.6442,
+      "step": 18769
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.835629124184543e-05,
+      "loss": 0.6408,
+      "step": 18770
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8353804146663546e-05,
+      "loss": 0.6592,
+      "step": 18771
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.835131712225962e-05,
+      "loss": 0.6851,
+      "step": 18772
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.834883016866013e-05,
+      "loss": 0.6317,
+      "step": 18773
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8346343285891555e-05,
+      "loss": 0.6099,
+      "step": 18774
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8343856473980388e-05,
+      "loss": 0.715,
+      "step": 18775
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8341369732953112e-05,
+      "loss": 0.641,
+      "step": 18776
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.83388830628362e-05,
+      "loss": 0.6519,
+      "step": 18777
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8336396463656154e-05,
+      "loss": 0.6688,
+      "step": 18778
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8333909935439418e-05,
+      "loss": 0.6863,
+      "step": 18779
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.833142347821251e-05,
+      "loss": 0.6768,
+      "step": 18780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8328937092001887e-05,
+      "loss": 0.6037,
+      "step": 18781
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8326450776834033e-05,
+      "loss": 0.6624,
+      "step": 18782
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8323964532735427e-05,
+      "loss": 0.7053,
+      "step": 18783
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8321478359732545e-05,
+      "loss": 0.7158,
+      "step": 18784
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8318992257851862e-05,
+      "loss": 0.6466,
+      "step": 18785
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8316506227119863e-05,
+      "loss": 0.6423,
+      "step": 18786
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8314020267563013e-05,
+      "loss": 0.8118,
+      "step": 18787
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8311534379207782e-05,
+      "loss": 0.6332,
+      "step": 18788
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8309048562080654e-05,
+      "loss": 0.7463,
+      "step": 18789
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8306562816208095e-05,
+      "loss": 0.6689,
+      "step": 18790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8304077141616577e-05,
+      "loss": 0.6366,
+      "step": 18791
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8301591538332575e-05,
+      "loss": 0.6936,
+      "step": 18792
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8299106006382555e-05,
+      "loss": 0.7327,
+      "step": 18793
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8296620545792986e-05,
+      "loss": 0.7398,
+      "step": 18794
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.829413515659034e-05,
+      "loss": 0.7184,
+      "step": 18795
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8291649838801088e-05,
+      "loss": 0.6701,
+      "step": 18796
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8289164592451692e-05,
+      "loss": 0.6816,
+      "step": 18797
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.828667941756862e-05,
+      "loss": 0.6762,
+      "step": 18798
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8284194314178327e-05,
+      "loss": 0.6217,
+      "step": 18799
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8281709282307295e-05,
+      "loss": 0.636,
+      "step": 18800
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.827922432198198e-05,
+      "loss": 0.6378,
+      "step": 18801
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.827673943322884e-05,
+      "loss": 0.6744,
+      "step": 18802
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8274254616074347e-05,
+      "loss": 0.6351,
+      "step": 18803
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8271769870544958e-05,
+      "loss": 0.7205,
+      "step": 18804
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8269285196667132e-05,
+      "loss": 0.7275,
+      "step": 18805
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8266800594467336e-05,
+      "loss": 0.6892,
+      "step": 18806
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8264316063972032e-05,
+      "loss": 0.6186,
+      "step": 18807
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8261831605207663e-05,
+      "loss": 0.6651,
+      "step": 18808
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8259347218200697e-05,
+      "loss": 0.6904,
+      "step": 18809
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8256862902977594e-05,
+      "loss": 0.7286,
+      "step": 18810
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8254378659564812e-05,
+      "loss": 0.6698,
+      "step": 18811
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.825189448798879e-05,
+      "loss": 0.6548,
+      "step": 18812
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8249410388276006e-05,
+      "loss": 0.7238,
+      "step": 18813
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.82469263604529e-05,
+      "loss": 0.6624,
+      "step": 18814
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8244442404545923e-05,
+      "loss": 0.7322,
+      "step": 18815
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8241958520581538e-05,
+      "loss": 0.639,
+      "step": 18816
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.823947470858619e-05,
+      "loss": 0.707,
+      "step": 18817
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.823699096858633e-05,
+      "loss": 0.6416,
+      "step": 18818
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.823450730060842e-05,
+      "loss": 0.687,
+      "step": 18819
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8232023704678897e-05,
+      "loss": 0.7344,
+      "step": 18820
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8229540180824206e-05,
+      "loss": 0.6895,
+      "step": 18821
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8227056729070806e-05,
+      "loss": 0.6663,
+      "step": 18822
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8224573349445134e-05,
+      "loss": 0.6667,
+      "step": 18823
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.822209004197365e-05,
+      "loss": 0.6787,
+      "step": 18824
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.821960680668279e-05,
+      "loss": 0.7155,
+      "step": 18825
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8217123643598993e-05,
+      "loss": 0.6283,
+      "step": 18826
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.821464055274872e-05,
+      "loss": 0.6472,
+      "step": 18827
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8212157534158405e-05,
+      "loss": 0.6957,
+      "step": 18828
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8209674587854487e-05,
+      "loss": 0.6965,
+      "step": 18829
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8207191713863413e-05,
+      "loss": 0.6701,
+      "step": 18830
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.820470891221163e-05,
+      "loss": 0.719,
+      "step": 18831
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.820222618292556e-05,
+      "loss": 0.6479,
+      "step": 18832
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8199743526031664e-05,
+      "loss": 0.6272,
+      "step": 18833
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.819726094155637e-05,
+      "loss": 0.6119,
+      "step": 18834
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8194778429526112e-05,
+      "loss": 0.7519,
+      "step": 18835
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8192295989967333e-05,
+      "loss": 0.6521,
+      "step": 18836
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.818981362290647e-05,
+      "loss": 0.6391,
+      "step": 18837
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8187331328369957e-05,
+      "loss": 0.6295,
+      "step": 18838
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.818484910638424e-05,
+      "loss": 0.6674,
+      "step": 18839
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.818236695697573e-05,
+      "loss": 0.6321,
+      "step": 18840
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8179884880170877e-05,
+      "loss": 0.6519,
+      "step": 18841
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.817740287599611e-05,
+      "loss": 0.6641,
+      "step": 18842
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8174920944477853e-05,
+      "loss": 0.6233,
+      "step": 18843
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.817243908564255e-05,
+      "loss": 0.6982,
+      "step": 18844
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8169957299516626e-05,
+      "loss": 0.574,
+      "step": 18845
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8167475586126505e-05,
+      "loss": 0.6576,
+      "step": 18846
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8164993945498627e-05,
+      "loss": 0.6509,
+      "step": 18847
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.816251237765941e-05,
+      "loss": 0.6655,
+      "step": 18848
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8160030882635288e-05,
+      "loss": 0.7043,
+      "step": 18849
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8157549460452685e-05,
+      "loss": 0.6649,
+      "step": 18850
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8155068111138025e-05,
+      "loss": 0.6185,
+      "step": 18851
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8152586834717727e-05,
+      "loss": 0.6855,
+      "step": 18852
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8150105631218227e-05,
+      "loss": 0.6194,
+      "step": 18853
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.814762450066595e-05,
+      "loss": 0.67,
+      "step": 18854
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8145143443087303e-05,
+      "loss": 0.6673,
+      "step": 18855
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.814266245850872e-05,
+      "loss": 0.6208,
+      "step": 18856
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.814018154695662e-05,
+      "loss": 0.6676,
+      "step": 18857
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8137700708457416e-05,
+      "loss": 0.6501,
+      "step": 18858
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8135219943037553e-05,
+      "loss": 0.6275,
+      "step": 18859
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.813273925072341e-05,
+      "loss": 0.7191,
+      "step": 18860
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.813025863154143e-05,
+      "loss": 0.6752,
+      "step": 18861
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8127778085518028e-05,
+      "loss": 0.6724,
+      "step": 18862
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8125297612679608e-05,
+      "loss": 0.6358,
+      "step": 18863
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.81228172130526e-05,
+      "loss": 0.6094,
+      "step": 18864
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8120336886663415e-05,
+      "loss": 0.6958,
+      "step": 18865
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.811785663353846e-05,
+      "loss": 0.6951,
+      "step": 18866
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8115376453704162e-05,
+      "loss": 0.6296,
+      "step": 18867
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8112896347186923e-05,
+      "loss": 0.6724,
+      "step": 18868
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.811041631401315e-05,
+      "loss": 0.6175,
+      "step": 18869
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8107936354209265e-05,
+      "loss": 0.6497,
+      "step": 18870
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8105456467801675e-05,
+      "loss": 0.6641,
+      "step": 18871
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8102976654816784e-05,
+      "loss": 0.6117,
+      "step": 18872
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8100496915281007e-05,
+      "loss": 0.5997,
+      "step": 18873
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.809801724922075e-05,
+      "loss": 0.5863,
+      "step": 18874
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8095537656662415e-05,
+      "loss": 0.6787,
+      "step": 18875
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8093058137632415e-05,
+      "loss": 0.6619,
+      "step": 18876
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8090578692157153e-05,
+      "loss": 0.7572,
+      "step": 18877
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.808809932026303e-05,
+      "loss": 0.679,
+      "step": 18878
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.808562002197647e-05,
+      "loss": 0.6293,
+      "step": 18879
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.808314079732384e-05,
+      "loss": 0.6937,
+      "step": 18880
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8080661646331565e-05,
+      "loss": 0.6791,
+      "step": 18881
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.807818256902604e-05,
+      "loss": 0.6366,
+      "step": 18882
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.807570356543367e-05,
+      "loss": 0.636,
+      "step": 18883
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8073224635580856e-05,
+      "loss": 0.6414,
+      "step": 18884
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8070745779494e-05,
+      "loss": 0.6434,
+      "step": 18885
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.806826699719948e-05,
+      "loss": 0.6449,
+      "step": 18886
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8065788288723716e-05,
+      "loss": 0.6314,
+      "step": 18887
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8063309654093097e-05,
+      "loss": 0.6567,
+      "step": 18888
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8060831093334015e-05,
+      "loss": 0.6427,
+      "step": 18889
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8058352606472875e-05,
+      "loss": 0.7214,
+      "step": 18890
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8055874193536064e-05,
+      "loss": 0.6295,
+      "step": 18891
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.805339585454997e-05,
+      "loss": 0.6675,
+      "step": 18892
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8050917589541e-05,
+      "loss": 0.6883,
+      "step": 18893
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.804843939853554e-05,
+      "loss": 0.569,
+      "step": 18894
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8045961281559975e-05,
+      "loss": 0.6593,
+      "step": 18895
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8043483238640702e-05,
+      "loss": 0.6176,
+      "step": 18896
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8041005269804113e-05,
+      "loss": 0.5913,
+      "step": 18897
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8038527375076585e-05,
+      "loss": 0.6224,
+      "step": 18898
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.803604955448453e-05,
+      "loss": 0.6104,
+      "step": 18899
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8033571808054307e-05,
+      "loss": 0.6482,
+      "step": 18900
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8031094135812315e-05,
+      "loss": 0.6621,
+      "step": 18901
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.802861653778494e-05,
+      "loss": 0.6454,
+      "step": 18902
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8026139013998568e-05,
+      "loss": 0.6211,
+      "step": 18903
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8023661564479576e-05,
+      "loss": 0.6947,
+      "step": 18904
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8021184189254363e-05,
+      "loss": 0.6597,
+      "step": 18905
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.801870688834929e-05,
+      "loss": 0.6422,
+      "step": 18906
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8016229661790756e-05,
+      "loss": 0.6369,
+      "step": 18907
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.801375250960514e-05,
+      "loss": 0.6521,
+      "step": 18908
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.801127543181881e-05,
+      "loss": 0.6781,
+      "step": 18909
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.800879842845816e-05,
+      "loss": 0.6094,
+      "step": 18910
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.800632149954956e-05,
+      "loss": 0.6565,
+      "step": 18911
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8003844645119387e-05,
+      "loss": 0.6402,
+      "step": 18912
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8001367865194028e-05,
+      "loss": 0.5531,
+      "step": 18913
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.799889115979985e-05,
+      "loss": 0.6303,
+      "step": 18914
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.799641452896323e-05,
+      "loss": 0.704,
+      "step": 18915
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.799393797271054e-05,
+      "loss": 0.6956,
+      "step": 18916
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7991461491068163e-05,
+      "loss": 0.6428,
+      "step": 18917
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7988985084062462e-05,
+      "loss": 0.6217,
+      "step": 18918
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.798650875171981e-05,
+      "loss": 0.6377,
+      "step": 18919
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7984032494066593e-05,
+      "loss": 0.6333,
+      "step": 18920
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7981556311129168e-05,
+      "loss": 0.6064,
+      "step": 18921
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7979080202933902e-05,
+      "loss": 0.6549,
+      "step": 18922
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7976604169507167e-05,
+      "loss": 0.7351,
+      "step": 18923
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7974128210875338e-05,
+      "loss": 0.6799,
+      "step": 18924
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.797165232706478e-05,
+      "loss": 0.6226,
+      "step": 18925
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7969176518101856e-05,
+      "loss": 0.6463,
+      "step": 18926
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7966700784012925e-05,
+      "loss": 0.639,
+      "step": 18927
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7964225124824367e-05,
+      "loss": 0.6799,
+      "step": 18928
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7961749540562544e-05,
+      "loss": 0.6292,
+      "step": 18929
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.795927403125381e-05,
+      "loss": 0.6309,
+      "step": 18930
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7956798596924533e-05,
+      "loss": 0.6593,
+      "step": 18931
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.795432323760108e-05,
+      "loss": 0.5729,
+      "step": 18932
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7951847953309804e-05,
+      "loss": 0.6483,
+      "step": 18933
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7949372744077066e-05,
+      "loss": 0.6291,
+      "step": 18934
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.794689760992923e-05,
+      "loss": 0.6058,
+      "step": 18935
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7944422550892652e-05,
+      "loss": 0.6179,
+      "step": 18936
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.794194756699369e-05,
+      "loss": 0.6961,
+      "step": 18937
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7939472658258706e-05,
+      "loss": 0.672,
+      "step": 18938
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7936997824714042e-05,
+      "loss": 0.6452,
+      "step": 18939
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7934523066386084e-05,
+      "loss": 0.6778,
+      "step": 18940
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7932048383301143e-05,
+      "loss": 0.6929,
+      "step": 18941
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7929573775485607e-05,
+      "loss": 0.6782,
+      "step": 18942
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7927099242965814e-05,
+      "loss": 0.7187,
+      "step": 18943
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7924624785768118e-05,
+      "loss": 0.7065,
+      "step": 18944
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7922150403918875e-05,
+      "loss": 0.6567,
+      "step": 18945
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7919676097444434e-05,
+      "loss": 0.6712,
+      "step": 18946
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7917201866371138e-05,
+      "loss": 0.6261,
+      "step": 18947
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.791472771072535e-05,
+      "loss": 0.6309,
+      "step": 18948
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7912253630533406e-05,
+      "loss": 0.7455,
+      "step": 18949
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7909779625821655e-05,
+      "loss": 0.6875,
+      "step": 18950
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.790730569661645e-05,
+      "loss": 0.6522,
+      "step": 18951
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7904831842944136e-05,
+      "loss": 0.6965,
+      "step": 18952
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.790235806483105e-05,
+      "loss": 0.6944,
+      "step": 18953
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.789988436230354e-05,
+      "loss": 0.6807,
+      "step": 18954
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.789741073538796e-05,
+      "loss": 0.6073,
+      "step": 18955
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7894937184110636e-05,
+      "loss": 0.6711,
+      "step": 18956
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.789246370849792e-05,
+      "loss": 0.5869,
+      "step": 18957
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7889990308576155e-05,
+      "loss": 0.6457,
+      "step": 18958
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7887516984371667e-05,
+      "loss": 0.6871,
+      "step": 18959
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7885043735910824e-05,
+      "loss": 0.6447,
+      "step": 18960
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7882570563219923e-05,
+      "loss": 0.6054,
+      "step": 18961
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.788009746632534e-05,
+      "loss": 0.6475,
+      "step": 18962
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7877624445253388e-05,
+      "loss": 0.5785,
+      "step": 18963
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.787515150003041e-05,
+      "loss": 0.6288,
+      "step": 18964
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7872678630682747e-05,
+      "loss": 0.6635,
+      "step": 18965
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.787020583723673e-05,
+      "loss": 0.6937,
+      "step": 18966
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7867733119718687e-05,
+      "loss": 0.7566,
+      "step": 18967
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.786526047815496e-05,
+      "loss": 0.6397,
+      "step": 18968
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.786278791257188e-05,
+      "loss": 0.6254,
+      "step": 18969
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.786031542299577e-05,
+      "loss": 0.6571,
+      "step": 18970
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7857843009452967e-05,
+      "loss": 0.6448,
+      "step": 18971
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7855370671969807e-05,
+      "loss": 0.6391,
+      "step": 18972
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.78528984105726e-05,
+      "loss": 0.6536,
+      "step": 18973
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7850426225287696e-05,
+      "loss": 0.6248,
+      "step": 18974
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7847954116141407e-05,
+      "loss": 0.6326,
+      "step": 18975
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7845482083160063e-05,
+      "loss": 0.6106,
+      "step": 18976
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7843010126369996e-05,
+      "loss": 0.6131,
+      "step": 18977
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7840538245797527e-05,
+      "loss": 0.6489,
+      "step": 18978
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.783806644146897e-05,
+      "loss": 0.7062,
+      "step": 18979
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7835594713410677e-05,
+      "loss": 0.6391,
+      "step": 18980
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7833123061648935e-05,
+      "loss": 0.6718,
+      "step": 18981
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7830651486210086e-05,
+      "loss": 0.651,
+      "step": 18982
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7828179987120442e-05,
+      "loss": 0.6481,
+      "step": 18983
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7825708564406326e-05,
+      "loss": 0.6433,
+      "step": 18984
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7823237218094064e-05,
+      "loss": 0.6972,
+      "step": 18985
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7820765948209967e-05,
+      "loss": 0.6863,
+      "step": 18986
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7818294754780345e-05,
+      "loss": 0.6623,
+      "step": 18987
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7815823637831536e-05,
+      "loss": 0.6332,
+      "step": 18988
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7813352597389837e-05,
+      "loss": 0.6588,
+      "step": 18989
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.781088163348157e-05,
+      "loss": 0.6031,
+      "step": 18990
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.780841074613305e-05,
+      "loss": 0.5872,
+      "step": 18991
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.780593993537059e-05,
+      "loss": 0.7457,
+      "step": 18992
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.78034692012205e-05,
+      "loss": 0.6245,
+      "step": 18993
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7800998543709096e-05,
+      "loss": 0.6601,
+      "step": 18994
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7798527962862686e-05,
+      "loss": 0.6321,
+      "step": 18995
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7796057458707582e-05,
+      "loss": 0.7412,
+      "step": 18996
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.779358703127009e-05,
+      "loss": 0.6282,
+      "step": 18997
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7791116680576526e-05,
+      "loss": 0.5921,
+      "step": 18998
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7788646406653186e-05,
+      "loss": 0.6963,
+      "step": 18999
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.77861762095264e-05,
+      "loss": 0.6405,
+      "step": 19000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7783706089222443e-05,
+      "loss": 0.5924,
+      "step": 19001
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.778123604576764e-05,
+      "loss": 0.6153,
+      "step": 19002
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7778766079188287e-05,
+      "loss": 0.6985,
+      "step": 19003
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.777629618951069e-05,
+      "loss": 0.6994,
+      "step": 19004
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7773826376761153e-05,
+      "loss": 0.6356,
+      "step": 19005
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7771356640965986e-05,
+      "loss": 0.6346,
+      "step": 19006
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7768886982151474e-05,
+      "loss": 0.6935,
+      "step": 19007
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7766417400343925e-05,
+      "loss": 0.5939,
+      "step": 19008
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7763947895569642e-05,
+      "loss": 0.6266,
+      "step": 19009
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.776147846785492e-05,
+      "loss": 0.6147,
+      "step": 19010
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.775900911722606e-05,
+      "loss": 0.6497,
+      "step": 19011
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7756539843709354e-05,
+      "loss": 0.575,
+      "step": 19012
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.77540706473311e-05,
+      "loss": 0.6733,
+      "step": 19013
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7751601528117596e-05,
+      "loss": 0.6296,
+      "step": 19014
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.774913248609514e-05,
+      "loss": 0.6956,
+      "step": 19015
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.774666352129001e-05,
+      "loss": 0.6247,
+      "step": 19016
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.774419463372852e-05,
+      "loss": 0.6649,
+      "step": 19017
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.774172582343695e-05,
+      "loss": 0.5874,
+      "step": 19018
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7739257090441584e-05,
+      "loss": 0.6299,
+      "step": 19019
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7736788434768734e-05,
+      "loss": 0.6514,
+      "step": 19020
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7734319856444682e-05,
+      "loss": 0.6139,
+      "step": 19021
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7731851355495704e-05,
+      "loss": 0.6505,
+      "step": 19022
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7729382931948096e-05,
+      "loss": 0.6396,
+      "step": 19023
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7726914585828142e-05,
+      "loss": 0.7179,
+      "step": 19024
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7724446317162138e-05,
+      "loss": 0.6339,
+      "step": 19025
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.772197812597636e-05,
+      "loss": 0.6113,
+      "step": 19026
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.77195100122971e-05,
+      "loss": 0.6153,
+      "step": 19027
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7717041976150634e-05,
+      "loss": 0.6266,
+      "step": 19028
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7714574017563256e-05,
+      "loss": 0.6785,
+      "step": 19029
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7712106136561233e-05,
+      "loss": 0.6146,
+      "step": 19030
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7709638333170857e-05,
+      "loss": 0.6056,
+      "step": 19031
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7707170607418413e-05,
+      "loss": 0.6145,
+      "step": 19032
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7704702959330172e-05,
+      "loss": 0.628,
+      "step": 19033
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.770223538893241e-05,
+      "loss": 0.6634,
+      "step": 19034
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7699767896251415e-05,
+      "loss": 0.633,
+      "step": 19035
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7697300481313463e-05,
+      "loss": 0.6551,
+      "step": 19036
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7694833144144817e-05,
+      "loss": 0.6013,
+      "step": 19037
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7692365884771773e-05,
+      "loss": 0.6187,
+      "step": 19038
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7689898703220593e-05,
+      "loss": 0.6721,
+      "step": 19039
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7687431599517548e-05,
+      "loss": 0.6376,
+      "step": 19040
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.768496457368893e-05,
+      "loss": 0.6113,
+      "step": 19041
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.768249762576099e-05,
+      "loss": 0.6186,
+      "step": 19042
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7680030755760006e-05,
+      "loss": 0.6779,
+      "step": 19043
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.767756396371225e-05,
+      "loss": 0.6403,
+      "step": 19044
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7675097249643996e-05,
+      "loss": 0.7043,
+      "step": 19045
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.767263061358151e-05,
+      "loss": 0.6657,
+      "step": 19046
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7670164055551057e-05,
+      "loss": 0.6948,
+      "step": 19047
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7667697575578907e-05,
+      "loss": 0.7026,
+      "step": 19048
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7665231173691326e-05,
+      "loss": 0.6301,
+      "step": 19049
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.766276484991458e-05,
+      "loss": 0.6494,
+      "step": 19050
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.766029860427493e-05,
+      "loss": 0.6219,
+      "step": 19051
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7657832436798648e-05,
+      "loss": 0.5793,
+      "step": 19052
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7655366347511995e-05,
+      "loss": 0.666,
+      "step": 19053
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7652900336441224e-05,
+      "loss": 0.613,
+      "step": 19054
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.765043440361261e-05,
+      "loss": 0.6169,
+      "step": 19055
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.764796854905241e-05,
+      "loss": 0.6609,
+      "step": 19056
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7645502772786875e-05,
+      "loss": 0.7019,
+      "step": 19057
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.764303707484227e-05,
+      "loss": 0.649,
+      "step": 19058
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7640571455244863e-05,
+      "loss": 0.6574,
+      "step": 19059
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.763810591402089e-05,
+      "loss": 0.6941,
+      "step": 19060
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.763564045119664e-05,
+      "loss": 0.6669,
+      "step": 19061
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.763317506679833e-05,
+      "loss": 0.6089,
+      "step": 19062
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.763070976085224e-05,
+      "loss": 0.6069,
+      "step": 19063
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7628244533384614e-05,
+      "loss": 0.6934,
+      "step": 19064
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7625779384421704e-05,
+      "loss": 0.6178,
+      "step": 19065
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.762331431398977e-05,
+      "loss": 0.7162,
+      "step": 19066
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7620849322115063e-05,
+      "loss": 0.6839,
+      "step": 19067
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7618384408823823e-05,
+      "loss": 0.7019,
+      "step": 19068
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7615919574142314e-05,
+      "loss": 0.6628,
+      "step": 19069
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.761345481809678e-05,
+      "loss": 0.6417,
+      "step": 19070
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.761099014071346e-05,
+      "loss": 0.6139,
+      "step": 19071
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7608525542018617e-05,
+      "loss": 0.6225,
+      "step": 19072
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7606061022038487e-05,
+      "loss": 0.6691,
+      "step": 19073
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7603596580799316e-05,
+      "loss": 0.6686,
+      "step": 19074
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.760113221832735e-05,
+      "loss": 0.5736,
+      "step": 19075
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7598667934648837e-05,
+      "loss": 0.6919,
+      "step": 19076
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.759620372979002e-05,
+      "loss": 0.5782,
+      "step": 19077
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.759373960377713e-05,
+      "loss": 0.7069,
+      "step": 19078
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7591275556636426e-05,
+      "loss": 0.5963,
+      "step": 19079
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.758881158839413e-05,
+      "loss": 0.7017,
+      "step": 19080
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.758634769907651e-05,
+      "loss": 0.6383,
+      "step": 19081
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7583883888709766e-05,
+      "loss": 0.6423,
+      "step": 19082
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7581420157320167e-05,
+      "loss": 0.6484,
+      "step": 19083
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7578956504933936e-05,
+      "loss": 0.6501,
+      "step": 19084
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7576492931577306e-05,
+      "loss": 0.5995,
+      "step": 19085
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7574029437276525e-05,
+      "loss": 0.6786,
+      "step": 19086
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7571566022057825e-05,
+      "loss": 0.6261,
+      "step": 19087
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.756910268594743e-05,
+      "loss": 0.7592,
+      "step": 19088
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7566639428971588e-05,
+      "loss": 0.6531,
+      "step": 19089
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.756417625115652e-05,
+      "loss": 0.7107,
+      "step": 19090
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7561713152528452e-05,
+      "loss": 0.6443,
+      "step": 19091
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7559250133113636e-05,
+      "loss": 0.6043,
+      "step": 19092
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7556787192938285e-05,
+      "loss": 0.7034,
+      "step": 19093
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7554324332028626e-05,
+      "loss": 0.6044,
+      "step": 19094
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7551861550410897e-05,
+      "loss": 0.7507,
+      "step": 19095
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.754939884811132e-05,
+      "loss": 0.7157,
+      "step": 19096
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7546936225156118e-05,
+      "loss": 0.6687,
+      "step": 19097
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7544473681571526e-05,
+      "loss": 0.6534,
+      "step": 19098
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7542011217383765e-05,
+      "loss": 0.6527,
+      "step": 19099
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7539548832619047e-05,
+      "loss": 0.6616,
+      "step": 19100
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7537086527303624e-05,
+      "loss": 0.7137,
+      "step": 19101
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.753462430146368e-05,
+      "loss": 0.6241,
+      "step": 19102
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7532162155125466e-05,
+      "loss": 0.6628,
+      "step": 19103
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.752970008831518e-05,
+      "loss": 0.6562,
+      "step": 19104
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7527238101059058e-05,
+      "loss": 0.678,
+      "step": 19105
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7524776193383314e-05,
+      "loss": 0.674,
+      "step": 19106
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7522314365314165e-05,
+      "loss": 0.6456,
+      "step": 19107
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7519852616877826e-05,
+      "loss": 0.6479,
+      "step": 19108
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7517390948100517e-05,
+      "loss": 0.6458,
+      "step": 19109
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7514929359008453e-05,
+      "loss": 0.605,
+      "step": 19110
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.751246784962784e-05,
+      "loss": 0.6352,
+      "step": 19111
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.751000641998491e-05,
+      "loss": 0.5974,
+      "step": 19112
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.750754507010586e-05,
+      "loss": 0.6174,
+      "step": 19113
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.75050838000169e-05,
+      "loss": 0.6235,
+      "step": 19114
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.750262260974425e-05,
+      "loss": 0.6447,
+      "step": 19115
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7500161499314124e-05,
+      "loss": 0.6691,
+      "step": 19116
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7497700468752716e-05,
+      "loss": 0.6161,
+      "step": 19117
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7495239518086247e-05,
+      "loss": 0.7118,
+      "step": 19118
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7492778647340925e-05,
+      "loss": 0.6579,
+      "step": 19119
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7490317856542946e-05,
+      "loss": 0.6398,
+      "step": 19120
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.748785714571854e-05,
+      "loss": 0.601,
+      "step": 19121
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7485396514893876e-05,
+      "loss": 0.668,
+      "step": 19122
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.748293596409518e-05,
+      "loss": 0.6389,
+      "step": 19123
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7480475493348656e-05,
+      "loss": 0.6816,
+      "step": 19124
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7478015102680494e-05,
+      "loss": 0.6305,
+      "step": 19125
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7475554792116915e-05,
+      "loss": 0.644,
+      "step": 19126
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7473094561684106e-05,
+      "loss": 0.6886,
+      "step": 19127
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7470634411408264e-05,
+      "loss": 0.6269,
+      "step": 19128
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.74681743413156e-05,
+      "loss": 0.6928,
+      "step": 19129
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.746571435143231e-05,
+      "loss": 0.6972,
+      "step": 19130
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7463254441784582e-05,
+      "loss": 0.6467,
+      "step": 19131
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.746079461239862e-05,
+      "loss": 0.6716,
+      "step": 19132
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.745833486330062e-05,
+      "loss": 0.6348,
+      "step": 19133
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.745587519451677e-05,
+      "loss": 0.5981,
+      "step": 19134
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7453415606073276e-05,
+      "loss": 0.689,
+      "step": 19135
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7450956097996324e-05,
+      "loss": 0.6933,
+      "step": 19136
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.74484966703121e-05,
+      "loss": 0.5964,
+      "step": 19137
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7446037323046806e-05,
+      "loss": 0.669,
+      "step": 19138
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.744357805622663e-05,
+      "loss": 0.6251,
+      "step": 19139
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7441118869877755e-05,
+      "loss": 0.7152,
+      "step": 19140
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7438659764026375e-05,
+      "loss": 0.6424,
+      "step": 19141
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7436200738698688e-05,
+      "loss": 0.6456,
+      "step": 19142
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7433741793920865e-05,
+      "loss": 0.6261,
+      "step": 19143
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7431282929719102e-05,
+      "loss": 0.7443,
+      "step": 19144
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7428824146119567e-05,
+      "loss": 0.6169,
+      "step": 19145
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7426365443148472e-05,
+      "loss": 0.6732,
+      "step": 19146
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7423906820831983e-05,
+      "loss": 0.6748,
+      "step": 19147
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.742144827919628e-05,
+      "loss": 0.6215,
+      "step": 19148
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7418989818267563e-05,
+      "loss": 0.6038,
+      "step": 19149
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7416531438071996e-05,
+      "loss": 0.6755,
+      "step": 19150
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7414073138635766e-05,
+      "loss": 0.6435,
+      "step": 19151
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.741161491998505e-05,
+      "loss": 0.6325,
+      "step": 19152
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.740915678214603e-05,
+      "loss": 0.723,
+      "step": 19153
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.740669872514488e-05,
+      "loss": 0.6295,
+      "step": 19154
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7404240749007777e-05,
+      "loss": 0.58,
+      "step": 19155
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.74017828537609e-05,
+      "loss": 0.6406,
+      "step": 19156
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7399325039430427e-05,
+      "loss": 0.6875,
+      "step": 19157
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.739686730604252e-05,
+      "loss": 0.6239,
+      "step": 19158
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7394409653623367e-05,
+      "loss": 0.5635,
+      "step": 19159
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.739195208219913e-05,
+      "loss": 0.6202,
+      "step": 19160
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7389494591795984e-05,
+      "loss": 0.5884,
+      "step": 19161
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7387037182440108e-05,
+      "loss": 0.7431,
+      "step": 19162
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7384579854157662e-05,
+      "loss": 0.6189,
+      "step": 19163
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7382122606974816e-05,
+      "loss": 0.711,
+      "step": 19164
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7379665440917735e-05,
+      "loss": 0.6947,
+      "step": 19165
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7377208356012588e-05,
+      "loss": 0.6499,
+      "step": 19166
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.737475135228555e-05,
+      "loss": 0.688,
+      "step": 19167
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7372294429762783e-05,
+      "loss": 0.6178,
+      "step": 19168
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.736983758847044e-05,
+      "loss": 0.6295,
+      "step": 19169
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.73673808284347e-05,
+      "loss": 0.6646,
+      "step": 19170
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7364924149681722e-05,
+      "loss": 0.6494,
+      "step": 19171
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.736246755223766e-05,
+      "loss": 0.607,
+      "step": 19172
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7360011036128686e-05,
+      "loss": 0.5981,
+      "step": 19173
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.735755460138096e-05,
+      "loss": 0.6537,
+      "step": 19174
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.735509824802063e-05,
+      "loss": 0.6283,
+      "step": 19175
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.735264197607387e-05,
+      "loss": 0.6974,
+      "step": 19176
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7350185785566823e-05,
+      "loss": 0.6826,
+      "step": 19177
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7347729676525655e-05,
+      "loss": 0.6954,
+      "step": 19178
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7345273648976523e-05,
+      "loss": 0.6628,
+      "step": 19179
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.734281770294558e-05,
+      "loss": 0.6783,
+      "step": 19180
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7340361838458972e-05,
+      "loss": 0.7151,
+      "step": 19181
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.733790605554288e-05,
+      "loss": 0.6687,
+      "step": 19182
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.733545035422342e-05,
+      "loss": 0.6614,
+      "step": 19183
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7332994734526764e-05,
+      "loss": 0.6282,
+      "step": 19184
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7330539196479056e-05,
+      "loss": 0.6252,
+      "step": 19185
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.732808374010645e-05,
+      "loss": 0.6522,
+      "step": 19186
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7325628365435098e-05,
+      "loss": 0.6416,
+      "step": 19187
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7323173072491143e-05,
+      "loss": 0.6179,
+      "step": 19188
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7320717861300728e-05,
+      "loss": 0.6585,
+      "step": 19189
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7318262731890015e-05,
+      "loss": 0.6642,
+      "step": 19190
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7315807684285138e-05,
+      "loss": 0.6275,
+      "step": 19191
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7313352718512236e-05,
+      "loss": 0.5832,
+      "step": 19192
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7310897834597473e-05,
+      "loss": 0.6325,
+      "step": 19193
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7308443032566973e-05,
+      "loss": 0.6192,
+      "step": 19194
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7305988312446884e-05,
+      "loss": 0.6769,
+      "step": 19195
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.730353367426335e-05,
+      "loss": 0.6812,
+      "step": 19196
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7301079118042514e-05,
+      "loss": 0.6558,
+      "step": 19197
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.72986246438105e-05,
+      "loss": 0.6202,
+      "step": 19198
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7296170251593468e-05,
+      "loss": 0.7092,
+      "step": 19199
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.729371594141755e-05,
+      "loss": 0.6453,
+      "step": 19200
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7291261713308866e-05,
+      "loss": 0.6054,
+      "step": 19201
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.728880756729358e-05,
+      "loss": 0.6391,
+      "step": 19202
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7286353503397802e-05,
+      "loss": 0.6641,
+      "step": 19203
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7283899521647676e-05,
+      "loss": 0.6908,
+      "step": 19204
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7281445622069338e-05,
+      "loss": 0.7268,
+      "step": 19205
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.727899180468891e-05,
+      "loss": 0.7204,
+      "step": 19206
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7276538069532542e-05,
+      "loss": 0.667,
+      "step": 19207
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7274084416626353e-05,
+      "loss": 0.6681,
+      "step": 19208
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7271630845996468e-05,
+      "loss": 0.6364,
+      "step": 19209
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7269177357669024e-05,
+      "loss": 0.7006,
+      "step": 19210
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.726672395167015e-05,
+      "loss": 0.6801,
+      "step": 19211
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.726427062802597e-05,
+      "loss": 0.6484,
+      "step": 19212
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7261817386762614e-05,
+      "loss": 0.655,
+      "step": 19213
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7259364227906206e-05,
+      "loss": 0.6668,
+      "step": 19214
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.725691115148286e-05,
+      "loss": 0.6906,
+      "step": 19215
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.725445815751872e-05,
+      "loss": 0.6069,
+      "step": 19216
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7252005246039893e-05,
+      "loss": 0.6259,
+      "step": 19217
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7249552417072505e-05,
+      "loss": 0.676,
+      "step": 19218
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.724709967064268e-05,
+      "loss": 0.5999,
+      "step": 19219
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.724464700677654e-05,
+      "loss": 0.71,
+      "step": 19220
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7242194425500197e-05,
+      "loss": 0.658,
+      "step": 19221
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7239741926839788e-05,
+      "loss": 0.6567,
+      "step": 19222
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7237289510821398e-05,
+      "loss": 0.6689,
+      "step": 19223
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7234837177471164e-05,
+      "loss": 0.7406,
+      "step": 19224
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7232384926815205e-05,
+      "loss": 0.617,
+      "step": 19225
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7229932758879623e-05,
+      "loss": 0.6145,
+      "step": 19226
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.722748067369055e-05,
+      "loss": 0.6883,
+      "step": 19227
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7225028671274084e-05,
+      "loss": 0.682,
+      "step": 19228
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7222576751656337e-05,
+      "loss": 0.6762,
+      "step": 19229
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7220124914863434e-05,
+      "loss": 0.6146,
+      "step": 19230
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7217673160921477e-05,
+      "loss": 0.7904,
+      "step": 19231
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.721522148985657e-05,
+      "loss": 0.609,
+      "step": 19232
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.721276990169483e-05,
+      "loss": 0.5876,
+      "step": 19233
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7210318396462366e-05,
+      "loss": 0.5955,
+      "step": 19234
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7207866974185276e-05,
+      "loss": 0.6199,
+      "step": 19235
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7205415634889677e-05,
+      "loss": 0.616,
+      "step": 19236
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.720296437860167e-05,
+      "loss": 0.6728,
+      "step": 19237
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7200513205347354e-05,
+      "loss": 0.6168,
+      "step": 19238
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.719806211515284e-05,
+      "loss": 0.6189,
+      "step": 19239
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.719561110804423e-05,
+      "loss": 0.6379,
+      "step": 19240
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.719316018404762e-05,
+      "loss": 0.6158,
+      "step": 19241
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7190709343189116e-05,
+      "loss": 0.5769,
+      "step": 19242
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7188258585494823e-05,
+      "loss": 0.6287,
+      "step": 19243
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.718580791099083e-05,
+      "loss": 0.6177,
+      "step": 19244
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7183357319703236e-05,
+      "loss": 0.6412,
+      "step": 19245
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7180906811658137e-05,
+      "loss": 0.6993,
+      "step": 19246
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7178456386881636e-05,
+      "loss": 0.6493,
+      "step": 19247
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7176006045399827e-05,
+      "loss": 0.6866,
+      "step": 19248
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7173555787238802e-05,
+      "loss": 0.6777,
+      "step": 19249
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.717110561242466e-05,
+      "loss": 0.6708,
+      "step": 19250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.716865552098349e-05,
+      "loss": 0.6652,
+      "step": 19251
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7166205512941376e-05,
+      "loss": 0.6278,
+      "step": 19252
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7163755588324426e-05,
+      "loss": 0.6025,
+      "step": 19253
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7161305747158718e-05,
+      "loss": 0.6353,
+      "step": 19254
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.715885598947034e-05,
+      "loss": 0.7131,
+      "step": 19255
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7156406315285396e-05,
+      "loss": 0.6519,
+      "step": 19256
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.715395672462996e-05,
+      "loss": 0.6727,
+      "step": 19257
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7151507217530115e-05,
+      "loss": 0.6627,
+      "step": 19258
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7149057794011955e-05,
+      "loss": 0.6491,
+      "step": 19259
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7146608454101568e-05,
+      "loss": 0.6752,
+      "step": 19260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.714415919782503e-05,
+      "loss": 0.6436,
+      "step": 19261
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7141710025208425e-05,
+      "loss": 0.6227,
+      "step": 19262
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7139260936277847e-05,
+      "loss": 0.6611,
+      "step": 19263
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.713681193105936e-05,
+      "loss": 0.6418,
+      "step": 19264
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7134363009579054e-05,
+      "loss": 0.6498,
+      "step": 19265
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7131914171863e-05,
+      "loss": 0.6366,
+      "step": 19266
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.712946541793729e-05,
+      "loss": 0.6908,
+      "step": 19267
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7127016747827995e-05,
+      "loss": 0.687,
+      "step": 19268
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.712456816156119e-05,
+      "loss": 0.7345,
+      "step": 19269
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.712211965916295e-05,
+      "loss": 0.6452,
+      "step": 19270
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.711967124065936e-05,
+      "loss": 0.6714,
+      "step": 19271
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7117222906076483e-05,
+      "loss": 0.688,
+      "step": 19272
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.711477465544039e-05,
+      "loss": 0.7218,
+      "step": 19273
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.711232648877717e-05,
+      "loss": 0.5771,
+      "step": 19274
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.710987840611288e-05,
+      "loss": 0.6152,
+      "step": 19275
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.710743040747359e-05,
+      "loss": 0.6119,
+      "step": 19276
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7104982492885383e-05,
+      "loss": 0.6484,
+      "step": 19277
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7102534662374314e-05,
+      "loss": 0.6623,
+      "step": 19278
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7100086915966458e-05,
+      "loss": 0.6279,
+      "step": 19279
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.709763925368788e-05,
+      "loss": 0.6165,
+      "step": 19280
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7095191675564646e-05,
+      "loss": 0.6447,
+      "step": 19281
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7092744181622817e-05,
+      "loss": 0.5963,
+      "step": 19282
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.709029677188848e-05,
+      "loss": 0.6215,
+      "step": 19283
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.708784944638766e-05,
+      "loss": 0.6446,
+      "step": 19284
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.708540220514645e-05,
+      "loss": 0.6669,
+      "step": 19285
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7082955048190902e-05,
+      "loss": 0.652,
+      "step": 19286
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7080507975547066e-05,
+      "loss": 0.6025,
+      "step": 19287
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.707806098724102e-05,
+      "loss": 0.6406,
+      "step": 19288
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7075614083298814e-05,
+      "loss": 0.6179,
+      "step": 19289
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7073167263746504e-05,
+      "loss": 0.6068,
+      "step": 19290
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7070720528610152e-05,
+      "loss": 0.7008,
+      "step": 19291
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7068273877915813e-05,
+      "loss": 0.7475,
+      "step": 19292
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.706582731168954e-05,
+      "loss": 0.6015,
+      "step": 19293
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7063380829957392e-05,
+      "loss": 0.6543,
+      "step": 19294
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7060934432745417e-05,
+      "loss": 0.6602,
+      "step": 19295
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.705848812007967e-05,
+      "loss": 0.6303,
+      "step": 19296
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.705604189198621e-05,
+      "loss": 0.5932,
+      "step": 19297
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7053595748491074e-05,
+      "loss": 0.694,
+      "step": 19298
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7051149689620315e-05,
+      "loss": 0.6324,
+      "step": 19299
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7048703715399995e-05,
+      "loss": 0.6813,
+      "step": 19300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.704625782585615e-05,
+      "loss": 0.7276,
+      "step": 19301
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7043812021014827e-05,
+      "loss": 0.5859,
+      "step": 19302
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.704136630090209e-05,
+      "loss": 0.6484,
+      "step": 19303
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7038920665543956e-05,
+      "loss": 0.6785,
+      "step": 19304
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7036475114966488e-05,
+      "loss": 0.7922,
+      "step": 19305
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7034029649195723e-05,
+      "loss": 0.5761,
+      "step": 19306
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7031584268257707e-05,
+      "loss": 0.6482,
+      "step": 19307
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7029138972178477e-05,
+      "loss": 0.6313,
+      "step": 19308
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7026693760984085e-05,
+      "loss": 0.6088,
+      "step": 19309
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7024248634700558e-05,
+      "loss": 0.6761,
+      "step": 19310
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.702180359335394e-05,
+      "loss": 0.606,
+      "step": 19311
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7019358636970274e-05,
+      "loss": 0.6095,
+      "step": 19312
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.701691376557559e-05,
+      "loss": 0.6665,
+      "step": 19313
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.701446897919593e-05,
+      "loss": 0.6269,
+      "step": 19314
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7012024277857332e-05,
+      "loss": 0.6166,
+      "step": 19315
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7009579661585813e-05,
+      "loss": 0.6724,
+      "step": 19316
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.700713513040743e-05,
+      "loss": 0.7466,
+      "step": 19317
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7004690684348204e-05,
+      "loss": 0.6313,
+      "step": 19318
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.700224632343416e-05,
+      "loss": 0.6768,
+      "step": 19319
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6999802047691348e-05,
+      "loss": 0.621,
+      "step": 19320
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6997357857145785e-05,
+      "loss": 0.6358,
+      "step": 19321
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6994913751823497e-05,
+      "loss": 0.6685,
+      "step": 19322
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6992469731750532e-05,
+      "loss": 0.6827,
+      "step": 19323
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.699002579695289e-05,
+      "loss": 0.6601,
+      "step": 19324
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6987581947456612e-05,
+      "loss": 0.6364,
+      "step": 19325
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6985138183287723e-05,
+      "loss": 0.6571,
+      "step": 19326
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.698269450447224e-05,
+      "loss": 0.6492,
+      "step": 19327
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.69802509110362e-05,
+      "loss": 0.6331,
+      "step": 19328
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.697780740300562e-05,
+      "loss": 0.6653,
+      "step": 19329
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6975363980406517e-05,
+      "loss": 0.6323,
+      "step": 19330
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6972920643264917e-05,
+      "loss": 0.6937,
+      "step": 19331
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.697047739160684e-05,
+      "loss": 0.6148,
+      "step": 19332
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6968034225458297e-05,
+      "loss": 0.5744,
+      "step": 19333
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6965591144845322e-05,
+      "loss": 0.6573,
+      "step": 19334
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6963148149793923e-05,
+      "loss": 0.6742,
+      "step": 19335
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6960705240330107e-05,
+      "loss": 0.6282,
+      "step": 19336
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.695826241647991e-05,
+      "loss": 0.6182,
+      "step": 19337
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6955819678269334e-05,
+      "loss": 0.7347,
+      "step": 19338
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6953377025724387e-05,
+      "loss": 0.6597,
+      "step": 19339
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6950934458871097e-05,
+      "loss": 0.6127,
+      "step": 19340
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.694849197773547e-05,
+      "loss": 0.7009,
+      "step": 19341
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.694604958234351e-05,
+      "loss": 0.646,
+      "step": 19342
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6943607272721235e-05,
+      "loss": 0.6574,
+      "step": 19343
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.694116504889466e-05,
+      "loss": 0.6252,
+      "step": 19344
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6938722910889773e-05,
+      "loss": 0.6178,
+      "step": 19345
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6936280858732598e-05,
+      "loss": 0.6825,
+      "step": 19346
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6933838892449127e-05,
+      "loss": 0.8064,
+      "step": 19347
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6931397012065385e-05,
+      "loss": 0.6132,
+      "step": 19348
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6928955217607362e-05,
+      "loss": 0.6498,
+      "step": 19349
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6926513509101066e-05,
+      "loss": 0.6357,
+      "step": 19350
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.69240718865725e-05,
+      "loss": 0.739,
+      "step": 19351
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6921630350047668e-05,
+      "loss": 0.6316,
+      "step": 19352
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6919188899552564e-05,
+      "loss": 0.5721,
+      "step": 19353
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6916747535113192e-05,
+      "loss": 0.7713,
+      "step": 19354
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6914306256755556e-05,
+      "loss": 0.6772,
+      "step": 19355
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6911865064505642e-05,
+      "loss": 0.7052,
+      "step": 19356
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.690942395838946e-05,
+      "loss": 0.6827,
+      "step": 19357
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6906982938433003e-05,
+      "loss": 0.6659,
+      "step": 19358
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6904542004662254e-05,
+      "loss": 0.5803,
+      "step": 19359
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6902101157103227e-05,
+      "loss": 0.6422,
+      "step": 19360
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.689966039578191e-05,
+      "loss": 0.6549,
+      "step": 19361
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6897219720724282e-05,
+      "loss": 0.6368,
+      "step": 19362
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.689477913195635e-05,
+      "loss": 0.6057,
+      "step": 19363
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6892338629504107e-05,
+      "loss": 0.6304,
+      "step": 19364
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6889898213393526e-05,
+      "loss": 0.6906,
+      "step": 19365
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6887457883650608e-05,
+      "loss": 0.6477,
+      "step": 19366
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.688501764030133e-05,
+      "loss": 0.6727,
+      "step": 19367
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6882577483371698e-05,
+      "loss": 0.7262,
+      "step": 19368
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6880137412887686e-05,
+      "loss": 0.6651,
+      "step": 19369
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6877697428875276e-05,
+      "loss": 0.6329,
+      "step": 19370
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.687525753136046e-05,
+      "loss": 0.6528,
+      "step": 19371
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6872817720369218e-05,
+      "loss": 0.7354,
+      "step": 19372
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.687037799592753e-05,
+      "loss": 0.6864,
+      "step": 19373
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6867938358061385e-05,
+      "loss": 0.5984,
+      "step": 19374
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6865498806796762e-05,
+      "loss": 0.585,
+      "step": 19375
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6863059342159636e-05,
+      "loss": 0.641,
+      "step": 19376
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6860619964175982e-05,
+      "loss": 0.673,
+      "step": 19377
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6858180672871792e-05,
+      "loss": 0.7163,
+      "step": 19378
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.685574146827303e-05,
+      "loss": 0.6331,
+      "step": 19379
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6853302350405674e-05,
+      "loss": 0.6247,
+      "step": 19380
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.685086331929571e-05,
+      "loss": 0.623,
+      "step": 19381
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.68484243749691e-05,
+      "loss": 0.6396,
+      "step": 19382
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6845985517451818e-05,
+      "loss": 0.6888,
+      "step": 19383
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6843546746769852e-05,
+      "loss": 0.6408,
+      "step": 19384
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6841108062949153e-05,
+      "loss": 0.6524,
+      "step": 19385
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6838669466015698e-05,
+      "loss": 0.717,
+      "step": 19386
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6836230955995457e-05,
+      "loss": 0.7251,
+      "step": 19387
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6833792532914407e-05,
+      "loss": 0.6293,
+      "step": 19388
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6831354196798504e-05,
+      "loss": 0.6308,
+      "step": 19389
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6828915947673724e-05,
+      "loss": 0.6743,
+      "step": 19390
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.682647778556602e-05,
+      "loss": 0.6633,
+      "step": 19391
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.682403971050137e-05,
+      "loss": 0.6029,
+      "step": 19392
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6821601722505736e-05,
+      "loss": 0.5831,
+      "step": 19393
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.681916382160507e-05,
+      "loss": 0.6461,
+      "step": 19394
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.681672600782535e-05,
+      "loss": 0.6636,
+      "step": 19395
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.681428828119253e-05,
+      "loss": 0.768,
+      "step": 19396
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6811850641732565e-05,
+      "loss": 0.5996,
+      "step": 19397
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6809413089471427e-05,
+      "loss": 0.66,
+      "step": 19398
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6806975624435062e-05,
+      "loss": 0.5959,
+      "step": 19399
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.680453824664943e-05,
+      "loss": 0.6313,
+      "step": 19400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6802100956140494e-05,
+      "loss": 0.6986,
+      "step": 19401
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6799663752934207e-05,
+      "loss": 0.6019,
+      "step": 19402
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6797226637056516e-05,
+      "loss": 0.7361,
+      "step": 19403
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6794789608533396e-05,
+      "loss": 0.7301,
+      "step": 19404
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6792352667390772e-05,
+      "loss": 0.6029,
+      "step": 19405
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6789915813654612e-05,
+      "loss": 0.6563,
+      "step": 19406
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6787479047350867e-05,
+      "loss": 0.6932,
+      "step": 19407
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6785042368505476e-05,
+      "loss": 0.6163,
+      "step": 19408
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6782605777144406e-05,
+      "loss": 0.6919,
+      "step": 19409
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6780169273293592e-05,
+      "loss": 0.6658,
+      "step": 19410
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.677773285697898e-05,
+      "loss": 0.6521,
+      "step": 19411
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6775296528226525e-05,
+      "loss": 0.635,
+      "step": 19412
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6772860287062173e-05,
+      "loss": 0.6846,
+      "step": 19413
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6770424133511854e-05,
+      "loss": 0.6581,
+      "step": 19414
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.676798806760153e-05,
+      "loss": 0.7279,
+      "step": 19415
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6765552089357134e-05,
+      "loss": 0.6012,
+      "step": 19416
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.676311619880461e-05,
+      "loss": 0.6237,
+      "step": 19417
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.676068039596989e-05,
+      "loss": 0.6931,
+      "step": 19418
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6758244680878932e-05,
+      "loss": 0.6612,
+      "step": 19419
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6755809053557657e-05,
+      "loss": 0.6731,
+      "step": 19420
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.675337351403201e-05,
+      "loss": 0.6292,
+      "step": 19421
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6750938062327937e-05,
+      "loss": 0.6991,
+      "step": 19422
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6748502698471354e-05,
+      "loss": 0.609,
+      "step": 19423
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6746067422488225e-05,
+      "loss": 0.6305,
+      "step": 19424
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6743632234404456e-05,
+      "loss": 0.6437,
+      "step": 19425
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.674119713424599e-05,
+      "loss": 0.6465,
+      "step": 19426
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6738762122038765e-05,
+      "loss": 0.6418,
+      "step": 19427
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.67363271978087e-05,
+      "loss": 0.6766,
+      "step": 19428
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6733892361581737e-05,
+      "loss": 0.6558,
+      "step": 19429
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6731457613383804e-05,
+      "loss": 0.6115,
+      "step": 19430
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.672902295324082e-05,
+      "loss": 0.6481,
+      "step": 19431
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6726588381178732e-05,
+      "loss": 0.6112,
+      "step": 19432
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.672415389722345e-05,
+      "loss": 0.6584,
+      "step": 19433
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6721719501400897e-05,
+      "loss": 0.6644,
+      "step": 19434
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6719285193737018e-05,
+      "loss": 0.693,
+      "step": 19435
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6716850974257715e-05,
+      "loss": 0.6651,
+      "step": 19436
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.671441684298892e-05,
+      "loss": 0.6418,
+      "step": 19437
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.671198279995656e-05,
+      "loss": 0.6527,
+      "step": 19438
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6709548845186552e-05,
+      "loss": 0.67,
+      "step": 19439
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.670711497870481e-05,
+      "loss": 0.7464,
+      "step": 19440
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6704681200537264e-05,
+      "loss": 0.626,
+      "step": 19441
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6702247510709828e-05,
+      "loss": 0.5811,
+      "step": 19442
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6699813909248413e-05,
+      "loss": 0.6327,
+      "step": 19443
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6697380396178955e-05,
+      "loss": 0.5916,
+      "step": 19444
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.669494697152734e-05,
+      "loss": 0.6595,
+      "step": 19445
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.66925136353195e-05,
+      "loss": 0.6114,
+      "step": 19446
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.669008038758135e-05,
+      "loss": 0.6562,
+      "step": 19447
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6687647228338793e-05,
+      "loss": 0.6923,
+      "step": 19448
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.668521415761775e-05,
+      "loss": 0.6177,
+      "step": 19449
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6682781175444132e-05,
+      "loss": 0.6535,
+      "step": 19450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6680348281843838e-05,
+      "loss": 0.6058,
+      "step": 19451
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6677915476842787e-05,
+      "loss": 0.5984,
+      "step": 19452
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6675482760466885e-05,
+      "loss": 0.6471,
+      "step": 19453
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.667305013274203e-05,
+      "loss": 0.6443,
+      "step": 19454
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6670617593694143e-05,
+      "loss": 0.6644,
+      "step": 19455
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6668185143349125e-05,
+      "loss": 0.6461,
+      "step": 19456
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6665752781732868e-05,
+      "loss": 0.6342,
+      "step": 19457
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6663320508871287e-05,
+      "loss": 0.6977,
+      "step": 19458
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6660888324790286e-05,
+      "loss": 0.7051,
+      "step": 19459
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6658456229515757e-05,
+      "loss": 0.6238,
+      "step": 19460
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6656024223073608e-05,
+      "loss": 0.6256,
+      "step": 19461
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6653592305489735e-05,
+      "loss": 0.682,
+      "step": 19462
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.665116047679003e-05,
+      "loss": 0.7386,
+      "step": 19463
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.66487287370004e-05,
+      "loss": 0.644,
+      "step": 19464
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.664629708614675e-05,
+      "loss": 0.6113,
+      "step": 19465
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.664386552425496e-05,
+      "loss": 0.6528,
+      "step": 19466
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6641434051350924e-05,
+      "loss": 0.6543,
+      "step": 19467
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6639002667460536e-05,
+      "loss": 0.5916,
+      "step": 19468
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.66365713726097e-05,
+      "loss": 0.6518,
+      "step": 19469
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.66341401668243e-05,
+      "loss": 0.6134,
+      "step": 19470
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.663170905013023e-05,
+      "loss": 0.6487,
+      "step": 19471
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6629278022553374e-05,
+      "loss": 0.6759,
+      "step": 19472
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6626847084119628e-05,
+      "loss": 0.6857,
+      "step": 19473
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6624416234854874e-05,
+      "loss": 0.6309,
+      "step": 19474
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6621985474785006e-05,
+      "loss": 0.6389,
+      "step": 19475
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6619554803935908e-05,
+      "loss": 0.6356,
+      "step": 19476
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6617124222333456e-05,
+      "loss": 0.6755,
+      "step": 19477
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6614693730003548e-05,
+      "loss": 0.6642,
+      "step": 19478
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.661226332697206e-05,
+      "loss": 0.7368,
+      "step": 19479
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.660983301326487e-05,
+      "loss": 0.7105,
+      "step": 19480
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.660740278890787e-05,
+      "loss": 0.5882,
+      "step": 19481
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6604972653926936e-05,
+      "loss": 0.6001,
+      "step": 19482
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6602542608347947e-05,
+      "loss": 0.6409,
+      "step": 19483
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6600112652196775e-05,
+      "loss": 0.6049,
+      "step": 19484
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6597682785499313e-05,
+      "loss": 0.7327,
+      "step": 19485
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6595253008281426e-05,
+      "loss": 0.6291,
+      "step": 19486
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6592823320568988e-05,
+      "loss": 0.5931,
+      "step": 19487
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6590393722387872e-05,
+      "loss": 0.6895,
+      "step": 19488
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6587964213763963e-05,
+      "loss": 0.6431,
+      "step": 19489
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.658553479472313e-05,
+      "loss": 0.6608,
+      "step": 19490
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6583105465291238e-05,
+      "loss": 0.6733,
+      "step": 19491
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.658067622549417e-05,
+      "loss": 0.5874,
+      "step": 19492
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6578247075357782e-05,
+      "loss": 0.7048,
+      "step": 19493
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.657581801490795e-05,
+      "loss": 0.6681,
+      "step": 19494
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.657338904417054e-05,
+      "loss": 0.6517,
+      "step": 19495
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.657096016317143e-05,
+      "loss": 0.6325,
+      "step": 19496
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.656853137193647e-05,
+      "loss": 0.7715,
+      "step": 19497
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6566102670491528e-05,
+      "loss": 0.6263,
+      "step": 19498
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6563674058862482e-05,
+      "loss": 0.5964,
+      "step": 19499
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6561245537075182e-05,
+      "loss": 0.6222,
+      "step": 19500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.655881710515549e-05,
+      "loss": 0.684,
+      "step": 19501
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6556388763129275e-05,
+      "loss": 0.654,
+      "step": 19502
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6553960511022392e-05,
+      "loss": 0.7269,
+      "step": 19503
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6551532348860697e-05,
+      "loss": 0.6366,
+      "step": 19504
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6549104276670066e-05,
+      "loss": 0.7257,
+      "step": 19505
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6546676294476337e-05,
+      "loss": 0.693,
+      "step": 19506
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6544248402305374e-05,
+      "loss": 0.6825,
+      "step": 19507
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6541820600183033e-05,
+      "loss": 0.6445,
+      "step": 19508
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6539392888135158e-05,
+      "loss": 0.6953,
+      "step": 19509
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.653696526618762e-05,
+      "loss": 0.6585,
+      "step": 19510
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6534537734366266e-05,
+      "loss": 0.6523,
+      "step": 19511
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.653211029269694e-05,
+      "loss": 0.6709,
+      "step": 19512
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.65296829412055e-05,
+      "loss": 0.6482,
+      "step": 19513
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6527255679917796e-05,
+      "loss": 0.6722,
+      "step": 19514
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6524828508859676e-05,
+      "loss": 0.6073,
+      "step": 19515
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6522401428056983e-05,
+      "loss": 0.661,
+      "step": 19516
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.651997443753557e-05,
+      "loss": 0.6387,
+      "step": 19517
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.651754753732128e-05,
+      "loss": 0.6226,
+      "step": 19518
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6515120727439957e-05,
+      "loss": 0.593,
+      "step": 19519
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6512694007917453e-05,
+      "loss": 0.6218,
+      "step": 19520
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6510267378779598e-05,
+      "loss": 0.6402,
+      "step": 19521
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6507840840052248e-05,
+      "loss": 0.6423,
+      "step": 19522
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6505414391761237e-05,
+      "loss": 0.7458,
+      "step": 19523
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.65029880339324e-05,
+      "loss": 0.6771,
+      "step": 19524
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6500561766591593e-05,
+      "loss": 0.653,
+      "step": 19525
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6498135589764633e-05,
+      "loss": 0.6351,
+      "step": 19526
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.649570950347737e-05,
+      "loss": 0.6216,
+      "step": 19527
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6493283507755635e-05,
+      "loss": 0.6078,
+      "step": 19528
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6490857602625264e-05,
+      "loss": 0.6235,
+      "step": 19529
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6488431788112103e-05,
+      "loss": 0.7386,
+      "step": 19530
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.648600606424197e-05,
+      "loss": 0.6171,
+      "step": 19531
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6483580431040702e-05,
+      "loss": 0.7292,
+      "step": 19532
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6481154888534134e-05,
+      "loss": 0.6783,
+      "step": 19533
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.64787294367481e-05,
+      "loss": 0.6652,
+      "step": 19534
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.647630407570841e-05,
+      "loss": 0.6955,
+      "step": 19535
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6473878805440917e-05,
+      "loss": 0.656,
+      "step": 19536
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.647145362597144e-05,
+      "loss": 0.61,
+      "step": 19537
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6469028537325793e-05,
+      "loss": 0.7011,
+      "step": 19538
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6466603539529824e-05,
+      "loss": 0.6739,
+      "step": 19539
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6464178632609346e-05,
+      "loss": 0.705,
+      "step": 19540
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6461753816590177e-05,
+      "loss": 0.6999,
+      "step": 19541
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.645932909149815e-05,
+      "loss": 0.5782,
+      "step": 19542
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.645690445735908e-05,
+      "loss": 0.606,
+      "step": 19543
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.645447991419879e-05,
+      "loss": 0.635,
+      "step": 19544
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6452055462043118e-05,
+      "loss": 0.6642,
+      "step": 19545
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6449631100917847e-05,
+      "loss": 0.6208,
+      "step": 19546
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.644720683084882e-05,
+      "loss": 0.7184,
+      "step": 19547
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6444782651861848e-05,
+      "loss": 0.641,
+      "step": 19548
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.644235856398274e-05,
+      "loss": 0.6588,
+      "step": 19549
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6439934567237326e-05,
+      "loss": 0.6427,
+      "step": 19550
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6437510661651412e-05,
+      "loss": 0.636,
+      "step": 19551
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6435086847250804e-05,
+      "loss": 0.6557,
+      "step": 19552
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6432663124061326e-05,
+      "loss": 0.6663,
+      "step": 19553
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6430239492108786e-05,
+      "loss": 0.6661,
+      "step": 19554
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6427815951418984e-05,
+      "loss": 0.5927,
+      "step": 19555
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6425392502017744e-05,
+      "loss": 0.6277,
+      "step": 19556
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.642296914393087e-05,
+      "loss": 0.6177,
+      "step": 19557
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6420545877184162e-05,
+      "loss": 0.7339,
+      "step": 19558
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6418122701803434e-05,
+      "loss": 0.6676,
+      "step": 19559
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6415699617814495e-05,
+      "loss": 0.6111,
+      "step": 19560
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6413276625243135e-05,
+      "loss": 0.6523,
+      "step": 19561
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.641085372411517e-05,
+      "loss": 0.7249,
+      "step": 19562
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6408430914456396e-05,
+      "loss": 0.5861,
+      "step": 19563
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6406008196292615e-05,
+      "loss": 0.6866,
+      "step": 19564
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6403585569649636e-05,
+      "loss": 0.6581,
+      "step": 19565
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6401163034553256e-05,
+      "loss": 0.6185,
+      "step": 19566
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.639874059102926e-05,
+      "loss": 0.6375,
+      "step": 19567
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.639631823910346e-05,
+      "loss": 0.6449,
+      "step": 19568
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6393895978801637e-05,
+      "loss": 0.6681,
+      "step": 19569
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6391473810149606e-05,
+      "loss": 0.6832,
+      "step": 19570
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.638905173317315e-05,
+      "loss": 0.6806,
+      "step": 19571
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6386629747898064e-05,
+      "loss": 0.604,
+      "step": 19572
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6384207854350144e-05,
+      "loss": 0.6466,
+      "step": 19573
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6381786052555177e-05,
+      "loss": 0.6173,
+      "step": 19574
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6379364342538954e-05,
+      "loss": 0.5547,
+      "step": 19575
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6376942724327273e-05,
+      "loss": 0.6486,
+      "step": 19576
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6374521197945916e-05,
+      "loss": 0.5938,
+      "step": 19577
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6372099763420667e-05,
+      "loss": 0.6799,
+      "step": 19578
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6369678420777325e-05,
+      "loss": 0.6733,
+      "step": 19579
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.636725717004166e-05,
+      "loss": 0.6547,
+      "step": 19580
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6364836011239468e-05,
+      "loss": 0.6073,
+      "step": 19581
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.636241494439653e-05,
+      "loss": 0.6284,
+      "step": 19582
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6359993969538635e-05,
+      "loss": 0.6779,
+      "step": 19583
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6357573086691546e-05,
+      "loss": 0.6013,
+      "step": 19584
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6355152295881064e-05,
+      "loss": 0.6158,
+      "step": 19585
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.635273159713297e-05,
+      "loss": 0.6319,
+      "step": 19586
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.635031099047303e-05,
+      "loss": 0.6433,
+      "step": 19587
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6347890475927026e-05,
+      "loss": 0.7035,
+      "step": 19588
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6345470053520725e-05,
+      "loss": 0.6143,
+      "step": 19589
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6343049723279922e-05,
+      "loss": 0.6801,
+      "step": 19590
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6340629485230386e-05,
+      "loss": 0.6114,
+      "step": 19591
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.633820933939788e-05,
+      "loss": 0.6106,
+      "step": 19592
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6335789285808195e-05,
+      "loss": 0.6243,
+      "step": 19593
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.633336932448709e-05,
+      "loss": 0.6906,
+      "step": 19594
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6330949455460336e-05,
+      "loss": 0.6646,
+      "step": 19595
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6328529678753715e-05,
+      "loss": 0.6424,
+      "step": 19596
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6326109994392984e-05,
+      "loss": 0.6405,
+      "step": 19597
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6323690402403917e-05,
+      "loss": 0.6337,
+      "step": 19598
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.632127090281228e-05,
+      "loss": 0.626,
+      "step": 19599
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.631885149564384e-05,
+      "loss": 0.6634,
+      "step": 19600
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.631643218092436e-05,
+      "loss": 0.6923,
+      "step": 19601
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.63140129586796e-05,
+      "loss": 0.648,
+      "step": 19602
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6311593828935335e-05,
+      "loss": 0.6761,
+      "step": 19603
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.630917479171732e-05,
+      "loss": 0.6524,
+      "step": 19604
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6306755847051312e-05,
+      "loss": 0.6667,
+      "step": 19605
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.630433699496309e-05,
+      "loss": 0.6297,
+      "step": 19606
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6301918235478388e-05,
+      "loss": 0.583,
+      "step": 19607
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6299499568622985e-05,
+      "loss": 0.6532,
+      "step": 19608
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6297080994422615e-05,
+      "loss": 0.6142,
+      "step": 19609
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.629466251290306e-05,
+      "loss": 0.6446,
+      "step": 19610
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6292244124090062e-05,
+      "loss": 0.7325,
+      "step": 19611
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.628982582800937e-05,
+      "loss": 0.6753,
+      "step": 19612
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6287407624686753e-05,
+      "loss": 0.5968,
+      "step": 19613
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6284989514147956e-05,
+      "loss": 0.651,
+      "step": 19614
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.628257149641873e-05,
+      "loss": 0.5948,
+      "step": 19615
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6280153571524814e-05,
+      "loss": 0.6693,
+      "step": 19616
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6277735739491978e-05,
+      "loss": 0.7161,
+      "step": 19617
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.627531800034596e-05,
+      "loss": 0.681,
+      "step": 19618
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.627290035411251e-05,
+      "loss": 0.6877,
+      "step": 19619
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.627048280081737e-05,
+      "loss": 0.6959,
+      "step": 19620
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6268065340486288e-05,
+      "loss": 0.6812,
+      "step": 19621
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6265647973145007e-05,
+      "loss": 0.7289,
+      "step": 19622
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6263230698819277e-05,
+      "loss": 0.644,
+      "step": 19623
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6260813517534834e-05,
+      "loss": 0.6691,
+      "step": 19624
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.625839642931742e-05,
+      "loss": 0.7235,
+      "step": 19625
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.625597943419279e-05,
+      "loss": 0.8006,
+      "step": 19626
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6253562532186656e-05,
+      "loss": 0.6682,
+      "step": 19627
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6251145723324772e-05,
+      "loss": 0.6467,
+      "step": 19628
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.624872900763288e-05,
+      "loss": 0.7224,
+      "step": 19629
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6246312385136703e-05,
+      "loss": 0.5503,
+      "step": 19630
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.624389585586199e-05,
+      "loss": 0.6289,
+      "step": 19631
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6241479419834472e-05,
+      "loss": 0.6282,
+      "step": 19632
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6239063077079873e-05,
+      "loss": 0.6383,
+      "step": 19633
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6236646827623938e-05,
+      "loss": 0.6766,
+      "step": 19634
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6234230671492396e-05,
+      "loss": 0.6405,
+      "step": 19635
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.623181460871097e-05,
+      "loss": 0.6615,
+      "step": 19636
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.62293986393054e-05,
+      "loss": 0.5967,
+      "step": 19637
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6226982763301406e-05,
+      "loss": 0.6134,
+      "step": 19638
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6224566980724722e-05,
+      "loss": 0.6355,
+      "step": 19639
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6222151291601075e-05,
+      "loss": 0.6536,
+      "step": 19640
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6219735695956184e-05,
+      "loss": 0.6471,
+      "step": 19641
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.621732019381577e-05,
+      "loss": 0.6226,
+      "step": 19642
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6214904785205577e-05,
+      "loss": 0.6462,
+      "step": 19643
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.621248947015131e-05,
+      "loss": 0.6028,
+      "step": 19644
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.621007424867869e-05,
+      "loss": 0.5932,
+      "step": 19645
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6207659120813456e-05,
+      "loss": 0.6746,
+      "step": 19646
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.62052440865813e-05,
+      "loss": 0.6405,
+      "step": 19647
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6202829146007964e-05,
+      "loss": 0.652,
+      "step": 19648
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6200414299119155e-05,
+      "loss": 0.6078,
+      "step": 19649
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6197999545940585e-05,
+      "loss": 0.6199,
+      "step": 19650
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6195584886497982e-05,
+      "loss": 0.6248,
+      "step": 19651
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6193170320817053e-05,
+      "loss": 0.6141,
+      "step": 19652
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6190755848923512e-05,
+      "loss": 0.6067,
+      "step": 19653
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6188341470843072e-05,
+      "loss": 0.6633,
+      "step": 19654
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.618592718660145e-05,
+      "loss": 0.6075,
+      "step": 19655
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6183512996224348e-05,
+      "loss": 0.6053,
+      "step": 19656
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6181098899737483e-05,
+      "loss": 0.6175,
+      "step": 19657
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.617868489716656e-05,
+      "loss": 0.6503,
+      "step": 19658
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6176270988537283e-05,
+      "loss": 0.6486,
+      "step": 19659
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6173857173875366e-05,
+      "loss": 0.6602,
+      "step": 19660
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6171443453206517e-05,
+      "loss": 0.6209,
+      "step": 19661
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6169029826556427e-05,
+      "loss": 0.7026,
+      "step": 19662
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6166616293950813e-05,
+      "loss": 0.6134,
+      "step": 19663
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6164202855415374e-05,
+      "loss": 0.6509,
+      "step": 19664
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6161789510975802e-05,
+      "loss": 0.6706,
+      "step": 19665
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6159376260657812e-05,
+      "loss": 0.5977,
+      "step": 19666
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6156963104487105e-05,
+      "loss": 0.6715,
+      "step": 19667
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6154550042489367e-05,
+      "loss": 0.6507,
+      "step": 19668
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.61521370746903e-05,
+      "loss": 0.5595,
+      "step": 19669
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6149724201115594e-05,
+      "loss": 0.6404,
+      "step": 19670
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6147311421790963e-05,
+      "loss": 0.6584,
+      "step": 19671
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6144898736742085e-05,
+      "loss": 0.6547,
+      "step": 19672
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6142486145994655e-05,
+      "loss": 0.63,
+      "step": 19673
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.614007364957438e-05,
+      "loss": 0.6178,
+      "step": 19674
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6137661247506937e-05,
+      "loss": 0.6004,
+      "step": 19675
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6135248939818016e-05,
+      "loss": 0.5867,
+      "step": 19676
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.613283672653332e-05,
+      "loss": 0.6382,
+      "step": 19677
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6130424607678522e-05,
+      "loss": 0.5703,
+      "step": 19678
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6128012583279317e-05,
+      "loss": 0.6245,
+      "step": 19679
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6125600653361402e-05,
+      "loss": 0.6056,
+      "step": 19680
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6123188817950446e-05,
+      "loss": 0.6986,
+      "step": 19681
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6120777077072135e-05,
+      "loss": 0.5972,
+      "step": 19682
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6118365430752163e-05,
+      "loss": 0.6705,
+      "step": 19683
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.611595387901621e-05,
+      "loss": 0.6043,
+      "step": 19684
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6113542421889946e-05,
+      "loss": 0.5817,
+      "step": 19685
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6111131059399063e-05,
+      "loss": 0.6261,
+      "step": 19686
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.610871979156925e-05,
+      "loss": 0.5841,
+      "step": 19687
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6106308618426165e-05,
+      "loss": 0.6335,
+      "step": 19688
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6103897539995493e-05,
+      "loss": 0.6405,
+      "step": 19689
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6101486556302905e-05,
+      "loss": 0.5916,
+      "step": 19690
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6099075667374093e-05,
+      "loss": 0.639,
+      "step": 19691
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.609666487323471e-05,
+      "loss": 0.6154,
+      "step": 19692
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6094254173910447e-05,
+      "loss": 0.6337,
+      "step": 19693
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6091843569426974e-05,
+      "loss": 0.7094,
+      "step": 19694
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6089433059809954e-05,
+      "loss": 0.677,
+      "step": 19695
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.608702264508506e-05,
+      "loss": 0.6169,
+      "step": 19696
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6084612325277963e-05,
+      "loss": 0.6273,
+      "step": 19697
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.608220210041434e-05,
+      "loss": 0.6889,
+      "step": 19698
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.607979197051984e-05,
+      "loss": 0.6591,
+      "step": 19699
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.607738193562015e-05,
+      "loss": 0.5963,
+      "step": 19700
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6074971995740923e-05,
+      "loss": 0.5978,
+      "step": 19701
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.607256215090782e-05,
+      "loss": 0.5972,
+      "step": 19702
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6070152401146516e-05,
+      "loss": 0.6823,
+      "step": 19703
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6067742746482666e-05,
+      "loss": 0.7355,
+      "step": 19704
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6065333186941927e-05,
+      "loss": 0.6499,
+      "step": 19705
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.606292372254997e-05,
+      "loss": 0.6791,
+      "step": 19706
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6060514353332463e-05,
+      "loss": 0.7048,
+      "step": 19707
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.605810507931504e-05,
+      "loss": 0.6507,
+      "step": 19708
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6055695900523365e-05,
+      "loss": 0.6886,
+      "step": 19709
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6053286816983097e-05,
+      "loss": 0.6613,
+      "step": 19710
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6050877828719898e-05,
+      "loss": 0.6224,
+      "step": 19711
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6048468935759418e-05,
+      "loss": 0.6464,
+      "step": 19712
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.60460601381273e-05,
+      "loss": 0.6434,
+      "step": 19713
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6043651435849215e-05,
+      "loss": 0.6392,
+      "step": 19714
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6041242828950803e-05,
+      "loss": 0.66,
+      "step": 19715
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6038834317457712e-05,
+      "loss": 0.6785,
+      "step": 19716
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6036425901395596e-05,
+      "loss": 0.6895,
+      "step": 19717
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6034017580790105e-05,
+      "loss": 0.6064,
+      "step": 19718
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6031609355666884e-05,
+      "loss": 0.5774,
+      "step": 19719
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6029201226051574e-05,
+      "loss": 0.5737,
+      "step": 19720
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.602679319196983e-05,
+      "loss": 0.629,
+      "step": 19721
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.602438525344729e-05,
+      "loss": 0.6552,
+      "step": 19722
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6021977410509592e-05,
+      "loss": 0.5721,
+      "step": 19723
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6019569663182386e-05,
+      "loss": 0.627,
+      "step": 19724
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6017162011491316e-05,
+      "loss": 0.6157,
+      "step": 19725
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.601475445546201e-05,
+      "loss": 0.6139,
+      "step": 19726
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6012346995120126e-05,
+      "loss": 0.6895,
+      "step": 19727
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6009939630491282e-05,
+      "loss": 0.6598,
+      "step": 19728
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6007532361601123e-05,
+      "loss": 0.6493,
+      "step": 19729
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6005125188475282e-05,
+      "loss": 0.6182,
+      "step": 19730
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.60027181111394e-05,
+      "loss": 0.7253,
+      "step": 19731
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6000311129619107e-05,
+      "loss": 0.6228,
+      "step": 19732
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.599790424394004e-05,
+      "loss": 0.6813,
+      "step": 19733
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.599549745412782e-05,
+      "loss": 0.7018,
+      "step": 19734
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.599309076020809e-05,
+      "loss": 0.6515,
+      "step": 19735
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5990684162206476e-05,
+      "loss": 0.6421,
+      "step": 19736
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5988277660148603e-05,
+      "loss": 0.6567,
+      "step": 19737
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5985871254060102e-05,
+      "loss": 0.6465,
+      "step": 19738
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5983464943966603e-05,
+      "loss": 0.7177,
+      "step": 19739
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5981058729893722e-05,
+      "loss": 0.6824,
+      "step": 19740
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5978652611867096e-05,
+      "loss": 0.6676,
+      "step": 19741
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5976246589912346e-05,
+      "loss": 0.6772,
+      "step": 19742
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.597384066405508e-05,
+      "loss": 0.6968,
+      "step": 19743
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5971434834320938e-05,
+      "loss": 0.6517,
+      "step": 19744
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5969029100735536e-05,
+      "loss": 0.6396,
+      "step": 19745
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5966623463324485e-05,
+      "loss": 0.6259,
+      "step": 19746
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.596421792211342e-05,
+      "loss": 0.6132,
+      "step": 19747
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.596181247712794e-05,
+      "loss": 0.6593,
+      "step": 19748
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5959407128393672e-05,
+      "loss": 0.6506,
+      "step": 19749
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5957001875936235e-05,
+      "loss": 0.6386,
+      "step": 19750
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5954596719781223e-05,
+      "loss": 0.6418,
+      "step": 19751
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5952191659954274e-05,
+      "loss": 0.6292,
+      "step": 19752
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5949786696480995e-05,
+      "loss": 0.6847,
+      "step": 19753
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.594738182938698e-05,
+      "loss": 0.71,
+      "step": 19754
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5944977058697863e-05,
+      "loss": 0.6282,
+      "step": 19755
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.594257238443924e-05,
+      "loss": 0.7078,
+      "step": 19756
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5940167806636723e-05,
+      "loss": 0.6382,
+      "step": 19757
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.593776332531592e-05,
+      "loss": 0.6533,
+      "step": 19758
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.593535894050243e-05,
+      "loss": 0.6386,
+      "step": 19759
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5932954652221867e-05,
+      "loss": 0.6176,
+      "step": 19760
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5930550460499832e-05,
+      "loss": 0.62,
+      "step": 19761
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5928146365361936e-05,
+      "loss": 0.6269,
+      "step": 19762
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.592574236683376e-05,
+      "loss": 0.5979,
+      "step": 19763
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.592333846494093e-05,
+      "loss": 0.6784,
+      "step": 19764
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.592093465970903e-05,
+      "loss": 0.6358,
+      "step": 19765
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.591853095116366e-05,
+      "loss": 0.5928,
+      "step": 19766
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.591612733933044e-05,
+      "loss": 0.6045,
+      "step": 19767
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5913723824234926e-05,
+      "loss": 0.6146,
+      "step": 19768
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5911320405902748e-05,
+      "loss": 0.728,
+      "step": 19769
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5908917084359493e-05,
+      "loss": 0.7135,
+      "step": 19770
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.590651385963074e-05,
+      "loss": 0.6284,
+      "step": 19771
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.59041107317421e-05,
+      "loss": 0.6126,
+      "step": 19772
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5901707700719162e-05,
+      "loss": 0.6324,
+      "step": 19773
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5899304766587507e-05,
+      "loss": 0.6029,
+      "step": 19774
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5896901929372735e-05,
+      "loss": 0.5988,
+      "step": 19775
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.589449918910043e-05,
+      "loss": 0.6724,
+      "step": 19776
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5892096545796175e-05,
+      "loss": 0.6143,
+      "step": 19777
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.588969399948557e-05,
+      "loss": 0.6009,
+      "step": 19778
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5887291550194193e-05,
+      "loss": 0.59,
+      "step": 19779
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5884889197947623e-05,
+      "loss": 0.6809,
+      "step": 19780
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5882486942771452e-05,
+      "loss": 0.6306,
+      "step": 19781
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5880084784691258e-05,
+      "loss": 0.608,
+      "step": 19782
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5877682723732624e-05,
+      "loss": 0.6147,
+      "step": 19783
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5875280759921134e-05,
+      "loss": 0.656,
+      "step": 19784
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5872878893282364e-05,
+      "loss": 0.7074,
+      "step": 19785
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5870477123841886e-05,
+      "loss": 0.6877,
+      "step": 19786
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5868075451625292e-05,
+      "loss": 0.6696,
+      "step": 19787
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5865673876658154e-05,
+      "loss": 0.6488,
+      "step": 19788
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5863272398966038e-05,
+      "loss": 0.659,
+      "step": 19789
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5860871018574524e-05,
+      "loss": 0.6724,
+      "step": 19790
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5858469735509184e-05,
+      "loss": 0.6587,
+      "step": 19791
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5856068549795592e-05,
+      "loss": 0.6697,
+      "step": 19792
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.585366746145932e-05,
+      "loss": 0.654,
+      "step": 19793
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.585126647052593e-05,
+      "loss": 0.6308,
+      "step": 19794
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5848865577021006e-05,
+      "loss": 0.6171,
+      "step": 19795
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5846464780970106e-05,
+      "loss": 0.672,
+      "step": 19796
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.584406408239879e-05,
+      "loss": 0.7096,
+      "step": 19797
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5841663481332642e-05,
+      "loss": 0.6731,
+      "step": 19798
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5839262977797218e-05,
+      "loss": 0.6426,
+      "step": 19799
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5836862571818075e-05,
+      "loss": 0.637,
+      "step": 19800
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.583446226342079e-05,
+      "loss": 0.665,
+      "step": 19801
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5832062052630914e-05,
+      "loss": 0.6757,
+      "step": 19802
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.582966193947401e-05,
+      "loss": 0.6443,
+      "step": 19803
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5827261923975638e-05,
+      "loss": 0.659,
+      "step": 19804
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5824862006161363e-05,
+      "loss": 0.7128,
+      "step": 19805
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.582246218605673e-05,
+      "loss": 0.639,
+      "step": 19806
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.582006246368731e-05,
+      "loss": 0.669,
+      "step": 19807
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5817662839078656e-05,
+      "loss": 0.6394,
+      "step": 19808
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5815263312256314e-05,
+      "loss": 0.5935,
+      "step": 19809
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5812863883245836e-05,
+      "loss": 0.6653,
+      "step": 19810
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.581046455207278e-05,
+      "loss": 0.6478,
+      "step": 19811
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.58080653187627e-05,
+      "loss": 0.6163,
+      "step": 19812
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5805666183341146e-05,
+      "loss": 0.6575,
+      "step": 19813
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5803267145833657e-05,
+      "loss": 0.5898,
+      "step": 19814
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5800868206265796e-05,
+      "loss": 0.5966,
+      "step": 19815
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5798469364663103e-05,
+      "loss": 0.6592,
+      "step": 19816
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5796070621051124e-05,
+      "loss": 0.6789,
+      "step": 19817
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5793671975455405e-05,
+      "loss": 0.6781,
+      "step": 19818
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5791273427901494e-05,
+      "loss": 0.5986,
+      "step": 19819
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5788874978414923e-05,
+      "loss": 0.6266,
+      "step": 19820
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5786476627021245e-05,
+      "loss": 0.6188,
+      "step": 19821
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5784078373746e-05,
+      "loss": 0.6706,
+      "step": 19822
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5781680218614718e-05,
+      "loss": 0.7299,
+      "step": 19823
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5779282161652952e-05,
+      "loss": 0.6095,
+      "step": 19824
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.577688420288623e-05,
+      "loss": 0.5916,
+      "step": 19825
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5774486342340093e-05,
+      "loss": 0.6781,
+      "step": 19826
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5772088580040073e-05,
+      "loss": 0.709,
+      "step": 19827
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5769690916011717e-05,
+      "loss": 0.6214,
+      "step": 19828
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5767293350280543e-05,
+      "loss": 0.674,
+      "step": 19829
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.576489588287209e-05,
+      "loss": 0.6374,
+      "step": 19830
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.576249851381188e-05,
+      "loss": 0.6543,
+      "step": 19831
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5760101243125464e-05,
+      "loss": 0.6114,
+      "step": 19832
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5757704070838356e-05,
+      "loss": 0.6212,
+      "step": 19833
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5755306996976083e-05,
+      "loss": 0.6443,
+      "step": 19834
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5752910021564185e-05,
+      "loss": 0.6267,
+      "step": 19835
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5750513144628182e-05,
+      "loss": 0.6047,
+      "step": 19836
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.574811636619359e-05,
+      "loss": 0.5865,
+      "step": 19837
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5745719686285948e-05,
+      "loss": 0.5766,
+      "step": 19838
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5743323104930775e-05,
+      "loss": 0.6371,
+      "step": 19839
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.574092662215359e-05,
+      "loss": 0.5958,
+      "step": 19840
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.573853023797991e-05,
+      "loss": 0.7081,
+      "step": 19841
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5736133952435264e-05,
+      "loss": 0.7376,
+      "step": 19842
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.573373776554517e-05,
+      "loss": 0.6054,
+      "step": 19843
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5731341677335136e-05,
+      "loss": 0.6332,
+      "step": 19844
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.572894568783069e-05,
+      "loss": 0.6578,
+      "step": 19845
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5726549797057343e-05,
+      "loss": 0.6514,
+      "step": 19846
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.572415400504061e-05,
+      "loss": 0.7511,
+      "step": 19847
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5721758311806012e-05,
+      "loss": 0.6291,
+      "step": 19848
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.571936271737905e-05,
+      "loss": 0.6493,
+      "step": 19849
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5716967221785245e-05,
+      "loss": 0.586,
+      "step": 19850
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5714571825050102e-05,
+      "loss": 0.6278,
+      "step": 19851
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.571217652719912e-05,
+      "loss": 0.63,
+      "step": 19852
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5709781328257832e-05,
+      "loss": 0.6381,
+      "step": 19853
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5707386228251727e-05,
+      "loss": 0.6831,
+      "step": 19854
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5704991227206317e-05,
+      "loss": 0.699,
+      "step": 19855
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5702596325147105e-05,
+      "loss": 0.5871,
+      "step": 19856
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5700201522099605e-05,
+      "loss": 0.6978,
+      "step": 19857
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5697806818089307e-05,
+      "loss": 0.6538,
+      "step": 19858
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5695412213141718e-05,
+      "loss": 0.6908,
+      "step": 19859
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5693017707282347e-05,
+      "loss": 0.7007,
+      "step": 19860
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5690623300536676e-05,
+      "loss": 0.621,
+      "step": 19861
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5688228992930225e-05,
+      "loss": 0.6579,
+      "step": 19862
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.568583478448848e-05,
+      "loss": 0.6414,
+      "step": 19863
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5683440675236934e-05,
+      "loss": 0.6606,
+      "step": 19864
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5681046665201093e-05,
+      "loss": 0.6066,
+      "step": 19865
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.567865275440645e-05,
+      "loss": 0.6239,
+      "step": 19866
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5676258942878492e-05,
+      "loss": 0.6348,
+      "step": 19867
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5673865230642725e-05,
+      "loss": 0.6501,
+      "step": 19868
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.567147161772462e-05,
+      "loss": 0.635,
+      "step": 19869
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5669078104149686e-05,
+      "loss": 0.6105,
+      "step": 19870
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.56666846899434e-05,
+      "loss": 0.6209,
+      "step": 19871
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5664291375131255e-05,
+      "loss": 0.6504,
+      "step": 19872
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5661898159738743e-05,
+      "loss": 0.7228,
+      "step": 19873
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5659505043791344e-05,
+      "loss": 0.6257,
+      "step": 19874
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.565711202731454e-05,
+      "loss": 0.6046,
+      "step": 19875
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.565471911033383e-05,
+      "loss": 0.6803,
+      "step": 19876
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5652326292874686e-05,
+      "loss": 0.6185,
+      "step": 19877
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5649933574962588e-05,
+      "loss": 0.6584,
+      "step": 19878
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5647540956623024e-05,
+      "loss": 0.64,
+      "step": 19879
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.564514843788147e-05,
+      "loss": 0.6481,
+      "step": 19880
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.56427560187634e-05,
+      "loss": 0.6478,
+      "step": 19881
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5640363699294307e-05,
+      "loss": 0.6369,
+      "step": 19882
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5637971479499655e-05,
+      "loss": 0.6373,
+      "step": 19883
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5635579359404918e-05,
+      "loss": 0.6192,
+      "step": 19884
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5633187339035582e-05,
+      "loss": 0.6401,
+      "step": 19885
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5630795418417114e-05,
+      "loss": 0.6496,
+      "step": 19886
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.562840359757498e-05,
+      "loss": 0.5732,
+      "step": 19887
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5626011876534664e-05,
+      "loss": 0.6543,
+      "step": 19888
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5623620255321635e-05,
+      "loss": 0.6511,
+      "step": 19889
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5621228733961352e-05,
+      "loss": 0.6866,
+      "step": 19890
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.561883731247929e-05,
+      "loss": 0.6573,
+      "step": 19891
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.561644599090091e-05,
+      "loss": 0.5774,
+      "step": 19892
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.561405476925169e-05,
+      "loss": 0.6555,
+      "step": 19893
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5611663647557085e-05,
+      "loss": 0.7227,
+      "step": 19894
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5609272625842558e-05,
+      "loss": 0.6003,
+      "step": 19895
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5606881704133586e-05,
+      "loss": 0.5848,
+      "step": 19896
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5604490882455616e-05,
+      "loss": 0.6423,
+      "step": 19897
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.560210016083411e-05,
+      "loss": 0.6805,
+      "step": 19898
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5599709539294538e-05,
+      "loss": 0.6506,
+      "step": 19899
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.559731901786235e-05,
+      "loss": 0.6584,
+      "step": 19900
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5594928596563003e-05,
+      "loss": 0.6603,
+      "step": 19901
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5592538275421962e-05,
+      "loss": 0.6484,
+      "step": 19902
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5590148054464676e-05,
+      "loss": 0.6864,
+      "step": 19903
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5587757933716597e-05,
+      "loss": 0.7108,
+      "step": 19904
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5585367913203183e-05,
+      "loss": 0.6028,
+      "step": 19905
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5582977992949887e-05,
+      "loss": 0.61,
+      "step": 19906
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.558058817298215e-05,
+      "loss": 0.5746,
+      "step": 19907
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5578198453325444e-05,
+      "loss": 0.6521,
+      "step": 19908
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.55758088340052e-05,
+      "loss": 0.6611,
+      "step": 19909
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5573419315046872e-05,
+      "loss": 0.7214,
+      "step": 19910
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5571029896475903e-05,
+      "loss": 0.6584,
+      "step": 19911
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5568640578317733e-05,
+      "loss": 0.6768,
+      "step": 19912
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5566251360597822e-05,
+      "loss": 0.6152,
+      "step": 19913
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.556386224334161e-05,
+      "loss": 0.6749,
+      "step": 19914
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5561473226574523e-05,
+      "loss": 0.6114,
+      "step": 19915
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.555908431032203e-05,
+      "loss": 0.6454,
+      "step": 19916
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.555669549460955e-05,
+      "loss": 0.6187,
+      "step": 19917
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.555430677946253e-05,
+      "loss": 0.5391,
+      "step": 19918
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5551918164906414e-05,
+      "loss": 0.5883,
+      "step": 19919
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5549529650966633e-05,
+      "loss": 0.6195,
+      "step": 19920
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5547141237668613e-05,
+      "loss": 0.6552,
+      "step": 19921
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.554475292503781e-05,
+      "loss": 0.5483,
+      "step": 19922
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5542364713099647e-05,
+      "loss": 0.7283,
+      "step": 19923
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5539976601879554e-05,
+      "loss": 0.6758,
+      "step": 19924
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.553758859140297e-05,
+      "loss": 0.6514,
+      "step": 19925
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.553520068169532e-05,
+      "loss": 0.6772,
+      "step": 19926
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5532812872782038e-05,
+      "loss": 0.6222,
+      "step": 19927
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.553042516468855e-05,
+      "loss": 0.5741,
+      "step": 19928
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5528037557440296e-05,
+      "loss": 0.6696,
+      "step": 19929
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5525650051062683e-05,
+      "loss": 0.6174,
+      "step": 19930
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5523262645581143e-05,
+      "loss": 0.6875,
+      "step": 19931
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5520875341021103e-05,
+      "loss": 0.6646,
+      "step": 19932
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5518488137407983e-05,
+      "loss": 0.6081,
+      "step": 19933
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5516101034767214e-05,
+      "loss": 0.6171,
+      "step": 19934
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5513714033124204e-05,
+      "loss": 0.5968,
+      "step": 19935
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5511327132504385e-05,
+      "loss": 0.6913,
+      "step": 19936
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5508940332933174e-05,
+      "loss": 0.65,
+      "step": 19937
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5506553634435978e-05,
+      "loss": 0.5819,
+      "step": 19938
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.550416703703823e-05,
+      "loss": 0.6023,
+      "step": 19939
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5501780540765335e-05,
+      "loss": 0.6781,
+      "step": 19940
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5499394145642707e-05,
+      "loss": 0.567,
+      "step": 19941
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.549700785169577e-05,
+      "loss": 0.5401,
+      "step": 19942
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5494621658949925e-05,
+      "loss": 0.5764,
+      "step": 19943
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.549223556743059e-05,
+      "loss": 0.584,
+      "step": 19944
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5489849577163174e-05,
+      "loss": 0.6016,
+      "step": 19945
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5487463688173086e-05,
+      "loss": 0.634,
+      "step": 19946
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.548507790048574e-05,
+      "loss": 0.6495,
+      "step": 19947
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.548269221412653e-05,
+      "loss": 0.5966,
+      "step": 19948
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.548030662912088e-05,
+      "loss": 0.6165,
+      "step": 19949
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.547792114549418e-05,
+      "loss": 0.6351,
+      "step": 19950
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5475535763271838e-05,
+      "loss": 0.6356,
+      "step": 19951
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.547315048247925e-05,
+      "loss": 0.6614,
+      "step": 19952
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.547076530314183e-05,
+      "loss": 0.7017,
+      "step": 19953
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5468380225284978e-05,
+      "loss": 0.6515,
+      "step": 19954
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5465995248934083e-05,
+      "loss": 0.6977,
+      "step": 19955
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.546361037411456e-05,
+      "loss": 0.5805,
+      "step": 19956
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.546122560085179e-05,
+      "loss": 0.6868,
+      "step": 19957
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.545884092917118e-05,
+      "loss": 0.5907,
+      "step": 19958
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5456456359098117e-05,
+      "loss": 0.6038,
+      "step": 19959
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5454071890658006e-05,
+      "loss": 0.6152,
+      "step": 19960
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5451687523876228e-05,
+      "loss": 0.6641,
+      "step": 19961
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.544930325877818e-05,
+      "loss": 0.6122,
+      "step": 19962
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.544691909538926e-05,
+      "loss": 0.6506,
+      "step": 19963
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5444535033734852e-05,
+      "loss": 0.5814,
+      "step": 19964
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.544215107384034e-05,
+      "loss": 0.6482,
+      "step": 19965
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5439767215731117e-05,
+      "loss": 0.6592,
+      "step": 19966
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.543738345943257e-05,
+      "loss": 0.7157,
+      "step": 19967
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5434999804970078e-05,
+      "loss": 0.6824,
+      "step": 19968
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5432616252369044e-05,
+      "loss": 0.6503,
+      "step": 19969
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5430232801654832e-05,
+      "loss": 0.6344,
+      "step": 19970
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5427849452852832e-05,
+      "loss": 0.7808,
+      "step": 19971
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.542546620598842e-05,
+      "loss": 0.6414,
+      "step": 19972
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5423083061086978e-05,
+      "loss": 0.7493,
+      "step": 19973
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5420700018173894e-05,
+      "loss": 0.7108,
+      "step": 19974
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.541831707727454e-05,
+      "loss": 0.6928,
+      "step": 19975
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.541593423841428e-05,
+      "loss": 0.6694,
+      "step": 19976
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5413551501618512e-05,
+      "loss": 0.6102,
+      "step": 19977
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5411168866912602e-05,
+      "loss": 0.616,
+      "step": 19978
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.540878633432191e-05,
+      "loss": 0.5874,
+      "step": 19979
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5406403903871833e-05,
+      "loss": 0.6734,
+      "step": 19980
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5404021575587725e-05,
+      "loss": 0.6899,
+      "step": 19981
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.540163934949496e-05,
+      "loss": 0.672,
+      "step": 19982
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5399257225618914e-05,
+      "loss": 0.6488,
+      "step": 19983
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5396875203984944e-05,
+      "loss": 0.6553,
+      "step": 19984
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5394493284618423e-05,
+      "loss": 0.6212,
+      "step": 19985
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.539211146754472e-05,
+      "loss": 0.63,
+      "step": 19986
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5389729752789197e-05,
+      "loss": 0.6068,
+      "step": 19987
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.538734814037721e-05,
+      "loss": 0.6379,
+      "step": 19988
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.538496663033414e-05,
+      "loss": 0.593,
+      "step": 19989
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.538258522268534e-05,
+      "loss": 0.6774,
+      "step": 19990
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5380203917456165e-05,
+      "loss": 0.6463,
+      "step": 19991
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5377822714671975e-05,
+      "loss": 0.6227,
+      "step": 19992
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5375441614358127e-05,
+      "loss": 0.6111,
+      "step": 19993
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5373060616539993e-05,
+      "loss": 0.652,
+      "step": 19994
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.537067972124291e-05,
+      "loss": 0.63,
+      "step": 19995
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5368298928492242e-05,
+      "loss": 0.6507,
+      "step": 19996
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5365918238313347e-05,
+      "loss": 0.646,
+      "step": 19997
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5363537650731575e-05,
+      "loss": 0.6402,
+      "step": 19998
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.536115716577227e-05,
+      "loss": 0.6427,
+      "step": 19999
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5358776783460794e-05,
+      "loss": 0.6805,
+      "step": 20000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.535639650382249e-05,
+      "loss": 0.72,
+      "step": 20001
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.535401632688271e-05,
+      "loss": 0.638,
+      "step": 20002
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.53516362526668e-05,
+      "loss": 0.6833,
+      "step": 20003
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5349256281200107e-05,
+      "loss": 0.697,
+      "step": 20004
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.534687641250797e-05,
+      "loss": 0.651,
+      "step": 20005
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5344496646615744e-05,
+      "loss": 0.6866,
+      "step": 20006
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5342116983548767e-05,
+      "loss": 0.6149,
+      "step": 20007
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5339737423332375e-05,
+      "loss": 0.6402,
+      "step": 20008
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5337357965991924e-05,
+      "loss": 0.6153,
+      "step": 20009
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.533497861155274e-05,
+      "loss": 0.6213,
+      "step": 20010
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.533259936004017e-05,
+      "loss": 0.5729,
+      "step": 20011
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5330220211479545e-05,
+      "loss": 0.6178,
+      "step": 20012
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5327841165896198e-05,
+      "loss": 0.7448,
+      "step": 20013
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5325462223315473e-05,
+      "loss": 0.6589,
+      "step": 20014
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5323083383762706e-05,
+      "loss": 0.6063,
+      "step": 20015
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.532070464726322e-05,
+      "loss": 0.6547,
+      "step": 20016
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5318326013842358e-05,
+      "loss": 0.5982,
+      "step": 20017
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5315947483525443e-05,
+      "loss": 0.6436,
+      "step": 20018
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5313569056337805e-05,
+      "loss": 0.6779,
+      "step": 20019
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5311190732304784e-05,
+      "loss": 0.694,
+      "step": 20020
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5308812511451698e-05,
+      "loss": 0.6708,
+      "step": 20021
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.530643439380387e-05,
+      "loss": 0.6387,
+      "step": 20022
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5304056379386634e-05,
+      "loss": 0.6659,
+      "step": 20023
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.530167846822531e-05,
+      "loss": 0.6003,
+      "step": 20024
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5299300660345216e-05,
+      "loss": 0.6856,
+      "step": 20025
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5296922955771692e-05,
+      "loss": 0.6591,
+      "step": 20026
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5294545354530042e-05,
+      "loss": 0.6442,
+      "step": 20027
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.529216785664559e-05,
+      "loss": 0.6716,
+      "step": 20028
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.528979046214366e-05,
+      "loss": 0.6182,
+      "step": 20029
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5287413171049573e-05,
+      "loss": 0.5965,
+      "step": 20030
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5285035983388632e-05,
+      "loss": 0.6295,
+      "step": 20031
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.528265889918616e-05,
+      "loss": 0.666,
+      "step": 20032
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5280281918467466e-05,
+      "loss": 0.6396,
+      "step": 20033
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5277905041257872e-05,
+      "loss": 0.6632,
+      "step": 20034
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.527552826758269e-05,
+      "loss": 0.6514,
+      "step": 20035
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5273151597467223e-05,
+      "loss": 0.5964,
+      "step": 20036
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.527077503093679e-05,
+      "loss": 0.6029,
+      "step": 20037
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.52683985680167e-05,
+      "loss": 0.6076,
+      "step": 20038
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5266022208732245e-05,
+      "loss": 0.6114,
+      "step": 20039
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.526364595310875e-05,
+      "loss": 0.648,
+      "step": 20040
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5261269801171518e-05,
+      "loss": 0.5911,
+      "step": 20041
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5258893752945847e-05,
+      "loss": 0.6121,
+      "step": 20042
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5256517808457044e-05,
+      "loss": 0.6466,
+      "step": 20043
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5254141967730411e-05,
+      "loss": 0.7164,
+      "step": 20044
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5251766230791247e-05,
+      "loss": 0.7766,
+      "step": 20045
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.524939059766486e-05,
+      "loss": 0.6123,
+      "step": 20046
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.524701506837654e-05,
+      "loss": 0.6128,
+      "step": 20047
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5244639642951587e-05,
+      "loss": 0.684,
+      "step": 20048
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.52422643214153e-05,
+      "loss": 0.6642,
+      "step": 20049
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.523988910379298e-05,
+      "loss": 0.7111,
+      "step": 20050
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.523751399010991e-05,
+      "loss": 0.6397,
+      "step": 20051
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5235138980391391e-05,
+      "loss": 0.693,
+      "step": 20052
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.523276407466271e-05,
+      "loss": 0.6938,
+      "step": 20053
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5230389272949163e-05,
+      "loss": 0.6453,
+      "step": 20054
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.522801457527604e-05,
+      "loss": 0.5691,
+      "step": 20055
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5225639981668626e-05,
+      "loss": 0.7168,
+      "step": 20056
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5223265492152217e-05,
+      "loss": 0.5949,
+      "step": 20057
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5220891106752092e-05,
+      "loss": 0.5756,
+      "step": 20058
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5218516825493533e-05,
+      "loss": 0.6772,
+      "step": 20059
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5216142648401844e-05,
+      "loss": 0.7168,
+      "step": 20060
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5213768575502291e-05,
+      "loss": 0.6143,
+      "step": 20061
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5211394606820156e-05,
+      "loss": 0.6485,
+      "step": 20062
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.520902074238073e-05,
+      "loss": 0.6208,
+      "step": 20063
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.520664698220929e-05,
+      "loss": 0.6349,
+      "step": 20064
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5204273326331114e-05,
+      "loss": 0.6311,
+      "step": 20065
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5201899774771477e-05,
+      "loss": 0.7185,
+      "step": 20066
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5199526327555664e-05,
+      "loss": 0.6731,
+      "step": 20067
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.519715298470894e-05,
+      "loss": 0.6075,
+      "step": 20068
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5194779746256586e-05,
+      "loss": 0.6129,
+      "step": 20069
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5192406612223881e-05,
+      "loss": 0.6182,
+      "step": 20070
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5190033582636087e-05,
+      "loss": 0.6495,
+      "step": 20071
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5187660657518483e-05,
+      "loss": 0.6238,
+      "step": 20072
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5185287836896327e-05,
+      "loss": 0.6445,
+      "step": 20073
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5182915120794905e-05,
+      "loss": 0.6327,
+      "step": 20074
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5180542509239474e-05,
+      "loss": 0.5959,
+      "step": 20075
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5178170002255304e-05,
+      "loss": 0.5985,
+      "step": 20076
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5175797599867656e-05,
+      "loss": 0.6174,
+      "step": 20077
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5173425302101805e-05,
+      "loss": 0.6749,
+      "step": 20078
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5171053108983007e-05,
+      "loss": 0.572,
+      "step": 20079
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5168681020536525e-05,
+      "loss": 0.5901,
+      "step": 20080
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.516630903678762e-05,
+      "loss": 0.6366,
+      "step": 20081
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.516393715776156e-05,
+      "loss": 0.6245,
+      "step": 20082
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5161565383483589e-05,
+      "loss": 0.6689,
+      "step": 20083
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5159193713978979e-05,
+      "loss": 0.6802,
+      "step": 20084
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5156822149272981e-05,
+      "loss": 0.7887,
+      "step": 20085
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5154450689390848e-05,
+      "loss": 0.6172,
+      "step": 20086
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5152079334357841e-05,
+      "loss": 0.5936,
+      "step": 20087
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.514970808419921e-05,
+      "loss": 0.654,
+      "step": 20088
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5147336938940205e-05,
+      "loss": 0.7323,
+      "step": 20089
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5144965898606084e-05,
+      "loss": 0.6591,
+      "step": 20090
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5142594963222095e-05,
+      "loss": 0.6856,
+      "step": 20091
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5140224132813486e-05,
+      "loss": 0.6698,
+      "step": 20092
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.51378534074055e-05,
+      "loss": 0.6563,
+      "step": 20093
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.513548278702338e-05,
+      "loss": 0.6644,
+      "step": 20094
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.513311227169239e-05,
+      "loss": 0.7023,
+      "step": 20095
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5130741861437762e-05,
+      "loss": 0.6665,
+      "step": 20096
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5128371556284737e-05,
+      "loss": 0.6348,
+      "step": 20097
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5126001356258564e-05,
+      "loss": 0.7035,
+      "step": 20098
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5123631261384486e-05,
+      "loss": 0.6092,
+      "step": 20099
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.512126127168773e-05,
+      "loss": 0.6914,
+      "step": 20100
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5118891387193552e-05,
+      "loss": 0.6676,
+      "step": 20101
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.511652160792718e-05,
+      "loss": 0.6178,
+      "step": 20102
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5114151933913848e-05,
+      "loss": 0.627,
+      "step": 20103
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5111782365178801e-05,
+      "loss": 0.6222,
+      "step": 20104
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5109412901747267e-05,
+      "loss": 0.6554,
+      "step": 20105
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5107043543644477e-05,
+      "loss": 0.6633,
+      "step": 20106
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5104674290895673e-05,
+      "loss": 0.5996,
+      "step": 20107
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.510230514352608e-05,
+      "loss": 0.6627,
+      "step": 20108
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5099936101560923e-05,
+      "loss": 0.6946,
+      "step": 20109
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5097567165025443e-05,
+      "loss": 0.6736,
+      "step": 20110
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5095198333944862e-05,
+      "loss": 0.6061,
+      "step": 20111
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5092829608344402e-05,
+      "loss": 0.6405,
+      "step": 20112
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5090460988249294e-05,
+      "loss": 0.6783,
+      "step": 20113
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5088092473684751e-05,
+      "loss": 0.6378,
+      "step": 20114
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5085724064676013e-05,
+      "loss": 0.6012,
+      "step": 20115
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5083355761248297e-05,
+      "loss": 0.6287,
+      "step": 20116
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5080987563426812e-05,
+      "loss": 0.6051,
+      "step": 20117
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5078619471236794e-05,
+      "loss": 0.7161,
+      "step": 20118
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5076251484703458e-05,
+      "loss": 0.644,
+      "step": 20119
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5073883603852009e-05,
+      "loss": 0.6302,
+      "step": 20120
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5071515828707683e-05,
+      "loss": 0.6042,
+      "step": 20121
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5069148159295682e-05,
+      "loss": 0.6262,
+      "step": 20122
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5066780595641219e-05,
+      "loss": 0.6174,
+      "step": 20123
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5064413137769517e-05,
+      "loss": 0.6048,
+      "step": 20124
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5062045785705785e-05,
+      "loss": 0.5922,
+      "step": 20125
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5059678539475227e-05,
+      "loss": 0.6653,
+      "step": 20126
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5057311399103064e-05,
+      "loss": 0.7179,
+      "step": 20127
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5054944364614496e-05,
+      "loss": 0.6306,
+      "step": 20128
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5052577436034725e-05,
+      "loss": 0.7218,
+      "step": 20129
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5050210613388976e-05,
+      "loss": 0.6531,
+      "step": 20130
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5047843896702445e-05,
+      "loss": 0.6417,
+      "step": 20131
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5045477286000331e-05,
+      "loss": 0.6075,
+      "step": 20132
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.504311078130784e-05,
+      "loss": 0.6682,
+      "step": 20133
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5040744382650168e-05,
+      "loss": 0.6383,
+      "step": 20134
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5038378090052526e-05,
+      "loss": 0.6634,
+      "step": 20135
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5036011903540109e-05,
+      "loss": 0.6435,
+      "step": 20136
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5033645823138115e-05,
+      "loss": 0.6675,
+      "step": 20137
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5031279848871743e-05,
+      "loss": 0.6483,
+      "step": 20138
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.502891398076619e-05,
+      "loss": 0.6729,
+      "step": 20139
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5026548218846648e-05,
+      "loss": 0.6287,
+      "step": 20140
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5024182563138312e-05,
+      "loss": 0.6943,
+      "step": 20141
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5021817013666377e-05,
+      "loss": 0.5873,
+      "step": 20142
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5019451570456031e-05,
+      "loss": 0.6734,
+      "step": 20143
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5017086233532468e-05,
+      "loss": 0.621,
+      "step": 20144
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5014721002920879e-05,
+      "loss": 0.6295,
+      "step": 20145
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5012355878646438e-05,
+      "loss": 0.6496,
+      "step": 20146
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5009990860734355e-05,
+      "loss": 0.6359,
+      "step": 20147
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5007625949209803e-05,
+      "loss": 0.7013,
+      "step": 20148
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.500526114409796e-05,
+      "loss": 0.6804,
+      "step": 20149
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5002896445424024e-05,
+      "loss": 0.6139,
+      "step": 20150
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5000531853213179e-05,
+      "loss": 0.7084,
+      "step": 20151
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.499816736749059e-05,
+      "loss": 0.6401,
+      "step": 20152
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.499580298828145e-05,
+      "loss": 0.6694,
+      "step": 20153
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4993438715610925e-05,
+      "loss": 0.721,
+      "step": 20154
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.499107454950421e-05,
+      "loss": 0.6042,
+      "step": 20155
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4988710489986474e-05,
+      "loss": 0.5988,
+      "step": 20156
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.498634653708289e-05,
+      "loss": 0.6378,
+      "step": 20157
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.498398269081864e-05,
+      "loss": 0.6982,
+      "step": 20158
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4981618951218892e-05,
+      "loss": 0.6859,
+      "step": 20159
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4979255318308816e-05,
+      "loss": 0.6076,
+      "step": 20160
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4976891792113595e-05,
+      "loss": 0.662,
+      "step": 20161
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4974528372658386e-05,
+      "loss": 0.6583,
+      "step": 20162
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4972165059968362e-05,
+      "loss": 0.6103,
+      "step": 20163
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4969801854068696e-05,
+      "loss": 0.6487,
+      "step": 20164
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.496743875498455e-05,
+      "loss": 0.6184,
+      "step": 20165
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4965075762741087e-05,
+      "loss": 0.6368,
+      "step": 20166
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4962712877363477e-05,
+      "loss": 0.653,
+      "step": 20167
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4960350098876885e-05,
+      "loss": 0.6344,
+      "step": 20168
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.495798742730646e-05,
+      "loss": 0.6423,
+      "step": 20169
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4955624862677382e-05,
+      "loss": 0.6702,
+      "step": 20170
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4953262405014801e-05,
+      "loss": 0.6879,
+      "step": 20171
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4950900054343875e-05,
+      "loss": 0.5765,
+      "step": 20172
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.494853781068976e-05,
+      "loss": 0.5854,
+      "step": 20173
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4946175674077611e-05,
+      "loss": 0.593,
+      "step": 20174
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4943813644532589e-05,
+      "loss": 0.5849,
+      "step": 20175
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4941451722079851e-05,
+      "loss": 0.5951,
+      "step": 20176
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4939089906744539e-05,
+      "loss": 0.6596,
+      "step": 20177
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4936728198551816e-05,
+      "loss": 0.6098,
+      "step": 20178
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4934366597526827e-05,
+      "loss": 0.6905,
+      "step": 20179
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4932005103694718e-05,
+      "loss": 0.6532,
+      "step": 20180
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4929643717080649e-05,
+      "loss": 0.6018,
+      "step": 20181
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.492728243770976e-05,
+      "loss": 0.5668,
+      "step": 20182
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.49249212656072e-05,
+      "loss": 0.5512,
+      "step": 20183
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4922560200798102e-05,
+      "loss": 0.6712,
+      "step": 20184
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4920199243307628e-05,
+      "loss": 0.6176,
+      "step": 20185
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.491783839316091e-05,
+      "loss": 0.606,
+      "step": 20186
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4915477650383092e-05,
+      "loss": 0.6197,
+      "step": 20187
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4913117014999315e-05,
+      "loss": 0.6266,
+      "step": 20188
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4910756487034716e-05,
+      "loss": 0.6388,
+      "step": 20189
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4908396066514435e-05,
+      "loss": 0.6529,
+      "step": 20190
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4906035753463618e-05,
+      "loss": 0.5921,
+      "step": 20191
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4903675547907386e-05,
+      "loss": 0.6127,
+      "step": 20192
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4901315449870879e-05,
+      "loss": 0.6909,
+      "step": 20193
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4898955459379226e-05,
+      "loss": 0.6098,
+      "step": 20194
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4896595576457573e-05,
+      "loss": 0.6527,
+      "step": 20195
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.489423580113104e-05,
+      "loss": 0.649,
+      "step": 20196
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4891876133424765e-05,
+      "loss": 0.6398,
+      "step": 20197
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4889516573363862e-05,
+      "loss": 0.5884,
+      "step": 20198
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4887157120973477e-05,
+      "loss": 0.6622,
+      "step": 20199
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.488479777627873e-05,
+      "loss": 0.6038,
+      "step": 20200
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4882438539304738e-05,
+      "loss": 0.6243,
+      "step": 20201
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4880079410076638e-05,
+      "loss": 0.621,
+      "step": 20202
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4877720388619551e-05,
+      "loss": 0.6565,
+      "step": 20203
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4875361474958589e-05,
+      "loss": 0.6556,
+      "step": 20204
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4873002669118884e-05,
+      "loss": 0.6864,
+      "step": 20205
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4870643971125558e-05,
+      "loss": 0.7042,
+      "step": 20206
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4868285381003716e-05,
+      "loss": 0.649,
+      "step": 20207
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4865926898778487e-05,
+      "loss": 0.5726,
+      "step": 20208
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4863568524474985e-05,
+      "loss": 0.6678,
+      "step": 20209
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4861210258118318e-05,
+      "loss": 0.6218,
+      "step": 20210
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4858852099733616e-05,
+      "loss": 0.6165,
+      "step": 20211
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4856494049345984e-05,
+      "loss": 0.6138,
+      "step": 20212
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4854136106980526e-05,
+      "loss": 0.6615,
+      "step": 20213
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4851778272662361e-05,
+      "loss": 0.629,
+      "step": 20214
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4849420546416592e-05,
+      "loss": 0.6073,
+      "step": 20215
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4847062928268335e-05,
+      "loss": 0.6321,
+      "step": 20216
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4844705418242693e-05,
+      "loss": 0.6379,
+      "step": 20217
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4842348016364768e-05,
+      "loss": 0.6678,
+      "step": 20218
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4839990722659675e-05,
+      "loss": 0.5913,
+      "step": 20219
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4837633537152512e-05,
+      "loss": 0.6743,
+      "step": 20220
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4835276459868381e-05,
+      "loss": 0.6173,
+      "step": 20221
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4832919490832386e-05,
+      "loss": 0.6636,
+      "step": 20222
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4830562630069625e-05,
+      "loss": 0.6706,
+      "step": 20223
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4828205877605194e-05,
+      "loss": 0.6019,
+      "step": 20224
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.48258492334642e-05,
+      "loss": 0.6991,
+      "step": 20225
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4823492697671737e-05,
+      "loss": 0.6605,
+      "step": 20226
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4821136270252888e-05,
+      "loss": 0.6572,
+      "step": 20227
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4818779951232769e-05,
+      "loss": 0.6877,
+      "step": 20228
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.481642374063646e-05,
+      "loss": 0.6254,
+      "step": 20229
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4814067638489049e-05,
+      "loss": 0.6777,
+      "step": 20230
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.481171164481564e-05,
+      "loss": 0.5809,
+      "step": 20231
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4809355759641325e-05,
+      "loss": 0.6344,
+      "step": 20232
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4806999982991174e-05,
+      "loss": 0.5508,
+      "step": 20233
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4804644314890288e-05,
+      "loss": 0.7513,
+      "step": 20234
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4802288755363741e-05,
+      "loss": 0.5843,
+      "step": 20235
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4799933304436635e-05,
+      "loss": 0.5806,
+      "step": 20236
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4797577962134051e-05,
+      "loss": 0.5688,
+      "step": 20237
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4795222728481056e-05,
+      "loss": 0.6224,
+      "step": 20238
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4792867603502752e-05,
+      "loss": 0.6592,
+      "step": 20239
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4790512587224211e-05,
+      "loss": 0.701,
+      "step": 20240
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4788157679670508e-05,
+      "loss": 0.6631,
+      "step": 20241
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.478580288086673e-05,
+      "loss": 0.5549,
+      "step": 20242
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.478344819083795e-05,
+      "loss": 0.7059,
+      "step": 20243
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.478109360960924e-05,
+      "loss": 0.5862,
+      "step": 20244
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4778739137205683e-05,
+      "loss": 0.5996,
+      "step": 20245
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.477638477365235e-05,
+      "loss": 0.6285,
+      "step": 20246
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4774030518974308e-05,
+      "loss": 0.6619,
+      "step": 20247
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4771676373196635e-05,
+      "loss": 0.5911,
+      "step": 20248
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.47693223363444e-05,
+      "loss": 0.6462,
+      "step": 20249
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4766968408442667e-05,
+      "loss": 0.6082,
+      "step": 20250
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.476461458951651e-05,
+      "loss": 0.6632,
+      "step": 20251
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4762260879591e-05,
+      "loss": 0.5957,
+      "step": 20252
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4759907278691194e-05,
+      "loss": 0.6176,
+      "step": 20253
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4757553786842155e-05,
+      "loss": 0.5833,
+      "step": 20254
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4755200404068942e-05,
+      "loss": 0.5819,
+      "step": 20255
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4752847130396635e-05,
+      "loss": 0.6287,
+      "step": 20256
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4750493965850285e-05,
+      "loss": 0.7794,
+      "step": 20257
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4748140910454944e-05,
+      "loss": 0.6148,
+      "step": 20258
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4745787964235686e-05,
+      "loss": 0.6027,
+      "step": 20259
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4743435127217558e-05,
+      "loss": 0.6324,
+      "step": 20260
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4741082399425616e-05,
+      "loss": 0.6663,
+      "step": 20261
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4738729780884918e-05,
+      "loss": 0.6117,
+      "step": 20262
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4736377271620524e-05,
+      "loss": 0.6585,
+      "step": 20263
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4734024871657476e-05,
+      "loss": 0.5852,
+      "step": 20264
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4731672581020834e-05,
+      "loss": 0.7187,
+      "step": 20265
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4729320399735644e-05,
+      "loss": 0.65,
+      "step": 20266
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.472696832782695e-05,
+      "loss": 0.6066,
+      "step": 20267
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4724616365319815e-05,
+      "loss": 0.5803,
+      "step": 20268
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4722264512239276e-05,
+      "loss": 0.6851,
+      "step": 20269
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4719912768610372e-05,
+      "loss": 0.7006,
+      "step": 20270
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4717561134458165e-05,
+      "loss": 0.6633,
+      "step": 20271
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4715209609807695e-05,
+      "loss": 0.6431,
+      "step": 20272
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.471285819468399e-05,
+      "loss": 0.6173,
+      "step": 20273
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4710506889112103e-05,
+      "loss": 0.6458,
+      "step": 20274
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4708155693117062e-05,
+      "loss": 0.6081,
+      "step": 20275
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4705804606723922e-05,
+      "loss": 0.6908,
+      "step": 20276
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4703453629957714e-05,
+      "loss": 0.6781,
+      "step": 20277
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4701102762843465e-05,
+      "loss": 0.5997,
+      "step": 20278
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4698752005406228e-05,
+      "loss": 0.6243,
+      "step": 20279
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4696401357671024e-05,
+      "loss": 0.761,
+      "step": 20280
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.469405081966289e-05,
+      "loss": 0.6278,
+      "step": 20281
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.469170039140686e-05,
+      "loss": 0.6364,
+      "step": 20282
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4689350072927963e-05,
+      "loss": 0.6936,
+      "step": 20283
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4686999864251222e-05,
+      "loss": 0.6385,
+      "step": 20284
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4684649765401679e-05,
+      "loss": 0.6625,
+      "step": 20285
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.468229977640435e-05,
+      "loss": 0.6319,
+      "step": 20286
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4679949897284262e-05,
+      "loss": 0.6237,
+      "step": 20287
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4677600128066447e-05,
+      "loss": 0.623,
+      "step": 20288
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4675250468775919e-05,
+      "loss": 0.6423,
+      "step": 20289
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.467290091943771e-05,
+      "loss": 0.6742,
+      "step": 20290
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4670551480076833e-05,
+      "loss": 0.6322,
+      "step": 20291
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4668202150718319e-05,
+      "loss": 0.5852,
+      "step": 20292
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4665852931387175e-05,
+      "loss": 0.6497,
+      "step": 20293
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4663503822108421e-05,
+      "loss": 0.6534,
+      "step": 20294
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4661154822907074e-05,
+      "loss": 0.6823,
+      "step": 20295
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4658805933808154e-05,
+      "loss": 0.6366,
+      "step": 20296
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4656457154836672e-05,
+      "loss": 0.7106,
+      "step": 20297
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.465410848601764e-05,
+      "loss": 0.6833,
+      "step": 20298
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4651759927376074e-05,
+      "loss": 0.6289,
+      "step": 20299
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4649411478936981e-05,
+      "loss": 0.6489,
+      "step": 20300
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4647063140725376e-05,
+      "loss": 0.6099,
+      "step": 20301
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4644714912766255e-05,
+      "loss": 0.6549,
+      "step": 20302
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.464236679508464e-05,
+      "loss": 0.5755,
+      "step": 20303
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4640018787705528e-05,
+      "loss": 0.6779,
+      "step": 20304
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4637670890653921e-05,
+      "loss": 0.6543,
+      "step": 20305
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4635323103954834e-05,
+      "loss": 0.6287,
+      "step": 20306
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4632975427633266e-05,
+      "loss": 0.6986,
+      "step": 20307
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4630627861714207e-05,
+      "loss": 0.645,
+      "step": 20308
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4628280406222672e-05,
+      "loss": 0.7266,
+      "step": 20309
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4625933061183655e-05,
+      "loss": 0.6555,
+      "step": 20310
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.462358582662215e-05,
+      "loss": 0.7161,
+      "step": 20311
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4621238702563156e-05,
+      "loss": 0.658,
+      "step": 20312
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4618891689031677e-05,
+      "loss": 0.6101,
+      "step": 20313
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4616544786052694e-05,
+      "loss": 0.6328,
+      "step": 20314
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4614197993651201e-05,
+      "loss": 0.5848,
+      "step": 20315
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4611851311852192e-05,
+      "loss": 0.6432,
+      "step": 20316
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4609504740680665e-05,
+      "loss": 0.5887,
+      "step": 20317
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4607158280161604e-05,
+      "loss": 0.6153,
+      "step": 20318
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4604811930319994e-05,
+      "loss": 0.6453,
+      "step": 20319
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4602465691180829e-05,
+      "loss": 0.5603,
+      "step": 20320
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4600119562769093e-05,
+      "loss": 0.6633,
+      "step": 20321
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4597773545109763e-05,
+      "loss": 0.6278,
+      "step": 20322
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4595427638227838e-05,
+      "loss": 0.6765,
+      "step": 20323
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4593081842148287e-05,
+      "loss": 0.5787,
+      "step": 20324
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4590736156896102e-05,
+      "loss": 0.5828,
+      "step": 20325
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4588390582496242e-05,
+      "loss": 0.6495,
+      "step": 20326
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4586045118973713e-05,
+      "loss": 0.6783,
+      "step": 20327
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4583699766353482e-05,
+      "loss": 0.6385,
+      "step": 20328
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4581354524660524e-05,
+      "loss": 0.5775,
+      "step": 20329
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4579009393919813e-05,
+      "loss": 0.6444,
+      "step": 20330
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4576664374156329e-05,
+      "loss": 0.6619,
+      "step": 20331
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4574319465395025e-05,
+      "loss": 0.7056,
+      "step": 20332
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4571974667660914e-05,
+      "loss": 0.6248,
+      "step": 20333
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4569629980978932e-05,
+      "loss": 0.6452,
+      "step": 20334
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4567285405374059e-05,
+      "loss": 0.6391,
+      "step": 20335
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4564940940871261e-05,
+      "loss": 0.7295,
+      "step": 20336
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4562596587495497e-05,
+      "loss": 0.7017,
+      "step": 20337
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4560252345271751e-05,
+      "loss": 0.6061,
+      "step": 20338
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.455790821422498e-05,
+      "loss": 0.6755,
+      "step": 20339
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.455556419438015e-05,
+      "loss": 0.6681,
+      "step": 20340
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4553220285762217e-05,
+      "loss": 0.5976,
+      "step": 20341
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4550876488396147e-05,
+      "loss": 0.5906,
+      "step": 20342
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.454853280230689e-05,
+      "loss": 0.7605,
+      "step": 20343
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4546189227519419e-05,
+      "loss": 0.6432,
+      "step": 20344
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4543845764058688e-05,
+      "loss": 0.6529,
+      "step": 20345
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4541502411949653e-05,
+      "loss": 0.6182,
+      "step": 20346
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4539159171217267e-05,
+      "loss": 0.6205,
+      "step": 20347
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4536816041886486e-05,
+      "loss": 0.6236,
+      "step": 20348
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.453447302398226e-05,
+      "loss": 0.6086,
+      "step": 20349
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4532130117529536e-05,
+      "loss": 0.6515,
+      "step": 20350
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4529787322553276e-05,
+      "loss": 0.6456,
+      "step": 20351
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4527444639078427e-05,
+      "loss": 0.6176,
+      "step": 20352
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4525102067129944e-05,
+      "loss": 0.602,
+      "step": 20353
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4522759606732745e-05,
+      "loss": 0.6094,
+      "step": 20354
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4520417257911806e-05,
+      "loss": 0.6159,
+      "step": 20355
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.451807502069206e-05,
+      "loss": 0.6239,
+      "step": 20356
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.451573289509845e-05,
+      "loss": 0.5881,
+      "step": 20357
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.451339088115592e-05,
+      "loss": 0.6183,
+      "step": 20358
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4511048978889413e-05,
+      "loss": 0.6234,
+      "step": 20359
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4508707188323855e-05,
+      "loss": 0.7202,
+      "step": 20360
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4506365509484204e-05,
+      "loss": 0.682,
+      "step": 20361
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4504023942395393e-05,
+      "loss": 0.6404,
+      "step": 20362
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4501682487082352e-05,
+      "loss": 0.5926,
+      "step": 20363
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4499341143570019e-05,
+      "loss": 0.6235,
+      "step": 20364
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.449699991188333e-05,
+      "loss": 0.6517,
+      "step": 20365
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4494658792047214e-05,
+      "loss": 0.6698,
+      "step": 20366
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4492317784086593e-05,
+      "loss": 0.6914,
+      "step": 20367
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.448997688802642e-05,
+      "loss": 0.6338,
+      "step": 20368
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4487636103891612e-05,
+      "loss": 0.6589,
+      "step": 20369
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4485295431707097e-05,
+      "loss": 0.7195,
+      "step": 20370
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.44829548714978e-05,
+      "loss": 0.5946,
+      "step": 20371
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4480614423288652e-05,
+      "loss": 0.6739,
+      "step": 20372
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4478274087104571e-05,
+      "loss": 0.6082,
+      "step": 20373
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4475933862970484e-05,
+      "loss": 0.702,
+      "step": 20374
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4473593750911314e-05,
+      "loss": 0.6095,
+      "step": 20375
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4471253750951977e-05,
+      "loss": 0.6051,
+      "step": 20376
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4468913863117395e-05,
+      "loss": 0.6627,
+      "step": 20377
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4466574087432477e-05,
+      "loss": 0.6817,
+      "step": 20378
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4464234423922163e-05,
+      "loss": 0.6349,
+      "step": 20379
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4461894872611353e-05,
+      "loss": 0.6451,
+      "step": 20380
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4459555433524969e-05,
+      "loss": 0.6344,
+      "step": 20381
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4457216106687914e-05,
+      "loss": 0.6443,
+      "step": 20382
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4454876892125113e-05,
+      "loss": 0.6695,
+      "step": 20383
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.445253778986146e-05,
+      "loss": 0.6219,
+      "step": 20384
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4450198799921882e-05,
+      "loss": 0.5949,
+      "step": 20385
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4447859922331287e-05,
+      "loss": 0.7467,
+      "step": 20386
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4445521157114573e-05,
+      "loss": 0.592,
+      "step": 20387
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.444318250429665e-05,
+      "loss": 0.6545,
+      "step": 20388
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4440843963902428e-05,
+      "loss": 0.5924,
+      "step": 20389
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4438505535956795e-05,
+      "loss": 0.5929,
+      "step": 20390
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4436167220484676e-05,
+      "loss": 0.6882,
+      "step": 20391
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4433829017510963e-05,
+      "loss": 0.6076,
+      "step": 20392
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4431490927060568e-05,
+      "loss": 0.7047,
+      "step": 20393
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4429152949158367e-05,
+      "loss": 0.6553,
+      "step": 20394
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.442681508382926e-05,
+      "loss": 0.6719,
+      "step": 20395
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4424477331098163e-05,
+      "loss": 0.6511,
+      "step": 20396
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4422139690989961e-05,
+      "loss": 0.6506,
+      "step": 20397
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.441980216352955e-05,
+      "loss": 0.5658,
+      "step": 20398
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4417464748741821e-05,
+      "loss": 0.5946,
+      "step": 20399
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4415127446651667e-05,
+      "loss": 0.6548,
+      "step": 20400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.441279025728397e-05,
+      "loss": 0.6382,
+      "step": 20401
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.441045318066364e-05,
+      "loss": 0.662,
+      "step": 20402
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4408116216815556e-05,
+      "loss": 0.6364,
+      "step": 20403
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4405779365764605e-05,
+      "loss": 0.6309,
+      "step": 20404
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4403442627535668e-05,
+      "loss": 0.6178,
+      "step": 20405
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4401106002153635e-05,
+      "loss": 0.6509,
+      "step": 20406
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4398769489643379e-05,
+      "loss": 0.6416,
+      "step": 20407
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.43964330900298e-05,
+      "loss": 0.6791,
+      "step": 20408
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4394096803337776e-05,
+      "loss": 0.6528,
+      "step": 20409
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4391760629592177e-05,
+      "loss": 0.6351,
+      "step": 20410
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4389424568817888e-05,
+      "loss": 0.5866,
+      "step": 20411
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4387088621039785e-05,
+      "loss": 0.6024,
+      "step": 20412
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4384752786282735e-05,
+      "loss": 0.6027,
+      "step": 20413
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4382417064571645e-05,
+      "loss": 0.6231,
+      "step": 20414
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4380081455931355e-05,
+      "loss": 0.7382,
+      "step": 20415
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4377745960386751e-05,
+      "loss": 0.7074,
+      "step": 20416
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4375410577962706e-05,
+      "loss": 0.5646,
+      "step": 20417
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4373075308684076e-05,
+      "loss": 0.5983,
+      "step": 20418
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.437074015257575e-05,
+      "loss": 0.6366,
+      "step": 20419
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4368405109662589e-05,
+      "loss": 0.588,
+      "step": 20420
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.436607017996946e-05,
+      "loss": 0.6165,
+      "step": 20421
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4363735363521227e-05,
+      "loss": 0.6131,
+      "step": 20422
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4361400660342755e-05,
+      "loss": 0.675,
+      "step": 20423
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4359066070458899e-05,
+      "loss": 0.7643,
+      "step": 20424
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4356731593894538e-05,
+      "loss": 0.5828,
+      "step": 20425
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4354397230674522e-05,
+      "loss": 0.65,
+      "step": 20426
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4352062980823716e-05,
+      "loss": 0.6213,
+      "step": 20427
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4349728844366971e-05,
+      "loss": 0.6641,
+      "step": 20428
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4347394821329153e-05,
+      "loss": 0.626,
+      "step": 20429
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4345060911735097e-05,
+      "loss": 0.6009,
+      "step": 20430
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4342727115609683e-05,
+      "loss": 0.5775,
+      "step": 20431
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.434039343297776e-05,
+      "loss": 0.6361,
+      "step": 20432
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4338059863864173e-05,
+      "loss": 0.6949,
+      "step": 20433
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4335726408293787e-05,
+      "loss": 0.6223,
+      "step": 20434
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4333393066291417e-05,
+      "loss": 0.6117,
+      "step": 20435
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4331059837881949e-05,
+      "loss": 0.569,
+      "step": 20436
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4328726723090213e-05,
+      "loss": 0.6685,
+      "step": 20437
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4326393721941061e-05,
+      "loss": 0.6292,
+      "step": 20438
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4324060834459336e-05,
+      "loss": 0.6157,
+      "step": 20439
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4321728060669879e-05,
+      "loss": 0.6903,
+      "step": 20440
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4319395400597524e-05,
+      "loss": 0.6591,
+      "step": 20441
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4317062854267139e-05,
+      "loss": 0.6628,
+      "step": 20442
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4314730421703546e-05,
+      "loss": 0.5956,
+      "step": 20443
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4312398102931584e-05,
+      "loss": 0.6108,
+      "step": 20444
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4310065897976099e-05,
+      "loss": 0.666,
+      "step": 20445
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.430773380686192e-05,
+      "loss": 0.6322,
+      "step": 20446
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4305401829613874e-05,
+      "loss": 0.6204,
+      "step": 20447
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4303069966256816e-05,
+      "loss": 0.5883,
+      "step": 20448
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4300738216815568e-05,
+      "loss": 0.6539,
+      "step": 20449
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4298406581314963e-05,
+      "loss": 0.7239,
+      "step": 20450
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4296075059779828e-05,
+      "loss": 0.6133,
+      "step": 20451
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4293743652235001e-05,
+      "loss": 0.6108,
+      "step": 20452
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4291412358705292e-05,
+      "loss": 0.6299,
+      "step": 20453
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4289081179215558e-05,
+      "loss": 0.576,
+      "step": 20454
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4286750113790598e-05,
+      "loss": 0.5667,
+      "step": 20455
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4284419162455243e-05,
+      "loss": 0.6567,
+      "step": 20456
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4282088325234321e-05,
+      "loss": 0.5711,
+      "step": 20457
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4279757602152638e-05,
+      "loss": 0.6199,
+      "step": 20458
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4277426993235043e-05,
+      "loss": 0.6616,
+      "step": 20459
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4275096498506337e-05,
+      "loss": 0.6135,
+      "step": 20460
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4272766117991343e-05,
+      "loss": 0.6636,
+      "step": 20461
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.427043585171488e-05,
+      "loss": 0.7215,
+      "step": 20462
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4268105699701756e-05,
+      "loss": 0.608,
+      "step": 20463
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4265775661976782e-05,
+      "loss": 0.6292,
+      "step": 20464
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4263445738564796e-05,
+      "loss": 0.6578,
+      "step": 20465
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4261115929490587e-05,
+      "loss": 0.6617,
+      "step": 20466
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.425878623477898e-05,
+      "loss": 0.6319,
+      "step": 20467
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4256456654454775e-05,
+      "loss": 0.6305,
+      "step": 20468
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4254127188542782e-05,
+      "loss": 0.6545,
+      "step": 20469
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4251797837067814e-05,
+      "loss": 0.638,
+      "step": 20470
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4249468600054661e-05,
+      "loss": 0.6327,
+      "step": 20471
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4247139477528154e-05,
+      "loss": 0.677,
+      "step": 20472
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4244810469513082e-05,
+      "loss": 0.6494,
+      "step": 20473
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4242481576034255e-05,
+      "loss": 0.6101,
+      "step": 20474
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4240152797116452e-05,
+      "loss": 0.6246,
+      "step": 20475
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.42378241327845e-05,
+      "loss": 0.6682,
+      "step": 20476
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4235495583063185e-05,
+      "loss": 0.6921,
+      "step": 20477
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4233167147977309e-05,
+      "loss": 0.672,
+      "step": 20478
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4230838827551662e-05,
+      "loss": 0.6454,
+      "step": 20479
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4228510621811047e-05,
+      "loss": 0.5864,
+      "step": 20480
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4226182530780256e-05,
+      "loss": 0.6337,
+      "step": 20481
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4223854554484068e-05,
+      "loss": 0.6663,
+      "step": 20482
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4221526692947295e-05,
+      "loss": 0.6293,
+      "step": 20483
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.421919894619472e-05,
+      "loss": 0.6211,
+      "step": 20484
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.421687131425113e-05,
+      "loss": 0.6175,
+      "step": 20485
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4214543797141316e-05,
+      "loss": 0.5946,
+      "step": 20486
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.421221639489006e-05,
+      "loss": 0.5972,
+      "step": 20487
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4209889107522143e-05,
+      "loss": 0.6558,
+      "step": 20488
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4207561935062363e-05,
+      "loss": 0.7377,
+      "step": 20489
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.42052348775355e-05,
+      "loss": 0.6124,
+      "step": 20490
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4202907934966331e-05,
+      "loss": 0.5854,
+      "step": 20491
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4200581107379634e-05,
+      "loss": 0.6274,
+      "step": 20492
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4198254394800195e-05,
+      "loss": 0.6324,
+      "step": 20493
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4195927797252787e-05,
+      "loss": 0.648,
+      "step": 20494
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.419360131476219e-05,
+      "loss": 0.6737,
+      "step": 20495
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4191274947353178e-05,
+      "loss": 0.6451,
+      "step": 20496
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4188948695050524e-05,
+      "loss": 0.6036,
+      "step": 20497
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4186622557879004e-05,
+      "loss": 0.634,
+      "step": 20498
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4184296535863378e-05,
+      "loss": 0.697,
+      "step": 20499
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.418197062902844e-05,
+      "loss": 0.653,
+      "step": 20500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4179644837398948e-05,
+      "loss": 0.5977,
+      "step": 20501
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4177319160999663e-05,
+      "loss": 0.5741,
+      "step": 20502
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4174993599855364e-05,
+      "loss": 0.6202,
+      "step": 20503
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.417266815399081e-05,
+      "loss": 0.6369,
+      "step": 20504
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4170342823430754e-05,
+      "loss": 0.6201,
+      "step": 20505
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4168017608199985e-05,
+      "loss": 0.6538,
+      "step": 20506
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4165692508323252e-05,
+      "loss": 0.6729,
+      "step": 20507
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4163367523825316e-05,
+      "loss": 0.608,
+      "step": 20508
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4161042654730936e-05,
+      "loss": 0.6611,
+      "step": 20509
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4158717901064873e-05,
+      "loss": 0.6436,
+      "step": 20510
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4156393262851869e-05,
+      "loss": 0.6803,
+      "step": 20511
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4154068740116704e-05,
+      "loss": 0.6134,
+      "step": 20512
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4151744332884125e-05,
+      "loss": 0.6204,
+      "step": 20513
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.414942004117889e-05,
+      "loss": 0.5966,
+      "step": 20514
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4147095865025736e-05,
+      "loss": 0.6253,
+      "step": 20515
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.414477180444941e-05,
+      "loss": 0.6193,
+      "step": 20516
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4142447859474686e-05,
+      "loss": 0.7462,
+      "step": 20517
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.41401240301263e-05,
+      "loss": 0.655,
+      "step": 20518
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4137800316429e-05,
+      "loss": 0.6487,
+      "step": 20519
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4135476718407531e-05,
+      "loss": 0.5801,
+      "step": 20520
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4133153236086638e-05,
+      "loss": 0.6152,
+      "step": 20521
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4130829869491058e-05,
+      "loss": 0.653,
+      "step": 20522
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4128506618645551e-05,
+      "loss": 0.5856,
+      "step": 20523
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4126183483574847e-05,
+      "loss": 0.6701,
+      "step": 20524
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.412386046430369e-05,
+      "loss": 0.6047,
+      "step": 20525
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4121537560856813e-05,
+      "loss": 0.6076,
+      "step": 20526
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4119214773258956e-05,
+      "loss": 0.604,
+      "step": 20527
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4116892101534847e-05,
+      "loss": 0.6442,
+      "step": 20528
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4114569545709242e-05,
+      "loss": 0.6131,
+      "step": 20529
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4112247105806858e-05,
+      "loss": 0.6477,
+      "step": 20530
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4109924781852435e-05,
+      "loss": 0.6221,
+      "step": 20531
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4107602573870704e-05,
+      "loss": 0.65,
+      "step": 20532
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.410528048188639e-05,
+      "loss": 0.5807,
+      "step": 20533
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4102958505924213e-05,
+      "loss": 0.6696,
+      "step": 20534
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4100636646008936e-05,
+      "loss": 0.579,
+      "step": 20535
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4098314902165246e-05,
+      "loss": 0.6549,
+      "step": 20536
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4095993274417888e-05,
+      "loss": 0.6801,
+      "step": 20537
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.409367176279158e-05,
+      "loss": 0.6568,
+      "step": 20538
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4091350367311037e-05,
+      "loss": 0.6339,
+      "step": 20539
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4089029088000998e-05,
+      "loss": 0.6181,
+      "step": 20540
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4086707924886177e-05,
+      "loss": 0.6187,
+      "step": 20541
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.408438687799129e-05,
+      "loss": 0.6213,
+      "step": 20542
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4082065947341058e-05,
+      "loss": 0.6062,
+      "step": 20543
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4079745132960193e-05,
+      "loss": 0.6546,
+      "step": 20544
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4077424434873404e-05,
+      "loss": 0.6912,
+      "step": 20545
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4075103853105423e-05,
+      "loss": 0.7232,
+      "step": 20546
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4072783387680955e-05,
+      "loss": 0.6301,
+      "step": 20547
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4070463038624703e-05,
+      "loss": 0.6471,
+      "step": 20548
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4068142805961394e-05,
+      "loss": 0.69,
+      "step": 20549
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4065822689715719e-05,
+      "loss": 0.6464,
+      "step": 20550
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4063502689912388e-05,
+      "loss": 0.6075,
+      "step": 20551
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4061182806576124e-05,
+      "loss": 0.6957,
+      "step": 20552
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4058863039731618e-05,
+      "loss": 0.6453,
+      "step": 20553
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4056543389403581e-05,
+      "loss": 0.5999,
+      "step": 20554
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4054223855616725e-05,
+      "loss": 0.6507,
+      "step": 20555
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4051904438395716e-05,
+      "loss": 0.601,
+      "step": 20556
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.404958513776529e-05,
+      "loss": 0.5824,
+      "step": 20557
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4047265953750133e-05,
+      "loss": 0.6404,
+      "step": 20558
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4044946886374943e-05,
+      "loss": 0.648,
+      "step": 20559
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.404262793566442e-05,
+      "loss": 0.6456,
+      "step": 20560
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4040309101643257e-05,
+      "loss": 0.5785,
+      "step": 20561
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4037990384336139e-05,
+      "loss": 0.6226,
+      "step": 20562
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4035671783767776e-05,
+      "loss": 0.6233,
+      "step": 20563
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4033353299962854e-05,
+      "loss": 0.5833,
+      "step": 20564
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4031034932946057e-05,
+      "loss": 0.6142,
+      "step": 20565
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4028716682742083e-05,
+      "loss": 0.5981,
+      "step": 20566
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4026398549375613e-05,
+      "loss": 0.7201,
+      "step": 20567
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4024080532871331e-05,
+      "loss": 0.641,
+      "step": 20568
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4021762633253932e-05,
+      "loss": 0.5766,
+      "step": 20569
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4019444850548097e-05,
+      "loss": 0.5845,
+      "step": 20570
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4017127184778512e-05,
+      "loss": 0.6392,
+      "step": 20571
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.401480963596985e-05,
+      "loss": 0.6015,
+      "step": 20572
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4012492204146799e-05,
+      "loss": 0.628,
+      "step": 20573
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4010174889334033e-05,
+      "loss": 0.6789,
+      "step": 20574
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.400785769155623e-05,
+      "loss": 0.6286,
+      "step": 20575
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4005540610838075e-05,
+      "loss": 0.601,
+      "step": 20576
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4003223647204234e-05,
+      "loss": 0.7075,
+      "step": 20577
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4000906800679386e-05,
+      "loss": 0.6641,
+      "step": 20578
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3998590071288193e-05,
+      "loss": 0.6694,
+      "step": 20579
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3996273459055348e-05,
+      "loss": 0.6486,
+      "step": 20580
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3993956964005505e-05,
+      "loss": 0.6087,
+      "step": 20581
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3991640586163343e-05,
+      "loss": 0.6109,
+      "step": 20582
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3989324325553526e-05,
+      "loss": 0.63,
+      "step": 20583
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3987008182200717e-05,
+      "loss": 0.6196,
+      "step": 20584
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3984692156129575e-05,
+      "loss": 0.6198,
+      "step": 20585
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3982376247364788e-05,
+      "loss": 0.5933,
+      "step": 20586
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3980060455930999e-05,
+      "loss": 0.6733,
+      "step": 20587
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3977744781852881e-05,
+      "loss": 0.5782,
+      "step": 20588
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3975429225155087e-05,
+      "loss": 0.6277,
+      "step": 20589
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.397311378586228e-05,
+      "loss": 0.6321,
+      "step": 20590
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3970798463999114e-05,
+      "loss": 0.6073,
+      "step": 20591
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.396848325959024e-05,
+      "loss": 0.6076,
+      "step": 20592
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3966168172660332e-05,
+      "loss": 0.5922,
+      "step": 20593
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3963853203234034e-05,
+      "loss": 0.6914,
+      "step": 20594
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3961538351336006e-05,
+      "loss": 0.6686,
+      "step": 20595
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3959223616990879e-05,
+      "loss": 0.6735,
+      "step": 20596
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3956909000223323e-05,
+      "loss": 0.6212,
+      "step": 20597
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3954594501057983e-05,
+      "loss": 0.6659,
+      "step": 20598
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3952280119519507e-05,
+      "loss": 0.6857,
+      "step": 20599
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.394996585563254e-05,
+      "loss": 0.6579,
+      "step": 20600
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3947651709421728e-05,
+      "loss": 0.7522,
+      "step": 20601
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3945337680911714e-05,
+      "loss": 0.6207,
+      "step": 20602
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3943023770127134e-05,
+      "loss": 0.6691,
+      "step": 20603
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3940709977092645e-05,
+      "loss": 0.6227,
+      "step": 20604
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3938396301832882e-05,
+      "loss": 0.639,
+      "step": 20605
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3936082744372484e-05,
+      "loss": 0.6379,
+      "step": 20606
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3933769304736085e-05,
+      "loss": 0.6296,
+      "step": 20607
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3931455982948327e-05,
+      "loss": 0.6147,
+      "step": 20608
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3929142779033832e-05,
+      "loss": 0.6434,
+      "step": 20609
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3926829693017258e-05,
+      "loss": 0.6508,
+      "step": 20610
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.392451672492322e-05,
+      "loss": 0.6445,
+      "step": 20611
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3922203874776358e-05,
+      "loss": 0.549,
+      "step": 20612
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3919891142601296e-05,
+      "loss": 0.6252,
+      "step": 20613
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.391757852842267e-05,
+      "loss": 0.6335,
+      "step": 20614
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.39152660322651e-05,
+      "loss": 0.6106,
+      "step": 20615
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3912953654153222e-05,
+      "loss": 0.6498,
+      "step": 20616
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3910641394111656e-05,
+      "loss": 0.674,
+      "step": 20617
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3908329252165025e-05,
+      "loss": 0.6112,
+      "step": 20618
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3906017228337953e-05,
+      "loss": 0.6309,
+      "step": 20619
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3903705322655052e-05,
+      "loss": 0.6891,
+      "step": 20620
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3901393535140967e-05,
+      "loss": 0.6208,
+      "step": 20621
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3899081865820296e-05,
+      "loss": 0.6424,
+      "step": 20622
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.389677031471767e-05,
+      "loss": 0.6676,
+      "step": 20623
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3894458881857694e-05,
+      "loss": 0.5885,
+      "step": 20624
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.389214756726499e-05,
+      "loss": 0.6076,
+      "step": 20625
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3889836370964163e-05,
+      "loss": 0.804,
+      "step": 20626
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3887525292979842e-05,
+      "loss": 0.6294,
+      "step": 20627
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.388521433333663e-05,
+      "loss": 0.6126,
+      "step": 20628
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3882903492059137e-05,
+      "loss": 0.6047,
+      "step": 20629
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3880592769171973e-05,
+      "loss": 0.6296,
+      "step": 20630
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3878282164699746e-05,
+      "loss": 0.5899,
+      "step": 20631
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.387597167866705e-05,
+      "loss": 0.6203,
+      "step": 20632
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3873661311098512e-05,
+      "loss": 0.5903,
+      "step": 20633
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3871351062018727e-05,
+      "loss": 0.6362,
+      "step": 20634
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3869040931452298e-05,
+      "loss": 0.6003,
+      "step": 20635
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3866730919423835e-05,
+      "loss": 0.6016,
+      "step": 20636
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3864421025957907e-05,
+      "loss": 0.6101,
+      "step": 20637
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3862111251079147e-05,
+      "loss": 0.6443,
+      "step": 20638
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3859801594812136e-05,
+      "loss": 0.7074,
+      "step": 20639
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3857492057181475e-05,
+      "loss": 0.6573,
+      "step": 20640
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3855182638211761e-05,
+      "loss": 0.6,
+      "step": 20641
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3852873337927586e-05,
+      "loss": 0.6163,
+      "step": 20642
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3850564156353535e-05,
+      "loss": 0.6475,
+      "step": 20643
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3848255093514212e-05,
+      "loss": 0.6126,
+      "step": 20644
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3845946149434203e-05,
+      "loss": 0.5976,
+      "step": 20645
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3843637324138098e-05,
+      "loss": 0.6163,
+      "step": 20646
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3841328617650478e-05,
+      "loss": 0.6228,
+      "step": 20647
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3839020029995936e-05,
+      "loss": 0.6412,
+      "step": 20648
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3836711561199044e-05,
+      "loss": 0.6507,
+      "step": 20649
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3834403211284407e-05,
+      "loss": 0.6519,
+      "step": 20650
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.38320949802766e-05,
+      "loss": 0.614,
+      "step": 20651
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.38297868682002e-05,
+      "loss": 0.6226,
+      "step": 20652
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3827478875079786e-05,
+      "loss": 0.6765,
+      "step": 20653
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.382517100093994e-05,
+      "loss": 0.6537,
+      "step": 20654
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3822863245805229e-05,
+      "loss": 0.6014,
+      "step": 20655
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3820555609700256e-05,
+      "loss": 0.6814,
+      "step": 20656
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3818248092649572e-05,
+      "loss": 0.6256,
+      "step": 20657
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3815940694677753e-05,
+      "loss": 0.6516,
+      "step": 20658
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.381363341580938e-05,
+      "loss": 0.5979,
+      "step": 20659
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3811326256069008e-05,
+      "loss": 0.6634,
+      "step": 20660
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3809019215481228e-05,
+      "loss": 0.6432,
+      "step": 20661
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.38067122940706e-05,
+      "loss": 0.6603,
+      "step": 20662
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3804405491861688e-05,
+      "loss": 0.6785,
+      "step": 20663
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3802098808879065e-05,
+      "loss": 0.6859,
+      "step": 20664
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3799792245147289e-05,
+      "loss": 0.7242,
+      "step": 20665
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3797485800690916e-05,
+      "loss": 0.6597,
+      "step": 20666
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3795179475534526e-05,
+      "loss": 0.6208,
+      "step": 20667
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3792873269702674e-05,
+      "loss": 0.6179,
+      "step": 20668
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.379056718321992e-05,
+      "loss": 0.6388,
+      "step": 20669
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3788261216110815e-05,
+      "loss": 0.6344,
+      "step": 20670
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3785955368399925e-05,
+      "loss": 0.5952,
+      "step": 20671
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3783649640111789e-05,
+      "loss": 0.6298,
+      "step": 20672
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3781344031270988e-05,
+      "loss": 0.6754,
+      "step": 20673
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3779038541902063e-05,
+      "loss": 0.6524,
+      "step": 20674
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.377673317202956e-05,
+      "loss": 0.6435,
+      "step": 20675
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3774427921678046e-05,
+      "loss": 0.5734,
+      "step": 20676
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3772122790872044e-05,
+      "loss": 0.6438,
+      "step": 20677
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3769817779636124e-05,
+      "loss": 0.6871,
+      "step": 20678
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3767512887994826e-05,
+      "loss": 0.6011,
+      "step": 20679
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.37652081159727e-05,
+      "loss": 0.63,
+      "step": 20680
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3762903463594285e-05,
+      "loss": 0.5865,
+      "step": 20681
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3760598930884121e-05,
+      "loss": 0.6252,
+      "step": 20682
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3758294517866749e-05,
+      "loss": 0.6236,
+      "step": 20683
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3755990224566728e-05,
+      "loss": 0.6161,
+      "step": 20684
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.375368605100858e-05,
+      "loss": 0.5922,
+      "step": 20685
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.375138199721685e-05,
+      "loss": 0.6155,
+      "step": 20686
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3749078063216075e-05,
+      "loss": 0.6388,
+      "step": 20687
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3746774249030786e-05,
+      "loss": 0.6297,
+      "step": 20688
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3744470554685507e-05,
+      "loss": 0.619,
+      "step": 20689
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3742166980204796e-05,
+      "loss": 0.6016,
+      "step": 20690
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3739863525613169e-05,
+      "loss": 0.5851,
+      "step": 20691
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3737560190935161e-05,
+      "loss": 0.6819,
+      "step": 20692
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3735256976195299e-05,
+      "loss": 0.6085,
+      "step": 20693
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3732953881418114e-05,
+      "loss": 0.6358,
+      "step": 20694
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3730650906628126e-05,
+      "loss": 0.6343,
+      "step": 20695
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3728348051849865e-05,
+      "loss": 0.5976,
+      "step": 20696
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3726045317107855e-05,
+      "loss": 0.6825,
+      "step": 20697
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3723742702426617e-05,
+      "loss": 0.5678,
+      "step": 20698
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3721440207830674e-05,
+      "loss": 0.6321,
+      "step": 20699
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3719137833344536e-05,
+      "loss": 0.6318,
+      "step": 20700
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3716835578992737e-05,
+      "loss": 0.672,
+      "step": 20701
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3714533444799793e-05,
+      "loss": 0.6174,
+      "step": 20702
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3712231430790217e-05,
+      "loss": 0.6538,
+      "step": 20703
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3709929536988519e-05,
+      "loss": 0.617,
+      "step": 20704
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3707627763419223e-05,
+      "loss": 0.7035,
+      "step": 20705
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3705326110106831e-05,
+      "loss": 0.6053,
+      "step": 20706
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3703024577075852e-05,
+      "loss": 0.5846,
+      "step": 20707
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.370072316435081e-05,
+      "loss": 0.6499,
+      "step": 20708
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3698421871956207e-05,
+      "loss": 0.6396,
+      "step": 20709
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3696120699916549e-05,
+      "loss": 0.6996,
+      "step": 20710
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3693819648256345e-05,
+      "loss": 0.6251,
+      "step": 20711
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3691518717000095e-05,
+      "loss": 0.6897,
+      "step": 20712
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3689217906172297e-05,
+      "loss": 0.6562,
+      "step": 20713
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.368691721579747e-05,
+      "loss": 0.5797,
+      "step": 20714
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3684616645900105e-05,
+      "loss": 0.6223,
+      "step": 20715
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3682316196504718e-05,
+      "loss": 0.6362,
+      "step": 20716
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3680015867635779e-05,
+      "loss": 0.6056,
+      "step": 20717
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3677715659317789e-05,
+      "loss": 0.5878,
+      "step": 20718
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3675415571575262e-05,
+      "loss": 0.6484,
+      "step": 20719
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3673115604432685e-05,
+      "loss": 0.5826,
+      "step": 20720
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3670815757914552e-05,
+      "loss": 0.6204,
+      "step": 20721
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3668516032045347e-05,
+      "loss": 0.6361,
+      "step": 20722
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3666216426849574e-05,
+      "loss": 0.6493,
+      "step": 20723
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3663916942351701e-05,
+      "loss": 0.6884,
+      "step": 20724
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.366161757857624e-05,
+      "loss": 0.606,
+      "step": 20725
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3659318335547671e-05,
+      "loss": 0.6074,
+      "step": 20726
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3657019213290476e-05,
+      "loss": 0.5968,
+      "step": 20727
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3654720211829136e-05,
+      "loss": 0.5981,
+      "step": 20728
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3652421331188142e-05,
+      "loss": 0.6592,
+      "step": 20729
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.365012257139196e-05,
+      "loss": 0.6438,
+      "step": 20730
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3647823932465093e-05,
+      "loss": 0.6883,
+      "step": 20731
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3645525414432008e-05,
+      "loss": 0.7057,
+      "step": 20732
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3643227017317184e-05,
+      "loss": 0.6765,
+      "step": 20733
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3640928741145098e-05,
+      "loss": 0.649,
+      "step": 20734
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3638630585940226e-05,
+      "loss": 0.6304,
+      "step": 20735
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3636332551727027e-05,
+      "loss": 0.6381,
+      "step": 20736
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3634034638530007e-05,
+      "loss": 0.6273,
+      "step": 20737
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.363173684637361e-05,
+      "loss": 0.6506,
+      "step": 20738
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3629439175282315e-05,
+      "loss": 0.6593,
+      "step": 20739
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3627141625280586e-05,
+      "loss": 0.7039,
+      "step": 20740
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3624844196392889e-05,
+      "loss": 0.6331,
+      "step": 20741
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.36225468886437e-05,
+      "loss": 0.6134,
+      "step": 20742
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3620249702057483e-05,
+      "loss": 0.6963,
+      "step": 20743
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3617952636658699e-05,
+      "loss": 0.7297,
+      "step": 20744
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3615655692471804e-05,
+      "loss": 0.6369,
+      "step": 20745
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3613358869521266e-05,
+      "loss": 0.5729,
+      "step": 20746
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3611062167831534e-05,
+      "loss": 0.6674,
+      "step": 20747
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3608765587427087e-05,
+      "loss": 0.583,
+      "step": 20748
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3606469128332369e-05,
+      "loss": 0.6649,
+      "step": 20749
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.360417279057184e-05,
+      "loss": 0.665,
+      "step": 20750
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3601876574169947e-05,
+      "loss": 0.6332,
+      "step": 20751
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3599580479151152e-05,
+      "loss": 0.6315,
+      "step": 20752
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3597284505539892e-05,
+      "loss": 0.6381,
+      "step": 20753
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3594988653360641e-05,
+      "loss": 0.63,
+      "step": 20754
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3592692922637837e-05,
+      "loss": 0.6494,
+      "step": 20755
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3590397313395925e-05,
+      "loss": 0.5739,
+      "step": 20756
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3588101825659366e-05,
+      "loss": 0.5968,
+      "step": 20757
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3585806459452578e-05,
+      "loss": 0.6144,
+      "step": 20758
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3583511214800027e-05,
+      "loss": 0.6835,
+      "step": 20759
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3581216091726152e-05,
+      "loss": 0.5728,
+      "step": 20760
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3578921090255391e-05,
+      "loss": 0.6209,
+      "step": 20761
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.357662621041219e-05,
+      "loss": 0.6325,
+      "step": 20762
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3574331452220985e-05,
+      "loss": 0.6227,
+      "step": 20763
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.35720368157062e-05,
+      "loss": 0.6658,
+      "step": 20764
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3569742300892296e-05,
+      "loss": 0.6277,
+      "step": 20765
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3567447907803695e-05,
+      "loss": 0.639,
+      "step": 20766
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3565153636464836e-05,
+      "loss": 0.59,
+      "step": 20767
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3562859486900149e-05,
+      "loss": 0.6313,
+      "step": 20768
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3560565459134061e-05,
+      "loss": 0.6541,
+      "step": 20769
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3558271553190998e-05,
+      "loss": 0.6836,
+      "step": 20770
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3555977769095407e-05,
+      "loss": 0.6424,
+      "step": 20771
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3553684106871705e-05,
+      "loss": 0.6513,
+      "step": 20772
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3551390566544317e-05,
+      "loss": 0.6331,
+      "step": 20773
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3549097148137669e-05,
+      "loss": 0.6511,
+      "step": 20774
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3546803851676184e-05,
+      "loss": 0.5838,
+      "step": 20775
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3544510677184274e-05,
+      "loss": 0.6213,
+      "step": 20776
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3542217624686392e-05,
+      "loss": 0.6348,
+      "step": 20777
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3539924694206929e-05,
+      "loss": 0.6313,
+      "step": 20778
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3537631885770307e-05,
+      "loss": 0.6278,
+      "step": 20779
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3535339199400948e-05,
+      "loss": 0.6144,
+      "step": 20780
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3533046635123254e-05,
+      "loss": 0.6283,
+      "step": 20781
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3530754192961661e-05,
+      "loss": 0.6737,
+      "step": 20782
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3528461872940572e-05,
+      "loss": 0.5985,
+      "step": 20783
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3526169675084405e-05,
+      "loss": 0.6342,
+      "step": 20784
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3523877599417561e-05,
+      "loss": 0.6129,
+      "step": 20785
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3521585645964454e-05,
+      "loss": 0.6608,
+      "step": 20786
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3519293814749481e-05,
+      "loss": 0.6863,
+      "step": 20787
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.351700210579707e-05,
+      "loss": 0.638,
+      "step": 20788
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3514710519131613e-05,
+      "loss": 0.6038,
+      "step": 20789
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3512419054777517e-05,
+      "loss": 0.6121,
+      "step": 20790
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3510127712759186e-05,
+      "loss": 0.6213,
+      "step": 20791
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3507836493101018e-05,
+      "loss": 0.6831,
+      "step": 20792
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3505545395827401e-05,
+      "loss": 0.6345,
+      "step": 20793
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3503254420962758e-05,
+      "loss": 0.63,
+      "step": 20794
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3500963568531478e-05,
+      "loss": 0.6348,
+      "step": 20795
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3498672838557952e-05,
+      "loss": 0.6059,
+      "step": 20796
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.349638223106659e-05,
+      "loss": 0.6391,
+      "step": 20797
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3494091746081755e-05,
+      "loss": 0.711,
+      "step": 20798
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3491801383627864e-05,
+      "loss": 0.6333,
+      "step": 20799
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3489511143729303e-05,
+      "loss": 0.6174,
+      "step": 20800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.348722102641046e-05,
+      "loss": 0.677,
+      "step": 20801
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3484931031695724e-05,
+      "loss": 0.7184,
+      "step": 20802
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.348264115960948e-05,
+      "loss": 0.6066,
+      "step": 20803
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3480351410176106e-05,
+      "loss": 0.5772,
+      "step": 20804
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3478061783420004e-05,
+      "loss": 0.6794,
+      "step": 20805
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3475772279365551e-05,
+      "loss": 0.5886,
+      "step": 20806
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3473482898037126e-05,
+      "loss": 0.6199,
+      "step": 20807
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3471193639459112e-05,
+      "loss": 0.6452,
+      "step": 20808
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3468904503655882e-05,
+      "loss": 0.6962,
+      "step": 20809
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3466615490651812e-05,
+      "loss": 0.5628,
+      "step": 20810
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3464326600471294e-05,
+      "loss": 0.6299,
+      "step": 20811
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3462037833138697e-05,
+      "loss": 0.6347,
+      "step": 20812
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3459749188678388e-05,
+      "loss": 0.6578,
+      "step": 20813
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3457460667114747e-05,
+      "loss": 0.6108,
+      "step": 20814
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.345517226847214e-05,
+      "loss": 0.5614,
+      "step": 20815
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.345288399277494e-05,
+      "loss": 0.5525,
+      "step": 20816
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3450595840047513e-05,
+      "loss": 0.601,
+      "step": 20817
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3448307810314234e-05,
+      "loss": 0.5865,
+      "step": 20818
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3446019903599461e-05,
+      "loss": 0.6172,
+      "step": 20819
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.344373211992756e-05,
+      "loss": 0.6231,
+      "step": 20820
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.344144445932289e-05,
+      "loss": 0.6348,
+      "step": 20821
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3439156921809824e-05,
+      "loss": 0.6978,
+      "step": 20822
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3436869507412719e-05,
+      "loss": 0.6432,
+      "step": 20823
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3434582216155933e-05,
+      "loss": 0.6139,
+      "step": 20824
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3432295048063826e-05,
+      "loss": 0.705,
+      "step": 20825
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3430008003160755e-05,
+      "loss": 0.6336,
+      "step": 20826
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3427721081471072e-05,
+      "loss": 0.5899,
+      "step": 20827
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3425434283019129e-05,
+      "loss": 0.6335,
+      "step": 20828
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.342314760782929e-05,
+      "loss": 0.6128,
+      "step": 20829
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3420861055925898e-05,
+      "loss": 0.63,
+      "step": 20830
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3418574627333308e-05,
+      "loss": 0.6007,
+      "step": 20831
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.341628832207587e-05,
+      "loss": 0.6331,
+      "step": 20832
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3414002140177925e-05,
+      "loss": 0.658,
+      "step": 20833
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3411716081663816e-05,
+      "loss": 0.7001,
+      "step": 20834
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3409430146557902e-05,
+      "loss": 0.5963,
+      "step": 20835
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3407144334884523e-05,
+      "loss": 0.6509,
+      "step": 20836
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3404858646668028e-05,
+      "loss": 0.7236,
+      "step": 20837
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3402573081932737e-05,
+      "loss": 0.6108,
+      "step": 20838
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3400287640702991e-05,
+      "loss": 0.6238,
+      "step": 20839
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3398002323003153e-05,
+      "loss": 0.6715,
+      "step": 20840
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3395717128857543e-05,
+      "loss": 0.5929,
+      "step": 20841
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3393432058290506e-05,
+      "loss": 0.6929,
+      "step": 20842
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3391147111326365e-05,
+      "loss": 0.6642,
+      "step": 20843
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3388862287989461e-05,
+      "loss": 0.6127,
+      "step": 20844
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3386577588304116e-05,
+      "loss": 0.5789,
+      "step": 20845
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3384293012294679e-05,
+      "loss": 0.6056,
+      "step": 20846
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3382008559985467e-05,
+      "loss": 0.6444,
+      "step": 20847
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3379724231400815e-05,
+      "loss": 0.6354,
+      "step": 20848
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3377440026565041e-05,
+      "loss": 0.5987,
+      "step": 20849
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3375155945502474e-05,
+      "loss": 0.6667,
+      "step": 20850
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3372871988237428e-05,
+      "loss": 0.5906,
+      "step": 20851
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.337058815479425e-05,
+      "loss": 0.587,
+      "step": 20852
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3368304445197244e-05,
+      "loss": 0.5885,
+      "step": 20853
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3366020859470737e-05,
+      "loss": 0.6368,
+      "step": 20854
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3363737397639043e-05,
+      "loss": 0.6461,
+      "step": 20855
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3361454059726483e-05,
+      "loss": 0.6302,
+      "step": 20856
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.335917084575736e-05,
+      "loss": 0.6492,
+      "step": 20857
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.335688775575602e-05,
+      "loss": 0.6452,
+      "step": 20858
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3354604789746747e-05,
+      "loss": 0.7391,
+      "step": 20859
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3352321947753862e-05,
+      "loss": 0.5981,
+      "step": 20860
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3350039229801681e-05,
+      "loss": 0.6248,
+      "step": 20861
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3347756635914494e-05,
+      "loss": 0.6925,
+      "step": 20862
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3345474166116637e-05,
+      "loss": 0.6558,
+      "step": 20863
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3343191820432404e-05,
+      "loss": 0.5793,
+      "step": 20864
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.33409095988861e-05,
+      "loss": 0.5981,
+      "step": 20865
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3338627501502032e-05,
+      "loss": 0.669,
+      "step": 20866
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.33363455283045e-05,
+      "loss": 0.6407,
+      "step": 20867
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3334063679317798e-05,
+      "loss": 0.6242,
+      "step": 20868
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3331781954566248e-05,
+      "loss": 0.6083,
+      "step": 20869
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3329500354074131e-05,
+      "loss": 0.6828,
+      "step": 20870
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3327218877865752e-05,
+      "loss": 0.6738,
+      "step": 20871
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3324937525965406e-05,
+      "loss": 0.6528,
+      "step": 20872
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3322656298397385e-05,
+      "loss": 0.6073,
+      "step": 20873
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3320375195185977e-05,
+      "loss": 0.6701,
+      "step": 20874
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3318094216355492e-05,
+      "loss": 0.5885,
+      "step": 20875
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3315813361930213e-05,
+      "loss": 0.6918,
+      "step": 20876
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3313532631934426e-05,
+      "loss": 0.5893,
+      "step": 20877
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.331125202639243e-05,
+      "loss": 0.6867,
+      "step": 20878
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3308971545328487e-05,
+      "loss": 0.6757,
+      "step": 20879
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3306691188766906e-05,
+      "loss": 0.558,
+      "step": 20880
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3304410956731966e-05,
+      "loss": 0.5751,
+      "step": 20881
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3302130849247948e-05,
+      "loss": 0.6416,
+      "step": 20882
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3299850866339137e-05,
+      "loss": 0.7485,
+      "step": 20883
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3297571008029808e-05,
+      "loss": 0.6035,
+      "step": 20884
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3295291274344234e-05,
+      "loss": 0.6254,
+      "step": 20885
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3293011665306716e-05,
+      "loss": 0.5952,
+      "step": 20886
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3290732180941508e-05,
+      "loss": 0.6506,
+      "step": 20887
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.32884528212729e-05,
+      "loss": 0.6177,
+      "step": 20888
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3286173586325157e-05,
+      "loss": 0.6674,
+      "step": 20889
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3283894476122555e-05,
+      "loss": 0.613,
+      "step": 20890
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3281615490689354e-05,
+      "loss": 0.6483,
+      "step": 20891
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3279336630049837e-05,
+      "loss": 0.637,
+      "step": 20892
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3277057894228276e-05,
+      "loss": 0.6326,
+      "step": 20893
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3274779283248928e-05,
+      "loss": 0.6178,
+      "step": 20894
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3272500797136066e-05,
+      "loss": 0.5787,
+      "step": 20895
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3270222435913948e-05,
+      "loss": 0.5981,
+      "step": 20896
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.326794419960683e-05,
+      "loss": 0.6782,
+      "step": 20897
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3265666088239004e-05,
+      "loss": 0.6181,
+      "step": 20898
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.32633881018347e-05,
+      "loss": 0.5885,
+      "step": 20899
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3261110240418184e-05,
+      "loss": 0.5769,
+      "step": 20900
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3258832504013721e-05,
+      "loss": 0.5944,
+      "step": 20901
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3256554892645553e-05,
+      "loss": 0.6076,
+      "step": 20902
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3254277406337956e-05,
+      "loss": 0.611,
+      "step": 20903
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.325200004511517e-05,
+      "loss": 0.6458,
+      "step": 20904
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3249722809001456e-05,
+      "loss": 0.5873,
+      "step": 20905
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3247445698021057e-05,
+      "loss": 0.5814,
+      "step": 20906
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.324516871219823e-05,
+      "loss": 0.6021,
+      "step": 20907
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3242891851557205e-05,
+      "loss": 0.6136,
+      "step": 20908
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3240615116122257e-05,
+      "loss": 0.5978,
+      "step": 20909
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.323833850591762e-05,
+      "loss": 0.5301,
+      "step": 20910
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3236062020967533e-05,
+      "loss": 0.6961,
+      "step": 20911
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3233785661296244e-05,
+      "loss": 0.628,
+      "step": 20912
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3231509426927995e-05,
+      "loss": 0.5831,
+      "step": 20913
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3229233317887019e-05,
+      "loss": 0.5717,
+      "step": 20914
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.322695733419757e-05,
+      "loss": 0.6861,
+      "step": 20915
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3224681475883876e-05,
+      "loss": 0.595,
+      "step": 20916
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3222405742970175e-05,
+      "loss": 0.6657,
+      "step": 20917
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3220130135480715e-05,
+      "loss": 0.6597,
+      "step": 20918
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3217854653439698e-05,
+      "loss": 0.6961,
+      "step": 20919
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3215579296871384e-05,
+      "loss": 0.5967,
+      "step": 20920
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3213304065799995e-05,
+      "loss": 0.6528,
+      "step": 20921
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3211028960249764e-05,
+      "loss": 0.6815,
+      "step": 20922
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3208753980244915e-05,
+      "loss": 0.6737,
+      "step": 20923
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3206479125809679e-05,
+      "loss": 0.5851,
+      "step": 20924
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.320420439696827e-05,
+      "loss": 0.5898,
+      "step": 20925
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3201929793744933e-05,
+      "loss": 0.6388,
+      "step": 20926
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.319965531616388e-05,
+      "loss": 0.6462,
+      "step": 20927
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3197380964249334e-05,
+      "loss": 0.607,
+      "step": 20928
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3195106738025517e-05,
+      "loss": 0.5784,
+      "step": 20929
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3192832637516645e-05,
+      "loss": 0.6311,
+      "step": 20930
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3190558662746939e-05,
+      "loss": 0.6724,
+      "step": 20931
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3188284813740598e-05,
+      "loss": 0.6003,
+      "step": 20932
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3186011090521868e-05,
+      "loss": 0.6472,
+      "step": 20933
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3183737493114945e-05,
+      "loss": 0.5828,
+      "step": 20934
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3181464021544043e-05,
+      "loss": 0.6375,
+      "step": 20935
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3179190675833377e-05,
+      "loss": 0.5916,
+      "step": 20936
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3176917456007154e-05,
+      "loss": 0.6832,
+      "step": 20937
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3174644362089578e-05,
+      "loss": 0.5824,
+      "step": 20938
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3172371394104866e-05,
+      "loss": 0.6746,
+      "step": 20939
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3170098552077214e-05,
+      "loss": 0.6047,
+      "step": 20940
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3167825836030827e-05,
+      "loss": 0.6197,
+      "step": 20941
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3165553245989918e-05,
+      "loss": 0.6294,
+      "step": 20942
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.316328078197867e-05,
+      "loss": 0.6366,
+      "step": 20943
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3161008444021305e-05,
+      "loss": 0.5936,
+      "step": 20944
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.315873623214201e-05,
+      "loss": 0.6134,
+      "step": 20945
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3156464146364988e-05,
+      "loss": 0.6363,
+      "step": 20946
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3154192186714432e-05,
+      "loss": 0.6668,
+      "step": 20947
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3151920353214536e-05,
+      "loss": 0.6689,
+      "step": 20948
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3149648645889487e-05,
+      "loss": 0.6147,
+      "step": 20949
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3147377064763495e-05,
+      "loss": 0.6533,
+      "step": 20950
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3145105609860741e-05,
+      "loss": 0.6608,
+      "step": 20951
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3142834281205413e-05,
+      "loss": 0.6171,
+      "step": 20952
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3140563078821704e-05,
+      "loss": 0.7351,
+      "step": 20953
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3138292002733798e-05,
+      "loss": 0.7417,
+      "step": 20954
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3136021052965868e-05,
+      "loss": 0.6139,
+      "step": 20955
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3133750229542125e-05,
+      "loss": 0.6064,
+      "step": 20956
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3131479532486738e-05,
+      "loss": 0.6875,
+      "step": 20957
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3129208961823885e-05,
+      "loss": 0.6935,
+      "step": 20958
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3126938517577762e-05,
+      "loss": 0.6731,
+      "step": 20959
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3124668199772513e-05,
+      "loss": 0.667,
+      "step": 20960
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3122398008432351e-05,
+      "loss": 0.6103,
+      "step": 20961
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3120127943581438e-05,
+      "loss": 0.6418,
+      "step": 20962
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3117858005243949e-05,
+      "loss": 0.6559,
+      "step": 20963
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3115588193444059e-05,
+      "loss": 0.6726,
+      "step": 20964
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3113318508205941e-05,
+      "loss": 0.578,
+      "step": 20965
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3111048949553751e-05,
+      "loss": 0.6574,
+      "step": 20966
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3108779517511682e-05,
+      "loss": 0.5822,
+      "step": 20967
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3106510212103895e-05,
+      "loss": 0.6068,
+      "step": 20968
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3104241033354553e-05,
+      "loss": 0.648,
+      "step": 20969
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.310197198128782e-05,
+      "loss": 0.7358,
+      "step": 20970
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3099703055927865e-05,
+      "loss": 0.5516,
+      "step": 20971
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3097434257298835e-05,
+      "loss": 0.6727,
+      "step": 20972
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3095165585424913e-05,
+      "loss": 0.6551,
+      "step": 20973
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3092897040330252e-05,
+      "loss": 0.5953,
+      "step": 20974
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3090628622039009e-05,
+      "loss": 0.6327,
+      "step": 20975
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3088360330575338e-05,
+      "loss": 0.6136,
+      "step": 20976
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3086092165963399e-05,
+      "loss": 0.6427,
+      "step": 20977
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3083824128227335e-05,
+      "loss": 0.6817,
+      "step": 20978
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3081556217391331e-05,
+      "loss": 0.6141,
+      "step": 20979
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3079288433479506e-05,
+      "loss": 0.6605,
+      "step": 20980
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3077020776516025e-05,
+      "loss": 0.6514,
+      "step": 20981
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3074753246525028e-05,
+      "loss": 0.7219,
+      "step": 20982
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3072485843530664e-05,
+      "loss": 0.6178,
+      "step": 20983
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3070218567557094e-05,
+      "loss": 0.6117,
+      "step": 20984
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3067951418628452e-05,
+      "loss": 0.6222,
+      "step": 20985
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3065684396768885e-05,
+      "loss": 0.5986,
+      "step": 20986
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.306341750200253e-05,
+      "loss": 0.6223,
+      "step": 20987
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3061150734353536e-05,
+      "loss": 0.6094,
+      "step": 20988
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3058884093846025e-05,
+      "loss": 0.5999,
+      "step": 20989
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.305661758050416e-05,
+      "loss": 0.5929,
+      "step": 20990
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3054351194352068e-05,
+      "loss": 0.6588,
+      "step": 20991
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3052084935413882e-05,
+      "loss": 0.6638,
+      "step": 20992
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.304981880371374e-05,
+      "loss": 0.6509,
+      "step": 20993
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.304755279927577e-05,
+      "loss": 0.6313,
+      "step": 20994
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3045286922124097e-05,
+      "loss": 0.6601,
+      "step": 20995
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.304302117228287e-05,
+      "loss": 0.6962,
+      "step": 20996
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.304075554977621e-05,
+      "loss": 0.6094,
+      "step": 20997
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3038490054628241e-05,
+      "loss": 0.5675,
+      "step": 20998
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3036224686863103e-05,
+      "loss": 0.6254,
+      "step": 20999
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3033959446504884e-05,
+      "loss": 0.5884,
+      "step": 21000
+    }
+  ],
+  "max_steps": 31385,
+  "num_train_epochs": 5,
+  "total_flos": 8.62762112472685e+18,
+  "trial_name": null,
+  "trial_params": null
+}