diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.12815255279885177,
+  "eval_steps": 500,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7079419299743807e-09,
+      "loss": 2.8359,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4158838599487614e-09,
+      "loss": 2.4531,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.123825789923143e-09,
+      "loss": 2.2266,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.831767719897523e-09,
+      "loss": 2.5566,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.539709649871904e-09,
+      "loss": 2.7461,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0247651579846285e-08,
+      "loss": 2.5,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1955593509820665e-08,
+      "loss": 2.4414,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3663535439795045e-08,
+      "loss": 2.5781,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5371477369769426e-08,
+      "loss": 2.8516,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.707941929974381e-08,
+      "loss": 2.3164,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8787361229718188e-08,
+      "loss": 2.6289,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.049530315969257e-08,
+      "loss": 2.1953,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.220324508966695e-08,
+      "loss": 2.8398,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.391118701964133e-08,
+      "loss": 2.748,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.561912894961571e-08,
+      "loss": 2.5137,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.732707087959009e-08,
+      "loss": 2.5234,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9035012809564473e-08,
+      "loss": 3.0508,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.074295473953885e-08,
+      "loss": 2.9062,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2450896669513235e-08,
+      "loss": 2.25,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.415883859948762e-08,
+      "loss": 2.2539,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5866780529461994e-08,
+      "loss": 2.1191,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7574722459436376e-08,
+      "loss": 2.3359,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.928266438941076e-08,
+      "loss": 2.041,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.099060631938514e-08,
+      "loss": 2.2305,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.269854824935952e-08,
+      "loss": 2.4023,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.44064901793339e-08,
+      "loss": 3.2969,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.611443210930828e-08,
+      "loss": 2.0156,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.782237403928266e-08,
+      "loss": 2.4844,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.953031596925705e-08,
+      "loss": 2.8926,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.123825789923142e-08,
+      "loss": 2.2285,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2946199829205806e-08,
+      "loss": 2.832,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.465414175918018e-08,
+      "loss": 2.7812,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6362083689154564e-08,
+      "loss": 1.9473,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.807002561912895e-08,
+      "loss": 2.3086,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.977796754910333e-08,
+      "loss": 2.4219,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.14859094790777e-08,
+      "loss": 2.8789,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.31938514090521e-08,
+      "loss": 2.0215,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.490179333902647e-08,
+      "loss": 2.2305,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.660973526900085e-08,
+      "loss": 2.3164,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.831767719897524e-08,
+      "loss": 2.2891,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.002561912894961e-08,
+      "loss": 2.0488,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.173356105892399e-08,
+      "loss": 2.1172,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.344150298889836e-08,
+      "loss": 2.4062,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.514944491887275e-08,
+      "loss": 2.1133,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.685738684884714e-08,
+      "loss": 2.084,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.856532877882152e-08,
+      "loss": 2.0781,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.02732707087959e-08,
+      "loss": 1.7383,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.198121263877028e-08,
+      "loss": 1.9531,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.368915456874466e-08,
+      "loss": 1.916,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.539709649871903e-08,
+      "loss": 1.9609,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.710503842869342e-08,
+      "loss": 2.1758,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.88129803586678e-08,
+      "loss": 2.0781,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.052092228864218e-08,
+      "loss": 2.0742,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.222886421861656e-08,
+      "loss": 1.9629,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.393680614859094e-08,
+      "loss": 1.9492,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.564474807856532e-08,
+      "loss": 1.75,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.73526900085397e-08,
+      "loss": 1.7578,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.90606319385141e-08,
+      "loss": 1.707,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0076857386848847e-07,
+      "loss": 1.9395,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0247651579846285e-07,
+      "loss": 1.6738,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0418445772843724e-07,
+      "loss": 1.8789,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0589239965841161e-07,
+      "loss": 1.7871,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0760034158838599e-07,
+      "loss": 1.6641,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0930828351836036e-07,
+      "loss": 1.6641,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1101622544833475e-07,
+      "loss": 1.7539,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1272416737830913e-07,
+      "loss": 1.6777,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.144321093082835e-07,
+      "loss": 1.4297,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.161400512382579e-07,
+      "loss": 1.5352,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1784799316823227e-07,
+      "loss": 1.6836,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1955593509820666e-07,
+      "loss": 1.373,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2126387702818105e-07,
+      "loss": 1.4453,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.229718189581554e-07,
+      "loss": 1.5762,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.246797608881298e-07,
+      "loss": 1.7363,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.263877028181042e-07,
+      "loss": 1.54,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2809564474807855e-07,
+      "loss": 1.4727,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2980358667805294e-07,
+      "loss": 1.5723,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3151152860802733e-07,
+      "loss": 1.5781,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.332194705380017e-07,
+      "loss": 1.8008,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3492741246797608e-07,
+      "loss": 1.4004,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3663535439795047e-07,
+      "loss": 1.6807,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3834329632792486e-07,
+      "loss": 1.4199,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4005123825789922e-07,
+      "loss": 1.2354,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.417591801878736e-07,
+      "loss": 1.2344,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4346712211784797e-07,
+      "loss": 1.6914,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4517506404782236e-07,
+      "loss": 1.5156,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4688300597779673e-07,
+      "loss": 1.3789,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4859094790777114e-07,
+      "loss": 1.6191,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.502988898377455e-07,
+      "loss": 1.4922,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.520068317677199e-07,
+      "loss": 1.4766,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5371477369769428e-07,
+      "loss": 1.165,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5542271562766865e-07,
+      "loss": 1.3555,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5713065755764303e-07,
+      "loss": 1.3516,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.588385994876174e-07,
+      "loss": 1.543,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.605465414175918e-07,
+      "loss": 1.416,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6225448334756618e-07,
+      "loss": 1.4863,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6396242527754057e-07,
+      "loss": 1.4414,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6567036720751493e-07,
+      "loss": 1.3887,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6737830913748932e-07,
+      "loss": 1.4336,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6908625106746368e-07,
+      "loss": 1.3652,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7079419299743807e-07,
+      "loss": 1.4316,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7250213492741248e-07,
+      "loss": 1.0703,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7421007685738685e-07,
+      "loss": 1.4082,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7591801878736124e-07,
+      "loss": 1.1934,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.776259607173356e-07,
+      "loss": 1.4219,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7933390264731e-07,
+      "loss": 1.3633,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8104184457728435e-07,
+      "loss": 1.4551,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8274978650725874e-07,
+      "loss": 1.0342,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8445772843723313e-07,
+      "loss": 1.0078,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8616567036720752e-07,
+      "loss": 1.3086,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8787361229718188e-07,
+      "loss": 1.3477,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8958155422715627e-07,
+      "loss": 1.2305,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9128949615713063e-07,
+      "loss": 1.3457,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9299743808710502e-07,
+      "loss": 1.2266,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.947053800170794e-07,
+      "loss": 0.9448,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.964133219470538e-07,
+      "loss": 1.0908,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.981212638770282e-07,
+      "loss": 1.1914,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9982920580700255e-07,
+      "loss": 1.2041,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0153714773697694e-07,
+      "loss": 1.498,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.032450896669513e-07,
+      "loss": 1.5176,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.049530315969257e-07,
+      "loss": 1.4189,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0666097352690006e-07,
+      "loss": 1.3984,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0836891545687447e-07,
+      "loss": 1.3066,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1007685738684883e-07,
+      "loss": 1.1553,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1178479931682322e-07,
+      "loss": 1.3281,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1349274124679759e-07,
+      "loss": 1.1719,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1520068317677198e-07,
+      "loss": 1.0166,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1690862510674636e-07,
+      "loss": 1.0801,
+      "step": 127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1861656703672073e-07,
+      "loss": 1.2656,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2032450896669514e-07,
+      "loss": 1.1465,
+      "step": 129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.220324508966695e-07,
+      "loss": 1.0635,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.237403928266439e-07,
+      "loss": 1.2129,
+      "step": 131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2544833475661826e-07,
+      "loss": 1.126,
+      "step": 132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2715627668659265e-07,
+      "loss": 1.2422,
+      "step": 133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.28864218616567e-07,
+      "loss": 1.1582,
+      "step": 134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.305721605465414e-07,
+      "loss": 1.0537,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.322801024765158e-07,
+      "loss": 1.2129,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3398804440649018e-07,
+      "loss": 1.1875,
+      "step": 137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3569598633646454e-07,
+      "loss": 0.9678,
+      "step": 138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3740392826643893e-07,
+      "loss": 1.2578,
+      "step": 139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.391118701964133e-07,
+      "loss": 1.1621,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.408198121263877e-07,
+      "loss": 1.2871,
+      "step": 141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.425277540563621e-07,
+      "loss": 1.0742,
+      "step": 142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4423569598633643e-07,
+      "loss": 1.0781,
+      "step": 143
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.459436379163108e-07,
+      "loss": 1.1885,
+      "step": 144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.476515798462852e-07,
+      "loss": 1.1855,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.493595217762596e-07,
+      "loss": 1.2354,
+      "step": 146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.51067463706234e-07,
+      "loss": 1.0176,
+      "step": 147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.527754056362084e-07,
+      "loss": 1.0557,
+      "step": 148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.544833475661827e-07,
+      "loss": 1.1113,
+      "step": 149
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.561912894961571e-07,
+      "loss": 1.4492,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.578992314261315e-07,
+      "loss": 1.1611,
+      "step": 151
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.596071733561059e-07,
+      "loss": 1.1807,
+      "step": 152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6131511528608027e-07,
+      "loss": 1.25,
+      "step": 153
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6302305721605466e-07,
+      "loss": 1.1367,
+      "step": 154
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6473099914602905e-07,
+      "loss": 1.2207,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.664389410760034e-07,
+      "loss": 1.2891,
+      "step": 156
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.681468830059778e-07,
+      "loss": 1.2412,
+      "step": 157
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6985482493595216e-07,
+      "loss": 0.9463,
+      "step": 158
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7156276686592655e-07,
+      "loss": 1.2891,
+      "step": 159
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7327070879590094e-07,
+      "loss": 1.2041,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.749786507258753e-07,
+      "loss": 1.0771,
+      "step": 161
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.766865926558497e-07,
+      "loss": 1.2969,
+      "step": 162
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7839453458582406e-07,
+      "loss": 1.4609,
+      "step": 163
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8010247651579845e-07,
+      "loss": 0.9658,
+      "step": 164
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8181041844577283e-07,
+      "loss": 1.0439,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.835183603757472e-07,
+      "loss": 1.1113,
+      "step": 166
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.852263023057216e-07,
+      "loss": 1.3906,
+      "step": 167
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8693424423569595e-07,
+      "loss": 1.0508,
+      "step": 168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.886421861656704e-07,
+      "loss": 1.1279,
+      "step": 169
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9035012809564473e-07,
+      "loss": 1.2852,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.920580700256191e-07,
+      "loss": 0.8525,
+      "step": 171
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9376601195559345e-07,
+      "loss": 1.25,
+      "step": 172
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.954739538855679e-07,
+      "loss": 1.0869,
+      "step": 173
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.971818958155423e-07,
+      "loss": 1.2266,
+      "step": 174
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.988898377455166e-07,
+      "loss": 1.4424,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.00597779675491e-07,
+      "loss": 0.9077,
+      "step": 176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.023057216054654e-07,
+      "loss": 1.1211,
+      "step": 177
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.040136635354398e-07,
+      "loss": 1.252,
+      "step": 178
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.057216054654141e-07,
+      "loss": 1.0664,
+      "step": 179
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0742954739538857e-07,
+      "loss": 1.0996,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0913748932536296e-07,
+      "loss": 1.1709,
+      "step": 181
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.108454312553373e-07,
+      "loss": 0.8936,
+      "step": 182
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125533731853117e-07,
+      "loss": 1.2891,
+      "step": 183
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1426131511528607e-07,
+      "loss": 1.0977,
+      "step": 184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1596925704526046e-07,
+      "loss": 1.4473,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.176771989752348e-07,
+      "loss": 1.3115,
+      "step": 186
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.193851409052092e-07,
+      "loss": 1.0586,
+      "step": 187
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.210930828351836e-07,
+      "loss": 1.3301,
+      "step": 188
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2280102476515796e-07,
+      "loss": 1.126,
+      "step": 189
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2450896669513235e-07,
+      "loss": 0.9668,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2621690862510674e-07,
+      "loss": 1.1221,
+      "step": 191
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2792485055508113e-07,
+      "loss": 1.291,
+      "step": 192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2963279248505547e-07,
+      "loss": 1.1934,
+      "step": 193
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3134073441502986e-07,
+      "loss": 1.1357,
+      "step": 194
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.330486763450043e-07,
+      "loss": 1.1934,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3475661827497863e-07,
+      "loss": 1.2578,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.36464560204953e-07,
+      "loss": 1.1631,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3817250213492736e-07,
+      "loss": 1.1602,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.398804440649018e-07,
+      "loss": 1.248,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4158838599487614e-07,
+      "loss": 1.1162,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4329632792485053e-07,
+      "loss": 1.1123,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4500426985482497e-07,
+      "loss": 1.1084,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.467122117847993e-07,
+      "loss": 1.2285,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.484201537147737e-07,
+      "loss": 1.1211,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5012809564474803e-07,
+      "loss": 1.0586,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5183603757472247e-07,
+      "loss": 1.1143,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.535439795046968e-07,
+      "loss": 1.2773,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.552519214346712e-07,
+      "loss": 1.0332,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.569598633646456e-07,
+      "loss": 1.0371,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5866780529462e-07,
+      "loss": 0.8853,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6037574722459437e-07,
+      "loss": 1.1816,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.620836891545687e-07,
+      "loss": 1.1865,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.637916310845431e-07,
+      "loss": 1.2402,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.654995730145175e-07,
+      "loss": 1.1846,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6720751494449187e-07,
+      "loss": 1.2129,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6891545687446626e-07,
+      "loss": 1.1211,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7062339880444065e-07,
+      "loss": 1.002,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7233134073441504e-07,
+      "loss": 0.915,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7403928266438937e-07,
+      "loss": 1.1201,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7574722459436376e-07,
+      "loss": 1.0869,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7745516652433815e-07,
+      "loss": 1.0088,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7916310845431254e-07,
+      "loss": 0.9395,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8087105038428693e-07,
+      "loss": 1.1514,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8257899231426127e-07,
+      "loss": 0.9434,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.842869342442357e-07,
+      "loss": 0.9873,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8599487617421004e-07,
+      "loss": 1.1006,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8770281810418443e-07,
+      "loss": 1.0166,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.894107600341588e-07,
+      "loss": 1.2129,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.911187019641332e-07,
+      "loss": 1.1572,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.928266438941076e-07,
+      "loss": 0.9033,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9453458582408194e-07,
+      "loss": 0.959,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.962425277540564e-07,
+      "loss": 1.1064,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.979504696840307e-07,
+      "loss": 1.1416,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.996584116140051e-07,
+      "loss": 1.3711,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0136635354397944e-07,
+      "loss": 1.043,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.030742954739539e-07,
+      "loss": 1.0303,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0478223740392827e-07,
+      "loss": 0.8579,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.064901793339026e-07,
+      "loss": 1.1992,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.08198121263877e-07,
+      "loss": 0.8105,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.099060631938514e-07,
+      "loss": 1.0303,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.116140051238258e-07,
+      "loss": 1.0039,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.133219470538001e-07,
+      "loss": 1.3086,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1502988898377455e-07,
+      "loss": 1.1338,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1673783091374894e-07,
+      "loss": 1.0273,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.184457728437233e-07,
+      "loss": 0.9795,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2015371477369767e-07,
+      "loss": 0.9775,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2186165670367206e-07,
+      "loss": 1.1836,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2356959863364645e-07,
+      "loss": 1.1309,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.252775405636208e-07,
+      "loss": 1.1484,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2698548249359517e-07,
+      "loss": 0.7554,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.286934244235696e-07,
+      "loss": 0.9375,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3040136635354395e-07,
+      "loss": 1.1152,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3210930828351834e-07,
+      "loss": 0.9209,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3381725021349273e-07,
+      "loss": 0.9922,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.355251921434671e-07,
+      "loss": 1.0566,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3723313407344145e-07,
+      "loss": 1.0693,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3894107600341584e-07,
+      "loss": 1.0098,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.406490179333903e-07,
+      "loss": 1.2002,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.423569598633646e-07,
+      "loss": 1.0488,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.44064901793339e-07,
+      "loss": 1.1133,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4577284372331335e-07,
+      "loss": 1.0498,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.474807856532878e-07,
+      "loss": 0.9395,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.491887275832621e-07,
+      "loss": 1.0898,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.508966695132365e-07,
+      "loss": 1.3242,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5260461144321096e-07,
+      "loss": 0.8994,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.543125533731853e-07,
+      "loss": 1.0664,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.560204953031597e-07,
+      "loss": 0.8447,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.57728437233134e-07,
+      "loss": 0.9893,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5943637916310846e-07,
+      "loss": 1.0898,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.611443210930828e-07,
+      "loss": 1.2002,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.628522630230572e-07,
+      "loss": 0.9106,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.645602049530316e-07,
+      "loss": 1.2363,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6626814688300596e-07,
+      "loss": 1.3057,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6797608881298035e-07,
+      "loss": 0.9131,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.696840307429547e-07,
+      "loss": 1.0869,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.713919726729291e-07,
+      "loss": 1.0244,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.730999146029035e-07,
+      "loss": 1.0195,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7480785653287786e-07,
+      "loss": 1.0977,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7651579846285225e-07,
+      "loss": 0.9365,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.782237403928266e-07,
+      "loss": 1.2754,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.79931682322801e-07,
+      "loss": 1.0928,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.816396242527754e-07,
+      "loss": 1.1465,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.833475661827497e-07,
+      "loss": 0.856,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.850555081127242e-07,
+      "loss": 1.1621,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.867634500426985e-07,
+      "loss": 1.0645,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.884713919726729e-07,
+      "loss": 1.0635,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.901793339026473e-07,
+      "loss": 1.1826,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.918872758326216e-07,
+      "loss": 0.9922,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.935952177625961e-07,
+      "loss": 1.2051,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953031596925704e-07,
+      "loss": 1.0996,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.970111016225449e-07,
+      "loss": 1.1465,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987190435525192e-07,
+      "loss": 0.959,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.004269854824936e-07,
+      "loss": 1.0752,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.02134927412468e-07,
+      "loss": 1.0166,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.038428693424423e-07,
+      "loss": 1.1826,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.055508112724168e-07,
+      "loss": 1.1201,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.072587532023911e-07,
+      "loss": 1.1641,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.089666951323654e-07,
+      "loss": 0.9453,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.106746370623399e-07,
+      "loss": 1.1025,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.123825789923142e-07,
+      "loss": 1.1543,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.140905209222886e-07,
+      "loss": 1.0498,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.15798462852263e-07,
+      "loss": 0.9502,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.175064047822374e-07,
+      "loss": 1.1162,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.192143467122118e-07,
+      "loss": 0.8975,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.209222886421861e-07,
+      "loss": 1.4521,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.226302305721605e-07,
+      "loss": 1.1406,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.243381725021349e-07,
+      "loss": 1.1289,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.260461144321093e-07,
+      "loss": 1.064,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.277540563620837e-07,
+      "loss": 0.9507,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.294619982920581e-07,
+      "loss": 1.1094,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.311699402220324e-07,
+      "loss": 0.9355,
+      "step": 311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.328778821520068e-07,
+      "loss": 1.1289,
+      "step": 312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.345858240819812e-07,
+      "loss": 1.2461,
+      "step": 313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.362937660119555e-07,
+      "loss": 1.1348,
+      "step": 314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3800170794193e-07,
+      "loss": 0.9053,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.397096498719043e-07,
+      "loss": 1.1006,
+      "step": 316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.414175918018788e-07,
+      "loss": 1.0391,
+      "step": 317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.431255337318531e-07,
+      "loss": 0.9883,
+      "step": 318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.448334756618274e-07,
+      "loss": 1.1436,
+      "step": 319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.465414175918019e-07,
+      "loss": 1.0342,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.482493595217762e-07,
+      "loss": 0.8926,
+      "step": 321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.499573014517506e-07,
+      "loss": 0.9258,
+      "step": 322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.51665243381725e-07,
+      "loss": 0.9902,
+      "step": 323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.533731853116994e-07,
+      "loss": 0.9766,
+      "step": 324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.550811272416738e-07,
+      "loss": 1.0791,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.567890691716481e-07,
+      "loss": 0.8574,
+      "step": 326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.584970111016226e-07,
+      "loss": 1.1416,
+      "step": 327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.602049530315969e-07,
+      "loss": 1.0605,
+      "step": 328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.619128949615712e-07,
+      "loss": 0.9629,
+      "step": 329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.636208368915457e-07,
+      "loss": 1.3828,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.653287788215201e-07,
+      "loss": 1.0889,
+      "step": 331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.670367207514944e-07,
+      "loss": 1.0586,
+      "step": 332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.687446626814688e-07,
+      "loss": 1.0518,
+      "step": 333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.704526046114432e-07,
+      "loss": 0.9277,
+      "step": 334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.721605465414176e-07,
+      "loss": 1.1211,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.738684884713919e-07,
+      "loss": 1.084,
+      "step": 336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.755764304013662e-07,
+      "loss": 0.9268,
+      "step": 337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.772843723313408e-07,
+      "loss": 1.1348,
+      "step": 338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.789923142613151e-07,
+      "loss": 1.0332,
+      "step": 339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.807002561912895e-07,
+      "loss": 0.8828,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.824081981212639e-07,
+      "loss": 0.8628,
+      "step": 341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.841161400512382e-07,
+      "loss": 1.0225,
+      "step": 342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.858240819812126e-07,
+      "loss": 0.6987,
+      "step": 343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.875320239111869e-07,
+      "loss": 0.9922,
+      "step": 344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.892399658411615e-07,
+      "loss": 0.9609,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.909479077711358e-07,
+      "loss": 1.0303,
+      "step": 346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.926558497011101e-07,
+      "loss": 1.0186,
+      "step": 347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.943637916310846e-07,
+      "loss": 0.9668,
+      "step": 348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.960717335610589e-07,
+      "loss": 1.0811,
+      "step": 349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.977796754910332e-07,
+      "loss": 1.0498,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.994876174210076e-07,
+      "loss": 1.2266,
+      "step": 351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.01195559350982e-07,
+      "loss": 1.1455,
+      "step": 352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.029035012809565e-07,
+      "loss": 0.8936,
+      "step": 353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.046114432109308e-07,
+      "loss": 0.9014,
+      "step": 354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.063193851409052e-07,
+      "loss": 0.9902,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.080273270708796e-07,
+      "loss": 1.1016,
+      "step": 356
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.097352690008539e-07,
+      "loss": 0.9785,
+      "step": 357
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.114432109308282e-07,
+      "loss": 0.9844,
+      "step": 358
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.131511528608027e-07,
+      "loss": 1.0771,
+      "step": 359
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.148590947907771e-07,
+      "loss": 0.9385,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.165670367207515e-07,
+      "loss": 0.9951,
+      "step": 361
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.182749786507259e-07,
+      "loss": 1.0596,
+      "step": 362
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.199829205807002e-07,
+      "loss": 0.8716,
+      "step": 363
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.216908625106746e-07,
+      "loss": 0.9834,
+      "step": 364
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.233988044406489e-07,
+      "loss": 1.1426,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.251067463706234e-07,
+      "loss": 1.0342,
+      "step": 366
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.268146883005978e-07,
+      "loss": 1.1494,
+      "step": 367
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.285226302305721e-07,
+      "loss": 1.1699,
+      "step": 368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.302305721605466e-07,
+      "loss": 1.1826,
+      "step": 369
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.319385140905209e-07,
+      "loss": 0.9854,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.336464560204953e-07,
+      "loss": 0.8887,
+      "step": 371
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.353543979504696e-07,
+      "loss": 0.7451,
+      "step": 372
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.37062339880444e-07,
+      "loss": 0.9766,
+      "step": 373
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.387702818104184e-07,
+      "loss": 1.3242,
+      "step": 374
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.404782237403928e-07,
+      "loss": 1.0547,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.421861656703673e-07,
+      "loss": 0.9727,
+      "step": 376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.438941076003416e-07,
+      "loss": 0.6865,
+      "step": 377
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.456020495303159e-07,
+      "loss": 0.9463,
+      "step": 378
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.473099914602903e-07,
+      "loss": 0.9258,
+      "step": 379
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.490179333902647e-07,
+      "loss": 1.1016,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.50725875320239e-07,
+      "loss": 0.9629,
+      "step": 381
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.524338172502135e-07,
+      "loss": 0.9453,
+      "step": 382
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.541417591801879e-07,
+      "loss": 0.8516,
+      "step": 383
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.558497011101623e-07,
+      "loss": 0.9092,
+      "step": 384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.575576430401366e-07,
+      "loss": 1.1934,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.592655849701109e-07,
+      "loss": 1.0176,
+      "step": 386
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.609735269000854e-07,
+      "loss": 0.9209,
+      "step": 387
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.626814688300597e-07,
+      "loss": 0.8428,
+      "step": 388
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.64389410760034e-07,
+      "loss": 1.083,
+      "step": 389
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.660973526900086e-07,
+      "loss": 1.0059,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.678052946199829e-07,
+      "loss": 0.9551,
+      "step": 391
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.695132365499573e-07,
+      "loss": 1.2363,
+      "step": 392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.712211784799316e-07,
+      "loss": 0.8877,
+      "step": 393
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.72929120409906e-07,
+      "loss": 1.2549,
+      "step": 394
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.746370623398804e-07,
+      "loss": 1.0254,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.763450042698547e-07,
+      "loss": 0.9619,
+      "step": 396
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.780529461998293e-07,
+      "loss": 0.8408,
+      "step": 397
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.797608881298036e-07,
+      "loss": 0.8213,
+      "step": 398
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.814688300597779e-07,
+      "loss": 0.9805,
+      "step": 399
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.831767719897523e-07,
+      "loss": 1.1094,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.848847139197267e-07,
+      "loss": 0.9609,
+      "step": 401
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.865926558497011e-07,
+      "loss": 1.1543,
+      "step": 402
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.883005977796754e-07,
+      "loss": 0.8228,
+      "step": 403
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.900085397096499e-07,
+      "loss": 0.957,
+      "step": 404
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.917164816396243e-07,
+      "loss": 0.9033,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.934244235695986e-07,
+      "loss": 1.1436,
+      "step": 406
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.951323654995729e-07,
+      "loss": 1.1338,
+      "step": 407
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.968403074295474e-07,
+      "loss": 0.9258,
+      "step": 408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.985482493595217e-07,
+      "loss": 1.0791,
+      "step": 409
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.002561912894961e-07,
+      "loss": 1.166,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.019641332194705e-07,
+      "loss": 1.021,
+      "step": 411
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.036720751494449e-07,
+      "loss": 1.0732,
+      "step": 412
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.053800170794193e-07,
+      "loss": 1.0522,
+      "step": 413
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.070879590093936e-07,
+      "loss": 1.0146,
+      "step": 414
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.087959009393681e-07,
+      "loss": 0.9131,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.105038428693424e-07,
+      "loss": 1.0723,
+      "step": 416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.122117847993167e-07,
+      "loss": 0.814,
+      "step": 417
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.139197267292912e-07,
+      "loss": 1.2051,
+      "step": 418
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.156276686592656e-07,
+      "loss": 0.9053,
+      "step": 419
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1733561058924e-07,
+      "loss": 0.8457,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.190435525192143e-07,
+      "loss": 1.0205,
+      "step": 421
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.207514944491887e-07,
+      "loss": 1.1445,
+      "step": 422
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.224594363791631e-07,
+      "loss": 0.9727,
+      "step": 423
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.241673783091374e-07,
+      "loss": 1.2812,
+      "step": 424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.258753202391118e-07,
+      "loss": 1.1436,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.275832621690862e-07,
+      "loss": 1.0098,
+      "step": 426
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.292912040990606e-07,
+      "loss": 0.9795,
+      "step": 427
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.30999146029035e-07,
+      "loss": 1.0107,
+      "step": 428
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.327070879590094e-07,
+      "loss": 0.9326,
+      "step": 429
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.344150298889837e-07,
+      "loss": 0.811,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.361229718189581e-07,
+      "loss": 1.0801,
+      "step": 431
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.378309137489325e-07,
+      "loss": 1.0322,
+      "step": 432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.395388556789069e-07,
+      "loss": 0.8447,
+      "step": 433
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.412467976088813e-07,
+      "loss": 0.916,
+      "step": 434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.429547395388556e-07,
+      "loss": 1.0225,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.446626814688301e-07,
+      "loss": 0.8467,
+      "step": 436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.463706233988044e-07,
+      "loss": 1.0645,
+      "step": 437
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.480785653287787e-07,
+      "loss": 1.2012,
+      "step": 438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.497865072587532e-07,
+      "loss": 1.2334,
+      "step": 439
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.514944491887275e-07,
+      "loss": 0.9316,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.53202391118702e-07,
+      "loss": 0.9658,
+      "step": 441
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.549103330486763e-07,
+      "loss": 0.9922,
+      "step": 442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.566182749786507e-07,
+      "loss": 1.0156,
+      "step": 443
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.583262169086251e-07,
+      "loss": 1.0127,
+      "step": 444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.600341588385994e-07,
+      "loss": 0.918,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.617421007685739e-07,
+      "loss": 1.0498,
+      "step": 446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.634500426985482e-07,
+      "loss": 0.959,
+      "step": 447
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.651579846285225e-07,
+      "loss": 0.9277,
+      "step": 448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.66865926558497e-07,
+      "loss": 0.9609,
+      "step": 449
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.685738684884714e-07,
+      "loss": 0.7563,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.702818104184458e-07,
+      "loss": 1.3008,
+      "step": 451
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.719897523484201e-07,
+      "loss": 1.2383,
+      "step": 452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.736976942783945e-07,
+      "loss": 0.9932,
+      "step": 453
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.754056362083689e-07,
+      "loss": 1.0752,
+      "step": 454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.771135781383432e-07,
+      "loss": 0.9736,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.788215200683176e-07,
+      "loss": 1.2686,
+      "step": 456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.805294619982921e-07,
+      "loss": 1.0439,
+      "step": 457
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.822374039282664e-07,
+      "loss": 1.0811,
+      "step": 458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.839453458582408e-07,
+      "loss": 1.1406,
+      "step": 459
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.856532877882152e-07,
+      "loss": 1.0122,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.873612297181895e-07,
+      "loss": 1.1104,
+      "step": 461
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.890691716481639e-07,
+      "loss": 1.0459,
+      "step": 462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.907771135781382e-07,
+      "loss": 0.9883,
+      "step": 463
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.924850555081128e-07,
+      "loss": 1.207,
+      "step": 464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.941929974380871e-07,
+      "loss": 0.9541,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.959009393680614e-07,
+      "loss": 1.0352,
+      "step": 466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.976088812980359e-07,
+      "loss": 0.9136,
+      "step": 467
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.993168232280102e-07,
+      "loss": 1.0059,
+      "step": 468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.010247651579845e-07,
+      "loss": 0.8223,
+      "step": 469
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.027327070879589e-07,
+      "loss": 0.8262,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.044406490179334e-07,
+      "loss": 1.0186,
+      "step": 471
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.061485909479078e-07,
+      "loss": 0.7524,
+      "step": 472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.078565328778821e-07,
+      "loss": 0.8809,
+      "step": 473
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.095644748078565e-07,
+      "loss": 0.8242,
+      "step": 474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.112724167378309e-07,
+      "loss": 1.0674,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.129803586678052e-07,
+      "loss": 1.0117,
+      "step": 476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.146883005977796e-07,
+      "loss": 1.0762,
+      "step": 477
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.16396242527754e-07,
+      "loss": 0.9668,
+      "step": 478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.181041844577284e-07,
+      "loss": 1.0527,
+      "step": 479
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.198121263877028e-07,
+      "loss": 1.0928,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.215200683176772e-07,
+      "loss": 1.1113,
+      "step": 481
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.232280102476516e-07,
+      "loss": 0.8477,
+      "step": 482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.249359521776259e-07,
+      "loss": 0.9634,
+      "step": 483
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.266438941076002e-07,
+      "loss": 0.9795,
+      "step": 484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.283518360375747e-07,
+      "loss": 1.0674,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.300597779675491e-07,
+      "loss": 1.0791,
+      "step": 486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.317677198975234e-07,
+      "loss": 1.0176,
+      "step": 487
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.334756618274979e-07,
+      "loss": 0.9053,
+      "step": 488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.351836037574722e-07,
+      "loss": 0.9131,
+      "step": 489
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.368915456874466e-07,
+      "loss": 1.2344,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.385994876174209e-07,
+      "loss": 1.0029,
+      "step": 491
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.403074295473953e-07,
+      "loss": 1.0283,
+      "step": 492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.420153714773698e-07,
+      "loss": 0.9668,
+      "step": 493
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.437233134073441e-07,
+      "loss": 1.2949,
+      "step": 494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.454312553373186e-07,
+      "loss": 1.0537,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.471391972672929e-07,
+      "loss": 1.0264,
+      "step": 496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.488471391972672e-07,
+      "loss": 0.9355,
+      "step": 497
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.505550811272416e-07,
+      "loss": 1.123,
+      "step": 498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.52263023057216e-07,
+      "loss": 0.9248,
+      "step": 499
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.539709649871903e-07,
+      "loss": 1.0684,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.556789069171648e-07,
+      "loss": 0.9834,
+      "step": 501
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.573868488471392e-07,
+      "loss": 0.9893,
+      "step": 502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.590947907771136e-07,
+      "loss": 1.1426,
+      "step": 503
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.608027327070879e-07,
+      "loss": 0.8242,
+      "step": 504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.625106746370622e-07,
+      "loss": 1.1387,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.642186165670367e-07,
+      "loss": 0.833,
+      "step": 506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.65926558497011e-07,
+      "loss": 1.1562,
+      "step": 507
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.676345004269855e-07,
+      "loss": 0.9736,
+      "step": 508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.693424423569599e-07,
+      "loss": 1.168,
+      "step": 509
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.710503842869342e-07,
+      "loss": 1.0205,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.727583262169086e-07,
+      "loss": 0.9839,
+      "step": 511
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.744662681468829e-07,
+      "loss": 0.9004,
+      "step": 512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.761742100768573e-07,
+      "loss": 1.1885,
+      "step": 513
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.778821520068317e-07,
+      "loss": 1.1719,
+      "step": 514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.79590093936806e-07,
+      "loss": 1.1523,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.812980358667806e-07,
+      "loss": 1.084,
+      "step": 516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.830059777967549e-07,
+      "loss": 1.209,
+      "step": 517
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.847139197267292e-07,
+      "loss": 0.8115,
+      "step": 518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.864218616567036e-07,
+      "loss": 1.0215,
+      "step": 519
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.88129803586678e-07,
+      "loss": 1.042,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.898377455166524e-07,
+      "loss": 1.0703,
+      "step": 521
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.915456874466267e-07,
+      "loss": 0.9746,
+      "step": 522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.932536293766012e-07,
+      "loss": 0.9473,
+      "step": 523
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.949615713065756e-07,
+      "loss": 0.9346,
+      "step": 524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.966695132365499e-07,
+      "loss": 0.9453,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.983774551665242e-07,
+      "loss": 1.0654,
+      "step": 526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.000853970964987e-07,
+      "loss": 1.0488,
+      "step": 527
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.01793339026473e-07,
+      "loss": 0.9424,
+      "step": 528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.035012809564474e-07,
+      "loss": 0.8389,
+      "step": 529
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.052092228864219e-07,
+      "loss": 1.1895,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.069171648163962e-07,
+      "loss": 1.0527,
+      "step": 531
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.086251067463706e-07,
+      "loss": 0.9111,
+      "step": 532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.103330486763449e-07,
+      "loss": 0.9941,
+      "step": 533
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.120409906063194e-07,
+      "loss": 1.0127,
+      "step": 534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.137489325362937e-07,
+      "loss": 1.0547,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.15456874466268e-07,
+      "loss": 0.9854,
+      "step": 536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.171648163962425e-07,
+      "loss": 0.9316,
+      "step": 537
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.188727583262169e-07,
+      "loss": 0.9326,
+      "step": 538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.205807002561913e-07,
+      "loss": 1.0537,
+      "step": 539
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.222886421861656e-07,
+      "loss": 0.917,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.2399658411614e-07,
+      "loss": 0.9658,
+      "step": 541
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.257045260461144e-07,
+      "loss": 1.0029,
+      "step": 542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.274124679760887e-07,
+      "loss": 0.9424,
+      "step": 543
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.291204099060631e-07,
+      "loss": 0.8638,
+      "step": 544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.308283518360376e-07,
+      "loss": 1.0195,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.325362937660119e-07,
+      "loss": 1.0195,
+      "step": 546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.342442356959863e-07,
+      "loss": 0.9619,
+      "step": 547
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.359521776259607e-07,
+      "loss": 1.2627,
+      "step": 548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.37660119555935e-07,
+      "loss": 1.0508,
+      "step": 549
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.393680614859094e-07,
+      "loss": 0.7354,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.410760034158838e-07,
+      "loss": 1.0186,
+      "step": 551
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.427839453458582e-07,
+      "loss": 1.0352,
+      "step": 552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.444918872758326e-07,
+      "loss": 0.9785,
+      "step": 553
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.46199829205807e-07,
+      "loss": 0.9395,
+      "step": 554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.479077711357814e-07,
+      "loss": 0.959,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.496157130657557e-07,
+      "loss": 0.9268,
+      "step": 556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.5132365499573e-07,
+      "loss": 0.9746,
+      "step": 557
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.530315969257045e-07,
+      "loss": 0.7124,
+      "step": 558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.547395388556788e-07,
+      "loss": 0.8438,
+      "step": 559
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.564474807856533e-07,
+      "loss": 1.1055,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.581554227156277e-07,
+      "loss": 0.8066,
+      "step": 561
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.59863364645602e-07,
+      "loss": 0.8984,
+      "step": 562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615713065755764e-07,
+      "loss": 0.9697,
+      "step": 563
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.632792485055508e-07,
+      "loss": 0.9766,
+      "step": 564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.64987190435525e-07,
+      "loss": 1.0391,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.666951323654995e-07,
+      "loss": 0.8633,
+      "step": 566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.68403074295474e-07,
+      "loss": 1.002,
+      "step": 567
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.701110162254484e-07,
+      "loss": 0.9912,
+      "step": 568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.718189581554226e-07,
+      "loss": 1.0107,
+      "step": 569
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.73526900085397e-07,
+      "loss": 0.9365,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.752348420153715e-07,
+      "loss": 0.9604,
+      "step": 571
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.769427839453457e-07,
+      "loss": 0.709,
+      "step": 572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.786507258753202e-07,
+      "loss": 0.8633,
+      "step": 573
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.803586678052946e-07,
+      "loss": 1.0596,
+      "step": 574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.82066609735269e-07,
+      "loss": 1.1348,
+      "step": 575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.837745516652433e-07,
+      "loss": 1.0977,
+      "step": 576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.854824935952177e-07,
+      "loss": 0.7944,
+      "step": 577
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.871904355251922e-07,
+      "loss": 0.9824,
+      "step": 578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.888983774551664e-07,
+      "loss": 1.0898,
+      "step": 579
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.906063193851408e-07,
+      "loss": 0.8818,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.923142613151153e-07,
+      "loss": 1.1143,
+      "step": 581
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.940222032450897e-07,
+      "loss": 0.8672,
+      "step": 582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.95730145175064e-07,
+      "loss": 1.0283,
+      "step": 583
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.974380871050384e-07,
+      "loss": 0.8799,
+      "step": 584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.991460290350128e-07,
+      "loss": 1.1914,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0008539709649873e-06,
+      "loss": 0.8599,
+      "step": 586
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0025619128949615e-06,
+      "loss": 1.0684,
+      "step": 587
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.004269854824936e-06,
+      "loss": 1.001,
+      "step": 588
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0059777967549104e-06,
+      "loss": 0.9033,
+      "step": 589
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0076857386848846e-06,
+      "loss": 1.1318,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.009393680614859e-06,
+      "loss": 0.8853,
+      "step": 591
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0111016225448335e-06,
+      "loss": 0.9932,
+      "step": 592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0128095644748077e-06,
+      "loss": 0.873,
+      "step": 593
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0145175064047822e-06,
+      "loss": 1.0547,
+      "step": 594
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0162254483347566e-06,
+      "loss": 0.9053,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0179333902647309e-06,
+      "loss": 0.9551,
+      "step": 596
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0196413321947053e-06,
+      "loss": 1.0186,
+      "step": 597
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0213492741246797e-06,
+      "loss": 0.9121,
+      "step": 598
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.023057216054654e-06,
+      "loss": 1.0146,
+      "step": 599
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0247651579846284e-06,
+      "loss": 1.1279,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0264730999146029e-06,
+      "loss": 0.9238,
+      "step": 601
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0281810418445773e-06,
+      "loss": 1.0566,
+      "step": 602
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0298889837745517e-06,
+      "loss": 0.897,
+      "step": 603
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.031596925704526e-06,
+      "loss": 0.9668,
+      "step": 604
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0333048676345004e-06,
+      "loss": 0.8691,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0350128095644749e-06,
+      "loss": 0.8613,
+      "step": 606
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.036720751494449e-06,
+      "loss": 1.1328,
+      "step": 607
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0384286934244235e-06,
+      "loss": 1.043,
+      "step": 608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.040136635354398e-06,
+      "loss": 0.7407,
+      "step": 609
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0418445772843722e-06,
+      "loss": 1.1055,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0435525192143466e-06,
+      "loss": 1.042,
+      "step": 611
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.045260461144321e-06,
+      "loss": 1.0234,
+      "step": 612
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0469684030742953e-06,
+      "loss": 1.0381,
+      "step": 613
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0486763450042698e-06,
+      "loss": 0.9014,
+      "step": 614
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.050384286934244e-06,
+      "loss": 0.9463,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0520922288642186e-06,
+      "loss": 1.1914,
+      "step": 616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.053800170794193e-06,
+      "loss": 0.9873,
+      "step": 617
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0555081127241673e-06,
+      "loss": 0.9131,
+      "step": 618
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0572160546541418e-06,
+      "loss": 1.1064,
+      "step": 619
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0589239965841162e-06,
+      "loss": 1.1885,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0606319385140904e-06,
+      "loss": 0.9678,
+      "step": 621
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0623398804440649e-06,
+      "loss": 0.9121,
+      "step": 622
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0640478223740393e-06,
+      "loss": 0.9619,
+      "step": 623
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0657557643040135e-06,
+      "loss": 1.1855,
+      "step": 624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.067463706233988e-06,
+      "loss": 1.0947,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0691716481639624e-06,
+      "loss": 1.0938,
+      "step": 626
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0708795900939367e-06,
+      "loss": 0.6758,
+      "step": 627
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.072587532023911e-06,
+      "loss": 0.9932,
+      "step": 628
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0742954739538853e-06,
+      "loss": 0.9453,
+      "step": 629
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.07600341588386e-06,
+      "loss": 0.9561,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0777113578138344e-06,
+      "loss": 0.7319,
+      "step": 631
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0794192997438087e-06,
+      "loss": 0.9922,
+      "step": 632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.081127241673783e-06,
+      "loss": 1.0449,
+      "step": 633
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0828351836037575e-06,
+      "loss": 1.0742,
+      "step": 634
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0845431255337318e-06,
+      "loss": 1.1143,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0862510674637062e-06,
+      "loss": 1.1494,
+      "step": 636
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0879590093936807e-06,
+      "loss": 0.8838,
+      "step": 637
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0896669513236549e-06,
+      "loss": 0.9619,
+      "step": 638
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0913748932536293e-06,
+      "loss": 0.915,
+      "step": 639
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0930828351836038e-06,
+      "loss": 0.9707,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.094790777113578e-06,
+      "loss": 1.0088,
+      "step": 641
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0964987190435524e-06,
+      "loss": 1.1182,
+      "step": 642
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0982066609735267e-06,
+      "loss": 1.1133,
+      "step": 643
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0999146029035011e-06,
+      "loss": 1.0312,
+      "step": 644
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1016225448334758e-06,
+      "loss": 0.9873,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.10333048676345e-06,
+      "loss": 1.1318,
+      "step": 646
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1050384286934244e-06,
+      "loss": 0.9844,
+      "step": 647
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1067463706233989e-06,
+      "loss": 0.9072,
+      "step": 648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1084543125533731e-06,
+      "loss": 1.0332,
+      "step": 649
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1101622544833476e-06,
+      "loss": 0.8164,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.111870196413322e-06,
+      "loss": 0.9629,
+      "step": 651
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1135781383432962e-06,
+      "loss": 0.6284,
+      "step": 652
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1152860802732707e-06,
+      "loss": 0.8906,
+      "step": 653
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1169940222032451e-06,
+      "loss": 1.0557,
+      "step": 654
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1187019641332193e-06,
+      "loss": 1.0605,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1204099060631938e-06,
+      "loss": 1.2363,
+      "step": 656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.122117847993168e-06,
+      "loss": 1.0537,
+      "step": 657
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1238257899231425e-06,
+      "loss": 0.9785,
+      "step": 658
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.125533731853117e-06,
+      "loss": 0.9521,
+      "step": 659
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1272416737830913e-06,
+      "loss": 1.0654,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1289496157130658e-06,
+      "loss": 0.6738,
+      "step": 661
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1306575576430402e-06,
+      "loss": 0.7402,
+      "step": 662
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1323654995730145e-06,
+      "loss": 1.084,
+      "step": 663
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.134073441502989e-06,
+      "loss": 0.9072,
+      "step": 664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1357813834329633e-06,
+      "loss": 1.0918,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1374893253629376e-06,
+      "loss": 0.8584,
+      "step": 666
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.139197267292912e-06,
+      "loss": 1.1641,
+      "step": 667
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1409052092228865e-06,
+      "loss": 0.8535,
+      "step": 668
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1426131511528607e-06,
+      "loss": 0.9209,
+      "step": 669
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1443210930828351e-06,
+      "loss": 0.9756,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1460290350128094e-06,
+      "loss": 1.1562,
+      "step": 671
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1477369769427838e-06,
+      "loss": 0.9678,
+      "step": 672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1494449188727582e-06,
+      "loss": 0.8994,
+      "step": 673
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1511528608027325e-06,
+      "loss": 0.9141,
+      "step": 674
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1528608027327071e-06,
+      "loss": 0.8994,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1545687446626816e-06,
+      "loss": 1.166,
+      "step": 676
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1562766865926558e-06,
+      "loss": 0.9707,
+      "step": 677
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1579846285226302e-06,
+      "loss": 1.1758,
+      "step": 678
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1596925704526047e-06,
+      "loss": 0.9434,
+      "step": 679
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.161400512382579e-06,
+      "loss": 1.0176,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1631084543125534e-06,
+      "loss": 1.2666,
+      "step": 681
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1648163962425278e-06,
+      "loss": 1.0596,
+      "step": 682
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.166524338172502e-06,
+      "loss": 0.9473,
+      "step": 683
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1682322801024765e-06,
+      "loss": 1.0088,
+      "step": 684
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1699402220324507e-06,
+      "loss": 0.9141,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1716481639624251e-06,
+      "loss": 1.0713,
+      "step": 686
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1733561058923996e-06,
+      "loss": 0.9824,
+      "step": 687
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1750640478223738e-06,
+      "loss": 0.9385,
+      "step": 688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1767719897523483e-06,
+      "loss": 1.0166,
+      "step": 689
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.178479931682323e-06,
+      "loss": 0.8662,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1801878736122971e-06,
+      "loss": 0.9346,
+      "step": 691
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1818958155422716e-06,
+      "loss": 0.7891,
+      "step": 692
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.183603757472246e-06,
+      "loss": 1.0342,
+      "step": 693
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1853116994022203e-06,
+      "loss": 0.8042,
+      "step": 694
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1870196413321947e-06,
+      "loss": 1.0547,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1887275832621691e-06,
+      "loss": 1.0205,
+      "step": 696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1904355251921434e-06,
+      "loss": 0.9878,
+      "step": 697
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1921434671221178e-06,
+      "loss": 0.9795,
+      "step": 698
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.193851409052092e-06,
+      "loss": 0.9092,
+      "step": 699
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1955593509820665e-06,
+      "loss": 0.8379,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.197267292912041e-06,
+      "loss": 0.9785,
+      "step": 701
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1989752348420152e-06,
+      "loss": 1.0195,
+      "step": 702
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2006831767719896e-06,
+      "loss": 0.8945,
+      "step": 703
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.202391118701964e-06,
+      "loss": 1.0625,
+      "step": 704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2040990606319385e-06,
+      "loss": 0.8818,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.205807002561913e-06,
+      "loss": 0.7749,
+      "step": 706
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2075149444918874e-06,
+      "loss": 1.0859,
+      "step": 707
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2092228864218616e-06,
+      "loss": 1.1035,
+      "step": 708
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.210930828351836e-06,
+      "loss": 1.083,
+      "step": 709
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2126387702818105e-06,
+      "loss": 1.1172,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2143467122117847e-06,
+      "loss": 0.8496,
+      "step": 711
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2160546541417592e-06,
+      "loss": 0.8213,
+      "step": 712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2177625960717334e-06,
+      "loss": 1.0283,
+      "step": 713
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2194705380017078e-06,
+      "loss": 1.0127,
+      "step": 714
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2211784799316823e-06,
+      "loss": 0.998,
+      "step": 715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2228864218616565e-06,
+      "loss": 1.0264,
+      "step": 716
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.224594363791631e-06,
+      "loss": 0.9131,
+      "step": 717
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2263023057216054e-06,
+      "loss": 0.9854,
+      "step": 718
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2280102476515798e-06,
+      "loss": 0.9233,
+      "step": 719
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2297181895815543e-06,
+      "loss": 1.0811,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2314261315115287e-06,
+      "loss": 0.8799,
+      "step": 721
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.233134073441503e-06,
+      "loss": 1.0508,
+      "step": 722
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2348420153714774e-06,
+      "loss": 1.0869,
+      "step": 723
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2365499573014518e-06,
+      "loss": 1.002,
+      "step": 724
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.238257899231426e-06,
+      "loss": 1.0117,
+      "step": 725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2399658411614005e-06,
+      "loss": 0.8574,
+      "step": 726
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2416737830913747e-06,
+      "loss": 1.3379,
+      "step": 727
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2433817250213492e-06,
+      "loss": 0.9717,
+      "step": 728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2450896669513236e-06,
+      "loss": 0.8809,
+      "step": 729
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2467976088812978e-06,
+      "loss": 0.8301,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2485055508112723e-06,
+      "loss": 0.7637,
+      "step": 731
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2502134927412467e-06,
+      "loss": 0.6958,
+      "step": 732
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.251921434671221e-06,
+      "loss": 1.1045,
+      "step": 733
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2536293766011956e-06,
+      "loss": 0.8911,
+      "step": 734
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25533731853117e-06,
+      "loss": 0.8438,
+      "step": 735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2570452604611443e-06,
+      "loss": 1.1592,
+      "step": 736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2587532023911187e-06,
+      "loss": 0.8154,
+      "step": 737
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2604611443210932e-06,
+      "loss": 0.9756,
+      "step": 738
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2621690862510674e-06,
+      "loss": 1.0635,
+      "step": 739
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2638770281810418e-06,
+      "loss": 0.9199,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.265584970111016e-06,
+      "loss": 0.7734,
+      "step": 741
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2672929120409905e-06,
+      "loss": 1.0293,
+      "step": 742
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.269000853970965e-06,
+      "loss": 0.9814,
+      "step": 743
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2707087959009392e-06,
+      "loss": 0.9409,
+      "step": 744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2724167378309136e-06,
+      "loss": 0.998,
+      "step": 745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.274124679760888e-06,
+      "loss": 1.0283,
+      "step": 746
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2758326216908623e-06,
+      "loss": 1.1074,
+      "step": 747
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2775405636208367e-06,
+      "loss": 0.6064,
+      "step": 748
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2792485055508114e-06,
+      "loss": 0.8022,
+      "step": 749
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2809564474807856e-06,
+      "loss": 1.0547,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.28266438941076e-06,
+      "loss": 0.9316,
+      "step": 751
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2843723313407345e-06,
+      "loss": 1.085,
+      "step": 752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2860802732707087e-06,
+      "loss": 0.8472,
+      "step": 753
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2877882152006832e-06,
+      "loss": 1.0645,
+      "step": 754
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2894961571306574e-06,
+      "loss": 1.0449,
+      "step": 755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2912040990606319e-06,
+      "loss": 1.0264,
+      "step": 756
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2929120409906063e-06,
+      "loss": 1.0244,
+      "step": 757
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2946199829205805e-06,
+      "loss": 0.8545,
+      "step": 758
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.296327924850555e-06,
+      "loss": 1.1943,
+      "step": 759
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2980358667805294e-06,
+      "loss": 0.998,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2997438087105036e-06,
+      "loss": 0.9805,
+      "step": 761
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.301451750640478e-06,
+      "loss": 1.1426,
+      "step": 762
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3031596925704525e-06,
+      "loss": 1.1133,
+      "step": 763
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.304867634500427e-06,
+      "loss": 1.1699,
+      "step": 764
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3065755764304014e-06,
+      "loss": 0.9077,
+      "step": 765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3082835183603758e-06,
+      "loss": 0.8193,
+      "step": 766
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.30999146029035e-06,
+      "loss": 0.7754,
+      "step": 767
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3116994022203245e-06,
+      "loss": 1.376,
+      "step": 768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.313407344150299e-06,
+      "loss": 1.0488,
+      "step": 769
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3151152860802732e-06,
+      "loss": 1.043,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3168232280102476e-06,
+      "loss": 0.9512,
+      "step": 771
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3185311699402219e-06,
+      "loss": 0.9775,
+      "step": 772
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3202391118701963e-06,
+      "loss": 1.0596,
+      "step": 773
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3219470538001707e-06,
+      "loss": 1.1182,
+      "step": 774
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.323654995730145e-06,
+      "loss": 0.9014,
+      "step": 775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3253629376601194e-06,
+      "loss": 1.0137,
+      "step": 776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3270708795900939e-06,
+      "loss": 1.1074,
+      "step": 777
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.328778821520068e-06,
+      "loss": 0.8018,
+      "step": 778
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3304867634500427e-06,
+      "loss": 1.0273,
+      "step": 779
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3321947053800172e-06,
+      "loss": 1.123,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3339026473099914e-06,
+      "loss": 1.1133,
+      "step": 781
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3356105892399659e-06,
+      "loss": 1.0449,
+      "step": 782
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3373185311699403e-06,
+      "loss": 1.001,
+      "step": 783
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3390264730999145e-06,
+      "loss": 0.8457,
+      "step": 784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.340734415029889e-06,
+      "loss": 1.0039,
+      "step": 785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3424423569598632e-06,
+      "loss": 0.959,
+      "step": 786
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3441502988898376e-06,
+      "loss": 1.0107,
+      "step": 787
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.345858240819812e-06,
+      "loss": 0.8447,
+      "step": 788
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3475661827497863e-06,
+      "loss": 1.0781,
+      "step": 789
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3492741246797608e-06,
+      "loss": 0.9902,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3509820666097352e-06,
+      "loss": 1.0771,
+      "step": 791
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3526900085397094e-06,
+      "loss": 1.0342,
+      "step": 792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3543979504696839e-06,
+      "loss": 1.0869,
+      "step": 793
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3561058923996585e-06,
+      "loss": 0.9365,
+      "step": 794
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3578138343296328e-06,
+      "loss": 1.0195,
+      "step": 795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3595217762596072e-06,
+      "loss": 1.04,
+      "step": 796
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3612297181895816e-06,
+      "loss": 0.9233,
+      "step": 797
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3629376601195559e-06,
+      "loss": 1.1133,
+      "step": 798
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3646456020495303e-06,
+      "loss": 0.9268,
+      "step": 799
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3663535439795045e-06,
+      "loss": 1.1152,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.368061485909479e-06,
+      "loss": 0.9072,
+      "step": 801
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3697694278394534e-06,
+      "loss": 0.9512,
+      "step": 802
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3714773697694277e-06,
+      "loss": 1.0371,
+      "step": 803
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3731853116994021e-06,
+      "loss": 0.8809,
+      "step": 804
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3748932536293765e-06,
+      "loss": 0.7393,
+      "step": 805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3766011955593508e-06,
+      "loss": 1.1621,
+      "step": 806
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3783091374893252e-06,
+      "loss": 0.9854,
+      "step": 807
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3800170794192999e-06,
+      "loss": 0.728,
+      "step": 808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.381725021349274e-06,
+      "loss": 0.8242,
+      "step": 809
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3834329632792485e-06,
+      "loss": 0.8809,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.385140905209223e-06,
+      "loss": 0.834,
+      "step": 811
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3868488471391972e-06,
+      "loss": 0.8906,
+      "step": 812
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3885567890691717e-06,
+      "loss": 0.9229,
+      "step": 813
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3902647309991459e-06,
+      "loss": 0.9111,
+      "step": 814
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3919726729291203e-06,
+      "loss": 0.9824,
+      "step": 815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3936806148590948e-06,
+      "loss": 1.04,
+      "step": 816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.395388556789069e-06,
+      "loss": 0.8945,
+      "step": 817
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3970964987190434e-06,
+      "loss": 1.0371,
+      "step": 818
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3988044406490179e-06,
+      "loss": 1.0391,
+      "step": 819
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4005123825789921e-06,
+      "loss": 0.9834,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4022203245089666e-06,
+      "loss": 1.0117,
+      "step": 821
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.403928266438941e-06,
+      "loss": 1.1016,
+      "step": 822
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4056362083689154e-06,
+      "loss": 0.9141,
+      "step": 823
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4073441502988899e-06,
+      "loss": 0.8418,
+      "step": 824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4090520922288643e-06,
+      "loss": 0.8779,
+      "step": 825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4107600341588386e-06,
+      "loss": 0.9258,
+      "step": 826
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.412467976088813e-06,
+      "loss": 1.0723,
+      "step": 827
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4141759180187872e-06,
+      "loss": 1.0674,
+      "step": 828
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4158838599487617e-06,
+      "loss": 1.2344,
+      "step": 829
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4175918018787361e-06,
+      "loss": 0.8555,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4192997438087103e-06,
+      "loss": 1.168,
+      "step": 831
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4210076857386848e-06,
+      "loss": 0.8145,
+      "step": 832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4227156276686592e-06,
+      "loss": 1.208,
+      "step": 833
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4244235695986335e-06,
+      "loss": 0.9678,
+      "step": 834
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.426131511528608e-06,
+      "loss": 1.0273,
+      "step": 835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4278394534585823e-06,
+      "loss": 1.0703,
+      "step": 836
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4295473953885566e-06,
+      "loss": 1.1152,
+      "step": 837
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4312553373185312e-06,
+      "loss": 1.001,
+      "step": 838
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4329632792485057e-06,
+      "loss": 1.0566,
+      "step": 839
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.43467122117848e-06,
+      "loss": 0.8623,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4363791631084543e-06,
+      "loss": 0.9131,
+      "step": 841
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4380871050384286e-06,
+      "loss": 1.0273,
+      "step": 842
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.439795046968403e-06,
+      "loss": 0.9912,
+      "step": 843
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4415029888983775e-06,
+      "loss": 1.0156,
+      "step": 844
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4432109308283517e-06,
+      "loss": 0.7725,
+      "step": 845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4449188727583261e-06,
+      "loss": 0.9414,
+      "step": 846
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4466268146883006e-06,
+      "loss": 0.8857,
+      "step": 847
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4483347566182748e-06,
+      "loss": 0.7764,
+      "step": 848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4500426985482492e-06,
+      "loss": 1.0166,
+      "step": 849
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4517506404782237e-06,
+      "loss": 0.8965,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.453458582408198e-06,
+      "loss": 0.9111,
+      "step": 851
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4551665243381724e-06,
+      "loss": 0.9004,
+      "step": 852
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.456874466268147e-06,
+      "loss": 1.082,
+      "step": 853
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4585824081981212e-06,
+      "loss": 0.8086,
+      "step": 854
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4602903501280957e-06,
+      "loss": 0.9932,
+      "step": 855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.46199829205807e-06,
+      "loss": 0.9355,
+      "step": 856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4637062339880444e-06,
+      "loss": 1.0879,
+      "step": 857
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4654141759180188e-06,
+      "loss": 1.082,
+      "step": 858
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.467122117847993e-06,
+      "loss": 0.834,
+      "step": 859
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4688300597779675e-06,
+      "loss": 0.8706,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.470538001707942e-06,
+      "loss": 1.0312,
+      "step": 861
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4722459436379161e-06,
+      "loss": 0.9141,
+      "step": 862
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4739538855678906e-06,
+      "loss": 1.3086,
+      "step": 863
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.475661827497865e-06,
+      "loss": 0.9463,
+      "step": 864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4773697694278393e-06,
+      "loss": 1.0088,
+      "step": 865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4790777113578137e-06,
+      "loss": 1.1201,
+      "step": 866
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.480785653287788e-06,
+      "loss": 1.0791,
+      "step": 867
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4824935952177626e-06,
+      "loss": 0.8369,
+      "step": 868
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.484201537147737e-06,
+      "loss": 0.9839,
+      "step": 869
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4859094790777113e-06,
+      "loss": 0.9873,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4876174210076857e-06,
+      "loss": 0.9912,
+      "step": 871
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4893253629376601e-06,
+      "loss": 1.0078,
+      "step": 872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4910333048676344e-06,
+      "loss": 0.9521,
+      "step": 873
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4927412467976088e-06,
+      "loss": 1.0254,
+      "step": 874
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4944491887275833e-06,
+      "loss": 1.04,
+      "step": 875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4961571306575575e-06,
+      "loss": 0.8369,
+      "step": 876
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.497865072587532e-06,
+      "loss": 1.1191,
+      "step": 877
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4995730145175064e-06,
+      "loss": 0.9336,
+      "step": 878
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5012809564474806e-06,
+      "loss": 0.957,
+      "step": 879
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.502988898377455e-06,
+      "loss": 0.8203,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5046968403074293e-06,
+      "loss": 0.9077,
+      "step": 881
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.506404782237404e-06,
+      "loss": 1.0049,
+      "step": 882
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5081127241673784e-06,
+      "loss": 1.0254,
+      "step": 883
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5098206660973526e-06,
+      "loss": 1.085,
+      "step": 884
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.511528608027327e-06,
+      "loss": 0.8916,
+      "step": 885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5132365499573015e-06,
+      "loss": 1.1279,
+      "step": 886
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5149444918872757e-06,
+      "loss": 1.1855,
+      "step": 887
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5166524338172502e-06,
+      "loss": 1.0742,
+      "step": 888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5183603757472246e-06,
+      "loss": 1.0605,
+      "step": 889
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5200683176771988e-06,
+      "loss": 0.9053,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5217762596071733e-06,
+      "loss": 1.0137,
+      "step": 891
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5234842015371477e-06,
+      "loss": 0.8271,
+      "step": 892
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.525192143467122e-06,
+      "loss": 0.9521,
+      "step": 893
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5269000853970964e-06,
+      "loss": 1.1836,
+      "step": 894
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5286080273270708e-06,
+      "loss": 1.1328,
+      "step": 895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.530315969257045e-06,
+      "loss": 1.1211,
+      "step": 896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5320239111870197e-06,
+      "loss": 0.8066,
+      "step": 897
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.533731853116994e-06,
+      "loss": 1.124,
+      "step": 898
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5354397950469684e-06,
+      "loss": 0.9336,
+      "step": 899
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5371477369769428e-06,
+      "loss": 1.0684,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.538855678906917e-06,
+      "loss": 0.8394,
+      "step": 901
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5405636208368915e-06,
+      "loss": 0.7148,
+      "step": 902
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.542271562766866e-06,
+      "loss": 0.8916,
+      "step": 903
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5439795046968402e-06,
+      "loss": 0.8721,
+      "step": 904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5456874466268146e-06,
+      "loss": 1.0615,
+      "step": 905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.547395388556789e-06,
+      "loss": 1.0742,
+      "step": 906
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5491033304867633e-06,
+      "loss": 1.106,
+      "step": 907
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5508112724167377e-06,
+      "loss": 0.6533,
+      "step": 908
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5525192143467122e-06,
+      "loss": 0.8271,
+      "step": 909
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5542271562766864e-06,
+      "loss": 0.957,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5559350982066608e-06,
+      "loss": 0.939,
+      "step": 911
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5576430401366353e-06,
+      "loss": 1.1309,
+      "step": 912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5593509820666097e-06,
+      "loss": 0.9258,
+      "step": 913
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5610589239965842e-06,
+      "loss": 0.8096,
+      "step": 914
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5627668659265584e-06,
+      "loss": 0.9619,
+      "step": 915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5644748078565328e-06,
+      "loss": 0.9287,
+      "step": 916
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5661827497865073e-06,
+      "loss": 0.8394,
+      "step": 917
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5678906917164815e-06,
+      "loss": 1.0352,
+      "step": 918
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.569598633646456e-06,
+      "loss": 0.7988,
+      "step": 919
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5713065755764304e-06,
+      "loss": 1.0566,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5730145175064046e-06,
+      "loss": 0.9941,
+      "step": 921
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.574722459436379e-06,
+      "loss": 0.9443,
+      "step": 922
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5764304013663535e-06,
+      "loss": 1.1211,
+      "step": 923
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5781383432963277e-06,
+      "loss": 0.8564,
+      "step": 924
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5798462852263022e-06,
+      "loss": 1.1045,
+      "step": 925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5815542271562764e-06,
+      "loss": 0.8726,
+      "step": 926
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.583262169086251e-06,
+      "loss": 1.0859,
+      "step": 927
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5849701110162255e-06,
+      "loss": 1.3301,
+      "step": 928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5866780529461997e-06,
+      "loss": 1.2305,
+      "step": 929
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5883859948761742e-06,
+      "loss": 0.9453,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5900939368061486e-06,
+      "loss": 1.1299,
+      "step": 931
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5918018787361229e-06,
+      "loss": 0.8857,
+      "step": 932
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5935098206660973e-06,
+      "loss": 0.9326,
+      "step": 933
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5952177625960717e-06,
+      "loss": 0.8394,
+      "step": 934
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.596925704526046e-06,
+      "loss": 1.1836,
+      "step": 935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5986336464560204e-06,
+      "loss": 1.0166,
+      "step": 936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6003415883859949e-06,
+      "loss": 1.0518,
+      "step": 937
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.602049530315969e-06,
+      "loss": 0.9209,
+      "step": 938
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6037574722459435e-06,
+      "loss": 0.8721,
+      "step": 939
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6054654141759178e-06,
+      "loss": 1.1602,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6071733561058922e-06,
+      "loss": 1.0078,
+      "step": 941
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6088812980358669e-06,
+      "loss": 0.8848,
+      "step": 942
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.610589239965841e-06,
+      "loss": 0.833,
+      "step": 943
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6122971818958155e-06,
+      "loss": 0.8809,
+      "step": 944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.61400512382579e-06,
+      "loss": 0.9521,
+      "step": 945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6157130657557642e-06,
+      "loss": 0.8433,
+      "step": 946
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6174210076857386e-06,
+      "loss": 0.6768,
+      "step": 947
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.619128949615713e-06,
+      "loss": 0.9404,
+      "step": 948
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6208368915456873e-06,
+      "loss": 0.8823,
+      "step": 949
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6225448334756618e-06,
+      "loss": 0.9521,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6242527754056362e-06,
+      "loss": 1.1934,
+      "step": 951
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6259607173356104e-06,
+      "loss": 1.041,
+      "step": 952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6276686592655849e-06,
+      "loss": 1.0215,
+      "step": 953
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.629376601195559e-06,
+      "loss": 0.8281,
+      "step": 954
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6310845431255335e-06,
+      "loss": 1.0557,
+      "step": 955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.632792485055508e-06,
+      "loss": 1.2266,
+      "step": 956
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6345004269854824e-06,
+      "loss": 1.0703,
+      "step": 957
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6362083689154569e-06,
+      "loss": 0.8506,
+      "step": 958
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6379163108454313e-06,
+      "loss": 1.1279,
+      "step": 959
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6396242527754055e-06,
+      "loss": 0.9316,
+      "step": 960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.64133219470538e-06,
+      "loss": 1.0254,
+      "step": 961
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6430401366353544e-06,
+      "loss": 0.8682,
+      "step": 962
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6447480785653287e-06,
+      "loss": 1.1543,
+      "step": 963
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.646456020495303e-06,
+      "loss": 0.8853,
+      "step": 964
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6481639624252775e-06,
+      "loss": 0.9795,
+      "step": 965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6498719043552518e-06,
+      "loss": 1.0537,
+      "step": 966
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6515798462852262e-06,
+      "loss": 1.0938,
+      "step": 967
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6532877882152004e-06,
+      "loss": 1.0791,
+      "step": 968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6549957301451749e-06,
+      "loss": 0.7156,
+      "step": 969
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6567036720751493e-06,
+      "loss": 1.0244,
+      "step": 970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6584116140051238e-06,
+      "loss": 1.0127,
+      "step": 971
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6601195559350982e-06,
+      "loss": 0.958,
+      "step": 972
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6618274978650727e-06,
+      "loss": 1.126,
+      "step": 973
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6635354397950469e-06,
+      "loss": 1.1924,
+      "step": 974
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6652433817250213e-06,
+      "loss": 1.1172,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6669513236549958e-06,
+      "loss": 0.9131,
+      "step": 976
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.66865926558497e-06,
+      "loss": 0.9766,
+      "step": 977
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6703672075149444e-06,
+      "loss": 0.7871,
+      "step": 978
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6720751494449189e-06,
+      "loss": 1.085,
+      "step": 979
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6737830913748931e-06,
+      "loss": 0.9561,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6754910333048676e-06,
+      "loss": 0.7715,
+      "step": 981
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6771989752348418e-06,
+      "loss": 0.8154,
+      "step": 982
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6789069171648162e-06,
+      "loss": 1.0205,
+      "step": 983
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6806148590947907e-06,
+      "loss": 1.1289,
+      "step": 984
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.682322801024765e-06,
+      "loss": 1.0391,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6840307429547396e-06,
+      "loss": 1.0186,
+      "step": 986
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.685738684884714e-06,
+      "loss": 0.959,
+      "step": 987
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6874466268146882e-06,
+      "loss": 1.0381,
+      "step": 988
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6891545687446627e-06,
+      "loss": 0.9844,
+      "step": 989
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6908625106746371e-06,
+      "loss": 1.123,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6925704526046113e-06,
+      "loss": 1.1426,
+      "step": 991
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6942783945345858e-06,
+      "loss": 0.9521,
+      "step": 992
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6959863364645602e-06,
+      "loss": 0.7764,
+      "step": 993
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6976942783945345e-06,
+      "loss": 1.0537,
+      "step": 994
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.699402220324509e-06,
+      "loss": 1.1094,
+      "step": 995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7011101622544831e-06,
+      "loss": 0.9736,
+      "step": 996
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7028181041844576e-06,
+      "loss": 0.9883,
+      "step": 997
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.704526046114432e-06,
+      "loss": 0.999,
+      "step": 998
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7062339880444062e-06,
+      "loss": 0.9941,
+      "step": 999
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7079419299743807e-06,
+      "loss": 0.9629,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7096498719043553e-06,
+      "loss": 0.9141,
+      "step": 1001
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7113578138343296e-06,
+      "loss": 0.759,
+      "step": 1002
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.713065755764304e-06,
+      "loss": 1.0723,
+      "step": 1003
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7147736976942785e-06,
+      "loss": 0.8672,
+      "step": 1004
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7164816396242527e-06,
+      "loss": 1.1592,
+      "step": 1005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7181895815542271e-06,
+      "loss": 1.0557,
+      "step": 1006
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7198975234842016e-06,
+      "loss": 0.8232,
+      "step": 1007
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7216054654141758e-06,
+      "loss": 0.8516,
+      "step": 1008
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7233134073441502e-06,
+      "loss": 0.8926,
+      "step": 1009
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7250213492741245e-06,
+      "loss": 0.8789,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.726729291204099e-06,
+      "loss": 1.1182,
+      "step": 1011
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7284372331340734e-06,
+      "loss": 0.8672,
+      "step": 1012
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7301451750640476e-06,
+      "loss": 1.0625,
+      "step": 1013
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.731853116994022e-06,
+      "loss": 1.0449,
+      "step": 1014
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7335610589239965e-06,
+      "loss": 0.8184,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.735269000853971e-06,
+      "loss": 1.082,
+      "step": 1016
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7369769427839454e-06,
+      "loss": 1.1465,
+      "step": 1017
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7386848847139198e-06,
+      "loss": 0.6377,
+      "step": 1018
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.740392826643894e-06,
+      "loss": 0.915,
+      "step": 1019
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7421007685738685e-06,
+      "loss": 1.0547,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.743808710503843e-06,
+      "loss": 1.2793,
+      "step": 1021
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7455166524338171e-06,
+      "loss": 0.8604,
+      "step": 1022
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7472245943637916e-06,
+      "loss": 0.9111,
+      "step": 1023
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7489325362937658e-06,
+      "loss": 0.9062,
+      "step": 1024
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7506404782237403e-06,
+      "loss": 1.0752,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7523484201537147e-06,
+      "loss": 1.123,
+      "step": 1026
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.754056362083689e-06,
+      "loss": 1.0762,
+      "step": 1027
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7557643040136634e-06,
+      "loss": 0.8467,
+      "step": 1028
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7574722459436378e-06,
+      "loss": 1.0283,
+      "step": 1029
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.759180187873612e-06,
+      "loss": 0.9092,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7608881298035867e-06,
+      "loss": 0.8926,
+      "step": 1031
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7625960717335611e-06,
+      "loss": 0.9033,
+      "step": 1032
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7643040136635354e-06,
+      "loss": 0.8647,
+      "step": 1033
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7660119555935098e-06,
+      "loss": 1.1143,
+      "step": 1034
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7677198975234843e-06,
+      "loss": 0.9775,
+      "step": 1035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7694278394534585e-06,
+      "loss": 0.9717,
+      "step": 1036
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.771135781383433e-06,
+      "loss": 0.9805,
+      "step": 1037
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7728437233134072e-06,
+      "loss": 0.8062,
+      "step": 1038
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7745516652433816e-06,
+      "loss": 0.7954,
+      "step": 1039
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.776259607173356e-06,
+      "loss": 0.9238,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7779675491033303e-06,
+      "loss": 0.9883,
+      "step": 1041
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7796754910333047e-06,
+      "loss": 1.0879,
+      "step": 1042
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7813834329632792e-06,
+      "loss": 1.127,
+      "step": 1043
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7830913748932534e-06,
+      "loss": 0.8008,
+      "step": 1044
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7847993168232278e-06,
+      "loss": 0.9463,
+      "step": 1045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7865072587532025e-06,
+      "loss": 0.96,
+      "step": 1046
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7882152006831767e-06,
+      "loss": 1.0117,
+      "step": 1047
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7899231426131512e-06,
+      "loss": 0.8945,
+      "step": 1048
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7916310845431256e-06,
+      "loss": 0.7495,
+      "step": 1049
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7933390264730998e-06,
+      "loss": 1.0996,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7950469684030743e-06,
+      "loss": 1.1709,
+      "step": 1051
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7967549103330485e-06,
+      "loss": 1.0645,
+      "step": 1052
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.798462852263023e-06,
+      "loss": 0.9971,
+      "step": 1053
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8001707941929974e-06,
+      "loss": 1.0469,
+      "step": 1054
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8018787361229716e-06,
+      "loss": 0.937,
+      "step": 1055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.803586678052946e-06,
+      "loss": 0.8828,
+      "step": 1056
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8052946199829205e-06,
+      "loss": 0.9277,
+      "step": 1057
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8070025619128947e-06,
+      "loss": 0.9805,
+      "step": 1058
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8087105038428692e-06,
+      "loss": 0.9165,
+      "step": 1059
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8104184457728438e-06,
+      "loss": 0.9912,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.812126387702818e-06,
+      "loss": 0.7705,
+      "step": 1061
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8138343296327925e-06,
+      "loss": 0.8242,
+      "step": 1062
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.815542271562767e-06,
+      "loss": 0.8877,
+      "step": 1063
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8172502134927412e-06,
+      "loss": 0.7451,
+      "step": 1064
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8189581554227156e-06,
+      "loss": 0.8818,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8206660973526898e-06,
+      "loss": 0.8135,
+      "step": 1066
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8223740392826643e-06,
+      "loss": 1.0557,
+      "step": 1067
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8240819812126387e-06,
+      "loss": 0.9028,
+      "step": 1068
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.825789923142613e-06,
+      "loss": 0.6206,
+      "step": 1069
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8274978650725874e-06,
+      "loss": 1.1514,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8292058070025618e-06,
+      "loss": 0.9014,
+      "step": 1071
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.830913748932536e-06,
+      "loss": 1.042,
+      "step": 1072
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8326216908625105e-06,
+      "loss": 0.9893,
+      "step": 1073
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.834329632792485e-06,
+      "loss": 0.8906,
+      "step": 1074
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8360375747224594e-06,
+      "loss": 0.9561,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8377455166524338e-06,
+      "loss": 0.9414,
+      "step": 1076
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8394534585824083e-06,
+      "loss": 1.0771,
+      "step": 1077
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8411614005123825e-06,
+      "loss": 1.0557,
+      "step": 1078
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.842869342442357e-06,
+      "loss": 1.0293,
+      "step": 1079
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8445772843723312e-06,
+      "loss": 0.9072,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8462852263023056e-06,
+      "loss": 1.0332,
+      "step": 1081
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.84799316823228e-06,
+      "loss": 0.9111,
+      "step": 1082
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8497011101622543e-06,
+      "loss": 0.9839,
+      "step": 1083
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8514090520922287e-06,
+      "loss": 1.0654,
+      "step": 1084
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8531169940222032e-06,
+      "loss": 0.8013,
+      "step": 1085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8548249359521774e-06,
+      "loss": 1.0381,
+      "step": 1086
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8565328778821519e-06,
+      "loss": 1.082,
+      "step": 1087
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8582408198121263e-06,
+      "loss": 0.9414,
+      "step": 1088
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8599487617421005e-06,
+      "loss": 1.0918,
+      "step": 1089
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8616567036720752e-06,
+      "loss": 0.8799,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8633646456020496e-06,
+      "loss": 0.9521,
+      "step": 1091
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8650725875320239e-06,
+      "loss": 1.0879,
+      "step": 1092
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8667805294619983e-06,
+      "loss": 1.041,
+      "step": 1093
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8684884713919725e-06,
+      "loss": 0.7607,
+      "step": 1094
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.870196413321947e-06,
+      "loss": 0.9458,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8719043552519214e-06,
+      "loss": 1.0664,
+      "step": 1096
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8736122971818956e-06,
+      "loss": 1.0898,
+      "step": 1097
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.87532023911187e-06,
+      "loss": 0.8857,
+      "step": 1098
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8770281810418445e-06,
+      "loss": 1.0527,
+      "step": 1099
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8787361229718188e-06,
+      "loss": 0.876,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8804440649017932e-06,
+      "loss": 1.0801,
+      "step": 1101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8821520068317676e-06,
+      "loss": 0.8174,
+      "step": 1102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8838599487617419e-06,
+      "loss": 0.9844,
+      "step": 1103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8855678906917163e-06,
+      "loss": 1.2305,
+      "step": 1104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.887275832621691e-06,
+      "loss": 1.0312,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8889837745516652e-06,
+      "loss": 0.8306,
+      "step": 1106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8906917164816396e-06,
+      "loss": 0.7549,
+      "step": 1107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.892399658411614e-06,
+      "loss": 0.9272,
+      "step": 1108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8941076003415883e-06,
+      "loss": 0.8765,
+      "step": 1109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8958155422715628e-06,
+      "loss": 1.0986,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.897523484201537e-06,
+      "loss": 1.1367,
+      "step": 1111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8992314261315114e-06,
+      "loss": 0.9619,
+      "step": 1112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9009393680614859e-06,
+      "loss": 1.0088,
+      "step": 1113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.90264730999146e-06,
+      "loss": 1.0298,
+      "step": 1114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9043552519214345e-06,
+      "loss": 0.9512,
+      "step": 1115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.906063193851409e-06,
+      "loss": 1.0615,
+      "step": 1116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9077711357813834e-06,
+      "loss": 0.9092,
+      "step": 1117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9094790777113577e-06,
+      "loss": 1.209,
+      "step": 1118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.911187019641332e-06,
+      "loss": 1.1641,
+      "step": 1119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9128949615713065e-06,
+      "loss": 1.0791,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9146029035012808e-06,
+      "loss": 0.7524,
+      "step": 1121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9163108454312554e-06,
+      "loss": 0.8574,
+      "step": 1122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9180187873612297e-06,
+      "loss": 0.9902,
+      "step": 1123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.919726729291204e-06,
+      "loss": 1.2393,
+      "step": 1124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9214346712211785e-06,
+      "loss": 0.958,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9231426131511528e-06,
+      "loss": 1.1992,
+      "step": 1126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.924850555081127e-06,
+      "loss": 0.9814,
+      "step": 1127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9265584970111017e-06,
+      "loss": 1.0049,
+      "step": 1128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.928266438941076e-06,
+      "loss": 1.1299,
+      "step": 1129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.92997438087105e-06,
+      "loss": 0.8892,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9316823228010248e-06,
+      "loss": 0.9189,
+      "step": 1131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.933390264730999e-06,
+      "loss": 0.7764,
+      "step": 1132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9350982066609732e-06,
+      "loss": 0.9619,
+      "step": 1133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.936806148590948e-06,
+      "loss": 0.9346,
+      "step": 1134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.938514090520922e-06,
+      "loss": 1.084,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9402220324508968e-06,
+      "loss": 0.9355,
+      "step": 1136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.941929974380871e-06,
+      "loss": 1.1738,
+      "step": 1137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9436379163108452e-06,
+      "loss": 1.1045,
+      "step": 1138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.94534585824082e-06,
+      "loss": 0.9688,
+      "step": 1139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.947053800170794e-06,
+      "loss": 0.9932,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9487617421007683e-06,
+      "loss": 0.9414,
+      "step": 1141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.950469684030743e-06,
+      "loss": 1.0459,
+      "step": 1142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9521776259607172e-06,
+      "loss": 1.126,
+      "step": 1143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9538855678906915e-06,
+      "loss": 0.7808,
+      "step": 1144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.955593509820666e-06,
+      "loss": 0.916,
+      "step": 1145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9573014517506403e-06,
+      "loss": 1.1494,
+      "step": 1146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9590093936806146e-06,
+      "loss": 0.9453,
+      "step": 1147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9607173356105892e-06,
+      "loss": 1.0137,
+      "step": 1148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9624252775405635e-06,
+      "loss": 0.9766,
+      "step": 1149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.964133219470538e-06,
+      "loss": 1.1035,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9658411614005123e-06,
+      "loss": 1.1084,
+      "step": 1151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9675491033304866e-06,
+      "loss": 0.9346,
+      "step": 1152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9692570452604612e-06,
+      "loss": 1.0039,
+      "step": 1153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9709649871904355e-06,
+      "loss": 1.1416,
+      "step": 1154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9726729291204097e-06,
+      "loss": 0.957,
+      "step": 1155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9743808710503843e-06,
+      "loss": 0.9189,
+      "step": 1156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9760888129803586e-06,
+      "loss": 0.7988,
+      "step": 1157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.977796754910333e-06,
+      "loss": 0.9365,
+      "step": 1158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9795046968403075e-06,
+      "loss": 0.6909,
+      "step": 1159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9812126387702817e-06,
+      "loss": 1.2227,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.982920580700256e-06,
+      "loss": 1.0127,
+      "step": 1161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9846285226302306e-06,
+      "loss": 1.0625,
+      "step": 1162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.986336464560205e-06,
+      "loss": 1.0898,
+      "step": 1163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9880444064901795e-06,
+      "loss": 0.9346,
+      "step": 1164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9897523484201537e-06,
+      "loss": 0.9248,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.991460290350128e-06,
+      "loss": 0.9502,
+      "step": 1166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9931682322801026e-06,
+      "loss": 0.7476,
+      "step": 1167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994876174210077e-06,
+      "loss": 1.0654,
+      "step": 1168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996584116140051e-06,
+      "loss": 0.998,
+      "step": 1169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982920580700257e-06,
+      "loss": 1.2603,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-06,
+      "loss": 0.9443,
+      "step": 1171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999996554496e-06,
+      "loss": 0.918,
+      "step": 1172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999986217985e-06,
+      "loss": 0.9023,
+      "step": 1173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999968990466e-06,
+      "loss": 1.0205,
+      "step": 1174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999994487194e-06,
+      "loss": 0.7886,
+      "step": 1175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999138624065e-06,
+      "loss": 1.0078,
+      "step": 1176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998759618665e-06,
+      "loss": 0.9609,
+      "step": 1177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998311703195e-06,
+      "loss": 0.959,
+      "step": 1178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999779487766e-06,
+      "loss": 0.8906,
+      "step": 1179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999997209142066e-06,
+      "loss": 1.1406,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999655449642e-06,
+      "loss": 1.041,
+      "step": 1181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999995830940713e-06,
+      "loss": 0.9463,
+      "step": 1182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999503847497e-06,
+      "loss": 0.8613,
+      "step": 1183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994177099176e-06,
+      "loss": 1.0049,
+      "step": 1184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993246813355e-06,
+      "loss": 0.9443,
+      "step": 1185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999922476175e-06,
+      "loss": 1.1504,
+      "step": 1186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999117951162e-06,
+      "loss": 0.9932,
+      "step": 1187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999004249573e-06,
+      "loss": 1.0068,
+      "step": 1188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998883656983e-06,
+      "loss": 1.0723,
+      "step": 1189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999987561733936e-06,
+      "loss": 0.9795,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998621798805e-06,
+      "loss": 1.0098,
+      "step": 1191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998480533218e-06,
+      "loss": 1.0654,
+      "step": 1192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999983323766343e-06,
+      "loss": 0.7603,
+      "step": 1193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999981773290545e-06,
+      "loss": 1.1143,
+      "step": 1194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999980153904797e-06,
+      "loss": 0.8125,
+      "step": 1195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997846560911e-06,
+      "loss": 0.999,
+      "step": 1196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999976708403496e-06,
+      "loss": 0.9111,
+      "step": 1197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999974882287964e-06,
+      "loss": 1.0264,
+      "step": 1198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999972987262533e-06,
+      "loss": 1.1074,
+      "step": 1199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999971023327215e-06,
+      "loss": 1.0342,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999968990482015e-06,
+      "loss": 0.7637,
+      "step": 1201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999966888726957e-06,
+      "loss": 1.0762,
+      "step": 1202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996471806205e-06,
+      "loss": 1.0103,
+      "step": 1203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996247848731e-06,
+      "loss": 0.9541,
+      "step": 1204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999960170002755e-06,
+      "loss": 1.1533,
+      "step": 1205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999577926084e-06,
+      "loss": 0.6733,
+      "step": 1206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999955346304255e-06,
+      "loss": 0.9844,
+      "step": 1207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999952831090346e-06,
+      "loss": 0.7695,
+      "step": 1208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995024696669e-06,
+      "loss": 0.916,
+      "step": 1209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999947593933296e-06,
+      "loss": 0.9863,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999944871990188e-06,
+      "loss": 0.9697,
+      "step": 1211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999942081137384e-06,
+      "loss": 0.7896,
+      "step": 1212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999939221374905e-06,
+      "loss": 1.0518,
+      "step": 1213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993629270277e-06,
+      "loss": 0.8398,
+      "step": 1214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999933295120998e-06,
+      "loss": 0.894,
+      "step": 1215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999930228629608e-06,
+      "loss": 0.6821,
+      "step": 1216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999927093228627e-06,
+      "loss": 1.0898,
+      "step": 1217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999992388891807e-06,
+      "loss": 1.0723,
+      "step": 1218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999920615697963e-06,
+      "loss": 0.8779,
+      "step": 1219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999917273568326e-06,
+      "loss": 0.877,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999913862529183e-06,
+      "loss": 1.0771,
+      "step": 1221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999910382580564e-06,
+      "loss": 0.8574,
+      "step": 1222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999990683372248e-06,
+      "loss": 0.9736,
+      "step": 1223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999903215954966e-06,
+      "loss": 0.9453,
+      "step": 1224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999989952927804e-06,
+      "loss": 0.8955,
+      "step": 1225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999989577369174e-06,
+      "loss": 1.0928,
+      "step": 1226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999891949196075e-06,
+      "loss": 0.9648,
+      "step": 1227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999988805579108e-06,
+      "loss": 0.833,
+      "step": 1228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999988409347678e-06,
+      "loss": 1.0234,
+      "step": 1229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999880062253203e-06,
+      "loss": 0.8438,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999987596212038e-06,
+      "loss": 1.0664,
+      "step": 1231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999871793078335e-06,
+      "loss": 0.8477,
+      "step": 1232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999867555127097e-06,
+      "loss": 0.8809,
+      "step": 1233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999863248266697e-06,
+      "loss": 1.1553,
+      "step": 1234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999858872497164e-06,
+      "loss": 1.0635,
+      "step": 1235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999854427818527e-06,
+      "loss": 0.8882,
+      "step": 1236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999849914230817e-06,
+      "loss": 0.9727,
+      "step": 1237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999845331734067e-06,
+      "loss": 1.084,
+      "step": 1238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999840680328307e-06,
+      "loss": 1.1387,
+      "step": 1239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999983596001357e-06,
+      "loss": 1.0928,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999831170789888e-06,
+      "loss": 1.1201,
+      "step": 1241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999982631265729e-06,
+      "loss": 1.0059,
+      "step": 1242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999821385615817e-06,
+      "loss": 0.7881,
+      "step": 1243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99998163896655e-06,
+      "loss": 1.0986,
+      "step": 1244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999981132480637e-06,
+      "loss": 0.9883,
+      "step": 1245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999806191038467e-06,
+      "loss": 0.8926,
+      "step": 1246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999800988361824e-06,
+      "loss": 1.0127,
+      "step": 1247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999795716776476e-06,
+      "loss": 1.0391,
+      "step": 1248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999979037628246e-06,
+      "loss": 0.9375,
+      "step": 1249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999978496687981e-06,
+      "loss": 0.8613,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999779488568572e-06,
+      "loss": 0.7905,
+      "step": 1251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999773941348776e-06,
+      "loss": 1.0049,
+      "step": 1252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999976832522046e-06,
+      "loss": 1.0547,
+      "step": 1253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999762640183664e-06,
+      "loss": 0.9321,
+      "step": 1254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999756886238434e-06,
+      "loss": 1.0684,
+      "step": 1255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99997510633848e-06,
+      "loss": 1.0352,
+      "step": 1256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999745171622806e-06,
+      "loss": 1.0361,
+      "step": 1257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999973921095249e-06,
+      "loss": 0.9629,
+      "step": 1258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99997331813739e-06,
+      "loss": 0.96,
+      "step": 1259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999727082887074e-06,
+      "loss": 0.9287,
+      "step": 1260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999720915492047e-06,
+      "loss": 0.9653,
+      "step": 1261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999714679188874e-06,
+      "loss": 0.8271,
+      "step": 1262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999970837397759e-06,
+      "loss": 0.8311,
+      "step": 1263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999701999858234e-06,
+      "loss": 1.1191,
+      "step": 1264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999969555683086e-06,
+      "loss": 1.1191,
+      "step": 1265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999968904489551e-06,
+      "loss": 1.0583,
+      "step": 1266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999968246405223e-06,
+      "loss": 1.0273,
+      "step": 1267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999675814301055e-06,
+      "loss": 0.769,
+      "step": 1268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999669095642045e-06,
+      "loss": 0.7451,
+      "step": 1269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999966230807524e-06,
+      "loss": 0.9219,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999655451600682e-06,
+      "loss": 1.041,
+      "step": 1271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999648526218424e-06,
+      "loss": 1.1357,
+      "step": 1272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999964153192851e-06,
+      "loss": 1.0693,
+      "step": 1273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999634468730996e-06,
+      "loss": 0.6895,
+      "step": 1274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999962733662592e-06,
+      "loss": 1.1719,
+      "step": 1275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999620135613343e-06,
+      "loss": 1.0264,
+      "step": 1276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99996128656933e-06,
+      "loss": 0.9707,
+      "step": 1277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999605526865854e-06,
+      "loss": 1.2891,
+      "step": 1278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999598119131052e-06,
+      "loss": 0.9844,
+      "step": 1279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999590642488943e-06,
+      "loss": 0.8994,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999583096939578e-06,
+      "loss": 1.1836,
+      "step": 1281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999957548248301e-06,
+      "loss": 0.9238,
+      "step": 1282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999567799119296e-06,
+      "loss": 1.0645,
+      "step": 1283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956004684848e-06,
+      "loss": 0.9746,
+      "step": 1284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999955222567062e-06,
+      "loss": 0.9033,
+      "step": 1285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999544335585773e-06,
+      "loss": 0.9121,
+      "step": 1286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999536376593992e-06,
+      "loss": 1.125,
+      "step": 1287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999528348695324e-06,
+      "loss": 0.793,
+      "step": 1288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999520251889835e-06,
+      "loss": 0.9561,
+      "step": 1289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999512086177576e-06,
+      "loss": 0.9316,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999503851558602e-06,
+      "loss": 0.9229,
+      "step": 1291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999495548032973e-06,
+      "loss": 0.9355,
+      "step": 1292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999487175600744e-06,
+      "loss": 1.207,
+      "step": 1293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999478734261974e-06,
+      "loss": 0.9131,
+      "step": 1294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999947022401672e-06,
+      "loss": 0.7646,
+      "step": 1295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999461644865042e-06,
+      "loss": 0.7559,
+      "step": 1296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999452996806996e-06,
+      "loss": 1.0391,
+      "step": 1297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999444279842646e-06,
+      "loss": 0.8662,
+      "step": 1298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999943549397205e-06,
+      "loss": 0.9121,
+      "step": 1299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999426639195267e-06,
+      "loss": 1.0693,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999941771551236e-06,
+      "loss": 0.8398,
+      "step": 1301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999940872292339e-06,
+      "loss": 0.9824,
+      "step": 1302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999939966142842e-06,
+      "loss": 1.1318,
+      "step": 1303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999390531027512e-06,
+      "loss": 0.8726,
+      "step": 1304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999381331720725e-06,
+      "loss": 1.002,
+      "step": 1305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999937206350813e-06,
+      "loss": 0.8428,
+      "step": 1306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999362726389783e-06,
+      "loss": 1.001,
+      "step": 1307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999353320365753e-06,
+      "loss": 0.7402,
+      "step": 1308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999343845436104e-06,
+      "loss": 0.9736,
+      "step": 1309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99993343016009e-06,
+      "loss": 0.8823,
+      "step": 1310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999324688860208e-06,
+      "loss": 0.9863,
+      "step": 1311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999315007214095e-06,
+      "loss": 0.9883,
+      "step": 1312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999305256662628e-06,
+      "loss": 1.1133,
+      "step": 1313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999295437205873e-06,
+      "loss": 0.916,
+      "step": 1314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999285548843898e-06,
+      "loss": 0.998,
+      "step": 1315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999275591576767e-06,
+      "loss": 0.7402,
+      "step": 1316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999926556540455e-06,
+      "loss": 1.0215,
+      "step": 1317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999255470327325e-06,
+      "loss": 1.0527,
+      "step": 1318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999245306345153e-06,
+      "loss": 0.7427,
+      "step": 1319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999235073458105e-06,
+      "loss": 1.0127,
+      "step": 1320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999224771666256e-06,
+      "loss": 1.1123,
+      "step": 1321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999921440096967e-06,
+      "loss": 0.728,
+      "step": 1322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999203961368424e-06,
+      "loss": 0.9824,
+      "step": 1323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999193452862585e-06,
+      "loss": 0.6914,
+      "step": 1324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999918287545223e-06,
+      "loss": 0.8369,
+      "step": 1325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999917222913743e-06,
+      "loss": 1.0996,
+      "step": 1326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999916151391826e-06,
+      "loss": 0.8389,
+      "step": 1327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999915072979479e-06,
+      "loss": 0.9609,
+      "step": 1328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99991398767671e-06,
+      "loss": 0.7471,
+      "step": 1329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999912895483526e-06,
+      "loss": 1.0107,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999117963999348e-06,
+      "loss": 0.9326,
+      "step": 1331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999106904259434e-06,
+      "loss": 0.876,
+      "step": 1332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999095775615604e-06,
+      "loss": 1.207,
+      "step": 1333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999084578067927e-06,
+      "loss": 0.9746,
+      "step": 1334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999073311616483e-06,
+      "loss": 1.1729,
+      "step": 1335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999061976261353e-06,
+      "loss": 1.0996,
+      "step": 1336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999050572002603e-06,
+      "loss": 0.9258,
+      "step": 1337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999903909884033e-06,
+      "loss": 0.9727,
+      "step": 1338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999027556774595e-06,
+      "loss": 1.0723,
+      "step": 1339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999901594580549e-06,
+      "loss": 0.7607,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999004265933086e-06,
+      "loss": 0.9971,
+      "step": 1341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998992517157475e-06,
+      "loss": 0.916,
+      "step": 1342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998980699478724e-06,
+      "loss": 1.124,
+      "step": 1343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998968812896926e-06,
+      "loss": 0.6675,
+      "step": 1344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998956857412157e-06,
+      "loss": 0.9834,
+      "step": 1345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99989448330245e-06,
+      "loss": 0.7178,
+      "step": 1346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999893273973404e-06,
+      "loss": 0.8975,
+      "step": 1347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999892057754086e-06,
+      "loss": 0.916,
+      "step": 1348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999890834644504e-06,
+      "loss": 0.9346,
+      "step": 1349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998896046446667e-06,
+      "loss": 1.1758,
+      "step": 1350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998883677545826e-06,
+      "loss": 0.9658,
+      "step": 1351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998871239742604e-06,
+      "loss": 0.8965,
+      "step": 1352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998858733037083e-06,
+      "loss": 0.8799,
+      "step": 1353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998846157429355e-06,
+      "loss": 1.085,
+      "step": 1354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998833512919498e-06,
+      "loss": 0.9307,
+      "step": 1355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998820799507606e-06,
+      "loss": 0.8916,
+      "step": 1356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998808017193764e-06,
+      "loss": 1.0518,
+      "step": 1357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998795165978057e-06,
+      "loss": 1.1328,
+      "step": 1358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999878224586058e-06,
+      "loss": 0.9863,
+      "step": 1359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998769256841423e-06,
+      "loss": 0.8193,
+      "step": 1360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998756198920666e-06,
+      "loss": 0.7339,
+      "step": 1361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998743072098403e-06,
+      "loss": 1.0117,
+      "step": 1362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999872987637473e-06,
+      "loss": 0.7773,
+      "step": 1363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998716611749734e-06,
+      "loss": 1.0908,
+      "step": 1364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998703278223506e-06,
+      "loss": 1.0107,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998689875796136e-06,
+      "loss": 0.8828,
+      "step": 1366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998676404467722e-06,
+      "loss": 0.8174,
+      "step": 1367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998662864238345e-06,
+      "loss": 1.1143,
+      "step": 1368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998649255108115e-06,
+      "loss": 0.9473,
+      "step": 1369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998635577077116e-06,
+      "loss": 0.9541,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998621830145442e-06,
+      "loss": 1.043,
+      "step": 1371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999860801431319e-06,
+      "loss": 1.0986,
+      "step": 1372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998594129580458e-06,
+      "loss": 1.0469,
+      "step": 1373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998580175947334e-06,
+      "loss": 1.0,
+      "step": 1374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999856615341392e-06,
+      "loss": 1.0068,
+      "step": 1375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998552061980313e-06,
+      "loss": 0.8867,
+      "step": 1376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998537901646607e-06,
+      "loss": 0.9561,
+      "step": 1377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99985236724129e-06,
+      "loss": 0.8076,
+      "step": 1378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999850937427929e-06,
+      "loss": 0.8545,
+      "step": 1379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999849500724588e-06,
+      "loss": 0.9077,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998480571312758e-06,
+      "loss": 0.8906,
+      "step": 1381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998466066480037e-06,
+      "loss": 0.9248,
+      "step": 1382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999845149274781e-06,
+      "loss": 0.9844,
+      "step": 1383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998436850116173e-06,
+      "loss": 0.7305,
+      "step": 1384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999842213858524e-06,
+      "loss": 1.2217,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99984073581551e-06,
+      "loss": 0.5537,
+      "step": 1386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999839250882586e-06,
+      "loss": 0.9404,
+      "step": 1387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998377590597626e-06,
+      "loss": 0.8486,
+      "step": 1388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999836260347049e-06,
+      "loss": 0.833,
+      "step": 1389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998347547444564e-06,
+      "loss": 1.1914,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999833242251995e-06,
+      "loss": 0.9512,
+      "step": 1391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999831722869675e-06,
+      "loss": 1.084,
+      "step": 1392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998301965975074e-06,
+      "loss": 0.9863,
+      "step": 1393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999828663435502e-06,
+      "loss": 1.0908,
+      "step": 1394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99982712338367e-06,
+      "loss": 1.0332,
+      "step": 1395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998255764420215e-06,
+      "loss": 0.877,
+      "step": 1396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998240226105676e-06,
+      "loss": 0.8955,
+      "step": 1397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999822461889319e-06,
+      "loss": 1.0156,
+      "step": 1398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999820894278286e-06,
+      "loss": 0.8506,
+      "step": 1399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998193197774795e-06,
+      "loss": 0.873,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998177383869108e-06,
+      "loss": 0.9932,
+      "step": 1401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998161501065904e-06,
+      "loss": 1.0742,
+      "step": 1402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999814554936529e-06,
+      "loss": 0.9756,
+      "step": 1403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998129528767387e-06,
+      "loss": 0.9453,
+      "step": 1404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998113439272297e-06,
+      "loss": 0.9238,
+      "step": 1405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998097280880123e-06,
+      "loss": 1.0498,
+      "step": 1406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998081053590996e-06,
+      "loss": 1.04,
+      "step": 1407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999806475740501e-06,
+      "loss": 1.165,
+      "step": 1408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998048392322287e-06,
+      "loss": 0.9717,
+      "step": 1409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998031958342937e-06,
+      "loss": 0.9297,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999801545546707e-06,
+      "loss": 0.9258,
+      "step": 1411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997998883694803e-06,
+      "loss": 1.0537,
+      "step": 1412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997982243026256e-06,
+      "loss": 1.0371,
+      "step": 1413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999796553346153e-06,
+      "loss": 1.0371,
+      "step": 1414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997948755000756e-06,
+      "loss": 1.0049,
+      "step": 1415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997931907644036e-06,
+      "loss": 0.9893,
+      "step": 1416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997914991391494e-06,
+      "loss": 0.9556,
+      "step": 1417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997898006243244e-06,
+      "loss": 1.1875,
+      "step": 1418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997880952199406e-06,
+      "loss": 0.6499,
+      "step": 1419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999786382926009e-06,
+      "loss": 0.7822,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997846637425424e-06,
+      "loss": 1.0264,
+      "step": 1421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997829376695513e-06,
+      "loss": 0.8823,
+      "step": 1422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997812047070492e-06,
+      "loss": 1.0352,
+      "step": 1423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999779464855047e-06,
+      "loss": 0.7891,
+      "step": 1424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999777718113557e-06,
+      "loss": 1.0459,
+      "step": 1425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999775964482591e-06,
+      "loss": 0.8213,
+      "step": 1426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997742039621613e-06,
+      "loss": 1.0518,
+      "step": 1427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99977243655228e-06,
+      "loss": 1.042,
+      "step": 1428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997706622529596e-06,
+      "loss": 0.6528,
+      "step": 1429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999768881064212e-06,
+      "loss": 1.2793,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997670929860492e-06,
+      "loss": 1.0957,
+      "step": 1431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999765298018484e-06,
+      "loss": 0.8369,
+      "step": 1432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997634961615284e-06,
+      "loss": 1.1738,
+      "step": 1433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999761687415195e-06,
+      "loss": 1.0186,
+      "step": 1434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997598717794966e-06,
+      "loss": 1.1865,
+      "step": 1435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997580492544454e-06,
+      "loss": 0.9385,
+      "step": 1436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997562198400534e-06,
+      "loss": 1.0156,
+      "step": 1437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997543835363338e-06,
+      "loss": 1.127,
+      "step": 1438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997525403432996e-06,
+      "loss": 0.9453,
+      "step": 1439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999750690260963e-06,
+      "loss": 1.0527,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999748833289337e-06,
+      "loss": 1.0205,
+      "step": 1441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997469694284336e-06,
+      "loss": 1.3848,
+      "step": 1442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997450986782667e-06,
+      "loss": 0.9551,
+      "step": 1443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999743221038849e-06,
+      "loss": 1.127,
+      "step": 1444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997413365101926e-06,
+      "loss": 0.9561,
+      "step": 1445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997394450923117e-06,
+      "loss": 1.1621,
+      "step": 1446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997375467852183e-06,
+      "loss": 0.9551,
+      "step": 1447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999735641588926e-06,
+      "loss": 1.041,
+      "step": 1448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999733729503448e-06,
+      "loss": 1.04,
+      "step": 1449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997318105287972e-06,
+      "loss": 0.9551,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999729884664987e-06,
+      "loss": 0.7969,
+      "step": 1451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997279519120304e-06,
+      "loss": 0.7788,
+      "step": 1452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997260122699414e-06,
+      "loss": 1.0234,
+      "step": 1453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997240657387326e-06,
+      "loss": 1.0186,
+      "step": 1454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997221123184177e-06,
+      "loss": 0.9375,
+      "step": 1455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997201520090102e-06,
+      "loss": 1.1514,
+      "step": 1456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997181848105237e-06,
+      "loss": 0.9395,
+      "step": 1457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997162107229717e-06,
+      "loss": 0.8613,
+      "step": 1458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997142297463673e-06,
+      "loss": 0.9316,
+      "step": 1459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999712241880725e-06,
+      "loss": 0.8242,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997102471260577e-06,
+      "loss": 1.0547,
+      "step": 1461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99970824548238e-06,
+      "loss": 0.8408,
+      "step": 1462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999706236949705e-06,
+      "loss": 0.8994,
+      "step": 1463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997042215280467e-06,
+      "loss": 0.7217,
+      "step": 1464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999702199217419e-06,
+      "loss": 0.7227,
+      "step": 1465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999700170017836e-06,
+      "loss": 0.7627,
+      "step": 1466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996981339293116e-06,
+      "loss": 0.9014,
+      "step": 1467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996960909518597e-06,
+      "loss": 1.0273,
+      "step": 1468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999694041085495e-06,
+      "loss": 0.8245,
+      "step": 1469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996919843302304e-06,
+      "loss": 0.9795,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996899206860814e-06,
+      "loss": 0.9043,
+      "step": 1471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996878501530613e-06,
+      "loss": 0.8486,
+      "step": 1472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999685772731185e-06,
+      "loss": 0.8555,
+      "step": 1473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996836884204657e-06,
+      "loss": 0.8545,
+      "step": 1474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999681597220919e-06,
+      "loss": 0.9551,
+      "step": 1475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999679499132559e-06,
+      "loss": 0.8506,
+      "step": 1476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996773941554e-06,
+      "loss": 0.8989,
+      "step": 1477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999675282289456e-06,
+      "loss": 0.8877,
+      "step": 1478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996731635347427e-06,
+      "loss": 0.6934,
+      "step": 1479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999671037891274e-06,
+      "loss": 1.0088,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996689053590643e-06,
+      "loss": 0.8633,
+      "step": 1481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999666765938129e-06,
+      "loss": 0.8779,
+      "step": 1482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999664619628482e-06,
+      "loss": 1.0723,
+      "step": 1483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999662466430139e-06,
+      "loss": 1.0791,
+      "step": 1484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996603063431143e-06,
+      "loss": 1.0684,
+      "step": 1485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999658139367423e-06,
+      "loss": 0.8223,
+      "step": 1486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996559655030795e-06,
+      "loss": 1.0176,
+      "step": 1487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996537847500995e-06,
+      "loss": 1.0225,
+      "step": 1488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999651597108498e-06,
+      "loss": 0.9043,
+      "step": 1489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999649402578289e-06,
+      "loss": 0.8159,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999647201159489e-06,
+      "loss": 1.0518,
+      "step": 1491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996449928521122e-06,
+      "loss": 1.0391,
+      "step": 1492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996427776561748e-06,
+      "loss": 1.0928,
+      "step": 1493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999640555571691e-06,
+      "loss": 0.8594,
+      "step": 1494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996383265986763e-06,
+      "loss": 1.124,
+      "step": 1495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996360907371467e-06,
+      "loss": 1.0605,
+      "step": 1496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996338479871177e-06,
+      "loss": 0.8867,
+      "step": 1497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999631598348604e-06,
+      "loss": 1.0967,
+      "step": 1498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996293418216206e-06,
+      "loss": 0.7886,
+      "step": 1499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999627078406185e-06,
+      "loss": 1.166,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999624808102311e-06,
+      "loss": 1.1758,
+      "step": 1501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999622530910015e-06,
+      "loss": 0.9106,
+      "step": 1502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999620246829313e-06,
+      "loss": 0.8818,
+      "step": 1503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99961795586022e-06,
+      "loss": 1.0947,
+      "step": 1504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996156580027517e-06,
+      "loss": 0.8105,
+      "step": 1505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999613353256925e-06,
+      "loss": 1.0605,
+      "step": 1506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996110416227547e-06,
+      "loss": 0.8506,
+      "step": 1507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996087231002576e-06,
+      "loss": 0.877,
+      "step": 1508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999606397689449e-06,
+      "loss": 0.958,
+      "step": 1509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999604065390345e-06,
+      "loss": 0.959,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999601726202962e-06,
+      "loss": 0.9902,
+      "step": 1511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999599380127316e-06,
+      "loss": 1.1445,
+      "step": 1512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999597027163423e-06,
+      "loss": 0.915,
+      "step": 1513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995946673112996e-06,
+      "loss": 1.0547,
+      "step": 1514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995923005709615e-06,
+      "loss": 0.8569,
+      "step": 1515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995899269424253e-06,
+      "loss": 0.9375,
+      "step": 1516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995875464257076e-06,
+      "loss": 1.0537,
+      "step": 1517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995851590208242e-06,
+      "loss": 0.9463,
+      "step": 1518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995827647277923e-06,
+      "loss": 1.0273,
+      "step": 1519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995803635466275e-06,
+      "loss": 0.835,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995779554773476e-06,
+      "loss": 1.1064,
+      "step": 1521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995755405199678e-06,
+      "loss": 0.9014,
+      "step": 1522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999573118674506e-06,
+      "loss": 0.8662,
+      "step": 1523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999570689940978e-06,
+      "loss": 1.1318,
+      "step": 1524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995682543194006e-06,
+      "loss": 0.8418,
+      "step": 1525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999565811809791e-06,
+      "loss": 1.0977,
+      "step": 1526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999563362412166e-06,
+      "loss": 0.9766,
+      "step": 1527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999560906126542e-06,
+      "loss": 1.0645,
+      "step": 1528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995584429529363e-06,
+      "loss": 0.9307,
+      "step": 1529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999555972891366e-06,
+      "loss": 0.9482,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999553495941848e-06,
+      "loss": 0.8428,
+      "step": 1531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999551012104399e-06,
+      "loss": 1.1367,
+      "step": 1532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995485213790365e-06,
+      "loss": 0.9805,
+      "step": 1533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995460237657778e-06,
+      "loss": 0.9844,
+      "step": 1534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99954351926464e-06,
+      "loss": 1.0449,
+      "step": 1535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995410078756397e-06,
+      "loss": 0.7256,
+      "step": 1536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995384895987954e-06,
+      "loss": 1.0244,
+      "step": 1537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995359644341235e-06,
+      "loss": 1.1416,
+      "step": 1538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999533432381642e-06,
+      "loss": 0.9023,
+      "step": 1539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995308934413675e-06,
+      "loss": 0.9844,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995283476133184e-06,
+      "loss": 0.9971,
+      "step": 1541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999525794897512e-06,
+      "loss": 0.9946,
+      "step": 1542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995232352939655e-06,
+      "loss": 0.8398,
+      "step": 1543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995206688026972e-06,
+      "loss": 0.7612,
+      "step": 1544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999518095423724e-06,
+      "loss": 1.0303,
+      "step": 1545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995155151570646e-06,
+      "loss": 1.043,
+      "step": 1546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999512928002736e-06,
+      "loss": 0.7935,
+      "step": 1547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995103339607558e-06,
+      "loss": 0.9727,
+      "step": 1548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999507733031143e-06,
+      "loss": 1.0918,
+      "step": 1549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995051252139147e-06,
+      "loss": 0.9775,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999502510509089e-06,
+      "loss": 1.0801,
+      "step": 1551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994998889166837e-06,
+      "loss": 0.9668,
+      "step": 1552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994972604367174e-06,
+      "loss": 0.8999,
+      "step": 1553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994946250692076e-06,
+      "loss": 1.0742,
+      "step": 1554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999491982814173e-06,
+      "loss": 1.0664,
+      "step": 1555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994893336716317e-06,
+      "loss": 0.9912,
+      "step": 1556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994866776416016e-06,
+      "loss": 1.0098,
+      "step": 1557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994840147241013e-06,
+      "loss": 1.0605,
+      "step": 1558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994813449191495e-06,
+      "loss": 0.9033,
+      "step": 1559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999478668226764e-06,
+      "loss": 1.2324,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999475984646963e-06,
+      "loss": 1.1172,
+      "step": 1561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994732941797658e-06,
+      "loss": 0.918,
+      "step": 1562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994705968251907e-06,
+      "loss": 0.7783,
+      "step": 1563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994678925832564e-06,
+      "loss": 1.1055,
+      "step": 1564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994651814539803e-06,
+      "loss": 0.8374,
+      "step": 1565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999462463437383e-06,
+      "loss": 1.0898,
+      "step": 1566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999459738533482e-06,
+      "loss": 1.1279,
+      "step": 1567
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994570067422962e-06,
+      "loss": 1.1543,
+      "step": 1568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994542680638453e-06,
+      "loss": 0.8213,
+      "step": 1569
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999451522498147e-06,
+      "loss": 1.0059,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999448770045221e-06,
+      "loss": 1.1279,
+      "step": 1571
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994460107050856e-06,
+      "loss": 0.8525,
+      "step": 1572
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99944324447776e-06,
+      "loss": 1.1152,
+      "step": 1573
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999440471363264e-06,
+      "loss": 0.8008,
+      "step": 1574
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994376913616156e-06,
+      "loss": 0.9697,
+      "step": 1575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994349044728352e-06,
+      "loss": 0.7773,
+      "step": 1576
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999432110696941e-06,
+      "loss": 0.9307,
+      "step": 1577
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999429310033952e-06,
+      "loss": 1.0098,
+      "step": 1578
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994265024838892e-06,
+      "loss": 1.125,
+      "step": 1579
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99942368804677e-06,
+      "loss": 1.125,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999420866722615e-06,
+      "loss": 1.1309,
+      "step": 1581
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999418038511443e-06,
+      "loss": 0.8936,
+      "step": 1582
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994152034132742e-06,
+      "loss": 0.7495,
+      "step": 1583
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999412361428127e-06,
+      "loss": 0.6406,
+      "step": 1584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999409512556022e-06,
+      "loss": 1.0166,
+      "step": 1585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994066567969787e-06,
+      "loss": 0.8389,
+      "step": 1586
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994037941510165e-06,
+      "loss": 0.8076,
+      "step": 1587
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994009246181547e-06,
+      "loss": 0.7783,
+      "step": 1588
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993980481984145e-06,
+      "loss": 0.9165,
+      "step": 1589
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999395164891814e-06,
+      "loss": 1.1572,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993922746983744e-06,
+      "loss": 0.7473,
+      "step": 1591
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999389377618115e-06,
+      "loss": 0.7969,
+      "step": 1592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999386473651055e-06,
+      "loss": 0.9785,
+      "step": 1593
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999383562797216e-06,
+      "loss": 1.0273,
+      "step": 1594
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993806450566177e-06,
+      "loss": 0.7676,
+      "step": 1595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999377720429279e-06,
+      "loss": 0.8628,
+      "step": 1596
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993747889152215e-06,
+      "loss": 1.1709,
+      "step": 1597
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993718505144645e-06,
+      "loss": 0.9492,
+      "step": 1598
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993689052270286e-06,
+      "loss": 1.0029,
+      "step": 1599
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999365953052934e-06,
+      "loss": 0.8291,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999362993992201e-06,
+      "loss": 1.042,
+      "step": 1601
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99936002804485e-06,
+      "loss": 1.0566,
+      "step": 1602
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999357055210902e-06,
+      "loss": 0.8452,
+      "step": 1603
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993540754903764e-06,
+      "loss": 1.0996,
+      "step": 1604
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993510888832947e-06,
+      "loss": 1.166,
+      "step": 1605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999348095389677e-06,
+      "loss": 0.9253,
+      "step": 1606
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999345095009544e-06,
+      "loss": 0.8589,
+      "step": 1607
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993420877429165e-06,
+      "loss": 1.2207,
+      "step": 1608
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993390735898147e-06,
+      "loss": 0.8809,
+      "step": 1609
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99933605255026e-06,
+      "loss": 1.0078,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999333024624273e-06,
+      "loss": 0.9238,
+      "step": 1611
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999329989811875e-06,
+      "loss": 0.8311,
+      "step": 1612
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999326948113086e-06,
+      "loss": 0.9629,
+      "step": 1613
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993238995279277e-06,
+      "loss": 1.1055,
+      "step": 1614
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993208440564207e-06,
+      "loss": 0.8184,
+      "step": 1615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993177816985863e-06,
+      "loss": 1.0654,
+      "step": 1616
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993147124544455e-06,
+      "loss": 0.8213,
+      "step": 1617
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993116363240193e-06,
+      "loss": 0.9639,
+      "step": 1618
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993085533073295e-06,
+      "loss": 0.9814,
+      "step": 1619
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993054634043965e-06,
+      "loss": 1.0938,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993023666152424e-06,
+      "loss": 0.8604,
+      "step": 1621
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999299262939888e-06,
+      "loss": 0.8472,
+      "step": 1622
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992961523783545e-06,
+      "loss": 0.9385,
+      "step": 1623
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999293034930664e-06,
+      "loss": 0.9717,
+      "step": 1624
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992899105968373e-06,
+      "loss": 0.7954,
+      "step": 1625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992867793768963e-06,
+      "loss": 0.9414,
+      "step": 1626
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999283641270863e-06,
+      "loss": 0.938,
+      "step": 1627
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999280496278758e-06,
+      "loss": 0.896,
+      "step": 1628
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992773444006043e-06,
+      "loss": 1.1504,
+      "step": 1629
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992741856364224e-06,
+      "loss": 1.0752,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992710199862344e-06,
+      "loss": 1.1104,
+      "step": 1631
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992678474500625e-06,
+      "loss": 1.1387,
+      "step": 1632
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999264668027928e-06,
+      "loss": 0.9512,
+      "step": 1633
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999261481719854e-06,
+      "loss": 0.9375,
+      "step": 1634
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992582885258605e-06,
+      "loss": 1.1729,
+      "step": 1635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992550884459715e-06,
+      "loss": 0.9688,
+      "step": 1636
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992518814802074e-06,
+      "loss": 0.8936,
+      "step": 1637
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992486676285915e-06,
+      "loss": 1.1064,
+      "step": 1638
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992454468911453e-06,
+      "loss": 0.9648,
+      "step": 1639
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992422192678913e-06,
+      "loss": 1.1211,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999238984758851e-06,
+      "loss": 0.8062,
+      "step": 1641
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992357433640484e-06,
+      "loss": 0.9414,
+      "step": 1642
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999232495083504e-06,
+      "loss": 1.1533,
+      "step": 1643
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999229239917241e-06,
+      "loss": 0.9775,
+      "step": 1644
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992259778652822e-06,
+      "loss": 1.0303,
+      "step": 1645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999222708927649e-06,
+      "loss": 1.043,
+      "step": 1646
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999219433104365e-06,
+      "loss": 1.0879,
+      "step": 1647
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999216150395452e-06,
+      "loss": 0.9023,
+      "step": 1648
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992128608009335e-06,
+      "loss": 1.0557,
+      "step": 1649
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999209564320831e-06,
+      "loss": 1.1309,
+      "step": 1650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992062609551682e-06,
+      "loss": 0.8105,
+      "step": 1651
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992029507039676e-06,
+      "loss": 1.1309,
+      "step": 1652
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991996335672517e-06,
+      "loss": 0.8496,
+      "step": 1653
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991963095450436e-06,
+      "loss": 0.9219,
+      "step": 1654
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999192978637366e-06,
+      "loss": 0.9717,
+      "step": 1655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991896408442424e-06,
+      "loss": 0.9385,
+      "step": 1656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991862961656954e-06,
+      "loss": 0.7324,
+      "step": 1657
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999182944601748e-06,
+      "loss": 1.084,
+      "step": 1658
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991795861524233e-06,
+      "loss": 0.8599,
+      "step": 1659
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991762208177447e-06,
+      "loss": 0.8389,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991728485977347e-06,
+      "loss": 0.9453,
+      "step": 1661
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999169469492418e-06,
+      "loss": 0.9941,
+      "step": 1662
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999166083501816e-06,
+      "loss": 1.2207,
+      "step": 1663
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991626906259534e-06,
+      "loss": 1.0576,
+      "step": 1664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999159290864853e-06,
+      "loss": 0.958,
+      "step": 1665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991558842185386e-06,
+      "loss": 0.8574,
+      "step": 1666
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999152470687033e-06,
+      "loss": 1.2197,
+      "step": 1667
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999149050270361e-06,
+      "loss": 0.9854,
+      "step": 1668
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991456229685446e-06,
+      "loss": 1.0791,
+      "step": 1669
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991421887816083e-06,
+      "loss": 1.1025,
+      "step": 1670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999138747709576e-06,
+      "loss": 1.0205,
+      "step": 1671
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999135299752471e-06,
+      "loss": 0.8145,
+      "step": 1672
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999131844910317e-06,
+      "loss": 1.0859,
+      "step": 1673
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991283831831376e-06,
+      "loss": 1.1084,
+      "step": 1674
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991249145709574e-06,
+      "loss": 0.9141,
+      "step": 1675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991214390737994e-06,
+      "loss": 0.6851,
+      "step": 1676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991179566916883e-06,
+      "loss": 1.0488,
+      "step": 1677
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999114467424648e-06,
+      "loss": 0.8306,
+      "step": 1678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999110971272702e-06,
+      "loss": 1.0273,
+      "step": 1679
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999107468235875e-06,
+      "loss": 0.915,
+      "step": 1680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991039583141906e-06,
+      "loss": 0.8135,
+      "step": 1681
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991004415076733e-06,
+      "loss": 0.8157,
+      "step": 1682
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990969178163477e-06,
+      "loss": 0.9941,
+      "step": 1683
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990933872402375e-06,
+      "loss": 0.8926,
+      "step": 1684
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999089849779367e-06,
+      "loss": 0.9482,
+      "step": 1685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999086305433761e-06,
+      "loss": 0.8691,
+      "step": 1686
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990827542034437e-06,
+      "loss": 1.2578,
+      "step": 1687
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990791960884395e-06,
+      "loss": 0.9639,
+      "step": 1688
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999075631088773e-06,
+      "loss": 1.1738,
+      "step": 1689
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999072059204469e-06,
+      "loss": 0.8311,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990684804355516e-06,
+      "loss": 1.0938,
+      "step": 1691
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999064894782046e-06,
+      "loss": 1.1533,
+      "step": 1692
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990613022439765e-06,
+      "loss": 0.9297,
+      "step": 1693
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990577028213684e-06,
+      "loss": 0.8062,
+      "step": 1694
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990540965142455e-06,
+      "loss": 0.6846,
+      "step": 1695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990504833226337e-06,
+      "loss": 0.9062,
+      "step": 1696
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990468632465574e-06,
+      "loss": 0.9473,
+      "step": 1697
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990432362860417e-06,
+      "loss": 0.9629,
+      "step": 1698
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999039602441111e-06,
+      "loss": 1.1035,
+      "step": 1699
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990359617117916e-06,
+      "loss": 1.0391,
+      "step": 1700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990323140981075e-06,
+      "loss": 1.0166,
+      "step": 1701
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990286596000844e-06,
+      "loss": 1.0264,
+      "step": 1702
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999024998217747e-06,
+      "loss": 1.0117,
+      "step": 1703
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990213299511206e-06,
+      "loss": 0.9824,
+      "step": 1704
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999017654800231e-06,
+      "loss": 0.687,
+      "step": 1705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990139727651034e-06,
+      "loss": 0.873,
+      "step": 1706
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990102838457624e-06,
+      "loss": 1.0879,
+      "step": 1707
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999006588042234e-06,
+      "loss": 0.8877,
+      "step": 1708
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990028853545443e-06,
+      "loss": 1.0547,
+      "step": 1709
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989991757827176e-06,
+      "loss": 0.8799,
+      "step": 1710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989954593267802e-06,
+      "loss": 0.9863,
+      "step": 1711
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989917359867574e-06,
+      "loss": 0.7686,
+      "step": 1712
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989880057626755e-06,
+      "loss": 0.6729,
+      "step": 1713
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998984268654559e-06,
+      "loss": 0.8477,
+      "step": 1714
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989805246624347e-06,
+      "loss": 0.8809,
+      "step": 1715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998976773786328e-06,
+      "loss": 0.6428,
+      "step": 1716
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998973016026265e-06,
+      "loss": 0.9922,
+      "step": 1717
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989692513822713e-06,
+      "loss": 1.0752,
+      "step": 1718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998965479854373e-06,
+      "loss": 1.0547,
+      "step": 1719
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998961701442596e-06,
+      "loss": 0.8896,
+      "step": 1720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998957916146966e-06,
+      "loss": 1.1797,
+      "step": 1721
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99895412396751e-06,
+      "loss": 0.9365,
+      "step": 1722
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989503249042535e-06,
+      "loss": 1.0645,
+      "step": 1723
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989465189572226e-06,
+      "loss": 1.0713,
+      "step": 1724
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989427061264435e-06,
+      "loss": 1.0586,
+      "step": 1725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998938886411943e-06,
+      "loss": 0.8027,
+      "step": 1726
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998935059813747e-06,
+      "loss": 1.1504,
+      "step": 1727
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998931226331882e-06,
+      "loss": 1.1152,
+      "step": 1728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998927385966374e-06,
+      "loss": 1.1221,
+      "step": 1729
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989235387172502e-06,
+      "loss": 0.9307,
+      "step": 1730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998919684584537e-06,
+      "loss": 0.7588,
+      "step": 1731
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989158235682604e-06,
+      "loss": 0.8721,
+      "step": 1732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998911955668447e-06,
+      "loss": 0.79,
+      "step": 1733
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998908080885124e-06,
+      "loss": 0.8198,
+      "step": 1734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989041992183182e-06,
+      "loss": 1.0439,
+      "step": 1735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989003106680554e-06,
+      "loss": 1.0967,
+      "step": 1736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988964152343637e-06,
+      "loss": 0.9146,
+      "step": 1737
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998892512917269e-06,
+      "loss": 0.9922,
+      "step": 1738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998888603716798e-06,
+      "loss": 0.9307,
+      "step": 1739
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998884687632979e-06,
+      "loss": 1.0156,
+      "step": 1740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988807646658375e-06,
+      "loss": 1.0254,
+      "step": 1741
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998876834815401e-06,
+      "loss": 0.9307,
+      "step": 1742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998872898081697e-06,
+      "loss": 0.9395,
+      "step": 1743
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998868954464752e-06,
+      "loss": 0.9395,
+      "step": 1744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988650039645937e-06,
+      "loss": 1.0781,
+      "step": 1745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998861046581249e-06,
+      "loss": 0.9092,
+      "step": 1746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998857082314745e-06,
+      "loss": 0.9619,
+      "step": 1747
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99885311116511e-06,
+      "loss": 0.8418,
+      "step": 1748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99884913313237e-06,
+      "loss": 1.0107,
+      "step": 1749
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988451482165534e-06,
+      "loss": 1.0693,
+      "step": 1750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988411564176877e-06,
+      "loss": 0.585,
+      "step": 1751
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988371577357996e-06,
+      "loss": 1.1025,
+      "step": 1752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998833152170917e-06,
+      "loss": 1.0127,
+      "step": 1753
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988291397230677e-06,
+      "loss": 1.0195,
+      "step": 1754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998825120392279e-06,
+      "loss": 1.0059,
+      "step": 1755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988210941785794e-06,
+      "loss": 0.813,
+      "step": 1756
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998817061081996e-06,
+      "loss": 0.9575,
+      "step": 1757
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988130211025566e-06,
+      "loss": 0.874,
+      "step": 1758
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998808974240289e-06,
+      "loss": 0.9082,
+      "step": 1759
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988049204952215e-06,
+      "loss": 0.9834,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988008598673813e-06,
+      "loss": 0.9121,
+      "step": 1761
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998796792356797e-06,
+      "loss": 0.9648,
+      "step": 1762
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987927179634966e-06,
+      "loss": 0.916,
+      "step": 1763
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998788636687508e-06,
+      "loss": 1.0762,
+      "step": 1764
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998784548528859e-06,
+      "loss": 0.7773,
+      "step": 1765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998780453487579e-06,
+      "loss": 0.8271,
+      "step": 1766
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998776351563694e-06,
+      "loss": 0.9336,
+      "step": 1767
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998772242757235e-06,
+      "loss": 1.0684,
+      "step": 1768
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987681270682284e-06,
+      "loss": 0.8691,
+      "step": 1769
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987640044967028e-06,
+      "loss": 1.0322,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998759875042687e-06,
+      "loss": 1.0107,
+      "step": 1771
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99875573870621e-06,
+      "loss": 0.9771,
+      "step": 1772
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998751595487299e-06,
+      "loss": 1.0117,
+      "step": 1773
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987474453859832e-06,
+      "loss": 1.1914,
+      "step": 1774
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987432884022914e-06,
+      "loss": 0.7246,
+      "step": 1775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998739124536252e-06,
+      "loss": 0.8457,
+      "step": 1776
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998734953787894e-06,
+      "loss": 0.998,
+      "step": 1777
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987307761572456e-06,
+      "loss": 0.9697,
+      "step": 1778
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998726591644336e-06,
+      "loss": 1.0547,
+      "step": 1779
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987224002491937e-06,
+      "loss": 0.8652,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998718201971848e-06,
+      "loss": 0.9814,
+      "step": 1781
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998713996812328e-06,
+      "loss": 0.999,
+      "step": 1782
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987097847706614e-06,
+      "loss": 0.9629,
+      "step": 1783
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987055658468786e-06,
+      "loss": 1.1191,
+      "step": 1784
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998701340041008e-06,
+      "loss": 1.0957,
+      "step": 1785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986971073530796e-06,
+      "loss": 1.041,
+      "step": 1786
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986928677831214e-06,
+      "loss": 0.9502,
+      "step": 1787
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998688621331163e-06,
+      "loss": 1.0273,
+      "step": 1788
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986843679972337e-06,
+      "loss": 0.9521,
+      "step": 1789
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986801077813633e-06,
+      "loss": 1.0566,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986758406835806e-06,
+      "loss": 0.7463,
+      "step": 1791
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998671566703915e-06,
+      "loss": 0.9502,
+      "step": 1792
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998667285842396e-06,
+      "loss": 0.915,
+      "step": 1793
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986629980990536e-06,
+      "loss": 1.0508,
+      "step": 1794
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986587034739165e-06,
+      "loss": 1.0264,
+      "step": 1795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998654401967015e-06,
+      "loss": 0.8252,
+      "step": 1796
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986500935783784e-06,
+      "loss": 0.9834,
+      "step": 1797
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986457783080363e-06,
+      "loss": 0.8604,
+      "step": 1798
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986414561560186e-06,
+      "loss": 0.7881,
+      "step": 1799
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998637127122355e-06,
+      "loss": 1.0645,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998632791207076e-06,
+      "loss": 0.8447,
+      "step": 1801
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99862844841021e-06,
+      "loss": 1.1025,
+      "step": 1802
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998624098731789e-06,
+      "loss": 1.124,
+      "step": 1803
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998619742171841e-06,
+      "loss": 0.9131,
+      "step": 1804
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998615378730397e-06,
+      "loss": 1.0791,
+      "step": 1805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986110084074867e-06,
+      "loss": 0.9512,
+      "step": 1806
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986066312031403e-06,
+      "loss": 0.9814,
+      "step": 1807
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998602247117388e-06,
+      "loss": 1.0273,
+      "step": 1808
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985978561502608e-06,
+      "loss": 0.9604,
+      "step": 1809
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985934583017877e-06,
+      "loss": 0.7681,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985890535719995e-06,
+      "loss": 0.8281,
+      "step": 1811
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985846419609266e-06,
+      "loss": 1.3066,
+      "step": 1812
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985802234685995e-06,
+      "loss": 1.0215,
+      "step": 1813
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985757980950484e-06,
+      "loss": 0.5566,
+      "step": 1814
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998571365840304e-06,
+      "loss": 1.0391,
+      "step": 1815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998566926704397e-06,
+      "loss": 0.8662,
+      "step": 1816
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985624806873576e-06,
+      "loss": 0.8203,
+      "step": 1817
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985580277892167e-06,
+      "loss": 0.9795,
+      "step": 1818
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998553568010005e-06,
+      "loss": 1.1699,
+      "step": 1819
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985491013497527e-06,
+      "loss": 1.043,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998544627808491e-06,
+      "loss": 1.208,
+      "step": 1821
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998540147386251e-06,
+      "loss": 0.9775,
+      "step": 1822
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998535660083063e-06,
+      "loss": 1.1777,
+      "step": 1823
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998531165898959e-06,
+      "loss": 0.8574,
+      "step": 1824
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985266648339685e-06,
+      "loss": 0.8652,
+      "step": 1825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985221568881233e-06,
+      "loss": 1.126,
+      "step": 1826
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985176420614546e-06,
+      "loss": 0.9805,
+      "step": 1827
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998513120353993e-06,
+      "loss": 0.7686,
+      "step": 1828
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99850859176577e-06,
+      "loss": 0.918,
+      "step": 1829
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985040562968168e-06,
+      "loss": 0.8945,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984995139471644e-06,
+      "loss": 1.0869,
+      "step": 1831
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984949647168448e-06,
+      "loss": 0.8994,
+      "step": 1832
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984904086058885e-06,
+      "loss": 1.0,
+      "step": 1833
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984858456143274e-06,
+      "loss": 1.1719,
+      "step": 1834
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984812757421927e-06,
+      "loss": 0.9268,
+      "step": 1835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998476698989516e-06,
+      "loss": 0.8589,
+      "step": 1836
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984721153563287e-06,
+      "loss": 1.1191,
+      "step": 1837
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984675248426626e-06,
+      "loss": 0.9199,
+      "step": 1838
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984629274485496e-06,
+      "loss": 1.0771,
+      "step": 1839
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984583231740205e-06,
+      "loss": 0.8047,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998453712019108e-06,
+      "loss": 1.0674,
+      "step": 1841
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998449093983843e-06,
+      "loss": 0.9473,
+      "step": 1842
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998444469068258e-06,
+      "loss": 0.9941,
+      "step": 1843
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984398372723844e-06,
+      "loss": 0.835,
+      "step": 1844
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984351985962546e-06,
+      "loss": 1.0693,
+      "step": 1845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984305530399003e-06,
+      "loss": 0.874,
+      "step": 1846
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984259006033536e-06,
+      "loss": 0.9849,
+      "step": 1847
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998421241286646e-06,
+      "loss": 1.0693,
+      "step": 1848
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984165750898104e-06,
+      "loss": 1.1006,
+      "step": 1849
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998411902012879e-06,
+      "loss": 1.0557,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984072220558834e-06,
+      "loss": 1.084,
+      "step": 1851
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984025352188557e-06,
+      "loss": 1.0947,
+      "step": 1852
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998397841501829e-06,
+      "loss": 1.3066,
+      "step": 1853
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998393140904835e-06,
+      "loss": 1.2227,
+      "step": 1854
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983884334279063e-06,
+      "loss": 1.0918,
+      "step": 1855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983837190710755e-06,
+      "loss": 0.9248,
+      "step": 1856
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983789978343746e-06,
+      "loss": 1.0264,
+      "step": 1857
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998374269717837e-06,
+      "loss": 1.168,
+      "step": 1858
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983695347214946e-06,
+      "loss": 0.8779,
+      "step": 1859
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983647928453804e-06,
+      "loss": 0.9697,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983600440895264e-06,
+      "loss": 1.0752,
+      "step": 1861
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983552884539663e-06,
+      "loss": 0.7705,
+      "step": 1862
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983505259387317e-06,
+      "loss": 1.3057,
+      "step": 1863
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998345756543857e-06,
+      "loss": 1.0225,
+      "step": 1864
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983409802693734e-06,
+      "loss": 1.0498,
+      "step": 1865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998336197115315e-06,
+      "loss": 0.9961,
+      "step": 1866
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998331407081714e-06,
+      "loss": 0.8359,
+      "step": 1867
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983266101686038e-06,
+      "loss": 0.9185,
+      "step": 1868
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983218063760175e-06,
+      "loss": 0.9854,
+      "step": 1869
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998316995703988e-06,
+      "loss": 0.9443,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983121781525484e-06,
+      "loss": 1.0449,
+      "step": 1871
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983073537217325e-06,
+      "loss": 1.1328,
+      "step": 1872
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983025224115725e-06,
+      "loss": 1.0195,
+      "step": 1873
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982976842221027e-06,
+      "loss": 0.7285,
+      "step": 1874
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982928391533557e-06,
+      "loss": 0.8398,
+      "step": 1875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998287987205366e-06,
+      "loss": 1.1318,
+      "step": 1876
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982831283781653e-06,
+      "loss": 1.1074,
+      "step": 1877
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982782626717883e-06,
+      "loss": 1.0146,
+      "step": 1878
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982733900862684e-06,
+      "loss": 1.0439,
+      "step": 1879
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982685106216387e-06,
+      "loss": 0.9531,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982636242779333e-06,
+      "loss": 0.876,
+      "step": 1881
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998258731055186e-06,
+      "loss": 0.959,
+      "step": 1882
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99825383095343e-06,
+      "loss": 0.9707,
+      "step": 1883
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982489239726995e-06,
+      "loss": 0.6416,
+      "step": 1884
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982440101130276e-06,
+      "loss": 0.9312,
+      "step": 1885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982390893744496e-06,
+      "loss": 1.0459,
+      "step": 1886
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998234161756998e-06,
+      "loss": 0.8423,
+      "step": 1887
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998229227260707e-06,
+      "loss": 0.9473,
+      "step": 1888
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998224285885611e-06,
+      "loss": 0.8452,
+      "step": 1889
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982193376317438e-06,
+      "loss": 0.96,
+      "step": 1890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99821438249914e-06,
+      "loss": 0.9814,
+      "step": 1891
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998209420487833e-06,
+      "loss": 0.9297,
+      "step": 1892
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982044515978574e-06,
+      "loss": 1.084,
+      "step": 1893
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998199475829248e-06,
+      "loss": 0.9932,
+      "step": 1894
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981944931820376e-06,
+      "loss": 0.8604,
+      "step": 1895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998189503656262e-06,
+      "loss": 0.8037,
+      "step": 1896
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981845072519546e-06,
+      "loss": 0.8389,
+      "step": 1897
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99817950396915e-06,
+      "loss": 0.9941,
+      "step": 1898
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981744938078834e-06,
+      "loss": 0.916,
+      "step": 1899
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981694767681886e-06,
+      "loss": 0.9014,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981644528501005e-06,
+      "loss": 1.002,
+      "step": 1901
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981594220536538e-06,
+      "loss": 0.9463,
+      "step": 1902
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998154384378883e-06,
+      "loss": 0.9277,
+      "step": 1903
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981493398258223e-06,
+      "loss": 0.9814,
+      "step": 1904
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981442883945074e-06,
+      "loss": 0.8418,
+      "step": 1905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998139230084972e-06,
+      "loss": 1.0977,
+      "step": 1906
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981341648972525e-06,
+      "loss": 1.1152,
+      "step": 1907
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981290928313824e-06,
+      "loss": 0.7014,
+      "step": 1908
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998124013887397e-06,
+      "loss": 0.8828,
+      "step": 1909
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998118928065332e-06,
+      "loss": 1.0664,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998113835365222e-06,
+      "loss": 0.9639,
+      "step": 1911
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981087357871016e-06,
+      "loss": 1.0498,
+      "step": 1912
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981036293310064e-06,
+      "loss": 1.082,
+      "step": 1913
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998098515996972e-06,
+      "loss": 0.7969,
+      "step": 1914
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980933957850325e-06,
+      "loss": 1.0264,
+      "step": 1915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998088268695224e-06,
+      "loss": 1.0986,
+      "step": 1916
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980831347275822e-06,
+      "loss": 0.8838,
+      "step": 1917
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980779938821417e-06,
+      "loss": 0.9014,
+      "step": 1918
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980728461589385e-06,
+      "loss": 0.8623,
+      "step": 1919
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980676915580073e-06,
+      "loss": 1.084,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998062530079384e-06,
+      "loss": 1.1387,
+      "step": 1921
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998057361723105e-06,
+      "loss": 1.0322,
+      "step": 1922
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980521864892047e-06,
+      "loss": 1.1797,
+      "step": 1923
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998047004377719e-06,
+      "loss": 0.9424,
+      "step": 1924
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980418153886843e-06,
+      "loss": 0.8262,
+      "step": 1925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998036619522136e-06,
+      "loss": 1.0312,
+      "step": 1926
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980314167781092e-06,
+      "loss": 1.0205,
+      "step": 1927
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980262071566406e-06,
+      "loss": 0.8232,
+      "step": 1928
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980209906577663e-06,
+      "loss": 0.9893,
+      "step": 1929
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980157672815214e-06,
+      "loss": 1.0996,
+      "step": 1930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998010537027942e-06,
+      "loss": 1.0566,
+      "step": 1931
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980052998970655e-06,
+      "loss": 1.0439,
+      "step": 1932
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998000055888926e-06,
+      "loss": 1.0957,
+      "step": 1933
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997994805003561e-06,
+      "loss": 0.9131,
+      "step": 1934
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997989547241006e-06,
+      "loss": 1.0869,
+      "step": 1935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979842826012974e-06,
+      "loss": 0.9053,
+      "step": 1936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997979011084472e-06,
+      "loss": 1.0254,
+      "step": 1937
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979737326905654e-06,
+      "loss": 1.207,
+      "step": 1938
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979684474196144e-06,
+      "loss": 0.9668,
+      "step": 1939
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979631552716554e-06,
+      "loss": 0.8672,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979578562467244e-06,
+      "loss": 1.041,
+      "step": 1941
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979525503448587e-06,
+      "loss": 0.7715,
+      "step": 1942
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979472375660943e-06,
+      "loss": 0.7002,
+      "step": 1943
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979419179104675e-06,
+      "loss": 1.0566,
+      "step": 1944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997936591378016e-06,
+      "loss": 1.0781,
+      "step": 1945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997931257968776e-06,
+      "loss": 1.0645,
+      "step": 1946
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997925917682784e-06,
+      "loss": 1.0869,
+      "step": 1947
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979205705200764e-06,
+      "loss": 0.9365,
+      "step": 1948
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997915216480691e-06,
+      "loss": 0.9248,
+      "step": 1949
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997909855564664e-06,
+      "loss": 0.7549,
+      "step": 1950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979044877720335e-06,
+      "loss": 1.0068,
+      "step": 1951
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997899113102835e-06,
+      "loss": 1.0469,
+      "step": 1952
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978937315571065e-06,
+      "loss": 1.0752,
+      "step": 1953
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997888343134884e-06,
+      "loss": 0.9922,
+      "step": 1954
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978829478362062e-06,
+      "loss": 1.0,
+      "step": 1955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978775456611094e-06,
+      "loss": 0.8057,
+      "step": 1956
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978721366096305e-06,
+      "loss": 0.5181,
+      "step": 1957
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978667206818073e-06,
+      "loss": 0.8486,
+      "step": 1958
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997861297877677e-06,
+      "loss": 0.9355,
+      "step": 1959
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997855868197277e-06,
+      "loss": 0.9873,
+      "step": 1960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978504316406447e-06,
+      "loss": 0.9863,
+      "step": 1961
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978449882078177e-06,
+      "loss": 0.9854,
+      "step": 1962
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997839537898833e-06,
+      "loss": 0.957,
+      "step": 1963
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978340807137287e-06,
+      "loss": 0.6782,
+      "step": 1964
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978286166525425e-06,
+      "loss": 0.915,
+      "step": 1965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997823145715311e-06,
+      "loss": 0.9185,
+      "step": 1966
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978176679020736e-06,
+      "loss": 1.3047,
+      "step": 1967
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997812183212866e-06,
+      "loss": 1.0977,
+      "step": 1968
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997806691647728e-06,
+      "loss": 1.0107,
+      "step": 1969
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978011932066965e-06,
+      "loss": 0.8652,
+      "step": 1970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997795687889809e-06,
+      "loss": 0.9199,
+      "step": 1971
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997790175697104e-06,
+      "loss": 0.998,
+      "step": 1972
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977846566286196e-06,
+      "loss": 0.8711,
+      "step": 1973
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977791306843938e-06,
+      "loss": 1.1494,
+      "step": 1974
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997773597864464e-06,
+      "loss": 0.9609,
+      "step": 1975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997768058168869e-06,
+      "loss": 1.125,
+      "step": 1976
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977625115976468e-06,
+      "loss": 1.1738,
+      "step": 1977
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977569581508355e-06,
+      "loss": 0.9297,
+      "step": 1978
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997751397828473e-06,
+      "loss": 1.1328,
+      "step": 1979
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997745830630599e-06,
+      "loss": 1.0459,
+      "step": 1980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99774025655725e-06,
+      "loss": 1.0352,
+      "step": 1981
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997734675608466e-06,
+      "loss": 0.9395,
+      "step": 1982
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997729087784285e-06,
+      "loss": 0.834,
+      "step": 1983
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977234930847447e-06,
+      "loss": 1.0449,
+      "step": 1984
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977178915098848e-06,
+      "loss": 0.9941,
+      "step": 1985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997712283059743e-06,
+      "loss": 0.6355,
+      "step": 1986
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997706667734358e-06,
+      "loss": 0.7588,
+      "step": 1987
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99770104553377e-06,
+      "loss": 1.0859,
+      "step": 1988
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997695416458016e-06,
+      "loss": 1.1895,
+      "step": 1989
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997689780507135e-06,
+      "loss": 0.9248,
+      "step": 1990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997684137681166e-06,
+      "loss": 0.7959,
+      "step": 1991
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976784879801488e-06,
+      "loss": 0.7959,
+      "step": 1992
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976728314041213e-06,
+      "loss": 1.1064,
+      "step": 1993
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997667167953122e-06,
+      "loss": 0.8691,
+      "step": 1994
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997661497627192e-06,
+      "loss": 0.7659,
+      "step": 1995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976558204263686e-06,
+      "loss": 1.1172,
+      "step": 1996
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997650136350691e-06,
+      "loss": 0.8516,
+      "step": 1997
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976444454001993e-06,
+      "loss": 0.9766,
+      "step": 1998
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997638747574932e-06,
+      "loss": 1.0088,
+      "step": 1999
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976330428749282e-06,
+      "loss": 1.0693,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997627331300228e-06,
+      "loss": 1.0771,
+      "step": 2001
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976216128508706e-06,
+      "loss": 0.9688,
+      "step": 2002
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976158875268946e-06,
+      "loss": 0.9004,
+      "step": 2003
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99761015532834e-06,
+      "loss": 0.9932,
+      "step": 2004
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976044162552467e-06,
+      "loss": 0.7998,
+      "step": 2005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997598670307653e-06,
+      "loss": 0.9551,
+      "step": 2006
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975929174856004e-06,
+      "loss": 0.9941,
+      "step": 2007
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997587157789127e-06,
+      "loss": 0.8818,
+      "step": 2008
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975813912182724e-06,
+      "loss": 0.9062,
+      "step": 2009
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975756177730778e-06,
+      "loss": 0.9902,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975698374535814e-06,
+      "loss": 0.8374,
+      "step": 2011
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997564050259824e-06,
+      "loss": 0.96,
+      "step": 2012
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975582561918453e-06,
+      "loss": 1.0664,
+      "step": 2013
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997552455249685e-06,
+      "loss": 0.7451,
+      "step": 2014
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975466474333834e-06,
+      "loss": 0.8682,
+      "step": 2015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975408327429803e-06,
+      "loss": 0.9102,
+      "step": 2016
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975350111785156e-06,
+      "loss": 0.8188,
+      "step": 2017
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975291827400295e-06,
+      "loss": 0.9688,
+      "step": 2018
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975233474275624e-06,
+      "loss": 0.9639,
+      "step": 2019
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975175052411543e-06,
+      "loss": 0.9707,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997511656180845e-06,
+      "loss": 0.9756,
+      "step": 2021
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975058002466765e-06,
+      "loss": 1.0352,
+      "step": 2022
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974999374386873e-06,
+      "loss": 0.7651,
+      "step": 2023
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997494067756919e-06,
+      "loss": 0.9248,
+      "step": 2024
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997488191201411e-06,
+      "loss": 1.1104,
+      "step": 2025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997482307772204e-06,
+      "loss": 1.0605,
+      "step": 2026
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974764174693395e-06,
+      "loss": 1.0762,
+      "step": 2027
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974705202928576e-06,
+      "loss": 0.9766,
+      "step": 2028
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997464616242799e-06,
+      "loss": 0.9561,
+      "step": 2029
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974587053192035e-06,
+      "loss": 1.0439,
+      "step": 2030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974527875221127e-06,
+      "loss": 0.8906,
+      "step": 2031
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974468628515676e-06,
+      "loss": 1.1406,
+      "step": 2032
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997440931307608e-06,
+      "loss": 1.1562,
+      "step": 2033
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974349928902757e-06,
+      "loss": 1.0664,
+      "step": 2034
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997429047599612e-06,
+      "loss": 0.8252,
+      "step": 2035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974230954356563e-06,
+      "loss": 0.9199,
+      "step": 2036
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974171363984507e-06,
+      "loss": 0.9434,
+      "step": 2037
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997411170488036e-06,
+      "loss": 0.9541,
+      "step": 2038
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997405197704454e-06,
+      "loss": 0.8125,
+      "step": 2039
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973992180477443e-06,
+      "loss": 0.8369,
+      "step": 2040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99739323151795e-06,
+      "loss": 0.9072,
+      "step": 2041
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973872381151107e-06,
+      "loss": 1.2539,
+      "step": 2042
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973812378392688e-06,
+      "loss": 0.9834,
+      "step": 2043
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997375230690465e-06,
+      "loss": 1.0635,
+      "step": 2044
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997369216668741e-06,
+      "loss": 1.2051,
+      "step": 2045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973631957741383e-06,
+      "loss": 1.2891,
+      "step": 2046
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997357168006698e-06,
+      "loss": 1.1719,
+      "step": 2047
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973511333664627e-06,
+      "loss": 0.959,
+      "step": 2048
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973450918534726e-06,
+      "loss": 0.897,
+      "step": 2049
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99733904346777e-06,
+      "loss": 0.8496,
+      "step": 2050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973329882093964e-06,
+      "loss": 0.8789,
+      "step": 2051
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997326926078394e-06,
+      "loss": 0.7881,
+      "step": 2052
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997320857074804e-06,
+      "loss": 0.9092,
+      "step": 2053
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973147811986683e-06,
+      "loss": 1.0195,
+      "step": 2054
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997308698450029e-06,
+      "loss": 0.9492,
+      "step": 2055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997302608828928e-06,
+      "loss": 0.958,
+      "step": 2056
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972965123354072e-06,
+      "loss": 0.9248,
+      "step": 2057
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972904089695087e-06,
+      "loss": 0.9355,
+      "step": 2058
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972842987312744e-06,
+      "loss": 0.9629,
+      "step": 2059
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972781816207463e-06,
+      "loss": 0.7534,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997272057637967e-06,
+      "loss": 0.9316,
+      "step": 2061
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997265926782978e-06,
+      "loss": 0.9932,
+      "step": 2062
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997259789055822e-06,
+      "loss": 1.1631,
+      "step": 2063
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972536444565413e-06,
+      "loss": 0.9307,
+      "step": 2064
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972474929851785e-06,
+      "loss": 1.1758,
+      "step": 2065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972413346417756e-06,
+      "loss": 0.8535,
+      "step": 2066
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972351694263753e-06,
+      "loss": 0.8877,
+      "step": 2067
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972289973390196e-06,
+      "loss": 0.9014,
+      "step": 2068
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972228183797516e-06,
+      "loss": 0.9512,
+      "step": 2069
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972166325486134e-06,
+      "loss": 1.0146,
+      "step": 2070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972104398456476e-06,
+      "loss": 1.0098,
+      "step": 2071
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997204240270898e-06,
+      "loss": 0.9961,
+      "step": 2072
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997198033824406e-06,
+      "loss": 1.0635,
+      "step": 2073
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971918205062145e-06,
+      "loss": 0.6246,
+      "step": 2074
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971856003163668e-06,
+      "loss": 0.9639,
+      "step": 2075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971793732549054e-06,
+      "loss": 0.9502,
+      "step": 2076
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997173139321874e-06,
+      "loss": 0.9287,
+      "step": 2077
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997166898517314e-06,
+      "loss": 1.1602,
+      "step": 2078
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971606508412705e-06,
+      "loss": 0.6318,
+      "step": 2079
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971543962937848e-06,
+      "loss": 0.9038,
+      "step": 2080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971481348749006e-06,
+      "loss": 1.1377,
+      "step": 2081
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971418665846615e-06,
+      "loss": 0.8838,
+      "step": 2082
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971355914231094e-06,
+      "loss": 1.0156,
+      "step": 2083
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971293093902893e-06,
+      "loss": 1.1309,
+      "step": 2084
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997123020486243e-06,
+      "loss": 0.9854,
+      "step": 2085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997116724711015e-06,
+      "loss": 1.0801,
+      "step": 2086
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997110422064648e-06,
+      "loss": 0.9336,
+      "step": 2087
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997104112547185e-06,
+      "loss": 0.6624,
+      "step": 2088
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970977961586704e-06,
+      "loss": 1.0576,
+      "step": 2089
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970914728991474e-06,
+      "loss": 0.9463,
+      "step": 2090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970851427686595e-06,
+      "loss": 0.9414,
+      "step": 2091
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970788057672505e-06,
+      "loss": 0.79,
+      "step": 2092
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997072461894964e-06,
+      "loss": 0.7668,
+      "step": 2093
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997066111151843e-06,
+      "loss": 1.0605,
+      "step": 2094
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997059753537932e-06,
+      "loss": 1.0146,
+      "step": 2095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997053389053275e-06,
+      "loss": 0.7148,
+      "step": 2096
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970470176979156e-06,
+      "loss": 0.9971,
+      "step": 2097
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970406394718975e-06,
+      "loss": 0.9053,
+      "step": 2098
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970342543752653e-06,
+      "loss": 0.9482,
+      "step": 2099
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997027862408062e-06,
+      "loss": 1.0898,
+      "step": 2100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970214635703324e-06,
+      "loss": 0.9189,
+      "step": 2101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970150578621203e-06,
+      "loss": 1.1445,
+      "step": 2102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970086452834698e-06,
+      "loss": 0.9629,
+      "step": 2103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970022258344253e-06,
+      "loss": 0.8818,
+      "step": 2104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996995799515031e-06,
+      "loss": 1.1348,
+      "step": 2105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996989366325331e-06,
+      "loss": 0.958,
+      "step": 2106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969829262653697e-06,
+      "loss": 1.0303,
+      "step": 2107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969764793351917e-06,
+      "loss": 1.0791,
+      "step": 2108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996970025534841e-06,
+      "loss": 0.9785,
+      "step": 2109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969635648643625e-06,
+      "loss": 0.8955,
+      "step": 2110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969570973238006e-06,
+      "loss": 1.1738,
+      "step": 2111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969506229131997e-06,
+      "loss": 1.1318,
+      "step": 2112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969441416326047e-06,
+      "loss": 0.9668,
+      "step": 2113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99693765348206e-06,
+      "loss": 1.0234,
+      "step": 2114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99693115846161e-06,
+      "loss": 0.9521,
+      "step": 2115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969246565713004e-06,
+      "loss": 1.1494,
+      "step": 2116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969181478111753e-06,
+      "loss": 0.6646,
+      "step": 2117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969116321812796e-06,
+      "loss": 1.0762,
+      "step": 2118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969051096816583e-06,
+      "loss": 0.7627,
+      "step": 2119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968985803123568e-06,
+      "loss": 1.0381,
+      "step": 2120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996892044073419e-06,
+      "loss": 0.8838,
+      "step": 2121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996885500964891e-06,
+      "loss": 1.2061,
+      "step": 2122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968789509868175e-06,
+      "loss": 0.9219,
+      "step": 2123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968723941392432e-06,
+      "loss": 1.1582,
+      "step": 2124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996865830422214e-06,
+      "loss": 1.0596,
+      "step": 2125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968592598357746e-06,
+      "loss": 1.1299,
+      "step": 2126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996852682379971e-06,
+      "loss": 0.8682,
+      "step": 2127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968460980548477e-06,
+      "loss": 1.1045,
+      "step": 2128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968395068604504e-06,
+      "loss": 0.957,
+      "step": 2129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968329087968247e-06,
+      "loss": 1.1406,
+      "step": 2130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996826303864016e-06,
+      "loss": 1.0205,
+      "step": 2131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968196920620693e-06,
+      "loss": 0.707,
+      "step": 2132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996813073391031e-06,
+      "loss": 0.7773,
+      "step": 2133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968064478509457e-06,
+      "loss": 0.9658,
+      "step": 2134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99679981544186e-06,
+      "loss": 0.9121,
+      "step": 2135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967931761638195e-06,
+      "loss": 1.0078,
+      "step": 2136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967865300168692e-06,
+      "loss": 1.0527,
+      "step": 2137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967798770010557e-06,
+      "loss": 0.9834,
+      "step": 2138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967732171164245e-06,
+      "loss": 0.877,
+      "step": 2139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967665503630215e-06,
+      "loss": 1.1191,
+      "step": 2140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967598767408928e-06,
+      "loss": 1.21,
+      "step": 2141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996753196250084e-06,
+      "loss": 1.082,
+      "step": 2142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967465088906416e-06,
+      "loss": 0.8076,
+      "step": 2143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967398146626115e-06,
+      "loss": 0.8706,
+      "step": 2144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967331135660395e-06,
+      "loss": 1.1289,
+      "step": 2145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967264056009727e-06,
+      "loss": 0.7769,
+      "step": 2146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996719690767456e-06,
+      "loss": 0.9902,
+      "step": 2147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996712969065537e-06,
+      "loss": 0.8926,
+      "step": 2148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967062404952607e-06,
+      "loss": 0.7456,
+      "step": 2149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966995050566746e-06,
+      "loss": 1.042,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966927627498246e-06,
+      "loss": 0.8359,
+      "step": 2151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996686013574757e-06,
+      "loss": 0.9746,
+      "step": 2152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996679257531519e-06,
+      "loss": 0.8081,
+      "step": 2153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966724946201567e-06,
+      "loss": 1.1191,
+      "step": 2154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966657248407163e-06,
+      "loss": 0.8574,
+      "step": 2155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996658948193245e-06,
+      "loss": 0.8428,
+      "step": 2156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99665216467779e-06,
+      "loss": 0.9443,
+      "step": 2157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966453742943966e-06,
+      "loss": 0.6816,
+      "step": 2158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996638577043113e-06,
+      "loss": 1.002,
+      "step": 2159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996631772923985e-06,
+      "loss": 1.0254,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966249619370597e-06,
+      "loss": 0.8799,
+      "step": 2161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996618144082385e-06,
+      "loss": 0.7119,
+      "step": 2162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966113193600065e-06,
+      "loss": 0.9795,
+      "step": 2163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996604487769972e-06,
+      "loss": 0.9365,
+      "step": 2164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965976493123285e-06,
+      "loss": 0.9502,
+      "step": 2165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996590803987123e-06,
+      "loss": 0.8838,
+      "step": 2166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996583951794403e-06,
+      "loss": 0.9932,
+      "step": 2167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965770927342158e-06,
+      "loss": 1.1201,
+      "step": 2168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965702268066073e-06,
+      "loss": 1.1016,
+      "step": 2169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965633540116267e-06,
+      "loss": 0.9053,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99655647434932e-06,
+      "loss": 1.0352,
+      "step": 2171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965495878197355e-06,
+      "loss": 1.1172,
+      "step": 2172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965426944229197e-06,
+      "loss": 1.1748,
+      "step": 2173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965357941589215e-06,
+      "loss": 0.729,
+      "step": 2174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965288870277866e-06,
+      "loss": 0.8936,
+      "step": 2175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965219730295643e-06,
+      "loss": 0.8086,
+      "step": 2176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965150521643013e-06,
+      "loss": 1.0576,
+      "step": 2177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965081244320458e-06,
+      "loss": 1.0488,
+      "step": 2178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965011898328454e-06,
+      "loss": 1.0312,
+      "step": 2179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964942483667472e-06,
+      "loss": 0.7231,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964873000338e-06,
+      "loss": 0.8984,
+      "step": 2181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964803448340515e-06,
+      "loss": 0.8037,
+      "step": 2182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964733827675487e-06,
+      "loss": 1.1738,
+      "step": 2183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996466413834341e-06,
+      "loss": 0.9873,
+      "step": 2184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964594380344754e-06,
+      "loss": 0.9072,
+      "step": 2185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996452455368e-06,
+      "loss": 1.1377,
+      "step": 2186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964454658349637e-06,
+      "loss": 1.0645,
+      "step": 2187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996438469435414e-06,
+      "loss": 0.7314,
+      "step": 2188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964314661693995e-06,
+      "loss": 0.9355,
+      "step": 2189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964244560369677e-06,
+      "loss": 0.9932,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964174390381676e-06,
+      "loss": 1.0088,
+      "step": 2191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996410415173048e-06,
+      "loss": 1.0576,
+      "step": 2192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996403384441656e-06,
+      "loss": 0.9736,
+      "step": 2193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996396346844041e-06,
+      "loss": 1.0098,
+      "step": 2194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963893023802513e-06,
+      "loss": 0.7744,
+      "step": 2195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996382251050335e-06,
+      "loss": 0.8379,
+      "step": 2196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996375192854342e-06,
+      "loss": 1.0088,
+      "step": 2197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996368127792319e-06,
+      "loss": 0.7959,
+      "step": 2198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963610558643166e-06,
+      "loss": 0.8691,
+      "step": 2199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996353977070382e-06,
+      "loss": 0.877,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963468914105647e-06,
+      "loss": 0.7087,
+      "step": 2201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963397988849135e-06,
+      "loss": 0.999,
+      "step": 2202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996332699493477e-06,
+      "loss": 0.9541,
+      "step": 2203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996325593236305e-06,
+      "loss": 0.8262,
+      "step": 2204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996318480113445e-06,
+      "loss": 1.1494,
+      "step": 2205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963113601249475e-06,
+      "loss": 1.0771,
+      "step": 2206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963042332708607e-06,
+      "loss": 0.9941,
+      "step": 2207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996297099551234e-06,
+      "loss": 0.8926,
+      "step": 2208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996289958966116e-06,
+      "loss": 0.9375,
+      "step": 2209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962828115155566e-06,
+      "loss": 0.8057,
+      "step": 2210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996275657199605e-06,
+      "loss": 0.9219,
+      "step": 2211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962684960183102e-06,
+      "loss": 1.1777,
+      "step": 2212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962613279717215e-06,
+      "loss": 0.876,
+      "step": 2213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996254153059889e-06,
+      "loss": 0.957,
+      "step": 2214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996246971282861e-06,
+      "loss": 0.9863,
+      "step": 2215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962397826406882e-06,
+      "loss": 0.7559,
+      "step": 2216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962325871334193e-06,
+      "loss": 0.999,
+      "step": 2217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996225384761104e-06,
+      "loss": 0.8477,
+      "step": 2218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962181755237916e-06,
+      "loss": 0.9551,
+      "step": 2219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996210959421533e-06,
+      "loss": 0.79,
+      "step": 2220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962037364543765e-06,
+      "loss": 0.4814,
+      "step": 2221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996196506622373e-06,
+      "loss": 1.0068,
+      "step": 2222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961892699255715e-06,
+      "loss": 0.9521,
+      "step": 2223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961820263640224e-06,
+      "loss": 1.0527,
+      "step": 2224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961747759377757e-06,
+      "loss": 0.9619,
+      "step": 2225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961675186468805e-06,
+      "loss": 1.0781,
+      "step": 2226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996160254491388e-06,
+      "loss": 1.0156,
+      "step": 2227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961529834713474e-06,
+      "loss": 1.1602,
+      "step": 2228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996145705586809e-06,
+      "loss": 0.9424,
+      "step": 2229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996138420837823e-06,
+      "loss": 1.041,
+      "step": 2230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99613112922444e-06,
+      "loss": 0.8896,
+      "step": 2231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961238307467096e-06,
+      "loss": 1.0098,
+      "step": 2232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961165254046822e-06,
+      "loss": 0.9756,
+      "step": 2233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961092131984083e-06,
+      "loss": 1.25,
+      "step": 2234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961018941279385e-06,
+      "loss": 0.8525,
+      "step": 2235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960945681933226e-06,
+      "loss": 0.6929,
+      "step": 2236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996087235394612e-06,
+      "loss": 1.0459,
+      "step": 2237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960798957318566e-06,
+      "loss": 1.1016,
+      "step": 2238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996072549205107e-06,
+      "loss": 0.9277,
+      "step": 2239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996065195814414e-06,
+      "loss": 1.0322,
+      "step": 2240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960578355598285e-06,
+      "loss": 0.8436,
+      "step": 2241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960504684414004e-06,
+      "loss": 0.8682,
+      "step": 2242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960430944591815e-06,
+      "loss": 0.9482,
+      "step": 2243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960357136132217e-06,
+      "loss": 1.0342,
+      "step": 2244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960283259035726e-06,
+      "loss": 0.8721,
+      "step": 2245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960209313302847e-06,
+      "loss": 1.1348,
+      "step": 2246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960135298934087e-06,
+      "loss": 0.8447,
+      "step": 2247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960061215929964e-06,
+      "loss": 0.8643,
+      "step": 2248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995998706429098e-06,
+      "loss": 0.9453,
+      "step": 2249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959912844017653e-06,
+      "loss": 1.2715,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995983855511049e-06,
+      "loss": 0.8945,
+      "step": 2251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959764197570007e-06,
+      "loss": 0.8193,
+      "step": 2252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959689771396707e-06,
+      "loss": 0.9775,
+      "step": 2253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995961527659112e-06,
+      "loss": 0.9116,
+      "step": 2254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995954071315374e-06,
+      "loss": 0.8623,
+      "step": 2255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995946608108509e-06,
+      "loss": 0.9756,
+      "step": 2256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995939138038569e-06,
+      "loss": 1.0811,
+      "step": 2257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959316611056045e-06,
+      "loss": 0.8818,
+      "step": 2258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959241773096674e-06,
+      "loss": 1.0742,
+      "step": 2259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959166866508093e-06,
+      "loss": 1.0498,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959091891290815e-06,
+      "loss": 1.0469,
+      "step": 2261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959016847445364e-06,
+      "loss": 0.8135,
+      "step": 2262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995894173497225e-06,
+      "loss": 0.8057,
+      "step": 2263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958866553871997e-06,
+      "loss": 1.0635,
+      "step": 2264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958791304145114e-06,
+      "loss": 1.0615,
+      "step": 2265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995871598579213e-06,
+      "loss": 0.8926,
+      "step": 2266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958640598813556e-06,
+      "loss": 1.0391,
+      "step": 2267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958565143209914e-06,
+      "loss": 0.9883,
+      "step": 2268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958489618981726e-06,
+      "loss": 0.8838,
+      "step": 2269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995841402612951e-06,
+      "loss": 0.9658,
+      "step": 2270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958338364653785e-06,
+      "loss": 1.1045,
+      "step": 2271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958262634555077e-06,
+      "loss": 0.8887,
+      "step": 2272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995818683583391e-06,
+      "loss": 1.0703,
+      "step": 2273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99581109684908e-06,
+      "loss": 0.8525,
+      "step": 2274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995803503252627e-06,
+      "loss": 0.9668,
+      "step": 2275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957959027940847e-06,
+      "loss": 1.1445,
+      "step": 2276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957882954735054e-06,
+      "loss": 0.7466,
+      "step": 2277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995780681290941e-06,
+      "loss": 1.0049,
+      "step": 2278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957730602464448e-06,
+      "loss": 0.9854,
+      "step": 2279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995765432340069e-06,
+      "loss": 1.1113,
+      "step": 2280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957577975718655e-06,
+      "loss": 0.9307,
+      "step": 2281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995750155941888e-06,
+      "loss": 0.8398,
+      "step": 2282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957425074501886e-06,
+      "loss": 0.9756,
+      "step": 2283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99573485209682e-06,
+      "loss": 0.7549,
+      "step": 2284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995727189881835e-06,
+      "loss": 0.8184,
+      "step": 2285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995719520805287e-06,
+      "loss": 1.0244,
+      "step": 2286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995711844867227e-06,
+      "loss": 1.0615,
+      "step": 2287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957041620677104e-06,
+      "loss": 0.7686,
+      "step": 2288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995696472406788e-06,
+      "loss": 1.1123,
+      "step": 2289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956887758845144e-06,
+      "loss": 1.1699,
+      "step": 2290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956810725009414e-06,
+      "loss": 0.698,
+      "step": 2291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995673362256123e-06,
+      "loss": 0.9854,
+      "step": 2292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995665645150111e-06,
+      "loss": 1.0225,
+      "step": 2293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956579211829603e-06,
+      "loss": 0.9326,
+      "step": 2294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956501903547236e-06,
+      "loss": 0.9395,
+      "step": 2295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956424526654535e-06,
+      "loss": 1.002,
+      "step": 2296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995634708115204e-06,
+      "loss": 0.9814,
+      "step": 2297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995626956704028e-06,
+      "loss": 0.8896,
+      "step": 2298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995619198431979e-06,
+      "loss": 0.9082,
+      "step": 2299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995611433299111e-06,
+      "loss": 0.8267,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956036613054766e-06,
+      "loss": 1.1543,
+      "step": 2301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955958824511303e-06,
+      "loss": 1.0537,
+      "step": 2302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995588096736125e-06,
+      "loss": 1.0791,
+      "step": 2303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955803041605146e-06,
+      "loss": 0.9893,
+      "step": 2304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955725047243527e-06,
+      "loss": 0.9727,
+      "step": 2305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955646984276933e-06,
+      "loss": 1.0098,
+      "step": 2306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955568852705904e-06,
+      "loss": 0.8047,
+      "step": 2307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955490652530966e-06,
+      "loss": 0.8506,
+      "step": 2308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995541238375267e-06,
+      "loss": 0.835,
+      "step": 2309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995533404637156e-06,
+      "loss": 1.0127,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995525564038816e-06,
+      "loss": 1.0547,
+      "step": 2311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995517716580302e-06,
+      "loss": 1.1191,
+      "step": 2312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995509862261668e-06,
+      "loss": 1.001,
+      "step": 2313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995502001082968e-06,
+      "loss": 0.9307,
+      "step": 2314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954941330442558e-06,
+      "loss": 1.0215,
+      "step": 2315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954862581455864e-06,
+      "loss": 0.9375,
+      "step": 2316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954783763870137e-06,
+      "loss": 0.9443,
+      "step": 2317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954704877685917e-06,
+      "loss": 0.6733,
+      "step": 2318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995462592290375e-06,
+      "loss": 0.8398,
+      "step": 2319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954546899524187e-06,
+      "loss": 0.8809,
+      "step": 2320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995446780754776e-06,
+      "loss": 0.8125,
+      "step": 2321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954388646975023e-06,
+      "loss": 0.9102,
+      "step": 2322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954309417806517e-06,
+      "loss": 1.0957,
+      "step": 2323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995423012004279e-06,
+      "loss": 1.0527,
+      "step": 2324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954150753684387e-06,
+      "loss": 0.8975,
+      "step": 2325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995407131873186e-06,
+      "loss": 0.9062,
+      "step": 2326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995399181518575e-06,
+      "loss": 0.957,
+      "step": 2327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953912243046607e-06,
+      "loss": 1.1621,
+      "step": 2328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953832602314977e-06,
+      "loss": 0.9873,
+      "step": 2329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953752892991416e-06,
+      "loss": 0.8926,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953673115076464e-06,
+      "loss": 1.1045,
+      "step": 2331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995359326857068e-06,
+      "loss": 0.833,
+      "step": 2332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953513353474607e-06,
+      "loss": 1.248,
+      "step": 2333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99534333697888e-06,
+      "loss": 0.9287,
+      "step": 2334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953353317513806e-06,
+      "loss": 0.9062,
+      "step": 2335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995327319665018e-06,
+      "loss": 0.9199,
+      "step": 2336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953193007198475e-06,
+      "loss": 1.2217,
+      "step": 2337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953112749159236e-06,
+      "loss": 0.999,
+      "step": 2338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995303242253303e-06,
+      "loss": 0.8936,
+      "step": 2339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952952027320397e-06,
+      "loss": 1.0186,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952871563521897e-06,
+      "loss": 1.043,
+      "step": 2341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952791031138085e-06,
+      "loss": 0.8491,
+      "step": 2342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952710430169515e-06,
+      "loss": 1.1406,
+      "step": 2343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952629760616743e-06,
+      "loss": 1.1162,
+      "step": 2344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952549022480323e-06,
+      "loss": 0.917,
+      "step": 2345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995246821576081e-06,
+      "loss": 1.0137,
+      "step": 2346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995238734045877e-06,
+      "loss": 0.8096,
+      "step": 2347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995230639657475e-06,
+      "loss": 0.7749,
+      "step": 2348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995222538410931e-06,
+      "loss": 1.0176,
+      "step": 2349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995214430306301e-06,
+      "loss": 0.8799,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995206315343641e-06,
+      "loss": 0.9961,
+      "step": 2351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951981935230065e-06,
+      "loss": 0.791,
+      "step": 2352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951900648444537e-06,
+      "loss": 1.0195,
+      "step": 2353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995181929308039e-06,
+      "loss": 0.6948,
+      "step": 2354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995173786913818e-06,
+      "loss": 1.0361,
+      "step": 2355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951656376618467e-06,
+      "loss": 0.8799,
+      "step": 2356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995157481552182e-06,
+      "loss": 0.6646,
+      "step": 2357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995149318584879e-06,
+      "loss": 0.9277,
+      "step": 2358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951411487599947e-06,
+      "loss": 0.96,
+      "step": 2359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995132972077585e-06,
+      "loss": 0.9971,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995124788537707e-06,
+      "loss": 0.9961,
+      "step": 2361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951165981404165e-06,
+      "loss": 1.1084,
+      "step": 2362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951084008857696e-06,
+      "loss": 0.7422,
+      "step": 2363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951001967738235e-06,
+      "loss": 1.248,
+      "step": 2364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995091985804634e-06,
+      "loss": 0.9854,
+      "step": 2365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995083767978259e-06,
+      "loss": 1.085,
+      "step": 2366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950755432947536e-06,
+      "loss": 1.1084,
+      "step": 2367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995067311754175e-06,
+      "loss": 0.9932,
+      "step": 2368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950590733565804e-06,
+      "loss": 1.1162,
+      "step": 2369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995050828102026e-06,
+      "loss": 1.0127,
+      "step": 2370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995042575990569e-06,
+      "loss": 0.7891,
+      "step": 2371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950343170222658e-06,
+      "loss": 0.9844,
+      "step": 2372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950260511971737e-06,
+      "loss": 1.1143,
+      "step": 2373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950177785153495e-06,
+      "loss": 1.0068,
+      "step": 2374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995009498976851e-06,
+      "loss": 1.1152,
+      "step": 2375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995001212581734e-06,
+      "loss": 0.9595,
+      "step": 2376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949929193300555e-06,
+      "loss": 0.9932,
+      "step": 2377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994984619221874e-06,
+      "loss": 0.9434,
+      "step": 2378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994976312257246e-06,
+      "loss": 1.0889,
+      "step": 2379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994967998436228e-06,
+      "loss": 0.9785,
+      "step": 2380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994959677758879e-06,
+      "loss": 1.1016,
+      "step": 2381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949513502252547e-06,
+      "loss": 0.9922,
+      "step": 2382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949430158354134e-06,
+      "loss": 0.8086,
+      "step": 2383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994934674589412e-06,
+      "loss": 1.002,
+      "step": 2384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994926326487309e-06,
+      "loss": 0.894,
+      "step": 2385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949179715291604e-06,
+      "loss": 0.8369,
+      "step": 2386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949096097150243e-06,
+      "loss": 1.0791,
+      "step": 2387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994901241044959e-06,
+      "loss": 0.998,
+      "step": 2388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994892865519022e-06,
+      "loss": 0.8984,
+      "step": 2389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948844831372705e-06,
+      "loss": 0.9697,
+      "step": 2390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994876093899762e-06,
+      "loss": 0.9067,
+      "step": 2391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948676978065556e-06,
+      "loss": 1.0605,
+      "step": 2392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994859294857708e-06,
+      "loss": 1.002,
+      "step": 2393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948508850532777e-06,
+      "loss": 1.1035,
+      "step": 2394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994842468393322e-06,
+      "loss": 0.8354,
+      "step": 2395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948340448779e-06,
+      "loss": 0.9199,
+      "step": 2396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948256145070685e-06,
+      "loss": 0.9971,
+      "step": 2397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948171772808866e-06,
+      "loss": 1.1768,
+      "step": 2398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948087331994116e-06,
+      "loss": 1.1006,
+      "step": 2399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948002822627025e-06,
+      "loss": 1.0352,
+      "step": 2400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994791824470817e-06,
+      "loss": 0.9961,
+      "step": 2401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947833598238135e-06,
+      "loss": 0.9492,
+      "step": 2402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99477488832175e-06,
+      "loss": 0.6167,
+      "step": 2403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947664099646858e-06,
+      "loss": 0.9932,
+      "step": 2404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947579247526783e-06,
+      "loss": 0.9785,
+      "step": 2405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947494326857866e-06,
+      "loss": 1.0459,
+      "step": 2406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947409337640693e-06,
+      "loss": 0.9824,
+      "step": 2407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947324279875842e-06,
+      "loss": 0.957,
+      "step": 2408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994723915356391e-06,
+      "loss": 0.9336,
+      "step": 2409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947153958705476e-06,
+      "loss": 1.2168,
+      "step": 2410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947068695301126e-06,
+      "loss": 1.0107,
+      "step": 2411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994698336335145e-06,
+      "loss": 0.9209,
+      "step": 2412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994689796285704e-06,
+      "loss": 1.1143,
+      "step": 2413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994681249381848e-06,
+      "loss": 1.2363,
+      "step": 2414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946726956236364e-06,
+      "loss": 0.7778,
+      "step": 2415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946641350111272e-06,
+      "loss": 1.0508,
+      "step": 2416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99465556754438e-06,
+      "loss": 0.9121,
+      "step": 2417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994646993223454e-06,
+      "loss": 0.876,
+      "step": 2418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994638412048408e-06,
+      "loss": 1.0195,
+      "step": 2419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994629824019301e-06,
+      "loss": 1.0898,
+      "step": 2420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994621229136192e-06,
+      "loss": 1.0039,
+      "step": 2421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946126273991415e-06,
+      "loss": 0.5532,
+      "step": 2422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994604018808207e-06,
+      "loss": 1.0195,
+      "step": 2423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945954033634494e-06,
+      "loss": 0.9448,
+      "step": 2424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994586781064927e-06,
+      "loss": 0.707,
+      "step": 2425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945781519127e-06,
+      "loss": 0.9785,
+      "step": 2426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945695159068267e-06,
+      "loss": 1.0322,
+      "step": 2427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994560873047368e-06,
+      "loss": 0.9736,
+      "step": 2428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945522233343827e-06,
+      "loss": 1.0117,
+      "step": 2429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945435667679302e-06,
+      "loss": 0.9004,
+      "step": 2430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945349033480706e-06,
+      "loss": 1.043,
+      "step": 2431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994526233074863e-06,
+      "loss": 0.9102,
+      "step": 2432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945175559483685e-06,
+      "loss": 0.9858,
+      "step": 2433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994508871968646e-06,
+      "loss": 1.1143,
+      "step": 2434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945001811357547e-06,
+      "loss": 1.0098,
+      "step": 2435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944914834497557e-06,
+      "loss": 0.8867,
+      "step": 2436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994482778910708e-06,
+      "loss": 1.0176,
+      "step": 2437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994474067518672e-06,
+      "loss": 1.0547,
+      "step": 2438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944653492737074e-06,
+      "loss": 1.25,
+      "step": 2439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944566241758753e-06,
+      "loss": 0.915,
+      "step": 2440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944478922252345e-06,
+      "loss": 0.8887,
+      "step": 2441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994439153421846e-06,
+      "loss": 1.0244,
+      "step": 2442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99443040776577e-06,
+      "loss": 0.8955,
+      "step": 2443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944216552570663e-06,
+      "loss": 0.9277,
+      "step": 2444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994412895895795e-06,
+      "loss": 0.874,
+      "step": 2445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944041296820178e-06,
+      "loss": 1.0215,
+      "step": 2446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994395356615794e-06,
+      "loss": 1.0742,
+      "step": 2447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994386576697184e-06,
+      "loss": 0.8887,
+      "step": 2448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943777899262484e-06,
+      "loss": 0.8193,
+      "step": 2449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994368996303048e-06,
+      "loss": 0.9482,
+      "step": 2450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943601958276436e-06,
+      "loss": 0.7939,
+      "step": 2451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943513885000956e-06,
+      "loss": 0.9443,
+      "step": 2452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994342574320465e-06,
+      "loss": 0.8525,
+      "step": 2453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943337532888116e-06,
+      "loss": 1.1572,
+      "step": 2454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994324925405197e-06,
+      "loss": 1.085,
+      "step": 2455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994316090669682e-06,
+      "loss": 0.8604,
+      "step": 2456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994307249082327e-06,
+      "loss": 0.8896,
+      "step": 2457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942984006431934e-06,
+      "loss": 0.7378,
+      "step": 2458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994289545352342e-06,
+      "loss": 1.0439,
+      "step": 2459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994280683209834e-06,
+      "loss": 0.9248,
+      "step": 2460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99427181421573e-06,
+      "loss": 0.8896,
+      "step": 2461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942629383700917e-06,
+      "loss": 1.2637,
+      "step": 2462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942540556729795e-06,
+      "loss": 1.0723,
+      "step": 2463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994245166124456e-06,
+      "loss": 1.0732,
+      "step": 2464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942362697245804e-06,
+      "loss": 0.7988,
+      "step": 2465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994227366473416e-06,
+      "loss": 0.9678,
+      "step": 2466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994218456371023e-06,
+      "loss": 0.895,
+      "step": 2467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994209539417463e-06,
+      "loss": 0.9727,
+      "step": 2468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942006156127975e-06,
+      "loss": 0.998,
+      "step": 2469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941916849570886e-06,
+      "loss": 1.1523,
+      "step": 2470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941827474503967e-06,
+      "loss": 0.9863,
+      "step": 2471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941738030927843e-06,
+      "loss": 1.0078,
+      "step": 2472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941648518843125e-06,
+      "loss": 1.1377,
+      "step": 2473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941558938250434e-06,
+      "loss": 1.0957,
+      "step": 2474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941469289150383e-06,
+      "loss": 1.0459,
+      "step": 2475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941379571543595e-06,
+      "loss": 0.9922,
+      "step": 2476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941289785430684e-06,
+      "loss": 1.0039,
+      "step": 2477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941199930812267e-06,
+      "loss": 0.9248,
+      "step": 2478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994111000768897e-06,
+      "loss": 1.1348,
+      "step": 2479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994102001606141e-06,
+      "loss": 0.9609,
+      "step": 2480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940929955930203e-06,
+      "loss": 0.8936,
+      "step": 2481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940839827295976e-06,
+      "loss": 0.8867,
+      "step": 2482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994074963015934e-06,
+      "loss": 0.9521,
+      "step": 2483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994065936452093e-06,
+      "loss": 1.0039,
+      "step": 2484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940569030381357e-06,
+      "loss": 0.8413,
+      "step": 2485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940478627741247e-06,
+      "loss": 0.8691,
+      "step": 2486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940388156601226e-06,
+      "loss": 0.9961,
+      "step": 2487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940297616961918e-06,
+      "loss": 0.9375,
+      "step": 2488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994020700882394e-06,
+      "loss": 1.0498,
+      "step": 2489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994011633218792e-06,
+      "loss": 1.123,
+      "step": 2490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940025587054485e-06,
+      "loss": 0.8787,
+      "step": 2491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993993477342426e-06,
+      "loss": 0.6821,
+      "step": 2492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939843891297865e-06,
+      "loss": 0.8354,
+      "step": 2493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939752940675935e-06,
+      "loss": 0.918,
+      "step": 2494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993966192155909e-06,
+      "loss": 0.9932,
+      "step": 2495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993957083394796e-06,
+      "loss": 0.8574,
+      "step": 2496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939479677843175e-06,
+      "loss": 0.833,
+      "step": 2497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939388453245355e-06,
+      "loss": 0.9707,
+      "step": 2498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993929716015514e-06,
+      "loss": 1.0742,
+      "step": 2499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993920579857315e-06,
+      "loss": 0.854,
+      "step": 2500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939114368500014e-06,
+      "loss": 0.8633,
+      "step": 2501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993902286993637e-06,
+      "loss": 1.1475,
+      "step": 2502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938931302882846e-06,
+      "loss": 0.9365,
+      "step": 2503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993883966734007e-06,
+      "loss": 1.1084,
+      "step": 2504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938747963308677e-06,
+      "loss": 1.1514,
+      "step": 2505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993865619078929e-06,
+      "loss": 0.8496,
+      "step": 2506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938564349782555e-06,
+      "loss": 0.9902,
+      "step": 2507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938472440289094e-06,
+      "loss": 1.0654,
+      "step": 2508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938380462309546e-06,
+      "loss": 0.9268,
+      "step": 2509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993828841584454e-06,
+      "loss": 0.7979,
+      "step": 2510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993819630089472e-06,
+      "loss": 0.9834,
+      "step": 2511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993810411746071e-06,
+      "loss": 0.8584,
+      "step": 2512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993801186554315e-06,
+      "loss": 0.917,
+      "step": 2513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937919545142673e-06,
+      "loss": 1.0713,
+      "step": 2514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993782715625992e-06,
+      "loss": 0.9795,
+      "step": 2515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937734698895524e-06,
+      "loss": 1.0176,
+      "step": 2516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937642173050123e-06,
+      "loss": 1.0273,
+      "step": 2517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937549578724357e-06,
+      "loss": 0.8877,
+      "step": 2518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937456915918858e-06,
+      "loss": 0.9473,
+      "step": 2519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937364184634272e-06,
+      "loss": 0.79,
+      "step": 2520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993727138487123e-06,
+      "loss": 1.0098,
+      "step": 2521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993717851663038e-06,
+      "loss": 0.8105,
+      "step": 2522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937085579912357e-06,
+      "loss": 1.1084,
+      "step": 2523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99369925747178e-06,
+      "loss": 1.0166,
+      "step": 2524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936899501047354e-06,
+      "loss": 0.9844,
+      "step": 2525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993680635890166e-06,
+      "loss": 0.9639,
+      "step": 2526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936713148281356e-06,
+      "loss": 1.0537,
+      "step": 2527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936619869187085e-06,
+      "loss": 0.9824,
+      "step": 2528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993652652161949e-06,
+      "loss": 0.9834,
+      "step": 2529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993643310557922e-06,
+      "loss": 0.8809,
+      "step": 2530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936339621066914e-06,
+      "loss": 0.9199,
+      "step": 2531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993624606808322e-06,
+      "loss": 0.9561,
+      "step": 2532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993615244662877e-06,
+      "loss": 0.9805,
+      "step": 2533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936058756704224e-06,
+      "loss": 0.9199,
+      "step": 2534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9935964998310223e-06,
+      "loss": 1.0049,
+      "step": 2535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993587117144741e-06,
+      "loss": 0.6753,
+      "step": 2536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935777276116434e-06,
+      "loss": 1.0479,
+      "step": 2537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935683312317942e-06,
+      "loss": 1.125,
+      "step": 2538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935589280052583e-06,
+      "loss": 0.874,
+      "step": 2539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935495179321e-06,
+      "loss": 1.0654,
+      "step": 2540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993540101012385e-06,
+      "loss": 0.874,
+      "step": 2541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935306772461773e-06,
+      "loss": 0.9121,
+      "step": 2542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993521246633542e-06,
+      "loss": 0.8154,
+      "step": 2543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935118091745443e-06,
+      "loss": 0.9971,
+      "step": 2544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935023648692495e-06,
+      "loss": 0.7935,
+      "step": 2545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934929137177224e-06,
+      "loss": 1.001,
+      "step": 2546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934834557200283e-06,
+      "loss": 0.9414,
+      "step": 2547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993473990876232e-06,
+      "loss": 0.959,
+      "step": 2548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934645191863987e-06,
+      "loss": 0.8457,
+      "step": 2549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934550406505937e-06,
+      "loss": 0.9072,
+      "step": 2550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993445555268883e-06,
+      "loss": 1.0381,
+      "step": 2551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934360630413313e-06,
+      "loss": 0.9355,
+      "step": 2552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934265639680042e-06,
+      "loss": 1.0068,
+      "step": 2553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934170580489675e-06,
+      "loss": 1.0029,
+      "step": 2554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993407545284286e-06,
+      "loss": 0.9121,
+      "step": 2555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933980256740254e-06,
+      "loss": 0.8259,
+      "step": 2556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933884992182517e-06,
+      "loss": 0.917,
+      "step": 2557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933789659170306e-06,
+      "loss": 1.0059,
+      "step": 2558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993369425770427e-06,
+      "loss": 1.2354,
+      "step": 2559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933598787785073e-06,
+      "loss": 0.9912,
+      "step": 2560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993350324941338e-06,
+      "loss": 0.8223,
+      "step": 2561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993340764258983e-06,
+      "loss": 0.9639,
+      "step": 2562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933311967315096e-06,
+      "loss": 0.9717,
+      "step": 2563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993321622358984e-06,
+      "loss": 0.9365,
+      "step": 2564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993312041141471e-06,
+      "loss": 0.77,
+      "step": 2565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933024530790374e-06,
+      "loss": 1.1982,
+      "step": 2566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993292858171749e-06,
+      "loss": 0.9932,
+      "step": 2567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993283256419672e-06,
+      "loss": 0.998,
+      "step": 2568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993273647822873e-06,
+      "loss": 1.0049,
+      "step": 2569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932640323814173e-06,
+      "loss": 1.1357,
+      "step": 2570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932544100953717e-06,
+      "loss": 1.0,
+      "step": 2571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932447809648028e-06,
+      "loss": 1.1094,
+      "step": 2572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932351449897765e-06,
+      "loss": 1.1562,
+      "step": 2573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993225502170359e-06,
+      "loss": 0.6772,
+      "step": 2574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932158525066173e-06,
+      "loss": 1.085,
+      "step": 2575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932061959986175e-06,
+      "loss": 1.2236,
+      "step": 2576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931965326464263e-06,
+      "loss": 0.8228,
+      "step": 2577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931868624501106e-06,
+      "loss": 1.0332,
+      "step": 2578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931771854097365e-06,
+      "loss": 1.0264,
+      "step": 2579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993167501525371e-06,
+      "loss": 0.9141,
+      "step": 2580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931578107970808e-06,
+      "loss": 1.0713,
+      "step": 2581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931481132249325e-06,
+      "loss": 0.999,
+      "step": 2582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931384088089927e-06,
+      "loss": 0.9131,
+      "step": 2583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931286975493295e-06,
+      "loss": 1.0488,
+      "step": 2584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993118979446008e-06,
+      "loss": 1.0127,
+      "step": 2585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931092544990967e-06,
+      "loss": 0.9746,
+      "step": 2586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993099522708662e-06,
+      "loss": 0.9346,
+      "step": 2587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993089784074771e-06,
+      "loss": 1.0518,
+      "step": 2588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930800385974905e-06,
+      "loss": 0.9082,
+      "step": 2589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993070286276888e-06,
+      "loss": 0.874,
+      "step": 2590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930605271130307e-06,
+      "loss": 0.791,
+      "step": 2591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993050761105986e-06,
+      "loss": 0.8486,
+      "step": 2592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930409882558207e-06,
+      "loss": 1.2598,
+      "step": 2593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930312085626026e-06,
+      "loss": 0.9111,
+      "step": 2594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993021422026399e-06,
+      "loss": 0.6909,
+      "step": 2595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930116286472772e-06,
+      "loss": 0.9863,
+      "step": 2596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930018284253046e-06,
+      "loss": 0.9404,
+      "step": 2597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929920213605493e-06,
+      "loss": 0.9746,
+      "step": 2598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992982207453078e-06,
+      "loss": 1.0322,
+      "step": 2599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929723867029592e-06,
+      "loss": 1.0596,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99296255911026e-06,
+      "loss": 1.0146,
+      "step": 2601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929527246750487e-06,
+      "loss": 0.9619,
+      "step": 2602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992942883397392e-06,
+      "loss": 1.0068,
+      "step": 2603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992933035277359e-06,
+      "loss": 1.0747,
+      "step": 2604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992923180315017e-06,
+      "loss": 1.0303,
+      "step": 2605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929133185104333e-06,
+      "loss": 0.8574,
+      "step": 2606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929034498636764e-06,
+      "loss": 1.0771,
+      "step": 2607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992893574374815e-06,
+      "loss": 1.0488,
+      "step": 2608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992883692043916e-06,
+      "loss": 0.8262,
+      "step": 2609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992873802871048e-06,
+      "loss": 0.8408,
+      "step": 2610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928639068562793e-06,
+      "loss": 0.9258,
+      "step": 2611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928540039996777e-06,
+      "loss": 0.6797,
+      "step": 2612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992844094301312e-06,
+      "loss": 0.7793,
+      "step": 2613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928341777612497e-06,
+      "loss": 1.0381,
+      "step": 2614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928242543795596e-06,
+      "loss": 1.0342,
+      "step": 2615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928143241563103e-06,
+      "loss": 1.0,
+      "step": 2616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99280438709157e-06,
+      "loss": 0.9092,
+      "step": 2617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992794443185407e-06,
+      "loss": 0.9019,
+      "step": 2618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99278449243789e-06,
+      "loss": 0.8799,
+      "step": 2619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992774534849088e-06,
+      "loss": 0.998,
+      "step": 2620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992764570419069e-06,
+      "loss": 1.001,
+      "step": 2621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927545991479016e-06,
+      "loss": 0.9893,
+      "step": 2622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927446210356546e-06,
+      "loss": 1.0596,
+      "step": 2623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927346360823975e-06,
+      "loss": 0.9316,
+      "step": 2624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992724644288198e-06,
+      "loss": 1.0391,
+      "step": 2625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927146456531256e-06,
+      "loss": 0.752,
+      "step": 2626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992704640177249e-06,
+      "loss": 0.9053,
+      "step": 2627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926946278606372e-06,
+      "loss": 0.8521,
+      "step": 2628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926846087033597e-06,
+      "loss": 0.5986,
+      "step": 2629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926745827054848e-06,
+      "loss": 0.8047,
+      "step": 2630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992664549867082e-06,
+      "loss": 0.5306,
+      "step": 2631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99265451018822e-06,
+      "loss": 0.918,
+      "step": 2632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926444636689682e-06,
+      "loss": 0.9893,
+      "step": 2633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992634410309396e-06,
+      "loss": 1.1016,
+      "step": 2634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992624350109573e-06,
+      "loss": 0.8965,
+      "step": 2635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926142830695677e-06,
+      "loss": 1.0635,
+      "step": 2636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926042091894496e-06,
+      "loss": 0.9795,
+      "step": 2637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992594128469289e-06,
+      "loss": 0.9688,
+      "step": 2638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925840409091542e-06,
+      "loss": 0.9688,
+      "step": 2639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925739465091154e-06,
+      "loss": 0.5957,
+      "step": 2640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925638452692428e-06,
+      "loss": 1.0,
+      "step": 2641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925537371896045e-06,
+      "loss": 1.0049,
+      "step": 2642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992543622270271e-06,
+      "loss": 0.9336,
+      "step": 2643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992533500511312e-06,
+      "loss": 1.0234,
+      "step": 2644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925233719127967e-06,
+      "loss": 1.1602,
+      "step": 2645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992513236474796e-06,
+      "loss": 1.0117,
+      "step": 2646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992503094197379e-06,
+      "loss": 0.8652,
+      "step": 2647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992492945080615e-06,
+      "loss": 0.8965,
+      "step": 2648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992482789124576e-06,
+      "loss": 1.082,
+      "step": 2649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924726263293294e-06,
+      "loss": 0.8281,
+      "step": 2650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924624566949467e-06,
+      "loss": 1.0605,
+      "step": 2651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992452280221498e-06,
+      "loss": 1.0293,
+      "step": 2652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924420969090527e-06,
+      "loss": 0.8633,
+      "step": 2653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924319067576817e-06,
+      "loss": 0.7598,
+      "step": 2654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924217097674547e-06,
+      "loss": 0.7446,
+      "step": 2655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924115059384425e-06,
+      "loss": 1.1201,
+      "step": 2656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992401295270715e-06,
+      "loss": 0.8218,
+      "step": 2657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923910777643425e-06,
+      "loss": 0.7305,
+      "step": 2658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923808534193957e-06,
+      "loss": 0.8691,
+      "step": 2659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992370622235945e-06,
+      "loss": 0.7676,
+      "step": 2660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992360384214061e-06,
+      "loss": 1.0156,
+      "step": 2661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992350139353814e-06,
+      "loss": 0.8999,
+      "step": 2662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923398876552748e-06,
+      "loss": 0.8232,
+      "step": 2663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923296291185132e-06,
+      "loss": 0.9248,
+      "step": 2664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923193637436013e-06,
+      "loss": 0.6572,
+      "step": 2665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923090915306094e-06,
+      "loss": 1.1377,
+      "step": 2666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922988124796077e-06,
+      "loss": 1.0811,
+      "step": 2667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992288526590667e-06,
+      "loss": 0.8804,
+      "step": 2668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922782338638594e-06,
+      "loss": 0.9697,
+      "step": 2669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922679342992545e-06,
+      "loss": 1.001,
+      "step": 2670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922576278969237e-06,
+      "loss": 0.9785,
+      "step": 2671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922473146569385e-06,
+      "loss": 0.9932,
+      "step": 2672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992236994579369e-06,
+      "loss": 1.1768,
+      "step": 2673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922266676642873e-06,
+      "loss": 0.7588,
+      "step": 2674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992216333911764e-06,
+      "loss": 1.083,
+      "step": 2675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922059933218706e-06,
+      "loss": 0.6689,
+      "step": 2676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992195645894678e-06,
+      "loss": 0.8896,
+      "step": 2677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921852916302576e-06,
+      "loss": 1.1504,
+      "step": 2678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992174930528681e-06,
+      "loss": 0.9834,
+      "step": 2679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921645625900196e-06,
+      "loss": 0.8516,
+      "step": 2680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921541878143448e-06,
+      "loss": 0.6638,
+      "step": 2681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992143806201728e-06,
+      "loss": 0.8564,
+      "step": 2682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99213341775224e-06,
+      "loss": 0.8521,
+      "step": 2683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992123022465954e-06,
+      "loss": 1.1074,
+      "step": 2684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921126203429403e-06,
+      "loss": 1.1094,
+      "step": 2685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921022113832714e-06,
+      "loss": 0.8513,
+      "step": 2686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920917955870183e-06,
+      "loss": 1.1465,
+      "step": 2687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992081372954253e-06,
+      "loss": 0.8242,
+      "step": 2688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920709434850476e-06,
+      "loss": 1.1416,
+      "step": 2689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920605071794744e-06,
+      "loss": 1.2217,
+      "step": 2690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992050064037604e-06,
+      "loss": 0.9141,
+      "step": 2691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920396140595096e-06,
+      "loss": 0.9512,
+      "step": 2692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920291572452626e-06,
+      "loss": 0.9072,
+      "step": 2693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992018693594935e-06,
+      "loss": 1.0381,
+      "step": 2694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920082231085994e-06,
+      "loss": 0.8955,
+      "step": 2695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919977457863273e-06,
+      "loss": 0.8379,
+      "step": 2696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919872616281914e-06,
+      "loss": 0.7734,
+      "step": 2697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919767706342637e-06,
+      "loss": 0.7817,
+      "step": 2698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919662728046167e-06,
+      "loss": 0.8916,
+      "step": 2699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919557681393227e-06,
+      "loss": 1.1318,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991945256638454e-06,
+      "loss": 0.9844,
+      "step": 2701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919347383020824e-06,
+      "loss": 0.7778,
+      "step": 2702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919242131302817e-06,
+      "loss": 1.0879,
+      "step": 2703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919136811231238e-06,
+      "loss": 0.78,
+      "step": 2704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991903142280681e-06,
+      "loss": 1.041,
+      "step": 2705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918925966030262e-06,
+      "loss": 0.8848,
+      "step": 2706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918820440902323e-06,
+      "loss": 0.9443,
+      "step": 2707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918714847423713e-06,
+      "loss": 0.9316,
+      "step": 2708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918609185595163e-06,
+      "loss": 1.0205,
+      "step": 2709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991850345541741e-06,
+      "loss": 0.9854,
+      "step": 2710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918397656891166e-06,
+      "loss": 0.9766,
+      "step": 2711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918291790017176e-06,
+      "loss": 0.8838,
+      "step": 2712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918185854796156e-06,
+      "loss": 1.0967,
+      "step": 2713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918079851228848e-06,
+      "loss": 1.0889,
+      "step": 2714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991797377931597e-06,
+      "loss": 1.0898,
+      "step": 2715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917867639058264e-06,
+      "loss": 0.9473,
+      "step": 2716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991776143045646e-06,
+      "loss": 0.9785,
+      "step": 2717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917655153511283e-06,
+      "loss": 0.7195,
+      "step": 2718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991754880822347e-06,
+      "loss": 0.9473,
+      "step": 2719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917442394593756e-06,
+      "loss": 1.1348,
+      "step": 2720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917335912622866e-06,
+      "loss": 1.1631,
+      "step": 2721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917229362311546e-06,
+      "loss": 0.8623,
+      "step": 2722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991712274366052e-06,
+      "loss": 0.7261,
+      "step": 2723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917016056670526e-06,
+      "loss": 0.959,
+      "step": 2724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916909301342296e-06,
+      "loss": 0.8516,
+      "step": 2725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916802477676574e-06,
+      "loss": 0.9932,
+      "step": 2726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991669558567409e-06,
+      "loss": 1.0205,
+      "step": 2727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916588625335583e-06,
+      "loss": 1.0449,
+      "step": 2728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991648159666179e-06,
+      "loss": 1.0859,
+      "step": 2729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916374499653443e-06,
+      "loss": 0.8633,
+      "step": 2730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916267334311288e-06,
+      "loss": 1.0225,
+      "step": 2731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991616010063606e-06,
+      "loss": 0.9414,
+      "step": 2732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916052798628495e-06,
+      "loss": 0.9434,
+      "step": 2733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915945428289336e-06,
+      "loss": 0.8975,
+      "step": 2734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991583798961932e-06,
+      "loss": 1.0264,
+      "step": 2735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915730482619197e-06,
+      "loss": 0.8662,
+      "step": 2736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991562290728969e-06,
+      "loss": 0.9883,
+      "step": 2737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991551526363156e-06,
+      "loss": 0.9414,
+      "step": 2738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915407551645536e-06,
+      "loss": 1.0029,
+      "step": 2739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915299771332363e-06,
+      "loss": 0.9795,
+      "step": 2740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915191922692786e-06,
+      "loss": 0.8477,
+      "step": 2741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915084005727547e-06,
+      "loss": 0.96,
+      "step": 2742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914976020437386e-06,
+      "loss": 0.7598,
+      "step": 2743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991486796682305e-06,
+      "loss": 0.9414,
+      "step": 2744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991475984488529e-06,
+      "loss": 0.915,
+      "step": 2745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991465165462484e-06,
+      "loss": 0.8906,
+      "step": 2746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991454339604245e-06,
+      "loss": 1.1064,
+      "step": 2747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914435069138865e-06,
+      "loss": 1.04,
+      "step": 2748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914326673914835e-06,
+      "loss": 0.5889,
+      "step": 2749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914218210371106e-06,
+      "loss": 0.9678,
+      "step": 2750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914109678508427e-06,
+      "loss": 0.9775,
+      "step": 2751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991400107832754e-06,
+      "loss": 0.8169,
+      "step": 2752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913892409829195e-06,
+      "loss": 1.0615,
+      "step": 2753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913783673014146e-06,
+      "loss": 1.1777,
+      "step": 2754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913674867883138e-06,
+      "loss": 0.9404,
+      "step": 2755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991356599443692e-06,
+      "loss": 1.2021,
+      "step": 2756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913457052676247e-06,
+      "loss": 0.9014,
+      "step": 2757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913348042601864e-06,
+      "loss": 1.0156,
+      "step": 2758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913238964214524e-06,
+      "loss": 0.6641,
+      "step": 2759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913129817514985e-06,
+      "loss": 0.8047,
+      "step": 2760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913020602503988e-06,
+      "loss": 0.896,
+      "step": 2761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912911319182295e-06,
+      "loss": 0.9785,
+      "step": 2762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991280196755065e-06,
+      "loss": 0.6899,
+      "step": 2763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912692547609816e-06,
+      "loss": 0.9307,
+      "step": 2764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912583059360537e-06,
+      "loss": 0.9072,
+      "step": 2765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991247350280358e-06,
+      "loss": 1.1201,
+      "step": 2766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991236387793969e-06,
+      "loss": 1.0303,
+      "step": 2767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912254184769627e-06,
+      "loss": 0.7358,
+      "step": 2768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912144423294147e-06,
+      "loss": 0.8237,
+      "step": 2769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912034593514e-06,
+      "loss": 0.9199,
+      "step": 2770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911924695429953e-06,
+      "loss": 0.7891,
+      "step": 2771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911814729042757e-06,
+      "loss": 1.0371,
+      "step": 2772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911704694353165e-06,
+      "loss": 0.709,
+      "step": 2773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991159459136195e-06,
+      "loss": 1.1152,
+      "step": 2774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911484420069855e-06,
+      "loss": 0.8335,
+      "step": 2775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991137418047765e-06,
+      "loss": 0.8057,
+      "step": 2776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991126387258609e-06,
+      "loss": 0.9956,
+      "step": 2777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911153496395933e-06,
+      "loss": 0.8457,
+      "step": 2778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911043051907943e-06,
+      "loss": 0.7085,
+      "step": 2779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910932539122882e-06,
+      "loss": 1.1826,
+      "step": 2780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910821958041512e-06,
+      "loss": 1.2158,
+      "step": 2781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910711308664588e-06,
+      "loss": 0.8164,
+      "step": 2782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910600590992883e-06,
+      "loss": 0.9238,
+      "step": 2783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991048980502715e-06,
+      "loss": 0.4988,
+      "step": 2784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910378950768157e-06,
+      "loss": 0.6453,
+      "step": 2785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991026802821667e-06,
+      "loss": 1.002,
+      "step": 2786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991015703737345e-06,
+      "loss": 1.0254,
+      "step": 2787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910045978239264e-06,
+      "loss": 0.8467,
+      "step": 2788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909934850814876e-06,
+      "loss": 0.7847,
+      "step": 2789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909823655101052e-06,
+      "loss": 0.9619,
+      "step": 2790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990971239109856e-06,
+      "loss": 1.1201,
+      "step": 2791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990960105880816e-06,
+      "loss": 1.1172,
+      "step": 2792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990948965823063e-06,
+      "loss": 0.9219,
+      "step": 2793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909378189366728e-06,
+      "loss": 1.1133,
+      "step": 2794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990926665221723e-06,
+      "loss": 0.7793,
+      "step": 2795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909155046782894e-06,
+      "loss": 0.9932,
+      "step": 2796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909043373064504e-06,
+      "loss": 0.8877,
+      "step": 2797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908931631062816e-06,
+      "loss": 0.9014,
+      "step": 2798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908819820778605e-06,
+      "loss": 0.9912,
+      "step": 2799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908707942212645e-06,
+      "loss": 1.1729,
+      "step": 2800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99085959953657e-06,
+      "loss": 0.854,
+      "step": 2801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908483980238546e-06,
+      "loss": 1.1289,
+      "step": 2802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908371896831958e-06,
+      "loss": 1.0498,
+      "step": 2803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99082597451467e-06,
+      "loss": 0.9443,
+      "step": 2804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990814752518355e-06,
+      "loss": 0.9658,
+      "step": 2805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990803523694328e-06,
+      "loss": 1.1309,
+      "step": 2806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907922880426668e-06,
+      "loss": 1.0303,
+      "step": 2807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907810455634482e-06,
+      "loss": 0.9609,
+      "step": 2808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99076979625675e-06,
+      "loss": 1.0527,
+      "step": 2809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907585401226495e-06,
+      "loss": 0.8833,
+      "step": 2810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907472771612244e-06,
+      "loss": 0.8955,
+      "step": 2811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907360073725527e-06,
+      "loss": 1.0625,
+      "step": 2812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907247307567113e-06,
+      "loss": 1.1543,
+      "step": 2813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907134473137783e-06,
+      "loss": 1.0146,
+      "step": 2814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907021570438318e-06,
+      "loss": 0.8799,
+      "step": 2815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906908599469488e-06,
+      "loss": 1.0352,
+      "step": 2816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990679556023208e-06,
+      "loss": 0.9443,
+      "step": 2817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906682452726868e-06,
+      "loss": 1.0381,
+      "step": 2818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990656927695463e-06,
+      "loss": 0.9854,
+      "step": 2819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906456032916152e-06,
+      "loss": 0.9873,
+      "step": 2820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906342720612204e-06,
+      "loss": 0.958,
+      "step": 2821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990622934004358e-06,
+      "loss": 1.0127,
+      "step": 2822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906115891211054e-06,
+      "loss": 1.1191,
+      "step": 2823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906002374115407e-06,
+      "loss": 0.9043,
+      "step": 2824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905888788757424e-06,
+      "loss": 0.9941,
+      "step": 2825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905775135137887e-06,
+      "loss": 1.0859,
+      "step": 2826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905661413257574e-06,
+      "loss": 0.6499,
+      "step": 2827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905547623117277e-06,
+      "loss": 1.0049,
+      "step": 2828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990543376471778e-06,
+      "loss": 1.1191,
+      "step": 2829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905319838059856e-06,
+      "loss": 1.0547,
+      "step": 2830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905205843144304e-06,
+      "loss": 1.2803,
+      "step": 2831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905091779971903e-06,
+      "loss": 0.9375,
+      "step": 2832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990497764854344e-06,
+      "loss": 0.8008,
+      "step": 2833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99048634488597e-06,
+      "loss": 1.1465,
+      "step": 2834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990474918092147e-06,
+      "loss": 1.0498,
+      "step": 2835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904634844729543e-06,
+      "loss": 0.7744,
+      "step": 2836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99045204402847e-06,
+      "loss": 0.8257,
+      "step": 2837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990440596758773e-06,
+      "loss": 1.0859,
+      "step": 2838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990429142663942e-06,
+      "loss": 0.895,
+      "step": 2839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990417681744057e-06,
+      "loss": 0.9785,
+      "step": 2840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990406213999196e-06,
+      "loss": 1.0342,
+      "step": 2841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903947394294386e-06,
+      "loss": 0.9424,
+      "step": 2842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990383258034863e-06,
+      "loss": 0.917,
+      "step": 2843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903717698155496e-06,
+      "loss": 0.8408,
+      "step": 2844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903602747715764e-06,
+      "loss": 1.0234,
+      "step": 2845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903487729030232e-06,
+      "loss": 0.9824,
+      "step": 2846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903372642099688e-06,
+      "loss": 1.1123,
+      "step": 2847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990325748692493e-06,
+      "loss": 0.8975,
+      "step": 2848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903142263506754e-06,
+      "loss": 0.9746,
+      "step": 2849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903026971845945e-06,
+      "loss": 0.7715,
+      "step": 2850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902911611943304e-06,
+      "loss": 1.1191,
+      "step": 2851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902796183799622e-06,
+      "loss": 0.6689,
+      "step": 2852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99026806874157e-06,
+      "loss": 0.9463,
+      "step": 2853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902565122792333e-06,
+      "loss": 0.9277,
+      "step": 2854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990244948993031e-06,
+      "loss": 0.9512,
+      "step": 2855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902333788830435e-06,
+      "loss": 1.0078,
+      "step": 2856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99022180194935e-06,
+      "loss": 1.0889,
+      "step": 2857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990210218192031e-06,
+      "loss": 0.9307,
+      "step": 2858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901986276111654e-06,
+      "loss": 1.0117,
+      "step": 2859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990187030206834e-06,
+      "loss": 1.0576,
+      "step": 2860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901754259791164e-06,
+      "loss": 0.8926,
+      "step": 2861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990163814928092e-06,
+      "loss": 1.0303,
+      "step": 2862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990152197053842e-06,
+      "loss": 0.8892,
+      "step": 2863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990140572356445e-06,
+      "loss": 0.9297,
+      "step": 2864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990128940835982e-06,
+      "loss": 1.0732,
+      "step": 2865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901173024925332e-06,
+      "loss": 0.8799,
+      "step": 2866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901056573261787e-06,
+      "loss": 1.0234,
+      "step": 2867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990094005336998e-06,
+      "loss": 1.0488,
+      "step": 2868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900823465250727e-06,
+      "loss": 0.9658,
+      "step": 2869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990070680890482e-06,
+      "loss": 0.8081,
+      "step": 2870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990059008433307e-06,
+      "loss": 0.8076,
+      "step": 2871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990047329153628e-06,
+      "loss": 0.8071,
+      "step": 2872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900356430515252e-06,
+      "loss": 0.9336,
+      "step": 2873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900239501270794e-06,
+      "loss": 1.0605,
+      "step": 2874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900122503803705e-06,
+      "loss": 0.9482,
+      "step": 2875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900005438114804e-06,
+      "loss": 0.9102,
+      "step": 2876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989988830420489e-06,
+      "loss": 0.8223,
+      "step": 2877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989977110207477e-06,
+      "loss": 1.0254,
+      "step": 2878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899653831725253e-06,
+      "loss": 0.7896,
+      "step": 2879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899536493157145e-06,
+      "loss": 1.1357,
+      "step": 2880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899419086371257e-06,
+      "loss": 1.0254,
+      "step": 2881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98993016113684e-06,
+      "loss": 1.0957,
+      "step": 2882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989918406814938e-06,
+      "loss": 0.7412,
+      "step": 2883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899066456715003e-06,
+      "loss": 1.082,
+      "step": 2884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898948777066086e-06,
+      "loss": 0.8096,
+      "step": 2885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989883102920344e-06,
+      "loss": 0.832,
+      "step": 2886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898713213127875e-06,
+      "loss": 0.9775,
+      "step": 2887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898595328840203e-06,
+      "loss": 1.0264,
+      "step": 2888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898477376341234e-06,
+      "loss": 0.9502,
+      "step": 2889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898359355631782e-06,
+      "loss": 0.9912,
+      "step": 2890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898241266712665e-06,
+      "loss": 1.0322,
+      "step": 2891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989812310958469e-06,
+      "loss": 1.0068,
+      "step": 2892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989800488424867e-06,
+      "loss": 0.875,
+      "step": 2893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897886590705426e-06,
+      "loss": 0.7861,
+      "step": 2894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897768228955772e-06,
+      "loss": 0.9414,
+      "step": 2895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897649799000527e-06,
+      "loss": 1.3027,
+      "step": 2896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897531300840494e-06,
+      "loss": 1.0273,
+      "step": 2897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897412734476504e-06,
+      "loss": 0.9131,
+      "step": 2898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897294099909367e-06,
+      "loss": 0.9678,
+      "step": 2899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98971753971399e-06,
+      "loss": 1.0059,
+      "step": 2900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989705662616892e-06,
+      "loss": 0.998,
+      "step": 2901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896937786997253e-06,
+      "loss": 0.9189,
+      "step": 2902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989681887962571e-06,
+      "loss": 0.7532,
+      "step": 2903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896699904055115e-06,
+      "loss": 1.0059,
+      "step": 2904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989658086028629e-06,
+      "loss": 0.9658,
+      "step": 2905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896461748320044e-06,
+      "loss": 0.71,
+      "step": 2906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989634256815721e-06,
+      "loss": 0.9922,
+      "step": 2907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98962233197986e-06,
+      "loss": 0.9512,
+      "step": 2908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896104003245043e-06,
+      "loss": 1.0264,
+      "step": 2909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989598461849736e-06,
+      "loss": 1.1621,
+      "step": 2910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895865165556373e-06,
+      "loss": 1.0566,
+      "step": 2911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98957456444229e-06,
+      "loss": 1.0469,
+      "step": 2912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989562605509777e-06,
+      "loss": 0.875,
+      "step": 2913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989550639758181e-06,
+      "loss": 0.96,
+      "step": 2914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989538667187584e-06,
+      "loss": 0.9727,
+      "step": 2915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989526687798068e-06,
+      "loss": 1.0088,
+      "step": 2916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989514701589717e-06,
+      "loss": 0.8945,
+      "step": 2917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895027085626123e-06,
+      "loss": 1.0049,
+      "step": 2918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894907087168367e-06,
+      "loss": 1.0801,
+      "step": 2919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894787020524735e-06,
+      "loss": 1.1777,
+      "step": 2920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989466688569605e-06,
+      "loss": 0.957,
+      "step": 2921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989454668268314e-06,
+      "loss": 1.0713,
+      "step": 2922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894426411486835e-06,
+      "loss": 0.8789,
+      "step": 2923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989430607210796e-06,
+      "loss": 1.0752,
+      "step": 2924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989418566454735e-06,
+      "loss": 0.7412,
+      "step": 2925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989406518880583e-06,
+      "loss": 1.0684,
+      "step": 2926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893944644884235e-06,
+      "loss": 0.9238,
+      "step": 2927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989382403278339e-06,
+      "loss": 0.8013,
+      "step": 2928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989370335250413e-06,
+      "loss": 0.9658,
+      "step": 2929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893582604047286e-06,
+      "loss": 0.8472,
+      "step": 2930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893461787413687e-06,
+      "loss": 0.7554,
+      "step": 2931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989334090260417e-06,
+      "loss": 1.1611,
+      "step": 2932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989321994961957e-06,
+      "loss": 0.8945,
+      "step": 2933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893098928460708e-06,
+      "loss": 1.0947,
+      "step": 2934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989297783912843e-06,
+      "loss": 1.2236,
+      "step": 2935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989285668162357e-06,
+      "loss": 0.9844,
+      "step": 2936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892735455946957e-06,
+      "loss": 1.002,
+      "step": 2937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989261416209943e-06,
+      "loss": 0.8232,
+      "step": 2938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892492800081823e-06,
+      "loss": 0.9883,
+      "step": 2939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892371369894975e-06,
+      "loss": 1.0098,
+      "step": 2940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892249871539724e-06,
+      "loss": 0.8428,
+      "step": 2941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98921283050169e-06,
+      "loss": 0.9111,
+      "step": 2942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989200667032735e-06,
+      "loss": 0.9883,
+      "step": 2943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98918849674719e-06,
+      "loss": 1.127,
+      "step": 2944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98917631964514e-06,
+      "loss": 1.0273,
+      "step": 2945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891641357266683e-06,
+      "loss": 0.8584,
+      "step": 2946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891519449918596e-06,
+      "loss": 1.292,
+      "step": 2947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989139747440797e-06,
+      "loss": 0.8364,
+      "step": 2948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891275430735646e-06,
+      "loss": 0.959,
+      "step": 2949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891153318902473e-06,
+      "loss": 1.0781,
+      "step": 2950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891031138909286e-06,
+      "loss": 1.0342,
+      "step": 2951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989090889075693e-06,
+      "loss": 0.9619,
+      "step": 2952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890786574446247e-06,
+      "loss": 0.9316,
+      "step": 2953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890664189978074e-06,
+      "loss": 1.0361,
+      "step": 2954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890541737353265e-06,
+      "loss": 0.8838,
+      "step": 2955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890419216572652e-06,
+      "loss": 0.8984,
+      "step": 2956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890296627637092e-06,
+      "loss": 1.1406,
+      "step": 2957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989017397054742e-06,
+      "loss": 0.8945,
+      "step": 2958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890051245304488e-06,
+      "loss": 0.7891,
+      "step": 2959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889928451909133e-06,
+      "loss": 0.9941,
+      "step": 2960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889805590362206e-06,
+      "loss": 0.9668,
+      "step": 2961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988968266066456e-06,
+      "loss": 0.7515,
+      "step": 2962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988955966281703e-06,
+      "loss": 0.9365,
+      "step": 2963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889436596820472e-06,
+      "loss": 0.8535,
+      "step": 2964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988931346267573e-06,
+      "loss": 0.9922,
+      "step": 2965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988919026038366e-06,
+      "loss": 0.96,
+      "step": 2966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98890669899451e-06,
+      "loss": 0.7402,
+      "step": 2967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888943651360906e-06,
+      "loss": 0.8906,
+      "step": 2968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888820244631925e-06,
+      "loss": 0.8296,
+      "step": 2969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888696769759012e-06,
+      "loss": 0.6973,
+      "step": 2970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888573226743014e-06,
+      "loss": 1.0098,
+      "step": 2971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888449615584783e-06,
+      "loss": 0.7231,
+      "step": 2972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988832593628517e-06,
+      "loss": 0.9717,
+      "step": 2973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988820218884503e-06,
+      "loss": 0.875,
+      "step": 2974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888078373265213e-06,
+      "loss": 1.04,
+      "step": 2975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887954489546573e-06,
+      "loss": 0.9883,
+      "step": 2976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887830537689964e-06,
+      "loss": 0.8262,
+      "step": 2977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887706517696237e-06,
+      "loss": 0.9004,
+      "step": 2978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887582429566257e-06,
+      "loss": 0.9844,
+      "step": 2979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887458273300866e-06,
+      "loss": 0.8457,
+      "step": 2980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988733404890093e-06,
+      "loss": 0.8315,
+      "step": 2981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887209756367296e-06,
+      "loss": 0.7615,
+      "step": 2982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887085395700823e-06,
+      "loss": 0.9316,
+      "step": 2983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886960966902373e-06,
+      "loss": 0.7651,
+      "step": 2984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98868364699728e-06,
+      "loss": 0.75,
+      "step": 2985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886711904912965e-06,
+      "loss": 1.1426,
+      "step": 2986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988658727172372e-06,
+      "loss": 0.9058,
+      "step": 2987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886462570405927e-06,
+      "loss": 1.042,
+      "step": 2988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886337800960447e-06,
+      "loss": 0.9932,
+      "step": 2989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988621296338814e-06,
+      "loss": 1.1201,
+      "step": 2990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886088057689865e-06,
+      "loss": 0.9629,
+      "step": 2991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885963083866477e-06,
+      "loss": 0.9619,
+      "step": 2992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988583804191885e-06,
+      "loss": 1.0273,
+      "step": 2993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988571293184784e-06,
+      "loss": 1.0088,
+      "step": 2994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98855877536543e-06,
+      "loss": 0.8584,
+      "step": 2995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98854625073391e-06,
+      "loss": 1.0186,
+      "step": 2996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885337192903105e-06,
+      "loss": 0.9092,
+      "step": 2997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885211810347184e-06,
+      "loss": 0.9131,
+      "step": 2998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885086359672182e-06,
+      "loss": 0.9131,
+      "step": 2999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988496084087898e-06,
+      "loss": 0.918,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988483525396844e-06,
+      "loss": 0.7183,
+      "step": 3001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884709598941427e-06,
+      "loss": 1.1016,
+      "step": 3002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98845838757988e-06,
+      "loss": 0.9834,
+      "step": 3003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884458084541435e-06,
+      "loss": 1.0723,
+      "step": 3004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884332225170195e-06,
+      "loss": 0.832,
+      "step": 3005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988420629768595e-06,
+      "loss": 1.1221,
+      "step": 3006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884080302089554e-06,
+      "loss": 1.0078,
+      "step": 3007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883954238381894e-06,
+      "loss": 1.084,
+      "step": 3008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883828106563828e-06,
+      "loss": 0.9556,
+      "step": 3009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988370190663623e-06,
+      "loss": 1.2598,
+      "step": 3010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883575638599966e-06,
+      "loss": 0.8438,
+      "step": 3011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883449302455907e-06,
+      "loss": 1.0752,
+      "step": 3012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883322898204927e-06,
+      "loss": 1.0166,
+      "step": 3013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988319642584789e-06,
+      "loss": 0.7588,
+      "step": 3014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883069885385673e-06,
+      "loss": 1.0791,
+      "step": 3015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882943276819153e-06,
+      "loss": 0.7661,
+      "step": 3016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988281660014919e-06,
+      "loss": 1.168,
+      "step": 3017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988268985537666e-06,
+      "loss": 0.9922,
+      "step": 3018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988256304250245e-06,
+      "loss": 0.7769,
+      "step": 3019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882436161527413e-06,
+      "loss": 1.0811,
+      "step": 3020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882309212452437e-06,
+      "loss": 1.0225,
+      "step": 3021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882182195278396e-06,
+      "loss": 0.8896,
+      "step": 3022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882055110006163e-06,
+      "loss": 1.1553,
+      "step": 3023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988192795663661e-06,
+      "loss": 0.8662,
+      "step": 3024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988180073517062e-06,
+      "loss": 0.9863,
+      "step": 3025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881673445609065e-06,
+      "loss": 0.9277,
+      "step": 3026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988154608795282e-06,
+      "loss": 0.9092,
+      "step": 3027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881418662202774e-06,
+      "loss": 0.9365,
+      "step": 3028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881291168359794e-06,
+      "loss": 1.1035,
+      "step": 3029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988116360642476e-06,
+      "loss": 1.2432,
+      "step": 3030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881035976398554e-06,
+      "loss": 1.0527,
+      "step": 3031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880908278282055e-06,
+      "loss": 1.0098,
+      "step": 3032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880780512076143e-06,
+      "loss": 0.9434,
+      "step": 3033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880652677781696e-06,
+      "loss": 0.9805,
+      "step": 3034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880524775399597e-06,
+      "loss": 1.0381,
+      "step": 3035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988039680493073e-06,
+      "loss": 1.0117,
+      "step": 3036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880268766375972e-06,
+      "loss": 1.0732,
+      "step": 3037
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880140659736207e-06,
+      "loss": 0.9932,
+      "step": 3038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988001248501232e-06,
+      "loss": 0.5859,
+      "step": 3039
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987988424220519e-06,
+      "loss": 0.9404,
+      "step": 3040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98797559313157e-06,
+      "loss": 0.8506,
+      "step": 3041
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879627552344743e-06,
+      "loss": 0.9756,
+      "step": 3042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879499105293194e-06,
+      "loss": 1.0381,
+      "step": 3043
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879370590161948e-06,
+      "loss": 0.8311,
+      "step": 3044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987924200695188e-06,
+      "loss": 0.958,
+      "step": 3045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987911335566388e-06,
+      "loss": 0.7563,
+      "step": 3046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987898463629884e-06,
+      "loss": 1.0742,
+      "step": 3047
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987885584885764e-06,
+      "loss": 0.7568,
+      "step": 3048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987872699334116e-06,
+      "loss": 0.8818,
+      "step": 3049
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987859806975031e-06,
+      "loss": 0.917,
+      "step": 3050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987846907808596e-06,
+      "loss": 0.8032,
+      "step": 3051
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878340018349007e-06,
+      "loss": 1.2471,
+      "step": 3052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987821089054034e-06,
+      "loss": 0.8525,
+      "step": 3053
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878081694660847e-06,
+      "loss": 0.8027,
+      "step": 3054
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877952430711415e-06,
+      "loss": 0.9883,
+      "step": 3055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987782309869294e-06,
+      "loss": 0.873,
+      "step": 3056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877693698606316e-06,
+      "loss": 0.9307,
+      "step": 3057
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877564230452424e-06,
+      "loss": 0.9541,
+      "step": 3058
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877434694232163e-06,
+      "loss": 1.0898,
+      "step": 3059
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877305089946426e-06,
+      "loss": 0.959,
+      "step": 3060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877175417596103e-06,
+      "loss": 1.0078,
+      "step": 3061
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987704567718209e-06,
+      "loss": 0.8857,
+      "step": 3062
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876915868705286e-06,
+      "loss": 0.8984,
+      "step": 3063
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876785992166575e-06,
+      "loss": 0.6851,
+      "step": 3064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876656047566857e-06,
+      "loss": 0.9609,
+      "step": 3065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987652603490703e-06,
+      "loss": 1.0332,
+      "step": 3066
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987639595418799e-06,
+      "loss": 0.96,
+      "step": 3067
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876265805410623e-06,
+      "loss": 1.0068,
+      "step": 3068
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987613558857584e-06,
+      "loss": 0.8779,
+      "step": 3069
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876005303684532e-06,
+      "loss": 1.0859,
+      "step": 3070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987587495073759e-06,
+      "loss": 0.8457,
+      "step": 3071
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875744529735927e-06,
+      "loss": 1.1035,
+      "step": 3072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987561404068043e-06,
+      "loss": 0.8066,
+      "step": 3073
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875483483572002e-06,
+      "loss": 0.9883,
+      "step": 3074
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875352858411546e-06,
+      "loss": 0.9014,
+      "step": 3075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987522216519996e-06,
+      "loss": 0.9658,
+      "step": 3076
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875091403938137e-06,
+      "loss": 1.0205,
+      "step": 3077
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874960574626985e-06,
+      "loss": 0.7451,
+      "step": 3078
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987482967726741e-06,
+      "loss": 1.0986,
+      "step": 3079
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874698711860306e-06,
+      "loss": 0.9277,
+      "step": 3080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874567678406578e-06,
+      "loss": 0.959,
+      "step": 3081
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987443657690713e-06,
+      "loss": 0.8799,
+      "step": 3082
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987430540736287e-06,
+      "loss": 0.8818,
+      "step": 3083
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987417416977469e-06,
+      "loss": 1.0889,
+      "step": 3084
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874042864143506e-06,
+      "loss": 1.002,
+      "step": 3085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873911490470215e-06,
+      "loss": 0.812,
+      "step": 3086
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873780048755725e-06,
+      "loss": 0.7695,
+      "step": 3087
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873648539000946e-06,
+      "loss": 1.0557,
+      "step": 3088
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873516961206776e-06,
+      "loss": 0.772,
+      "step": 3089
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987338531537413e-06,
+      "loss": 0.8403,
+      "step": 3090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873253601503904e-06,
+      "loss": 0.8594,
+      "step": 3091
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987312181959702e-06,
+      "loss": 0.7607,
+      "step": 3092
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872989969654377e-06,
+      "loss": 1.0889,
+      "step": 3093
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872858051676885e-06,
+      "loss": 0.9922,
+      "step": 3094
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987272606566545e-06,
+      "loss": 1.0107,
+      "step": 3095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872594011620993e-06,
+      "loss": 0.8613,
+      "step": 3096
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987246188954441e-06,
+      "loss": 1.0117,
+      "step": 3097
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987232969943662e-06,
+      "loss": 0.9609,
+      "step": 3098
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872197441298534e-06,
+      "loss": 0.9365,
+      "step": 3099
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872065115131057e-06,
+      "loss": 0.8657,
+      "step": 3100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871932720935103e-06,
+      "loss": 0.9648,
+      "step": 3101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871800258711586e-06,
+      "loss": 0.8789,
+      "step": 3102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871667728461427e-06,
+      "loss": 0.8555,
+      "step": 3103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987153513018552e-06,
+      "loss": 0.9023,
+      "step": 3104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98714024638848e-06,
+      "loss": 1.0361,
+      "step": 3105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871269729560164e-06,
+      "loss": 0.6875,
+      "step": 3106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871136927212538e-06,
+      "loss": 0.9385,
+      "step": 3107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871004056842833e-06,
+      "loss": 0.9941,
+      "step": 3108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987087111845196e-06,
+      "loss": 0.604,
+      "step": 3109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870738112040845e-06,
+      "loss": 1.0596,
+      "step": 3110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870605037610396e-06,
+      "loss": 0.8696,
+      "step": 3111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870471895161533e-06,
+      "loss": 0.9336,
+      "step": 3112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870338684695174e-06,
+      "loss": 0.9424,
+      "step": 3113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870205406212235e-06,
+      "loss": 1.085,
+      "step": 3114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987007205971364e-06,
+      "loss": 0.9434,
+      "step": 3115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98699386452003e-06,
+      "loss": 1.0039,
+      "step": 3116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869805162673137e-06,
+      "loss": 0.8975,
+      "step": 3117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869671612133074e-06,
+      "loss": 0.9746,
+      "step": 3118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869537993581035e-06,
+      "loss": 1.0059,
+      "step": 3119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869404307017925e-06,
+      "loss": 1.1191,
+      "step": 3120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869270552444685e-06,
+      "loss": 1.0371,
+      "step": 3121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986913672986222e-06,
+      "loss": 1.1426,
+      "step": 3122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869002839271464e-06,
+      "loss": 1.0234,
+      "step": 3123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986886888067333e-06,
+      "loss": 0.9551,
+      "step": 3124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986873485406875e-06,
+      "loss": 1.0537,
+      "step": 3125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868600759458644e-06,
+      "loss": 0.8486,
+      "step": 3126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986846659684393e-06,
+      "loss": 0.8633,
+      "step": 3127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868332366225544e-06,
+      "loss": 0.8975,
+      "step": 3128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98681980676044e-06,
+      "loss": 0.8994,
+      "step": 3129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986806370098144e-06,
+      "loss": 0.7173,
+      "step": 3130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867929266357564e-06,
+      "loss": 0.9727,
+      "step": 3131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986779476373372e-06,
+      "loss": 1.0381,
+      "step": 3132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986766019311083e-06,
+      "loss": 1.0117,
+      "step": 3133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867525554489814e-06,
+      "loss": 0.8042,
+      "step": 3134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867390847871607e-06,
+      "loss": 0.957,
+      "step": 3135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867256073257136e-06,
+      "loss": 0.875,
+      "step": 3136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986712123064733e-06,
+      "loss": 1.0723,
+      "step": 3137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986698632004311e-06,
+      "loss": 1.0518,
+      "step": 3138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986685134144542e-06,
+      "loss": 1.0625,
+      "step": 3139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986671629485518e-06,
+      "loss": 0.6519,
+      "step": 3140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866581180273324e-06,
+      "loss": 0.7412,
+      "step": 3141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986644599770078e-06,
+      "loss": 0.8452,
+      "step": 3142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986631074713849e-06,
+      "loss": 0.8408,
+      "step": 3143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986617542858737e-06,
+      "loss": 1.1816,
+      "step": 3144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986604004204836e-06,
+      "loss": 1.04,
+      "step": 3145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98659045875224e-06,
+      "loss": 0.9365,
+      "step": 3146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986576906501041e-06,
+      "loss": 0.917,
+      "step": 3147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865633474513338e-06,
+      "loss": 1.0195,
+      "step": 3148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865497816032107e-06,
+      "loss": 1.0205,
+      "step": 3149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865362089567657e-06,
+      "loss": 0.9844,
+      "step": 3150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865226295120926e-06,
+      "loss": 0.8721,
+      "step": 3151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986509043269284e-06,
+      "loss": 0.958,
+      "step": 3152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864954502284344e-06,
+      "loss": 1.1016,
+      "step": 3153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864818503896372e-06,
+      "loss": 0.9072,
+      "step": 3154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986468243752986e-06,
+      "loss": 1.0371,
+      "step": 3155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986454630318575e-06,
+      "loss": 0.9043,
+      "step": 3156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864410100864974e-06,
+      "loss": 1.0107,
+      "step": 3157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864273830568474e-06,
+      "loss": 0.9404,
+      "step": 3158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864137492297187e-06,
+      "loss": 0.7251,
+      "step": 3159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986400108605205e-06,
+      "loss": 1.1445,
+      "step": 3160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863864611834018e-06,
+      "loss": 0.9609,
+      "step": 3161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863728069644016e-06,
+      "loss": 0.7861,
+      "step": 3162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863591459482988e-06,
+      "loss": 0.8867,
+      "step": 3163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863454781351877e-06,
+      "loss": 1.2627,
+      "step": 3164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863318035251623e-06,
+      "loss": 0.8125,
+      "step": 3165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986318122118317e-06,
+      "loss": 0.9907,
+      "step": 3166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863044339147463e-06,
+      "loss": 1.0605,
+      "step": 3167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862907389145444e-06,
+      "loss": 0.8779,
+      "step": 3168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862770371178053e-06,
+      "loss": 0.8242,
+      "step": 3169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986263328524624e-06,
+      "loss": 0.835,
+      "step": 3170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986249613135094e-06,
+      "loss": 1.1562,
+      "step": 3171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986235890949311e-06,
+      "loss": 0.8447,
+      "step": 3172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862221619673688e-06,
+      "loss": 0.8057,
+      "step": 3173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862084261893627e-06,
+      "loss": 0.8965,
+      "step": 3174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861946836153864e-06,
+      "loss": 1.126,
+      "step": 3175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986180934245535e-06,
+      "loss": 1.1123,
+      "step": 3176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986167178079904e-06,
+      "loss": 0.9248,
+      "step": 3177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986153415118587e-06,
+      "loss": 0.9414,
+      "step": 3178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861396453616793e-06,
+      "loss": 1.0205,
+      "step": 3179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986125868809276e-06,
+      "loss": 0.9067,
+      "step": 3180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986112085461472e-06,
+      "loss": 0.9961,
+      "step": 3181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986098295318362e-06,
+      "loss": 0.7607,
+      "step": 3182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860844983800415e-06,
+      "loss": 1.1641,
+      "step": 3183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860706946466048e-06,
+      "loss": 0.9355,
+      "step": 3184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860568841181478e-06,
+      "loss": 0.8818,
+      "step": 3185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860430667947654e-06,
+      "loss": 0.7275,
+      "step": 3186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860292426765524e-06,
+      "loss": 1.0098,
+      "step": 3187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986015411763605e-06,
+      "loss": 0.8857,
+      "step": 3188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986001574056018e-06,
+      "loss": 0.7896,
+      "step": 3189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985987729553886e-06,
+      "loss": 0.7422,
+      "step": 3190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859738782573053e-06,
+      "loss": 0.9219,
+      "step": 3191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859600201663716e-06,
+      "loss": 0.9688,
+      "step": 3192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859461552811795e-06,
+      "loss": 0.666,
+      "step": 3193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985932283601825e-06,
+      "loss": 0.8091,
+      "step": 3194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985918405128404e-06,
+      "loss": 0.8887,
+      "step": 3195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859045198610114e-06,
+      "loss": 0.9834,
+      "step": 3196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985890627799744e-06,
+      "loss": 1.0039,
+      "step": 3197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858767289446965e-06,
+      "loss": 0.9893,
+      "step": 3198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858628232959646e-06,
+      "loss": 0.9639,
+      "step": 3199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985848910853645e-06,
+      "loss": 0.9043,
+      "step": 3200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985834991617833e-06,
+      "loss": 0.7871,
+      "step": 3201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858210655886246e-06,
+      "loss": 0.7026,
+      "step": 3202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985807132766116e-06,
+      "loss": 1.1328,
+      "step": 3203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985793193150403e-06,
+      "loss": 0.8682,
+      "step": 3204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985779246741581e-06,
+      "loss": 0.916,
+      "step": 3205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857652935397475e-06,
+      "loss": 0.6958,
+      "step": 3206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857513335449976e-06,
+      "loss": 0.7896,
+      "step": 3207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857373667574276e-06,
+      "loss": 1.0303,
+      "step": 3208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857233931771344e-06,
+      "loss": 0.9863,
+      "step": 3209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985709412804214e-06,
+      "loss": 0.7676,
+      "step": 3210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985695425638762e-06,
+      "loss": 0.8838,
+      "step": 3211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856814316808756e-06,
+      "loss": 1.0674,
+      "step": 3212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985667430930651e-06,
+      "loss": 0.8101,
+      "step": 3213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856534233881846e-06,
+      "loss": 0.6655,
+      "step": 3214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856394090535734e-06,
+      "loss": 0.9443,
+      "step": 3215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985625387926913e-06,
+      "loss": 0.8696,
+      "step": 3216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985611360008301e-06,
+      "loss": 0.9023,
+      "step": 3217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855973252978338e-06,
+      "loss": 0.9639,
+      "step": 3218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985583283795608e-06,
+      "loss": 0.8926,
+      "step": 3219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98556923550172e-06,
+      "loss": 0.7139,
+      "step": 3220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985555180416267e-06,
+      "loss": 0.8086,
+      "step": 3221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985541118539346e-06,
+      "loss": 0.7598,
+      "step": 3222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855270498710536e-06,
+      "loss": 0.9248,
+      "step": 3223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985512974411487e-06,
+      "loss": 0.6973,
+      "step": 3224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985498892160743e-06,
+      "loss": 0.7969,
+      "step": 3225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854848031189183e-06,
+      "loss": 0.8638,
+      "step": 3226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985470707286111e-06,
+      "loss": 1.0127,
+      "step": 3227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985456604662417e-06,
+      "loss": 1.1738,
+      "step": 3228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854424952479343e-06,
+      "loss": 0.8657,
+      "step": 3229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98542837904276e-06,
+      "loss": 0.7275,
+      "step": 3230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854142560469914e-06,
+      "loss": 0.8442,
+      "step": 3231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854001262607255e-06,
+      "loss": 0.9961,
+      "step": 3232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98538598968406e-06,
+      "loss": 1.125,
+      "step": 3233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853718463170925e-06,
+      "loss": 1.0234,
+      "step": 3234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853576961599198e-06,
+      "loss": 1.0391,
+      "step": 3235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853435392126395e-06,
+      "loss": 1.0684,
+      "step": 3236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98532937547535e-06,
+      "loss": 0.8672,
+      "step": 3237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985315204948148e-06,
+      "loss": 1.0742,
+      "step": 3238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853010276311318e-06,
+      "loss": 0.9111,
+      "step": 3239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852868435243986e-06,
+      "loss": 0.8818,
+      "step": 3240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852726526280463e-06,
+      "loss": 1.0518,
+      "step": 3241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852584549421727e-06,
+      "loss": 0.8701,
+      "step": 3242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852442504668755e-06,
+      "loss": 0.9092,
+      "step": 3243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985230039202253e-06,
+      "loss": 0.9404,
+      "step": 3244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985215821148403e-06,
+      "loss": 0.8145,
+      "step": 3245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852015963054226e-06,
+      "loss": 0.8525,
+      "step": 3246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985187364673411e-06,
+      "loss": 0.8701,
+      "step": 3247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985173126252466e-06,
+      "loss": 0.9111,
+      "step": 3248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851588810426853e-06,
+      "loss": 1.0,
+      "step": 3249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851446290441675e-06,
+      "loss": 1.0293,
+      "step": 3250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851303702570102e-06,
+      "loss": 0.8877,
+      "step": 3251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851161046813125e-06,
+      "loss": 0.9023,
+      "step": 3252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985101832317172e-06,
+      "loss": 0.8955,
+      "step": 3253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850875531646877e-06,
+      "loss": 0.6221,
+      "step": 3254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850732672239576e-06,
+      "loss": 0.8716,
+      "step": 3255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98505897449508e-06,
+      "loss": 0.9668,
+      "step": 3256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850446749781533e-06,
+      "loss": 0.8262,
+      "step": 3257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985030368673277e-06,
+      "loss": 0.8501,
+      "step": 3258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850160555805483e-06,
+      "loss": 0.9023,
+      "step": 3259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985001735700067e-06,
+      "loss": 1.0498,
+      "step": 3260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849874090319306e-06,
+      "loss": 0.9395,
+      "step": 3261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849730755762394e-06,
+      "loss": 1.0752,
+      "step": 3262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849587353330906e-06,
+      "loss": 0.8926,
+      "step": 3263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849443883025842e-06,
+      "loss": 1.0176,
+      "step": 3264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849300344848185e-06,
+      "loss": 0.9248,
+      "step": 3265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984915673879892e-06,
+      "loss": 0.7703,
+      "step": 3266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849013064879047e-06,
+      "loss": 0.9951,
+      "step": 3267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984886932308955e-06,
+      "loss": 1.0205,
+      "step": 3268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984872551343142e-06,
+      "loss": 1.1758,
+      "step": 3269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848581635905647e-06,
+      "loss": 1.0186,
+      "step": 3270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984843769051322e-06,
+      "loss": 0.7812,
+      "step": 3271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984829367725514e-06,
+      "loss": 0.8828,
+      "step": 3272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848149596132393e-06,
+      "loss": 0.8276,
+      "step": 3273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984800544714597e-06,
+      "loss": 0.8643,
+      "step": 3274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984786123029687e-06,
+      "loss": 1.0312,
+      "step": 3275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847716945586085e-06,
+      "loss": 0.998,
+      "step": 3276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847572593014604e-06,
+      "loss": 0.9189,
+      "step": 3277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984742817258343e-06,
+      "loss": 1.2344,
+      "step": 3278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984728368429355e-06,
+      "loss": 1.0449,
+      "step": 3279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847139128145966e-06,
+      "loss": 0.8521,
+      "step": 3280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984699450414167e-06,
+      "loss": 1.0049,
+      "step": 3281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846849812281664e-06,
+      "loss": 0.8672,
+      "step": 3282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984670505256694e-06,
+      "loss": 0.8438,
+      "step": 3283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846560224998496e-06,
+      "loss": 1.0605,
+      "step": 3284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984641532957733e-06,
+      "loss": 1.0947,
+      "step": 3285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846270366304443e-06,
+      "loss": 0.5449,
+      "step": 3286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846125335180835e-06,
+      "loss": 1.2373,
+      "step": 3287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98459802362075e-06,
+      "loss": 1.0996,
+      "step": 3288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984583506938544e-06,
+      "loss": 0.9277,
+      "step": 3289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984568983471566e-06,
+      "loss": 0.917,
+      "step": 3290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845544532199155e-06,
+      "loss": 0.9365,
+      "step": 3291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984539916183693e-06,
+      "loss": 1.1699,
+      "step": 3292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984525372362998e-06,
+      "loss": 0.8091,
+      "step": 3293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845108217579314e-06,
+      "loss": 0.8423,
+      "step": 3294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984496264368594e-06,
+      "loss": 0.7661,
+      "step": 3295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844817001950843e-06,
+      "loss": 0.8066,
+      "step": 3296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984467129237504e-06,
+      "loss": 0.8682,
+      "step": 3297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844525514959536e-06,
+      "loss": 0.8389,
+      "step": 3298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844379669705336e-06,
+      "loss": 0.8477,
+      "step": 3299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844233756613436e-06,
+      "loss": 1.127,
+      "step": 3300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984408777568485e-06,
+      "loss": 0.9746,
+      "step": 3301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843941726920577e-06,
+      "loss": 1.208,
+      "step": 3302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984379561032163e-06,
+      "loss": 1.166,
+      "step": 3303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984364942588901e-06,
+      "loss": 0.8877,
+      "step": 3304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984350317362373e-06,
+      "loss": 1.1562,
+      "step": 3305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843356853526795e-06,
+      "loss": 0.9668,
+      "step": 3306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843210465599215e-06,
+      "loss": 0.8662,
+      "step": 3307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843064009841995e-06,
+      "loss": 0.9365,
+      "step": 3308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842917486256146e-06,
+      "loss": 0.9268,
+      "step": 3309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984277089484268e-06,
+      "loss": 0.9473,
+      "step": 3310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842624235602606e-06,
+      "loss": 0.8711,
+      "step": 3311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842477508536934e-06,
+      "loss": 1.1396,
+      "step": 3312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984233071364667e-06,
+      "loss": 0.9805,
+      "step": 3313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984218385093284e-06,
+      "loss": 0.8716,
+      "step": 3314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984203692039644e-06,
+      "loss": 0.8896,
+      "step": 3315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984188992203849e-06,
+      "loss": 0.8955,
+      "step": 3316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984174285586e-06,
+      "loss": 0.9785,
+      "step": 3317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841595721861993e-06,
+      "loss": 0.8145,
+      "step": 3318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984144852004547e-06,
+      "loss": 0.8926,
+      "step": 3319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984130125041145e-06,
+      "loss": 0.9863,
+      "step": 3320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984115391296095e-06,
+      "loss": 0.9766,
+      "step": 3321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841006507694987e-06,
+      "loss": 0.9287,
+      "step": 3322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984085903461457e-06,
+      "loss": 1.1035,
+      "step": 3323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984071149372072e-06,
+      "loss": 1.0615,
+      "step": 3324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840563885014456e-06,
+      "loss": 0.9932,
+      "step": 3325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984041620849679e-06,
+      "loss": 0.8818,
+      "step": 3326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840268464168738e-06,
+      "loss": 0.9248,
+      "step": 3327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840120652031325e-06,
+      "loss": 0.9883,
+      "step": 3328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839972772085566e-06,
+      "loss": 0.9463,
+      "step": 3329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983982482433248e-06,
+      "loss": 1.0312,
+      "step": 3330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983967680877309e-06,
+      "loss": 0.8433,
+      "step": 3331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983952872540841e-06,
+      "loss": 0.8613,
+      "step": 3332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983938057423946e-06,
+      "loss": 1.041,
+      "step": 3333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983923235526727e-06,
+      "loss": 0.874,
+      "step": 3334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983908406849285e-06,
+      "loss": 0.8486,
+      "step": 3335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838935713917234e-06,
+      "loss": 0.7642,
+      "step": 3336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983878729154143e-06,
+      "loss": 0.9082,
+      "step": 3337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838638801366476e-06,
+      "loss": 0.8296,
+      "step": 3338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838490243393383e-06,
+      "loss": 1.1025,
+      "step": 3339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983834161762318e-06,
+      "loss": 0.9434,
+      "step": 3340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983819292405689e-06,
+      "loss": 0.9658,
+      "step": 3341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983804416269554e-06,
+      "loss": 0.7695,
+      "step": 3342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837895333540153e-06,
+      "loss": 0.8457,
+      "step": 3343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983774643659175e-06,
+      "loss": 0.9414,
+      "step": 3344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983759747185137e-06,
+      "loss": 1.1035,
+      "step": 3345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837448439320025e-06,
+      "loss": 1.0527,
+      "step": 3346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837299338998753e-06,
+      "loss": 0.9824,
+      "step": 3347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837150170888575e-06,
+      "loss": 0.9062,
+      "step": 3348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837000934990523e-06,
+      "loss": 0.9463,
+      "step": 3349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983685163130562e-06,
+      "loss": 0.9443,
+      "step": 3350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98367022598349e-06,
+      "loss": 1.0566,
+      "step": 3351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836552820579387e-06,
+      "loss": 0.8633,
+      "step": 3352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983640331354012e-06,
+      "loss": 1.0088,
+      "step": 3353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983625373871812e-06,
+      "loss": 1.1172,
+      "step": 3354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983610409611442e-06,
+      "loss": 0.9316,
+      "step": 3355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835954385730056e-06,
+      "loss": 0.8877,
+      "step": 3356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835804607566056e-06,
+      "loss": 0.8691,
+      "step": 3357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983565476162345e-06,
+      "loss": 0.9854,
+      "step": 3358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835504847903274e-06,
+      "loss": 0.8721,
+      "step": 3359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983535486640656e-06,
+      "loss": 0.8423,
+      "step": 3360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983520481713434e-06,
+      "loss": 1.0791,
+      "step": 3361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983505470008765e-06,
+      "loss": 0.8301,
+      "step": 3362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983490451526752e-06,
+      "loss": 1.0576,
+      "step": 3363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834754262674995e-06,
+      "loss": 1.0068,
+      "step": 3364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98346039423111e-06,
+      "loss": 0.9512,
+      "step": 3365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834453554176874e-06,
+      "loss": 0.998,
+      "step": 3366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834303098273357e-06,
+      "loss": 0.9092,
+      "step": 3367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834152574601584e-06,
+      "loss": 0.8926,
+      "step": 3368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983400198316259e-06,
+      "loss": 0.9404,
+      "step": 3369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983385132395741e-06,
+      "loss": 0.8965,
+      "step": 3370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833700596987093e-06,
+      "loss": 0.9102,
+      "step": 3371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983354980225267e-06,
+      "loss": 1.0439,
+      "step": 3372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833398939755176e-06,
+      "loss": 0.9551,
+      "step": 3373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833248009495658e-06,
+      "loss": 0.9688,
+      "step": 3374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833097011475155e-06,
+      "loss": 0.7656,
+      "step": 3375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98329459456947e-06,
+      "loss": 0.8447,
+      "step": 3376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832794812155353e-06,
+      "loss": 1.0879,
+      "step": 3377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832643610858133e-06,
+      "loss": 1.123,
+      "step": 3378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832492341804095e-06,
+      "loss": 0.9346,
+      "step": 3379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983234100499428e-06,
+      "loss": 0.9829,
+      "step": 3380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983218960042972e-06,
+      "loss": 0.8594,
+      "step": 3381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832038128111473e-06,
+      "loss": 0.8994,
+      "step": 3382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983188658804058e-06,
+      "loss": 1.0859,
+      "step": 3383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983173498021808e-06,
+      "loss": 1.0703,
+      "step": 3384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983158330464502e-06,
+      "loss": 1.0625,
+      "step": 3385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983143156132244e-06,
+      "loss": 1.0137,
+      "step": 3386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831279750251396e-06,
+      "loss": 0.9873,
+      "step": 3387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983112787143293e-06,
+      "loss": 0.9355,
+      "step": 3388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830975924868083e-06,
+      "loss": 0.7539,
+      "step": 3389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830823910557914e-06,
+      "loss": 0.8418,
+      "step": 3390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830671828503456e-06,
+      "loss": 0.7686,
+      "step": 3391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830519678705767e-06,
+      "loss": 0.9814,
+      "step": 3392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983036746116589e-06,
+      "loss": 0.8325,
+      "step": 3393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983021517588488e-06,
+      "loss": 0.7979,
+      "step": 3394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983006282286378e-06,
+      "loss": 1.0898,
+      "step": 3395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982991040210365e-06,
+      "loss": 0.9238,
+      "step": 3396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829757913605526e-06,
+      "loss": 0.9756,
+      "step": 3397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829605357370466e-06,
+      "loss": 0.9873,
+      "step": 3398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829452733399523e-06,
+      "loss": 0.6626,
+      "step": 3399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982930004169375e-06,
+      "loss": 0.9766,
+      "step": 3400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982914728225419e-06,
+      "loss": 0.6255,
+      "step": 3401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828994455081907e-06,
+      "loss": 0.9238,
+      "step": 3402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828841560177947e-06,
+      "loss": 1.0459,
+      "step": 3403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828688597543365e-06,
+      "loss": 1.1904,
+      "step": 3404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828535567179216e-06,
+      "loss": 0.897,
+      "step": 3405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828382469086554e-06,
+      "loss": 1.0635,
+      "step": 3406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828229303266437e-06,
+      "loss": 0.9521,
+      "step": 3407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828076069719916e-06,
+      "loss": 0.9014,
+      "step": 3408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827922768448046e-06,
+      "loss": 0.9609,
+      "step": 3409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827769399451886e-06,
+      "loss": 0.9697,
+      "step": 3410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827615962732498e-06,
+      "loss": 0.8291,
+      "step": 3411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982746245829093e-06,
+      "loss": 0.8745,
+      "step": 3412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827308886128244e-06,
+      "loss": 1.0225,
+      "step": 3413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827155246245497e-06,
+      "loss": 0.7676,
+      "step": 3414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827001538643752e-06,
+      "loss": 0.9854,
+      "step": 3415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826847763324066e-06,
+      "loss": 1.0684,
+      "step": 3416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826693920287496e-06,
+      "loss": 0.8672,
+      "step": 3417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98265400095351e-06,
+      "loss": 0.9668,
+      "step": 3418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826386031067946e-06,
+      "loss": 0.8311,
+      "step": 3419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826231984887096e-06,
+      "loss": 1.0596,
+      "step": 3420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98260778709936e-06,
+      "loss": 0.8618,
+      "step": 3421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982592368938853e-06,
+      "loss": 0.8994,
+      "step": 3422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825769440072947e-06,
+      "loss": 0.998,
+      "step": 3423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825615123047913e-06,
+      "loss": 0.8701,
+      "step": 3424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825460738314486e-06,
+      "loss": 1.0703,
+      "step": 3425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825306285873744e-06,
+      "loss": 0.9551,
+      "step": 3426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982515176572673e-06,
+      "loss": 1.0049,
+      "step": 3427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824997177874533e-06,
+      "loss": 0.9814,
+      "step": 3428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824842522318195e-06,
+      "loss": 0.9756,
+      "step": 3429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98246877990588e-06,
+      "loss": 0.9854,
+      "step": 3430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824533008097407e-06,
+      "loss": 0.8311,
+      "step": 3431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824378149435083e-06,
+      "loss": 0.7773,
+      "step": 3432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824223223072893e-06,
+      "loss": 1.0459,
+      "step": 3433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982406822901191e-06,
+      "loss": 1.1885,
+      "step": 3434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823913167253195e-06,
+      "loss": 0.8291,
+      "step": 3435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823758037797818e-06,
+      "loss": 0.7935,
+      "step": 3436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823602840646856e-06,
+      "loss": 0.668,
+      "step": 3437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823447575801368e-06,
+      "loss": 0.8765,
+      "step": 3438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982329224326243e-06,
+      "loss": 1.0137,
+      "step": 3439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982313684303111e-06,
+      "loss": 1.083,
+      "step": 3440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982298137510848e-06,
+      "loss": 0.9062,
+      "step": 3441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822825839495613e-06,
+      "loss": 0.9541,
+      "step": 3442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982267023619358e-06,
+      "loss": 0.9253,
+      "step": 3443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822514565203448e-06,
+      "loss": 0.9795,
+      "step": 3444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822358826526298e-06,
+      "loss": 1.0156,
+      "step": 3445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98222030201632e-06,
+      "loss": 0.8994,
+      "step": 3446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982204714611522e-06,
+      "loss": 0.854,
+      "step": 3447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821891204383444e-06,
+      "loss": 1.0479,
+      "step": 3448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982173519496894e-06,
+      "loss": 1.0957,
+      "step": 3449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821579117872787e-06,
+      "loss": 1.2402,
+      "step": 3450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821422973096054e-06,
+      "loss": 0.9775,
+      "step": 3451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821266760639826e-06,
+      "loss": 0.9121,
+      "step": 3452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982111048050517e-06,
+      "loss": 1.1973,
+      "step": 3453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982095413269317e-06,
+      "loss": 0.9146,
+      "step": 3454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98207977172049e-06,
+      "loss": 0.7881,
+      "step": 3455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982064123404144e-06,
+      "loss": 0.8799,
+      "step": 3456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820484683203867e-06,
+      "loss": 0.9336,
+      "step": 3457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820328064693257e-06,
+      "loss": 0.8975,
+      "step": 3458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820171378510695e-06,
+      "loss": 0.8398,
+      "step": 3459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820014624657257e-06,
+      "loss": 0.8477,
+      "step": 3460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819857803134025e-06,
+      "loss": 0.9717,
+      "step": 3461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819700913942077e-06,
+      "loss": 1.125,
+      "step": 3462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819543957082497e-06,
+      "loss": 0.9927,
+      "step": 3463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819386932556363e-06,
+      "loss": 0.9736,
+      "step": 3464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981922984036476e-06,
+      "loss": 0.8535,
+      "step": 3465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819072680508776e-06,
+      "loss": 0.8262,
+      "step": 3466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981891545298948e-06,
+      "loss": 1.2754,
+      "step": 3467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818758157807967e-06,
+      "loss": 0.6885,
+      "step": 3468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818600794965317e-06,
+      "loss": 0.8311,
+      "step": 3469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818443364462616e-06,
+      "loss": 0.7017,
+      "step": 3470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818285866300946e-06,
+      "loss": 1.0674,
+      "step": 3471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818128300481394e-06,
+      "loss": 1.0508,
+      "step": 3472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817970667005047e-06,
+      "loss": 1.0576,
+      "step": 3473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981781296587299e-06,
+      "loss": 0.8584,
+      "step": 3474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981765519708631e-06,
+      "loss": 1.0112,
+      "step": 3475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817497360646087e-06,
+      "loss": 0.9346,
+      "step": 3476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817339456553424e-06,
+      "loss": 0.9209,
+      "step": 3477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817181484809396e-06,
+      "loss": 0.8174,
+      "step": 3478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817023445415103e-06,
+      "loss": 0.9424,
+      "step": 3479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816865338371622e-06,
+      "loss": 0.9639,
+      "step": 3480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816707163680046e-06,
+      "loss": 0.9839,
+      "step": 3481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981654892134147e-06,
+      "loss": 1.0654,
+      "step": 3482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981639061135698e-06,
+      "loss": 0.9609,
+      "step": 3483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816232233727668e-06,
+      "loss": 0.9932,
+      "step": 3484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816073788454626e-06,
+      "loss": 0.853,
+      "step": 3485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815915275538944e-06,
+      "loss": 0.8066,
+      "step": 3486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815756694981718e-06,
+      "loss": 1.3145,
+      "step": 3487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815598046784036e-06,
+      "loss": 0.8789,
+      "step": 3488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815439330946996e-06,
+      "loss": 0.9424,
+      "step": 3489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981528054747169e-06,
+      "loss": 1.0273,
+      "step": 3490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981512169635921e-06,
+      "loss": 1.2324,
+      "step": 3491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981496277761065e-06,
+      "loss": 0.9609,
+      "step": 3492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814803791227114e-06,
+      "loss": 0.8066,
+      "step": 3493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814644737209684e-06,
+      "loss": 0.9458,
+      "step": 3494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814485615559468e-06,
+      "loss": 1.0811,
+      "step": 3495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814326426277554e-06,
+      "loss": 0.9805,
+      "step": 3496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981416716936504e-06,
+      "loss": 1.1035,
+      "step": 3497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981400784482303e-06,
+      "loss": 0.9629,
+      "step": 3498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981384845265262e-06,
+      "loss": 0.7427,
+      "step": 3499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98136889928549e-06,
+      "loss": 1.1133,
+      "step": 3500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813529465430985e-06,
+      "loss": 1.1309,
+      "step": 3501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813369870381957e-06,
+      "loss": 0.9766,
+      "step": 3502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813210207708923e-06,
+      "loss": 1.0459,
+      "step": 3503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813050477412984e-06,
+      "loss": 0.9219,
+      "step": 3504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981289067949524e-06,
+      "loss": 0.8691,
+      "step": 3505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812730813956792e-06,
+      "loss": 0.9092,
+      "step": 3506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981257088079874e-06,
+      "loss": 1.0566,
+      "step": 3507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812410880022194e-06,
+      "loss": 0.9473,
+      "step": 3508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812250811628247e-06,
+      "loss": 0.7129,
+      "step": 3509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812090675618003e-06,
+      "loss": 0.7993,
+      "step": 3510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981193047199257e-06,
+      "loss": 1.1602,
+      "step": 3511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981177020075305e-06,
+      "loss": 1.0273,
+      "step": 3512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981160986190055e-06,
+      "loss": 0.9531,
+      "step": 3513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811449455436167e-06,
+      "loss": 0.8975,
+      "step": 3514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811288981361017e-06,
+      "loss": 1.0117,
+      "step": 3515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811128439676196e-06,
+      "loss": 0.5906,
+      "step": 3516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981096783038282e-06,
+      "loss": 1.0488,
+      "step": 3517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810807153481987e-06,
+      "loss": 0.6719,
+      "step": 3518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810646408974806e-06,
+      "loss": 0.9736,
+      "step": 3519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981048559686239e-06,
+      "loss": 0.9824,
+      "step": 3520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810324717145844e-06,
+      "loss": 1.1797,
+      "step": 3521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810163769826275e-06,
+      "loss": 1.0225,
+      "step": 3522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810002754904795e-06,
+      "loss": 0.9707,
+      "step": 3523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980984167238251e-06,
+      "loss": 1.0215,
+      "step": 3524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980968052226053e-06,
+      "loss": 0.7119,
+      "step": 3525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980951930453997e-06,
+      "loss": 0.8613,
+      "step": 3526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809358019221937e-06,
+      "loss": 0.6729,
+      "step": 3527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809196666307545e-06,
+      "loss": 0.9814,
+      "step": 3528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809035245797903e-06,
+      "loss": 0.8408,
+      "step": 3529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980887375769413e-06,
+      "loss": 0.9688,
+      "step": 3530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808712201997327e-06,
+      "loss": 1.0371,
+      "step": 3531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980855057870862e-06,
+      "loss": 0.8848,
+      "step": 3532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808388887829115e-06,
+      "loss": 0.9922,
+      "step": 3533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808227129359925e-06,
+      "loss": 1.127,
+      "step": 3534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980806530330217e-06,
+      "loss": 0.8213,
+      "step": 3535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980790340965696e-06,
+      "loss": 0.8018,
+      "step": 3536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980774144842542e-06,
+      "loss": 0.9824,
+      "step": 3537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807579419608657e-06,
+      "loss": 0.8525,
+      "step": 3538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807417323207787e-06,
+      "loss": 0.6411,
+      "step": 3539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807255159223933e-06,
+      "loss": 0.8428,
+      "step": 3540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980709292765821e-06,
+      "loss": 0.9795,
+      "step": 3541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806930628511732e-06,
+      "loss": 0.957,
+      "step": 3542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806768261785625e-06,
+      "loss": 0.8936,
+      "step": 3543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806605827481002e-06,
+      "loss": 1.041,
+      "step": 3544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806443325598985e-06,
+      "loss": 1.0479,
+      "step": 3545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980628075614069e-06,
+      "loss": 1.0342,
+      "step": 3546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806118119107246e-06,
+      "loss": 1.1104,
+      "step": 3547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980595541449976e-06,
+      "loss": 0.9688,
+      "step": 3548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805792642319366e-06,
+      "loss": 0.9521,
+      "step": 3549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805629802567184e-06,
+      "loss": 0.9111,
+      "step": 3550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980546689524433e-06,
+      "loss": 0.8965,
+      "step": 3551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805303920351926e-06,
+      "loss": 0.9395,
+      "step": 3552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805140877891106e-06,
+      "loss": 1.0391,
+      "step": 3553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980497776786298e-06,
+      "loss": 0.8467,
+      "step": 3554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980481459026868e-06,
+      "loss": 0.9619,
+      "step": 3555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804651345109327e-06,
+      "loss": 1.1328,
+      "step": 3556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804488032386048e-06,
+      "loss": 0.8101,
+      "step": 3557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980432465209997e-06,
+      "loss": 1.0,
+      "step": 3558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804161204252217e-06,
+      "loss": 0.8623,
+      "step": 3559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803997688843915e-06,
+      "loss": 0.8398,
+      "step": 3560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980383410587619e-06,
+      "loss": 1.0117,
+      "step": 3561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980367045535017e-06,
+      "loss": 0.7671,
+      "step": 3562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803506737266978e-06,
+      "loss": 0.9961,
+      "step": 3563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803342951627755e-06,
+      "loss": 1.1045,
+      "step": 3564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980317909843362e-06,
+      "loss": 1.0381,
+      "step": 3565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98030151776857e-06,
+      "loss": 0.9795,
+      "step": 3566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980285118938513e-06,
+      "loss": 0.9883,
+      "step": 3567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980268713353304e-06,
+      "loss": 0.998,
+      "step": 3568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802523010130557e-06,
+      "loss": 0.832,
+      "step": 3569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802358819178815e-06,
+      "loss": 1.1631,
+      "step": 3570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802194560678943e-06,
+      "loss": 0.8584,
+      "step": 3571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802030234632078e-06,
+      "loss": 1.1465,
+      "step": 3572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980186584103934e-06,
+      "loss": 1.0527,
+      "step": 3573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980170137990188e-06,
+      "loss": 1.1035,
+      "step": 3574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801536851220817e-06,
+      "loss": 0.7163,
+      "step": 3575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801372254997293e-06,
+      "loss": 0.9834,
+      "step": 3576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801207591232436e-06,
+      "loss": 0.9434,
+      "step": 3577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980104285992738e-06,
+      "loss": 0.9004,
+      "step": 3578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980087806108327e-06,
+      "loss": 0.9707,
+      "step": 3579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800713194701233e-06,
+      "loss": 0.916,
+      "step": 3580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800548260782406e-06,
+      "loss": 0.9883,
+      "step": 3581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980038325932793e-06,
+      "loss": 1.0332,
+      "step": 3582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800218190338933e-06,
+      "loss": 0.8501,
+      "step": 3583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800053053816567e-06,
+      "loss": 0.769,
+      "step": 3584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799887849761954e-06,
+      "loss": 0.9287,
+      "step": 3585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979972257817624e-06,
+      "loss": 1.1221,
+      "step": 3586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979955723906057e-06,
+      "loss": 0.9307,
+      "step": 3587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979939183241607e-06,
+      "loss": 0.9766,
+      "step": 3588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979922635824389e-06,
+      "loss": 0.7964,
+      "step": 3589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979906081654517e-06,
+      "loss": 0.9414,
+      "step": 3590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798895207321045e-06,
+      "loss": 0.7886,
+      "step": 3591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798729530572657e-06,
+      "loss": 0.8467,
+      "step": 3592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798563786301153e-06,
+      "loss": 0.9126,
+      "step": 3593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979839797450767e-06,
+      "loss": 0.8506,
+      "step": 3594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798232095193353e-06,
+      "loss": 0.9785,
+      "step": 3595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798066148359344e-06,
+      "loss": 1.0801,
+      "step": 3596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979790013400679e-06,
+      "loss": 1.1357,
+      "step": 3597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979773405213683e-06,
+      "loss": 0.9746,
+      "step": 3598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979756790275061e-06,
+      "loss": 1.0059,
+      "step": 3599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979740168584928e-06,
+      "loss": 0.7754,
+      "step": 3600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797235401433973e-06,
+      "loss": 0.8628,
+      "step": 3601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979706904950585e-06,
+      "loss": 0.9824,
+      "step": 3602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796902630066045e-06,
+      "loss": 0.8779,
+      "step": 3603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796736143115714e-06,
+      "loss": 0.9307,
+      "step": 3604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796569588656e-06,
+      "loss": 0.8379,
+      "step": 3605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796402966688046e-06,
+      "loss": 0.8311,
+      "step": 3606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796236277213014e-06,
+      "loss": 0.7515,
+      "step": 3607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979606952023204e-06,
+      "loss": 0.9229,
+      "step": 3608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795902695746274e-06,
+      "loss": 0.6699,
+      "step": 3609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795735803756866e-06,
+      "loss": 1.2305,
+      "step": 3610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795568844264976e-06,
+      "loss": 0.8516,
+      "step": 3611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979540181727174e-06,
+      "loss": 0.9365,
+      "step": 3612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979523472277832e-06,
+      "loss": 0.9209,
+      "step": 3613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795067560785867e-06,
+      "loss": 1.0586,
+      "step": 3614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794900331295524e-06,
+      "loss": 1.0186,
+      "step": 3615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979473303430845e-06,
+      "loss": 1.168,
+      "step": 3616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97945656698258e-06,
+      "loss": 1.0156,
+      "step": 3617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979439823784872e-06,
+      "loss": 1.0459,
+      "step": 3618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979423073837837e-06,
+      "loss": 0.9873,
+      "step": 3619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97940631714159e-06,
+      "loss": 1.1172,
+      "step": 3620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979389553696247e-06,
+      "loss": 1.1152,
+      "step": 3621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979372783501923e-06,
+      "loss": 1.1348,
+      "step": 3622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793560065587338e-06,
+      "loss": 1.0254,
+      "step": 3623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793392228667947e-06,
+      "loss": 0.9062,
+      "step": 3624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979322432426222e-06,
+      "loss": 0.8447,
+      "step": 3625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793056352371303e-06,
+      "loss": 0.6501,
+      "step": 3626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792888312996367e-06,
+      "loss": 0.7432,
+      "step": 3627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979272020613856e-06,
+      "loss": 0.8154,
+      "step": 3628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979255203179905e-06,
+      "loss": 0.8408,
+      "step": 3629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979238378997899e-06,
+      "loss": 1.1689,
+      "step": 3630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979221548067953e-06,
+      "loss": 1.0488,
+      "step": 3631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979204710390185e-06,
+      "loss": 1.3066,
+      "step": 3632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979187865964709e-06,
+      "loss": 0.8789,
+      "step": 3633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791710147916426e-06,
+      "loss": 0.8496,
+      "step": 3634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979154156871101e-06,
+      "loss": 0.9893,
+      "step": 3635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979137292203201e-06,
+      "loss": 1.083,
+      "step": 3636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791204207880585e-06,
+      "loss": 0.7617,
+      "step": 3637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97910354262579e-06,
+      "loss": 1.0557,
+      "step": 3638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979086657716511e-06,
+      "loss": 1.0049,
+      "step": 3639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979069766060339e-06,
+      "loss": 1.1094,
+      "step": 3640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790528676573895e-06,
+      "loss": 0.8398,
+      "step": 3641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979035962507779e-06,
+      "loss": 0.8857,
+      "step": 3642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979019050611625e-06,
+      "loss": 0.7837,
+      "step": 3643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979002131969043e-06,
+      "loss": 0.7483,
+      "step": 3644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97898520658015e-06,
+      "loss": 0.9727,
+      "step": 3645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978968274445063e-06,
+      "loss": 1.0391,
+      "step": 3646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789513355638976e-06,
+      "loss": 0.8623,
+      "step": 3647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978934389936771e-06,
+      "loss": 1.0908,
+      "step": 3648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789174375638002e-06,
+      "loss": 0.8848,
+      "step": 3649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789004784451023e-06,
+      "loss": 0.9619,
+      "step": 3650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788835125807933e-06,
+      "loss": 0.9473,
+      "step": 3651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788665399709912e-06,
+      "loss": 0.9629,
+      "step": 3652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788495606158122e-06,
+      "loss": 0.9316,
+      "step": 3653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978832574515373e-06,
+      "loss": 0.9873,
+      "step": 3654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978815581669792e-06,
+      "loss": 0.8789,
+      "step": 3655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787985820791846e-06,
+      "loss": 0.9248,
+      "step": 3656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787815757436693e-06,
+      "loss": 1.127,
+      "step": 3657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787645626633627e-06,
+      "loss": 0.8652,
+      "step": 3658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978747542838382e-06,
+      "loss": 0.998,
+      "step": 3659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787305162688444e-06,
+      "loss": 0.96,
+      "step": 3660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978713482954868e-06,
+      "loss": 1.165,
+      "step": 3661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978696442896569e-06,
+      "loss": 0.9854,
+      "step": 3662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786793960940656e-06,
+      "loss": 1.0596,
+      "step": 3663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786623425474753e-06,
+      "loss": 1.0039,
+      "step": 3664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786452822569154e-06,
+      "loss": 0.9346,
+      "step": 3665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786282152225032e-06,
+      "loss": 0.9014,
+      "step": 3666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786111414443565e-06,
+      "loss": 1.0381,
+      "step": 3667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785940609225934e-06,
+      "loss": 0.957,
+      "step": 3668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978576973657331e-06,
+      "loss": 0.9443,
+      "step": 3669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785598796486873e-06,
+      "loss": 0.9033,
+      "step": 3670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97854277889678e-06,
+      "loss": 0.9331,
+      "step": 3671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785256714017272e-06,
+      "loss": 0.8008,
+      "step": 3672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785085571636466e-06,
+      "loss": 1.0693,
+      "step": 3673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978491436182656e-06,
+      "loss": 0.7695,
+      "step": 3674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978474308458874e-06,
+      "loss": 0.8545,
+      "step": 3675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784571739924174e-06,
+      "loss": 1.1025,
+      "step": 3676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784400327834055e-06,
+      "loss": 0.7803,
+      "step": 3677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784228848319557e-06,
+      "loss": 1.0234,
+      "step": 3678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978405730138187e-06,
+      "loss": 0.9766,
+      "step": 3679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978388568702216e-06,
+      "loss": 0.7783,
+      "step": 3680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783714005241626e-06,
+      "loss": 0.9697,
+      "step": 3681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783542256041444e-06,
+      "loss": 1.0332,
+      "step": 3682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783370439422797e-06,
+      "loss": 0.9961,
+      "step": 3683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783198555386874e-06,
+      "loss": 0.9785,
+      "step": 3684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783026603934854e-06,
+      "loss": 0.7075,
+      "step": 3685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782854585067923e-06,
+      "loss": 1.1289,
+      "step": 3686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978268249878727e-06,
+      "loss": 0.6968,
+      "step": 3687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782510345094075e-06,
+      "loss": 1.0898,
+      "step": 3688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782338123989527e-06,
+      "loss": 1.0859,
+      "step": 3689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782165835474815e-06,
+      "loss": 0.79,
+      "step": 3690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978199347955112e-06,
+      "loss": 0.915,
+      "step": 3691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781821056219638e-06,
+      "loss": 0.7925,
+      "step": 3692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781648565481556e-06,
+      "loss": 0.8066,
+      "step": 3693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781476007338054e-06,
+      "loss": 1.0156,
+      "step": 3694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978130338179033e-06,
+      "loss": 1.124,
+      "step": 3695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978113068883957e-06,
+      "loss": 0.9717,
+      "step": 3696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780957928486967e-06,
+      "loss": 1.168,
+      "step": 3697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780785100733708e-06,
+      "loss": 0.6641,
+      "step": 3698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780612205580983e-06,
+      "loss": 0.9961,
+      "step": 3699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780439243029986e-06,
+      "loss": 0.7305,
+      "step": 3700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780266213081907e-06,
+      "loss": 0.8867,
+      "step": 3701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978009311573794e-06,
+      "loss": 0.8936,
+      "step": 3702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.977991995099928e-06,
+      "loss": 0.7646,
+      "step": 3703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779746718867114e-06,
+      "loss": 0.856,
+      "step": 3704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779573419342643e-06,
+      "loss": 0.792,
+      "step": 3705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779400052427053e-06,
+      "loss": 0.8894,
+      "step": 3706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779226618121548e-06,
+      "loss": 0.8242,
+      "step": 3707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779053116427316e-06,
+      "loss": 0.8091,
+      "step": 3708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778879547345557e-06,
+      "loss": 0.9805,
+      "step": 3709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778705910877466e-06,
+      "loss": 0.7881,
+      "step": 3710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977853220702424e-06,
+      "loss": 0.8789,
+      "step": 3711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977835843578707e-06,
+      "loss": 0.9951,
+      "step": 3712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778184597167157e-06,
+      "loss": 0.8535,
+      "step": 3713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778010691165707e-06,
+      "loss": 1.0898,
+      "step": 3714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777836717783908e-06,
+      "loss": 0.9424,
+      "step": 3715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777662677022963e-06,
+      "loss": 0.793,
+      "step": 3716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777488568884066e-06,
+      "loss": 0.832,
+      "step": 3717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777314393368425e-06,
+      "loss": 1.0088,
+      "step": 3718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977714015047724e-06,
+      "loss": 1.0449,
+      "step": 3719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776965840211703e-06,
+      "loss": 0.9414,
+      "step": 3720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977679146257302e-06,
+      "loss": 0.6611,
+      "step": 3721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776617017562395e-06,
+      "loss": 0.7129,
+      "step": 3722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977644250518103e-06,
+      "loss": 1.1094,
+      "step": 3723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776267925430126e-06,
+      "loss": 0.9404,
+      "step": 3724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776093278310884e-06,
+      "loss": 1.0654,
+      "step": 3725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977591856382451e-06,
+      "loss": 1.0811,
+      "step": 3726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775743781972205e-06,
+      "loss": 0.7852,
+      "step": 3727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775568932755184e-06,
+      "loss": 0.8833,
+      "step": 3728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775394016174635e-06,
+      "loss": 0.8271,
+      "step": 3729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977521903223177e-06,
+      "loss": 0.9199,
+      "step": 3730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775043980927804e-06,
+      "loss": 1.0264,
+      "step": 3731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774868862263935e-06,
+      "loss": 1.0127,
+      "step": 3732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977469367624137e-06,
+      "loss": 1.0791,
+      "step": 3733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774518422861317e-06,
+      "loss": 0.9199,
+      "step": 3734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774343102124986e-06,
+      "loss": 0.9575,
+      "step": 3735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977416771403358e-06,
+      "loss": 0.7998,
+      "step": 3736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977399225858831e-06,
+      "loss": 0.9448,
+      "step": 3737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773816735790383e-06,
+      "loss": 0.8604,
+      "step": 3738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773641145641015e-06,
+      "loss": 1.0996,
+      "step": 3739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977346548814141e-06,
+      "loss": 0.9238,
+      "step": 3740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773289763292783e-06,
+      "loss": 1.0986,
+      "step": 3741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977311397109634e-06,
+      "loss": 1.0186,
+      "step": 3742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772938111553295e-06,
+      "loss": 0.9746,
+      "step": 3743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977276218466486e-06,
+      "loss": 1.0029,
+      "step": 3744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977258619043225e-06,
+      "loss": 0.9932,
+      "step": 3745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977241012885667e-06,
+      "loss": 0.6455,
+      "step": 3746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772233999939345e-06,
+      "loss": 0.8579,
+      "step": 3747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772057803681473e-06,
+      "loss": 0.9912,
+      "step": 3748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977188154008428e-06,
+      "loss": 0.999,
+      "step": 3749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771705209148978e-06,
+      "loss": 0.8613,
+      "step": 3750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771528810876785e-06,
+      "loss": 1.0586,
+      "step": 3751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771352345268907e-06,
+      "loss": 0.8877,
+      "step": 3752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977117581232657e-06,
+      "loss": 0.7451,
+      "step": 3753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770999212050986e-06,
+      "loss": 0.9238,
+      "step": 3754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770822544443373e-06,
+      "loss": 0.9521,
+      "step": 3755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770645809504946e-06,
+      "loss": 0.822,
+      "step": 3756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977046900723693e-06,
+      "loss": 0.8496,
+      "step": 3757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770292137640532e-06,
+      "loss": 1.1211,
+      "step": 3758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977011520071698e-06,
+      "loss": 0.9736,
+      "step": 3759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769938196467486e-06,
+      "loss": 0.6973,
+      "step": 3760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976976112489328e-06,
+      "loss": 1.0176,
+      "step": 3761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769583985995574e-06,
+      "loss": 1.0283,
+      "step": 3762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976940677977559e-06,
+      "loss": 0.5742,
+      "step": 3763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769229506234553e-06,
+      "loss": 0.9648,
+      "step": 3764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769052165373682e-06,
+      "loss": 0.9326,
+      "step": 3765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97688747571942e-06,
+      "loss": 0.8574,
+      "step": 3766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768697281697322e-06,
+      "loss": 0.8145,
+      "step": 3767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976851973888428e-06,
+      "loss": 1.1182,
+      "step": 3768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768342128756295e-06,
+      "loss": 1.0312,
+      "step": 3769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976816445131459e-06,
+      "loss": 1.002,
+      "step": 3770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976798670656039e-06,
+      "loss": 1.0244,
+      "step": 3771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767808894494923e-06,
+      "loss": 0.9922,
+      "step": 3772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767631015119405e-06,
+      "loss": 1.0879,
+      "step": 3773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767453068435075e-06,
+      "loss": 0.9453,
+      "step": 3774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767275054443147e-06,
+      "loss": 1.0098,
+      "step": 3775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767096973144855e-06,
+      "loss": 0.77,
+      "step": 3776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766918824541423e-06,
+      "loss": 1.0498,
+      "step": 3777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976674060863408e-06,
+      "loss": 1.0684,
+      "step": 3778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766562325424053e-06,
+      "loss": 0.9658,
+      "step": 3779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976638397491257e-06,
+      "loss": 1.0195,
+      "step": 3780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766205557100868e-06,
+      "loss": 1.0234,
+      "step": 3781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766027071990164e-06,
+      "loss": 0.9033,
+      "step": 3782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765848519581692e-06,
+      "loss": 1.0156,
+      "step": 3783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765669899876686e-06,
+      "loss": 1.0908,
+      "step": 3784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976549121287638e-06,
+      "loss": 0.7427,
+      "step": 3785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976531245858199e-06,
+      "loss": 0.6909,
+      "step": 3786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765133636994768e-06,
+      "loss": 0.8516,
+      "step": 3787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976495474811593e-06,
+      "loss": 0.9834,
+      "step": 3788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976477579194672e-06,
+      "loss": 0.8945,
+      "step": 3789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976459676848836e-06,
+      "loss": 0.9639,
+      "step": 3790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764417677742095e-06,
+      "loss": 0.7886,
+      "step": 3791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764238519709154e-06,
+      "loss": 0.5181,
+      "step": 3792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976405929439077e-06,
+      "loss": 1.1582,
+      "step": 3793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976388000178818e-06,
+      "loss": 0.9111,
+      "step": 3794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976370064190262e-06,
+      "loss": 1.0,
+      "step": 3795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976352121473532e-06,
+      "loss": 0.855,
+      "step": 3796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976334172028753e-06,
+      "loss": 0.9062,
+      "step": 3797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976316215856047e-06,
+      "loss": 0.7915,
+      "step": 3798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976298252955539e-06,
+      "loss": 1.0098,
+      "step": 3799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762802833273526e-06,
+      "loss": 0.8057,
+      "step": 3800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976262306971611e-06,
+      "loss": 1.0459,
+      "step": 3801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762443238884386e-06,
+      "loss": 0.9053,
+      "step": 3802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762263340779595e-06,
+      "loss": 1.0508,
+      "step": 3803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976208337540297e-06,
+      "loss": 0.9121,
+      "step": 3804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761903342755755e-06,
+      "loss": 0.8271,
+      "step": 3805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976172324283919e-06,
+      "loss": 0.7041,
+      "step": 3806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976154307565452e-06,
+      "loss": 1.0488,
+      "step": 3807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761362841202977e-06,
+      "loss": 0.9189,
+      "step": 3808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761182539485812e-06,
+      "loss": 0.8799,
+      "step": 3809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761002170504263e-06,
+      "loss": 1.0928,
+      "step": 3810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760821734259577e-06,
+      "loss": 1.0576,
+      "step": 3811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976064123075299e-06,
+      "loss": 1.0557,
+      "step": 3812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760460659985755e-06,
+      "loss": 0.9053,
+      "step": 3813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760280021959115e-06,
+      "loss": 0.918,
+      "step": 3814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760099316674306e-06,
+      "loss": 0.9824,
+      "step": 3815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759918544132577e-06,
+      "loss": 0.9766,
+      "step": 3816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975973770433518e-06,
+      "loss": 1.0664,
+      "step": 3817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759556797283357e-06,
+      "loss": 1.0635,
+      "step": 3818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975937582297835e-06,
+      "loss": 0.8203,
+      "step": 3819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759194781421416e-06,
+      "loss": 0.8423,
+      "step": 3820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759013672613795e-06,
+      "loss": 0.8413,
+      "step": 3821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758832496556735e-06,
+      "loss": 0.7852,
+      "step": 3822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975865125325149e-06,
+      "loss": 0.686,
+      "step": 3823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97584699426993e-06,
+      "loss": 1.1602,
+      "step": 3824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758288564901427e-06,
+      "loss": 0.8428,
+      "step": 3825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975810711985911e-06,
+      "loss": 0.9307,
+      "step": 3826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757925607573604e-06,
+      "loss": 1.0664,
+      "step": 3827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757744028046155e-06,
+      "loss": 1.0195,
+      "step": 3828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757562381278025e-06,
+      "loss": 1.1914,
+      "step": 3829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757380667270454e-06,
+      "loss": 0.8672,
+      "step": 3830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97571988860247e-06,
+      "loss": 0.9707,
+      "step": 3831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757017037542014e-06,
+      "loss": 0.7891,
+      "step": 3832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756835121823652e-06,
+      "loss": 0.9746,
+      "step": 3833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975665313887087e-06,
+      "loss": 1.124,
+      "step": 3834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975647108868491e-06,
+      "loss": 1.0225,
+      "step": 3835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756288971267033e-06,
+      "loss": 1.0117,
+      "step": 3836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756106786618497e-06,
+      "loss": 0.8232,
+      "step": 3837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755924534740558e-06,
+      "loss": 1.0967,
+      "step": 3838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975574221563447e-06,
+      "loss": 0.9443,
+      "step": 3839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975555982930149e-06,
+      "loss": 0.8965,
+      "step": 3840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755377375742866e-06,
+      "loss": 0.8413,
+      "step": 3841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755194854959867e-06,
+      "loss": 0.8203,
+      "step": 3842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755012266953746e-06,
+      "loss": 1.0498,
+      "step": 3843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975482961172576e-06,
+      "loss": 1.0635,
+      "step": 3844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754646889277177e-06,
+      "loss": 1.124,
+      "step": 3845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975446409960924e-06,
+      "loss": 1.1738,
+      "step": 3846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975428124272322e-06,
+      "loss": 1.1123,
+      "step": 3847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754098318620377e-06,
+      "loss": 0.8955,
+      "step": 3848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753915327301966e-06,
+      "loss": 1.0967,
+      "step": 3849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975373226876925e-06,
+      "loss": 0.8691,
+      "step": 3850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753549143023496e-06,
+      "loss": 0.9639,
+      "step": 3851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753365950065957e-06,
+      "loss": 1.1211,
+      "step": 3852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97531826898979e-06,
+      "loss": 1.0068,
+      "step": 3853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975299936252059e-06,
+      "loss": 1.0361,
+      "step": 3854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752815967935285e-06,
+      "loss": 0.9033,
+      "step": 3855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752632506143253e-06,
+      "loss": 1.1719,
+      "step": 3856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752448977145755e-06,
+      "loss": 0.6787,
+      "step": 3857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752265380944057e-06,
+      "loss": 0.9475,
+      "step": 3858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752081717539426e-06,
+      "loss": 0.8574,
+      "step": 3859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751897986933128e-06,
+      "loss": 0.9268,
+      "step": 3860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751714189126425e-06,
+      "loss": 0.8833,
+      "step": 3861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751530324120583e-06,
+      "loss": 0.8853,
+      "step": 3862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751346391916877e-06,
+      "loss": 0.8984,
+      "step": 3863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751162392516565e-06,
+      "loss": 0.9351,
+      "step": 3864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750978325920923e-06,
+      "loss": 0.7559,
+      "step": 3865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750794192131215e-06,
+      "loss": 1.1113,
+      "step": 3866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975060999114871e-06,
+      "loss": 0.7617,
+      "step": 3867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975042572297468e-06,
+      "loss": 0.8345,
+      "step": 3868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975024138761039e-06,
+      "loss": 0.6475,
+      "step": 3869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975005698505711e-06,
+      "loss": 1.0059,
+      "step": 3870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974987251531612e-06,
+      "loss": 1.041,
+      "step": 3871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749687978388683e-06,
+      "loss": 0.6875,
+      "step": 3872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749503374276067e-06,
+      "loss": 0.7793,
+      "step": 3873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749318702979554e-06,
+      "loss": 0.8242,
+      "step": 3874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749133964500416e-06,
+      "loss": 0.918,
+      "step": 3875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748949158839918e-06,
+      "loss": 1.001,
+      "step": 3876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974876428599934e-06,
+      "loss": 1.085,
+      "step": 3877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974857934597995e-06,
+      "loss": 0.8164,
+      "step": 3878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748394338783028e-06,
+      "loss": 0.7246,
+      "step": 3879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974820926440985e-06,
+      "loss": 1.04,
+      "step": 3880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748024122861684e-06,
+      "loss": 0.8857,
+      "step": 3881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974783891413981e-06,
+      "loss": 0.9473,
+      "step": 3882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747653638245505e-06,
+      "loss": 0.9448,
+      "step": 3883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747468295180045e-06,
+      "loss": 0.9111,
+      "step": 3884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747282884944708e-06,
+      "loss": 0.9209,
+      "step": 3885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747097407540775e-06,
+      "loss": 1.1074,
+      "step": 3886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746911862969515e-06,
+      "loss": 0.8193,
+      "step": 3887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746726251232213e-06,
+      "loss": 0.8555,
+      "step": 3888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974654057233015e-06,
+      "loss": 0.9639,
+      "step": 3889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97463548262646e-06,
+      "loss": 0.9346,
+      "step": 3890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974616901303684e-06,
+      "loss": 0.8291,
+      "step": 3891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974598313264816e-06,
+      "loss": 0.8179,
+      "step": 3892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745797185099837e-06,
+      "loss": 1.0742,
+      "step": 3893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745611170393155e-06,
+      "loss": 1.0947,
+      "step": 3894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974542508852939e-06,
+      "loss": 0.7578,
+      "step": 3895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745238939509826e-06,
+      "loss": 0.6943,
+      "step": 3896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745052723335742e-06,
+      "loss": 0.5601,
+      "step": 3897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744866440008434e-06,
+      "loss": 0.8633,
+      "step": 3898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974468008952918e-06,
+      "loss": 1.1123,
+      "step": 3899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744493671899253e-06,
+      "loss": 0.8223,
+      "step": 3900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974430718711995e-06,
+      "loss": 0.8613,
+      "step": 3901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744120635192557e-06,
+      "loss": 0.585,
+      "step": 3902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974393401611835e-06,
+      "loss": 0.7217,
+      "step": 3903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974374732989862e-06,
+      "loss": 0.9531,
+      "step": 3904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974356057653466e-06,
+      "loss": 0.8887,
+      "step": 3905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974337375602774e-06,
+      "loss": 1.1221,
+      "step": 3906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743186868379164e-06,
+      "loss": 1.1699,
+      "step": 3907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742999913590212e-06,
+      "loss": 0.8867,
+      "step": 3908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742812891662176e-06,
+      "loss": 0.7773,
+      "step": 3909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742625802596343e-06,
+      "loss": 1.0039,
+      "step": 3910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742438646394e-06,
+      "loss": 1.0684,
+      "step": 3911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974225142305644e-06,
+      "loss": 0.729,
+      "step": 3912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974206413258495e-06,
+      "loss": 0.5249,
+      "step": 3913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741876774980827e-06,
+      "loss": 1.0332,
+      "step": 3914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741689350245353e-06,
+      "loss": 0.9268,
+      "step": 3915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741501858379825e-06,
+      "loss": 0.7861,
+      "step": 3916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741314299385534e-06,
+      "loss": 0.9463,
+      "step": 3917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741126673263776e-06,
+      "loss": 0.9219,
+      "step": 3918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740938980015835e-06,
+      "loss": 0.9414,
+      "step": 3919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740751219643014e-06,
+      "loss": 0.8174,
+      "step": 3920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97405633921466e-06,
+      "loss": 0.8809,
+      "step": 3921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740375497527893e-06,
+      "loss": 0.7959,
+      "step": 3922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740187535788185e-06,
+      "loss": 1.0479,
+      "step": 3923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973999950692877e-06,
+      "loss": 0.9932,
+      "step": 3924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973981141095094e-06,
+      "loss": 0.8838,
+      "step": 3925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739623247856e-06,
+      "loss": 0.8867,
+      "step": 3926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739435017645243e-06,
+      "loss": 0.9346,
+      "step": 3927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739246720319967e-06,
+      "loss": 1.1387,
+      "step": 3928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973905835588147e-06,
+      "loss": 1.126,
+      "step": 3929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973886992433104e-06,
+      "loss": 1.0039,
+      "step": 3930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973868142566999e-06,
+      "loss": 0.8716,
+      "step": 3931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738492859899613e-06,
+      "loss": 1.0859,
+      "step": 3932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973830422702121e-06,
+      "loss": 0.8306,
+      "step": 3933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973811552703607e-06,
+      "loss": 1.1133,
+      "step": 3934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737926759945508e-06,
+      "loss": 0.6754,
+      "step": 3935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737737925750823e-06,
+      "loss": 0.9805,
+      "step": 3936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737549024453307e-06,
+      "loss": 0.8906,
+      "step": 3937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973736005605427e-06,
+      "loss": 0.791,
+      "step": 3938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973717102055501e-06,
+      "loss": 0.8301,
+      "step": 3939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973698191795683e-06,
+      "loss": 0.8838,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973679274826104e-06,
+      "loss": 0.8711,
+      "step": 3941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973660351146893e-06,
+      "loss": 0.8984,
+      "step": 3942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736414207581813e-06,
+      "loss": 1.0605,
+      "step": 3943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973622483660099e-06,
+      "loss": 0.9648,
+      "step": 3944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736035398527774e-06,
+      "loss": 0.9365,
+      "step": 3945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973584589336346e-06,
+      "loss": 0.8994,
+      "step": 3946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735656321109362e-06,
+      "loss": 1.0713,
+      "step": 3947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735466681766785e-06,
+      "loss": 1.0352,
+      "step": 3948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973527697533703e-06,
+      "loss": 0.7363,
+      "step": 3949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735087201821406e-06,
+      "loss": 0.875,
+      "step": 3950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734897361221225e-06,
+      "loss": 0.9492,
+      "step": 3951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734707453537793e-06,
+      "loss": 0.959,
+      "step": 3952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734517478772418e-06,
+      "loss": 0.9004,
+      "step": 3953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734327436926408e-06,
+      "loss": 0.7739,
+      "step": 3954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734137328001076e-06,
+      "loss": 1.2988,
+      "step": 3955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733947151997723e-06,
+      "loss": 0.8765,
+      "step": 3956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733756908917674e-06,
+      "loss": 0.8779,
+      "step": 3957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733566598762234e-06,
+      "loss": 0.9619,
+      "step": 3958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733376221532707e-06,
+      "loss": 1.0342,
+      "step": 3959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733185777230415e-06,
+      "loss": 1.0391,
+      "step": 3960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973299526585667e-06,
+      "loss": 0.6128,
+      "step": 3961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732804687412776e-06,
+      "loss": 0.8564,
+      "step": 3962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973261404190005e-06,
+      "loss": 1.1143,
+      "step": 3963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973242332931981e-06,
+      "loss": 0.9326,
+      "step": 3964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732232549673365e-06,
+      "loss": 1.0537,
+      "step": 3965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732041702962037e-06,
+      "loss": 0.8398,
+      "step": 3966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731850789187134e-06,
+      "loss": 0.9482,
+      "step": 3967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973165980834997e-06,
+      "loss": 0.8462,
+      "step": 3968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731468760451867e-06,
+      "loss": 0.6079,
+      "step": 3969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731277645494137e-06,
+      "loss": 0.6143,
+      "step": 3970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731086463478104e-06,
+      "loss": 0.9658,
+      "step": 3971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730895214405077e-06,
+      "loss": 0.915,
+      "step": 3972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973070389827638e-06,
+      "loss": 0.9629,
+      "step": 3973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973051251509333e-06,
+      "loss": 0.8066,
+      "step": 3974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973032106485724e-06,
+      "loss": 0.9629,
+      "step": 3975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973012954756944e-06,
+      "loss": 1.0137,
+      "step": 3976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972993796323124e-06,
+      "loss": 0.8545,
+      "step": 3977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972974631184396e-06,
+      "loss": 0.8154,
+      "step": 3978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972955459340893e-06,
+      "loss": 1.0625,
+      "step": 3979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729362807927467e-06,
+      "loss": 0.9111,
+      "step": 3980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972917095540089e-06,
+      "loss": 1.0439,
+      "step": 3981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972897903583052e-06,
+      "loss": 0.8623,
+      "step": 3982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728787049217684e-06,
+      "loss": 0.876,
+      "step": 3983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97285949955637e-06,
+      "loss": 0.96,
+      "step": 3984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97284028748699e-06,
+      "loss": 0.7756,
+      "step": 3985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728210687137597e-06,
+      "loss": 0.9248,
+      "step": 3986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972801843236812e-06,
+      "loss": 0.8652,
+      "step": 3987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727826110562804e-06,
+      "loss": 1.0176,
+      "step": 3988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727633721722955e-06,
+      "loss": 1.0732,
+      "step": 3989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727441265849914e-06,
+      "loss": 0.9336,
+      "step": 3990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727248742945e-06,
+      "loss": 0.9473,
+      "step": 3991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972705615300954e-06,
+      "loss": 1.0498,
+      "step": 3992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726863496044864e-06,
+      "loss": 0.8486,
+      "step": 3993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97266707720523e-06,
+      "loss": 1.0439,
+      "step": 3994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726477981033168e-06,
+      "loss": 1.0156,
+      "step": 3995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726285122988807e-06,
+      "loss": 0.9375,
+      "step": 3996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726092197920542e-06,
+      "loss": 0.9775,
+      "step": 3997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725899205829703e-06,
+      "loss": 0.9082,
+      "step": 3998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725706146717614e-06,
+      "loss": 0.9697,
+      "step": 3999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972551302058562e-06,
+      "loss": 0.9521,
+      "step": 4000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725319827435034e-06,
+      "loss": 1.0049,
+      "step": 4001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725126567267195e-06,
+      "loss": 0.999,
+      "step": 4002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724933240083442e-06,
+      "loss": 0.8613,
+      "step": 4003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97247398458851e-06,
+      "loss": 1.0322,
+      "step": 4004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724546384673493e-06,
+      "loss": 0.8623,
+      "step": 4005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724352856449973e-06,
+      "loss": 1.0869,
+      "step": 4006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972415926121586e-06,
+      "loss": 1.0791,
+      "step": 4007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972396559897249e-06,
+      "loss": 0.8252,
+      "step": 4008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723771869721206e-06,
+      "loss": 0.9951,
+      "step": 4009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972357807346333e-06,
+      "loss": 0.9033,
+      "step": 4010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972338421020021e-06,
+      "loss": 0.9199,
+      "step": 4011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972319027993317e-06,
+      "loss": 0.8018,
+      "step": 4012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972299628266356e-06,
+      "loss": 1.0283,
+      "step": 4013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97228022183927e-06,
+      "loss": 0.8584,
+      "step": 4014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722608087121945e-06,
+      "loss": 0.8135,
+      "step": 4015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972241388885262e-06,
+      "loss": 1.1162,
+      "step": 4016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972221962358607e-06,
+      "loss": 0.7607,
+      "step": 4017
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722025291323626e-06,
+      "loss": 1.0449,
+      "step": 4018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721830892066637e-06,
+      "loss": 0.7881,
+      "step": 4019
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972163642581643e-06,
+      "loss": 0.7793,
+      "step": 4020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972144189257436e-06,
+      "loss": 0.8369,
+      "step": 4021
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972124729234176e-06,
+      "loss": 0.915,
+      "step": 4022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721052625119975e-06,
+      "loss": 0.9043,
+      "step": 4023
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720857890910338e-06,
+      "loss": 0.9961,
+      "step": 4024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972066308971419e-06,
+      "loss": 0.9932,
+      "step": 4025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972046822153289e-06,
+      "loss": 0.8838,
+      "step": 4026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720273286367765e-06,
+      "loss": 0.6909,
+      "step": 4027
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720078284220167e-06,
+      "loss": 1.1221,
+      "step": 4028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719883215091433e-06,
+      "loss": 0.9902,
+      "step": 4029
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971968807898291e-06,
+      "loss": 0.7812,
+      "step": 4030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971949287589594e-06,
+      "loss": 1.0889,
+      "step": 4031
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719297605831878e-06,
+      "loss": 0.8228,
+      "step": 4032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971910226879206e-06,
+      "loss": 0.9961,
+      "step": 4033
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718906864777838e-06,
+      "loss": 1.0088,
+      "step": 4034
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971871139379055e-06,
+      "loss": 0.8945,
+      "step": 4035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971851585583155e-06,
+      "loss": 0.8877,
+      "step": 4036
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718320250902183e-06,
+      "loss": 0.9531,
+      "step": 4037
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97181245790038e-06,
+      "loss": 0.873,
+      "step": 4038
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717928840137744e-06,
+      "loss": 0.8281,
+      "step": 4039
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971773303430537e-06,
+      "loss": 0.6963,
+      "step": 4040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971753716150802e-06,
+      "loss": 1.0762,
+      "step": 4041
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717341221747056e-06,
+      "loss": 1.0059,
+      "step": 4042
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717145215023818e-06,
+      "loss": 0.8906,
+      "step": 4043
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971694914133965e-06,
+      "loss": 1.123,
+      "step": 4044
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971675300069592e-06,
+      "loss": 1.1104,
+      "step": 4045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716556793093974e-06,
+      "loss": 1.0596,
+      "step": 4046
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716360518535156e-06,
+      "loss": 0.833,
+      "step": 4047
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716164177020828e-06,
+      "loss": 1.042,
+      "step": 4048
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715967768552336e-06,
+      "loss": 0.915,
+      "step": 4049
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715771293131034e-06,
+      "loss": 0.9639,
+      "step": 4050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715574750758284e-06,
+      "loss": 0.9453,
+      "step": 4051
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715378141435435e-06,
+      "loss": 0.9893,
+      "step": 4052
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971518146516384e-06,
+      "loss": 0.9365,
+      "step": 4053
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714984721944856e-06,
+      "loss": 0.9258,
+      "step": 4054
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714787911779844e-06,
+      "loss": 1.1602,
+      "step": 4055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714591034670147e-06,
+      "loss": 1.1182,
+      "step": 4056
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714394090617136e-06,
+      "loss": 0.8574,
+      "step": 4057
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714197079622158e-06,
+      "loss": 0.9287,
+      "step": 4058
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714000001686577e-06,
+      "loss": 0.9258,
+      "step": 4059
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713802856811744e-06,
+      "loss": 0.9521,
+      "step": 4060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971360564499903e-06,
+      "loss": 0.874,
+      "step": 4061
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971340836624978e-06,
+      "loss": 0.793,
+      "step": 4062
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713211020565362e-06,
+      "loss": 1.1494,
+      "step": 4063
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971301360794713e-06,
+      "loss": 0.79,
+      "step": 4064
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712816128396454e-06,
+      "loss": 0.8965,
+      "step": 4065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712618581914684e-06,
+      "loss": 1.0332,
+      "step": 4066
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971242096850319e-06,
+      "loss": 0.8359,
+      "step": 4067
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712223288163326e-06,
+      "loss": 0.8838,
+      "step": 4068
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971202554089646e-06,
+      "loss": 0.9189,
+      "step": 4069
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971182772670395e-06,
+      "loss": 0.7852,
+      "step": 4070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711629845587163e-06,
+      "loss": 0.8828,
+      "step": 4071
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971143189754746e-06,
+      "loss": 0.8164,
+      "step": 4072
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971123388258621e-06,
+      "loss": 0.8032,
+      "step": 4073
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711035800704773e-06,
+      "loss": 0.8345,
+      "step": 4074
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710837651904515e-06,
+      "loss": 0.9355,
+      "step": 4075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97106394361868e-06,
+      "loss": 1.042,
+      "step": 4076
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710441153552993e-06,
+      "loss": 1.0791,
+      "step": 4077
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971024280400447e-06,
+      "loss": 0.7861,
+      "step": 4078
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710044387542583e-06,
+      "loss": 0.8501,
+      "step": 4079
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970984590416871e-06,
+      "loss": 0.5171,
+      "step": 4080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709647353884217e-06,
+      "loss": 0.9697,
+      "step": 4081
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970944873669047e-06,
+      "loss": 0.7319,
+      "step": 4082
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709250052588838e-06,
+      "loss": 0.7734,
+      "step": 4083
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970905130158069e-06,
+      "loss": 0.8809,
+      "step": 4084
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708852483667393e-06,
+      "loss": 0.8789,
+      "step": 4085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970865359885033e-06,
+      "loss": 0.9688,
+      "step": 4086
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970845464713085e-06,
+      "loss": 0.7573,
+      "step": 4087
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708255628510344e-06,
+      "loss": 1.1367,
+      "step": 4088
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970805654299017e-06,
+      "loss": 1.0801,
+      "step": 4089
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970785739057171e-06,
+      "loss": 1.0654,
+      "step": 4090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707658171256326e-06,
+      "loss": 0.8672,
+      "step": 4091
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707458885045395e-06,
+      "loss": 1.0713,
+      "step": 4092
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707259531940295e-06,
+      "loss": 0.9482,
+      "step": 4093
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97070601119424e-06,
+      "loss": 0.8809,
+      "step": 4094
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9706860625053078e-06,
+      "loss": 1.1396,
+      "step": 4095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97066610712737e-06,
+      "loss": 0.9102,
+      "step": 4096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970646145060565e-06,
+      "loss": 0.8091,
+      "step": 4097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970626176305031e-06,
+      "loss": 0.8896,
+      "step": 4098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970606200860904e-06,
+      "loss": 0.8682,
+      "step": 4099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970586218728322e-06,
+      "loss": 0.7229,
+      "step": 4100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705662299074235e-06,
+      "loss": 0.9893,
+      "step": 4101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705462343983454e-06,
+      "loss": 1.0615,
+      "step": 4102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970526232201226e-06,
+      "loss": 1.0566,
+      "step": 4103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970506223316203e-06,
+      "loss": 0.8394,
+      "step": 4104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704862077434146e-06,
+      "loss": 0.8198,
+      "step": 4105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970466185482998e-06,
+      "loss": 0.7607,
+      "step": 4106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704461565350923e-06,
+      "loss": 0.7827,
+      "step": 4107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704261208998337e-06,
+      "loss": 0.874,
+      "step": 4108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704060785773623e-06,
+      "loss": 0.9248,
+      "step": 4109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703860295678147e-06,
+      "loss": 1.0449,
+      "step": 4110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97036597387133e-06,
+      "loss": 1.0264,
+      "step": 4111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703459114880458e-06,
+      "loss": 0.8535,
+      "step": 4112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970325842418101e-06,
+      "loss": 0.9912,
+      "step": 4113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970305766661633e-06,
+      "loss": 0.917,
+      "step": 4114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702856842187807e-06,
+      "loss": 0.8306,
+      "step": 4115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970265595089683e-06,
+      "loss": 0.9561,
+      "step": 4116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970245499274477e-06,
+      "loss": 1.0127,
+      "step": 4117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970225396773302e-06,
+      "loss": 0.9302,
+      "step": 4118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702052875862967e-06,
+      "loss": 0.6108,
+      "step": 4119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701851717135997e-06,
+      "loss": 1.166,
+      "step": 4120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970165049155349e-06,
+      "loss": 0.8994,
+      "step": 4121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970144919911684e-06,
+      "loss": 0.9668,
+      "step": 4122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701247839827427e-06,
+      "loss": 0.875,
+      "step": 4123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970104641368664e-06,
+      "loss": 0.5879,
+      "step": 4124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970084492069587e-06,
+      "loss": 1.083,
+      "step": 4125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970064336085651e-06,
+      "loss": 0.9561,
+      "step": 4126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700441734169937e-06,
+      "loss": 0.8184,
+      "step": 4127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970024004063755e-06,
+      "loss": 0.8857,
+      "step": 4128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970003828026073e-06,
+      "loss": 0.8442,
+      "step": 4129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969983645304088e-06,
+      "loss": 0.8438,
+      "step": 4130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969963455897938e-06,
+      "loss": 0.9004,
+      "step": 4131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699432598077627e-06,
+      "loss": 0.9053,
+      "step": 4132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969923057033701e-06,
+      "loss": 1.0908,
+      "step": 4133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969902847575892e-06,
+      "loss": 0.8467,
+      "step": 4134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698826314344756e-06,
+      "loss": 1.0869,
+      "step": 4135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96986240860959e-06,
+      "loss": 1.0703,
+      "step": 4136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969842179101376e-06,
+      "loss": 1.1797,
+      "step": 4137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698219429099713e-06,
+      "loss": 0.8652,
+      "step": 4138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969801700035517e-06,
+      "loss": 0.7227,
+      "step": 4139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697814504781514e-06,
+      "loss": 0.8379,
+      "step": 4140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969761194238015e-06,
+      "loss": 0.8564,
+      "step": 4141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697409313152465e-06,
+      "loss": 0.9795,
+      "step": 4142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969720661709986e-06,
+      "loss": 0.8955,
+      "step": 4143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969700385422373e-06,
+      "loss": 1.0068,
+      "step": 4144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696801024525472e-06,
+      "loss": 1.0479,
+      "step": 4145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969659812800649e-06,
+      "loss": 1.0615,
+      "step": 4146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969639516466817e-06,
+      "loss": 0.9736,
+      "step": 4147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696192134511923e-06,
+      "loss": 0.8096,
+      "step": 4148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969598903753914e-06,
+      "loss": 0.9844,
+      "step": 4149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969578587375123e-06,
+      "loss": 0.9385,
+      "step": 4150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969558264314958e-06,
+      "loss": 1.0068,
+      "step": 4151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695379345735596e-06,
+      "loss": 0.8867,
+      "step": 4152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695175981510684e-06,
+      "loss": 0.9043,
+      "step": 4153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969497255047624e-06,
+      "loss": 0.8574,
+      "step": 4154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694769052633666e-06,
+      "loss": 1.0039,
+      "step": 4155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694565487984364e-06,
+      "loss": 0.9297,
+      "step": 4156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969436185652974e-06,
+      "loss": 0.9844,
+      "step": 4157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694158158271192e-06,
+      "loss": 0.96,
+      "step": 4158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969395439321013e-06,
+      "loss": 1.0469,
+      "step": 4159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693750561347954e-06,
+      "loss": 0.9756,
+      "step": 4160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969354666268607e-06,
+      "loss": 1.0986,
+      "step": 4161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969334269722588e-06,
+      "loss": 0.7256,
+      "step": 4162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693138664968797e-06,
+      "loss": 0.9614,
+      "step": 4163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692934565916215e-06,
+      "loss": 0.9912,
+      "step": 4164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969273040006956e-06,
+      "loss": 1.1162,
+      "step": 4165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692526167430215e-06,
+      "loss": 0.8979,
+      "step": 4166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96923218679996e-06,
+      "loss": 0.877,
+      "step": 4167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692117501779127e-06,
+      "loss": 0.9385,
+      "step": 4168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691913068770198e-06,
+      "loss": 0.731,
+      "step": 4169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691708568974224e-06,
+      "loss": 1.1953,
+      "step": 4170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691504002392608e-06,
+      "loss": 1.0586,
+      "step": 4171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691299369026767e-06,
+      "loss": 0.9844,
+      "step": 4172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969109466887811e-06,
+      "loss": 0.9229,
+      "step": 4173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969088990194805e-06,
+      "loss": 1.0283,
+      "step": 4174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690685068237986e-06,
+      "loss": 0.8535,
+      "step": 4175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690480167749346e-06,
+      "loss": 0.875,
+      "step": 4176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969027520048353e-06,
+      "loss": 0.9121,
+      "step": 4177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690070166441954e-06,
+      "loss": 0.9619,
+      "step": 4178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968986506562603e-06,
+      "loss": 0.9756,
+      "step": 4179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689659898037173e-06,
+      "loss": 0.9092,
+      "step": 4180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96894546636768e-06,
+      "loss": 0.7373,
+      "step": 4181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968924936254632e-06,
+      "loss": 0.9287,
+      "step": 4182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689043994647148e-06,
+      "loss": 0.9824,
+      "step": 4183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688838559980702e-06,
+      "loss": 0.9951,
+      "step": 4184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688633058548397e-06,
+      "loss": 0.9268,
+      "step": 4185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968842749035165e-06,
+      "loss": 0.9668,
+      "step": 4186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688221855391873e-06,
+      "loss": 0.8438,
+      "step": 4187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688016153670483e-06,
+      "loss": 0.7729,
+      "step": 4188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687810385188904e-06,
+      "loss": 0.7739,
+      "step": 4189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968760454994855e-06,
+      "loss": 0.8457,
+      "step": 4190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968739864795084e-06,
+      "loss": 0.6318,
+      "step": 4191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687192679197194e-06,
+      "loss": 0.7949,
+      "step": 4192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686986643689033e-06,
+      "loss": 0.9941,
+      "step": 4193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968678054142777e-06,
+      "loss": 0.7578,
+      "step": 4194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686574372414827e-06,
+      "loss": 0.7505,
+      "step": 4195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968636813665163e-06,
+      "loss": 1.0195,
+      "step": 4196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686161834139594e-06,
+      "loss": 0.9824,
+      "step": 4197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685955464880145e-06,
+      "loss": 1.0107,
+      "step": 4198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685749028874706e-06,
+      "loss": 0.9053,
+      "step": 4199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685542526124698e-06,
+      "loss": 0.7881,
+      "step": 4200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968533595663154e-06,
+      "loss": 0.7734,
+      "step": 4201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968512932039666e-06,
+      "loss": 1.0234,
+      "step": 4202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968492261742148e-06,
+      "loss": 0.9424,
+      "step": 4203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684715847707425e-06,
+      "loss": 0.9141,
+      "step": 4204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968450901125592e-06,
+      "loss": 1.0635,
+      "step": 4205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684302108068393e-06,
+      "loss": 1.0596,
+      "step": 4206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968409513814626e-06,
+      "loss": 0.8389,
+      "step": 4207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683888101490964e-06,
+      "loss": 1.0693,
+      "step": 4208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683680998103912e-06,
+      "loss": 0.8076,
+      "step": 4209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683473827986547e-06,
+      "loss": 0.9678,
+      "step": 4210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968326659114029e-06,
+      "loss": 1.0166,
+      "step": 4211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968305928756657e-06,
+      "loss": 0.877,
+      "step": 4212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682851917266815e-06,
+      "loss": 0.9639,
+      "step": 4213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968264448024245e-06,
+      "loss": 0.9414,
+      "step": 4214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968243697649492e-06,
+      "loss": 0.9199,
+      "step": 4215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682229406025634e-06,
+      "loss": 0.8164,
+      "step": 4216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968202176883603e-06,
+      "loss": 0.9258,
+      "step": 4217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968181406492755e-06,
+      "loss": 0.8662,
+      "step": 4218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968160629430161e-06,
+      "loss": 0.894,
+      "step": 4219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968139845695965e-06,
+      "loss": 1.1016,
+      "step": 4220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681190552903096e-06,
+      "loss": 0.8984,
+      "step": 4221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968098258213339e-06,
+      "loss": 1.0225,
+      "step": 4222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680774544651956e-06,
+      "loss": 0.7485,
+      "step": 4223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680566440460237e-06,
+      "loss": 0.7822,
+      "step": 4224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968035826955966e-06,
+      "loss": 0.8774,
+      "step": 4225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680150031951657e-06,
+      "loss": 1.0625,
+      "step": 4226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679941727637667e-06,
+      "loss": 1.0195,
+      "step": 4227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967973335661913e-06,
+      "loss": 0.8022,
+      "step": 4228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679524918897473e-06,
+      "loss": 0.606,
+      "step": 4229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967931641447414e-06,
+      "loss": 1.0977,
+      "step": 4230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967910784335056e-06,
+      "loss": 0.9805,
+      "step": 4231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967889920552818e-06,
+      "loss": 0.9307,
+      "step": 4232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967869050100843e-06,
+      "loss": 0.9131,
+      "step": 4233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967848172979275e-06,
+      "loss": 0.999,
+      "step": 4234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967827289188258e-06,
+      "loss": 0.9873,
+      "step": 4235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967806398727936e-06,
+      "loss": 0.8105,
+      "step": 4236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677855015984528e-06,
+      "loss": 0.9258,
+      "step": 4237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967764597799952e-06,
+      "loss": 0.8042,
+      "step": 4238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967743687332578e-06,
+      "loss": 0.959,
+      "step": 4239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677227701964757e-06,
+      "loss": 0.9199,
+      "step": 4240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967701846391788e-06,
+      "loss": 0.8174,
+      "step": 4241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676809159186595e-06,
+      "loss": 1.0625,
+      "step": 4242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676599787772346e-06,
+      "loss": 0.9053,
+      "step": 4243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676390349676576e-06,
+      "loss": 1.0273,
+      "step": 4244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676180844900725e-06,
+      "loss": 0.8516,
+      "step": 4245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675971273446237e-06,
+      "loss": 0.8772,
+      "step": 4246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967576163531456e-06,
+      "loss": 1.1572,
+      "step": 4247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967555193050714e-06,
+      "loss": 0.917,
+      "step": 4248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967534215902541e-06,
+      "loss": 1.0107,
+      "step": 4249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967513232087083e-06,
+      "loss": 1.0303,
+      "step": 4250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967492241604484e-06,
+      "loss": 0.9229,
+      "step": 4251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967471244454889e-06,
+      "loss": 1.0859,
+      "step": 4252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674502406384415e-06,
+      "loss": 1.0605,
+      "step": 4253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674292301552878e-06,
+      "loss": 0.9468,
+      "step": 4254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674082130055714e-06,
+      "loss": 1.291,
+      "step": 4255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967387189189438e-06,
+      "loss": 1.082,
+      "step": 4256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673661587070326e-06,
+      "loss": 1.0449,
+      "step": 4257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967345121558499e-06,
+      "loss": 0.9941,
+      "step": 4258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967324077743983e-06,
+      "loss": 1.1279,
+      "step": 4259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673030272636295e-06,
+      "loss": 0.7354,
+      "step": 4260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672819701175837e-06,
+      "loss": 0.9336,
+      "step": 4261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672609063059907e-06,
+      "loss": 0.9531,
+      "step": 4262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967239835828995e-06,
+      "loss": 0.7742,
+      "step": 4263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672187586867426e-06,
+      "loss": 0.752,
+      "step": 4264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671976748793784e-06,
+      "loss": 0.8975,
+      "step": 4265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671765844070478e-06,
+      "loss": 0.9912,
+      "step": 4266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967155487269896e-06,
+      "loss": 0.7002,
+      "step": 4267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967134383468068e-06,
+      "loss": 0.9795,
+      "step": 4268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96711327300171e-06,
+      "loss": 1.043,
+      "step": 4269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670921558709673e-06,
+      "loss": 0.7002,
+      "step": 4270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967071032075985e-06,
+      "loss": 1.1182,
+      "step": 4271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967049901616909e-06,
+      "loss": 0.9805,
+      "step": 4272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967028764493885e-06,
+      "loss": 1.2305,
+      "step": 4273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670076207070583e-06,
+      "loss": 0.8701,
+      "step": 4274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669864702565746e-06,
+      "loss": 1.0898,
+      "step": 4275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96696531314258e-06,
+      "loss": 0.7871,
+      "step": 4276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669441493652204e-06,
+      "loss": 0.9797,
+      "step": 4277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966922978924641e-06,
+      "loss": 0.9766,
+      "step": 4278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966901801820988e-06,
+      "loss": 0.8799,
+      "step": 4279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668806180544074e-06,
+      "loss": 1.0547,
+      "step": 4280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668594276250456e-06,
+      "loss": 0.7957,
+      "step": 4281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966838230533048e-06,
+      "loss": 0.9912,
+      "step": 4282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668170267785606e-06,
+      "loss": 1.0234,
+      "step": 4283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667958163617297e-06,
+      "loss": 0.9336,
+      "step": 4284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667745992827017e-06,
+      "loss": 0.918,
+      "step": 4285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667533755416224e-06,
+      "loss": 0.7212,
+      "step": 4286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667321451386387e-06,
+      "loss": 0.5596,
+      "step": 4287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667109080738963e-06,
+      "loss": 0.9756,
+      "step": 4288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966689664347542e-06,
+      "loss": 0.8882,
+      "step": 4289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966668413959721e-06,
+      "loss": 1.0371,
+      "step": 4290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966647156910582e-06,
+      "loss": 0.8398,
+      "step": 4291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666258932002692e-06,
+      "loss": 0.9688,
+      "step": 4292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666046228289306e-06,
+      "loss": 0.9736,
+      "step": 4293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665833457967115e-06,
+      "loss": 1.0684,
+      "step": 4294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96656206210376e-06,
+      "loss": 0.7754,
+      "step": 4295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665407717502217e-06,
+      "loss": 0.8623,
+      "step": 4296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966519474736244e-06,
+      "loss": 1.0107,
+      "step": 4297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664981710619727e-06,
+      "loss": 1.0234,
+      "step": 4298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664768607275556e-06,
+      "loss": 1.043,
+      "step": 4299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966455543733139e-06,
+      "loss": 0.8408,
+      "step": 4300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664342200788697e-06,
+      "loss": 0.9243,
+      "step": 4301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664128897648953e-06,
+      "loss": 0.9922,
+      "step": 4302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663915527913623e-06,
+      "loss": 0.9316,
+      "step": 4303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966370209158418e-06,
+      "loss": 1.0244,
+      "step": 4304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966348858866209e-06,
+      "loss": 0.9316,
+      "step": 4305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663275019148827e-06,
+      "loss": 1.0107,
+      "step": 4306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663061383045864e-06,
+      "loss": 1.0469,
+      "step": 4307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966284768035467e-06,
+      "loss": 0.7871,
+      "step": 4308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662633911076723e-06,
+      "loss": 0.9307,
+      "step": 4309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966242007521349e-06,
+      "loss": 1.0068,
+      "step": 4310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966220617276645e-06,
+      "loss": 0.9229,
+      "step": 4311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966199220373707e-06,
+      "loss": 0.7954,
+      "step": 4312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966177816812683e-06,
+      "loss": 1.0498,
+      "step": 4313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661564065937205e-06,
+      "loss": 0.8594,
+      "step": 4314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966134989716967e-06,
+      "loss": 0.832,
+      "step": 4315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96611356618257e-06,
+      "loss": 1.0068,
+      "step": 4316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966092135990677e-06,
+      "loss": 0.981,
+      "step": 4317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966070699141436e-06,
+      "loss": 0.915,
+      "step": 4318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660492556349943e-06,
+      "loss": 0.7178,
+      "step": 4319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660278054714995e-06,
+      "loss": 0.9521,
+      "step": 4320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660063486511003e-06,
+      "loss": 0.8682,
+      "step": 4321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965984885173944e-06,
+      "loss": 0.7876,
+      "step": 4322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659634150401784e-06,
+      "loss": 1.0752,
+      "step": 4323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659419382499518e-06,
+      "loss": 0.873,
+      "step": 4324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965920454803412e-06,
+      "loss": 0.9727,
+      "step": 4325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965898964700707e-06,
+      "loss": 1.0068,
+      "step": 4326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965877467941985e-06,
+      "loss": 0.8013,
+      "step": 4327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658559645273937e-06,
+      "loss": 0.7939,
+      "step": 4328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965834454457082e-06,
+      "loss": 0.8613,
+      "step": 4329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658129377311975e-06,
+      "loss": 0.8359,
+      "step": 4330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657914143498893e-06,
+      "loss": 0.75,
+      "step": 4331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657698843133047e-06,
+      "loss": 0.833,
+      "step": 4332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965748347621593e-06,
+      "loss": 0.7847,
+      "step": 4333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657268042749015e-06,
+      "loss": 0.9785,
+      "step": 4334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657052542733793e-06,
+      "loss": 0.9102,
+      "step": 4335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656836976171754e-06,
+      "loss": 1.335,
+      "step": 4336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656621343064376e-06,
+      "loss": 0.9014,
+      "step": 4337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965640564341315e-06,
+      "loss": 0.9253,
+      "step": 4338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656189877219556e-06,
+      "loss": 0.9023,
+      "step": 4339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965597404448509e-06,
+      "loss": 0.8857,
+      "step": 4340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655758145211226e-06,
+      "loss": 0.7158,
+      "step": 4341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655542179399467e-06,
+      "loss": 1.0186,
+      "step": 4342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655326147051287e-06,
+      "loss": 1.1797,
+      "step": 4343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655110048168186e-06,
+      "loss": 0.8462,
+      "step": 4344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654893882751647e-06,
+      "loss": 1.0371,
+      "step": 4345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965467765080316e-06,
+      "loss": 0.749,
+      "step": 4346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965446135232422e-06,
+      "loss": 0.8506,
+      "step": 4347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654244987316315e-06,
+      "loss": 1.0898,
+      "step": 4348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654028555780934e-06,
+      "loss": 0.7068,
+      "step": 4349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653812057719564e-06,
+      "loss": 0.9102,
+      "step": 4350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965359549313371e-06,
+      "loss": 0.9521,
+      "step": 4351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965337886202485e-06,
+      "loss": 0.9463,
+      "step": 4352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653162164394487e-06,
+      "loss": 0.7939,
+      "step": 4353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965294540024411e-06,
+      "loss": 0.915,
+      "step": 4354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965272856957521e-06,
+      "loss": 0.9463,
+      "step": 4355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652511672389286e-06,
+      "loss": 1.0742,
+      "step": 4356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652294708687833e-06,
+      "loss": 0.8271,
+      "step": 4357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652077678472347e-06,
+      "loss": 1.1113,
+      "step": 4358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651860581744315e-06,
+      "loss": 1.0225,
+      "step": 4359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965164341850524e-06,
+      "loss": 0.8398,
+      "step": 4360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965142618875662e-06,
+      "loss": 0.8193,
+      "step": 4361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651208892499945e-06,
+      "loss": 1.0479,
+      "step": 4362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650991529736723e-06,
+      "loss": 0.8623,
+      "step": 4363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965077410046844e-06,
+      "loss": 0.8809,
+      "step": 4364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96505566046966e-06,
+      "loss": 0.7996,
+      "step": 4365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650339042422705e-06,
+      "loss": 0.9121,
+      "step": 4366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965012141364825e-06,
+      "loss": 0.6411,
+      "step": 4367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649903718374735e-06,
+      "loss": 0.8315,
+      "step": 4368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964968595660366e-06,
+      "loss": 1.0449,
+      "step": 4369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964946812833653e-06,
+      "loss": 0.9277,
+      "step": 4370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649250233574833e-06,
+      "loss": 1.0576,
+      "step": 4371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649032272320087e-06,
+      "loss": 1.083,
+      "step": 4372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648814244573783e-06,
+      "loss": 1.0371,
+      "step": 4373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964859615033743e-06,
+      "loss": 0.9893,
+      "step": 4374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648377989612527e-06,
+      "loss": 0.9736,
+      "step": 4375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648159762400578e-06,
+      "loss": 0.9004,
+      "step": 4376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647941468703085e-06,
+      "loss": 1.0762,
+      "step": 4377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647723108521557e-06,
+      "loss": 1.0322,
+      "step": 4378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96475046818575e-06,
+      "loss": 0.9692,
+      "step": 4379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964728618871241e-06,
+      "loss": 0.8643,
+      "step": 4380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96470676290878e-06,
+      "loss": 1.0488,
+      "step": 4381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964684900298517e-06,
+      "loss": 1.001,
+      "step": 4382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646630310406036e-06,
+      "loss": 0.9492,
+      "step": 4383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96464115513519e-06,
+      "loss": 0.6587,
+      "step": 4384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646192725824265e-06,
+      "loss": 0.958,
+      "step": 4385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645973833824647e-06,
+      "loss": 0.9258,
+      "step": 4386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964575487535455e-06,
+      "loss": 0.6914,
+      "step": 4387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964553585041548e-06,
+      "loss": 1.1113,
+      "step": 4388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964531675900895e-06,
+      "loss": 1.0508,
+      "step": 4389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645097601136475e-06,
+      "loss": 0.998,
+      "step": 4390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644878376799555e-06,
+      "loss": 1.001,
+      "step": 4391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644659085999704e-06,
+      "loss": 1.04,
+      "step": 4392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644439728738437e-06,
+      "loss": 1.0615,
+      "step": 4393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644220305017263e-06,
+      "loss": 0.9648,
+      "step": 4394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644000814837693e-06,
+      "loss": 1.0234,
+      "step": 4395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964378125820124e-06,
+      "loss": 1.0264,
+      "step": 4396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643561635109414e-06,
+      "loss": 0.8701,
+      "step": 4397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964334194556374e-06,
+      "loss": 0.8916,
+      "step": 4398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964312218956572e-06,
+      "loss": 1.0205,
+      "step": 4399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964290236711687e-06,
+      "loss": 0.7769,
+      "step": 4400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964268247821871e-06,
+      "loss": 0.6143,
+      "step": 4401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642462522872753e-06,
+      "loss": 1.0361,
+      "step": 4402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964224250108051e-06,
+      "loss": 0.9717,
+      "step": 4403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642022412843504e-06,
+      "loss": 0.9336,
+      "step": 4404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641802258163245e-06,
+      "loss": 1.0068,
+      "step": 4405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641582037041257e-06,
+      "loss": 1.0283,
+      "step": 4406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641361749479054e-06,
+      "loss": 1.0059,
+      "step": 4407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641141395478157e-06,
+      "loss": 0.8994,
+      "step": 4408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964092097504008e-06,
+      "loss": 1.207,
+      "step": 4409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640700488166344e-06,
+      "loss": 0.9023,
+      "step": 4410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640479934858463e-06,
+      "loss": 1.0703,
+      "step": 4411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640259315117967e-06,
+      "loss": 0.9844,
+      "step": 4412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964003862894637e-06,
+      "loss": 1.0625,
+      "step": 4413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639817876345194e-06,
+      "loss": 0.9414,
+      "step": 4414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639597057315963e-06,
+      "loss": 1.0703,
+      "step": 4415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963937617186019e-06,
+      "loss": 0.8267,
+      "step": 4416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963915521997941e-06,
+      "loss": 0.6824,
+      "step": 4417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638934201675133e-06,
+      "loss": 1.0137,
+      "step": 4418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638713116948886e-06,
+      "loss": 1.0693,
+      "step": 4419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96384919658022e-06,
+      "loss": 1.04,
+      "step": 4420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963827074823659e-06,
+      "loss": 0.833,
+      "step": 4421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638049464253584e-06,
+      "loss": 1.0342,
+      "step": 4422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963782811385471e-06,
+      "loss": 1.1064,
+      "step": 4423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637606697041483e-06,
+      "loss": 0.8237,
+      "step": 4424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637385213815437e-06,
+      "loss": 0.7256,
+      "step": 4425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96371636641781e-06,
+      "loss": 1.0205,
+      "step": 4426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963694204813099e-06,
+      "loss": 0.876,
+      "step": 4427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636720365675647e-06,
+      "loss": 0.9658,
+      "step": 4428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963649861681359e-06,
+      "loss": 0.6602,
+      "step": 4429
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963627680154634e-06,
+      "loss": 0.7637,
+      "step": 4430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963605491987544e-06,
+      "loss": 1.0029,
+      "step": 4431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635832971802414e-06,
+      "loss": 1.0508,
+      "step": 4432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963561095732879e-06,
+      "loss": 0.9639,
+      "step": 4433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635388876456094e-06,
+      "loss": 0.6597,
+      "step": 4434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635166729185863e-06,
+      "loss": 0.9248,
+      "step": 4435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634944515519625e-06,
+      "loss": 0.707,
+      "step": 4436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963472223545891e-06,
+      "loss": 1.0088,
+      "step": 4437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963449988900526e-06,
+      "loss": 0.8848,
+      "step": 4438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634277476160188e-06,
+      "loss": 1.0073,
+      "step": 4439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963405499692524e-06,
+      "loss": 1.0625,
+      "step": 4440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633832451301947e-06,
+      "loss": 0.8135,
+      "step": 4441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963360983929184e-06,
+      "loss": 0.7588,
+      "step": 4442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633387160896455e-06,
+      "loss": 0.9355,
+      "step": 4443
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963316441611733e-06,
+      "loss": 0.6941,
+      "step": 4444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632941604955993e-06,
+      "loss": 0.9043,
+      "step": 4445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632718727413986e-06,
+      "loss": 1.0498,
+      "step": 4446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632495783492835e-06,
+      "loss": 0.9453,
+      "step": 4447
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632272773194088e-06,
+      "loss": 0.957,
+      "step": 4448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632049696519276e-06,
+      "loss": 0.9434,
+      "step": 4449
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963182655346994e-06,
+      "loss": 0.7002,
+      "step": 4450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631603344047606e-06,
+      "loss": 0.9961,
+      "step": 4451
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631380068253827e-06,
+      "loss": 0.9492,
+      "step": 4452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631156726090133e-06,
+      "loss": 0.793,
+      "step": 4453
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630933317558066e-06,
+      "loss": 0.9873,
+      "step": 4454
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630709842659163e-06,
+      "loss": 0.9268,
+      "step": 4455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630486301394967e-06,
+      "loss": 1.0215,
+      "step": 4456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963026269376702e-06,
+      "loss": 1.002,
+      "step": 4457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963003901977686e-06,
+      "loss": 0.9766,
+      "step": 4458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962981527942602e-06,
+      "loss": 0.8613,
+      "step": 4459
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962959147271606e-06,
+      "loss": 0.6514,
+      "step": 4460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629367599648505e-06,
+      "loss": 0.7637,
+      "step": 4461
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962914366022491e-06,
+      "loss": 1.084,
+      "step": 4462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962891965444681e-06,
+      "loss": 0.8633,
+      "step": 4463
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628695582315756e-06,
+      "loss": 1.0732,
+      "step": 4464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962847144383328e-06,
+      "loss": 0.9131,
+      "step": 4465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962824723900094e-06,
+      "loss": 0.9961,
+      "step": 4466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962802296782028e-06,
+      "loss": 0.6755,
+      "step": 4467
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627798630292836e-06,
+      "loss": 0.8174,
+      "step": 4468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627574226420162e-06,
+      "loss": 0.8286,
+      "step": 4469
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96273497562038e-06,
+      "loss": 0.9111,
+      "step": 4470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627125219645294e-06,
+      "loss": 0.9121,
+      "step": 4471
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96269006167462e-06,
+      "loss": 1.0605,
+      "step": 4472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962667594750806e-06,
+      "loss": 0.7646,
+      "step": 4473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626451211932423e-06,
+      "loss": 0.998,
+      "step": 4474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962622641002084e-06,
+      "loss": 0.9463,
+      "step": 4475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626001541774856e-06,
+      "loss": 0.8877,
+      "step": 4476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625776607196023e-06,
+      "loss": 0.9512,
+      "step": 4477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625551606285887e-06,
+      "loss": 0.877,
+      "step": 4478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625326539046007e-06,
+      "loss": 0.9922,
+      "step": 4479
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625101405477925e-06,
+      "loss": 1.1797,
+      "step": 4480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96248762055832e-06,
+      "loss": 0.9199,
+      "step": 4481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962465093936338e-06,
+      "loss": 0.8574,
+      "step": 4482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962442560682002e-06,
+      "loss": 0.8701,
+      "step": 4483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624200207954663e-06,
+      "loss": 1.0703,
+      "step": 4484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9623974742768874e-06,
+      "loss": 0.8379,
+      "step": 4485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9623749211264205e-06,
+      "loss": 0.957,
+      "step": 4486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623523613442206e-06,
+      "loss": 0.835,
+      "step": 4487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962329794930443e-06,
+      "loss": 0.8184,
+      "step": 4488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962307221885244e-06,
+      "loss": 0.8652,
+      "step": 4489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962284642208778e-06,
+      "loss": 0.9028,
+      "step": 4490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962262055901202e-06,
+      "loss": 0.9019,
+      "step": 4491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622394629626706e-06,
+      "loss": 0.9492,
+      "step": 4492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622168633933396e-06,
+      "loss": 0.918,
+      "step": 4493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621942571933648e-06,
+      "loss": 0.8682,
+      "step": 4494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621716443629025e-06,
+      "loss": 0.8271,
+      "step": 4495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962149024902108e-06,
+      "loss": 1.0273,
+      "step": 4496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621263988111375e-06,
+      "loss": 1.0195,
+      "step": 4497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962103766090147e-06,
+      "loss": 0.9287,
+      "step": 4498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620811267392914e-06,
+      "loss": 0.8516,
+      "step": 4499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620584807587276e-06,
+      "loss": 1.0,
+      "step": 4500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962035828148612e-06,
+      "loss": 0.9326,
+      "step": 4501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620131689090996e-06,
+      "loss": 0.7256,
+      "step": 4502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961990503040348e-06,
+      "loss": 0.9453,
+      "step": 4503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961967830542512e-06,
+      "loss": 1.0264,
+      "step": 4504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619451514157485e-06,
+      "loss": 0.6426,
+      "step": 4505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619224656602138e-06,
+      "loss": 0.9912,
+      "step": 4506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961899773276064e-06,
+      "loss": 0.9922,
+      "step": 4507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618770742634555e-06,
+      "loss": 0.7822,
+      "step": 4508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618543686225454e-06,
+      "loss": 0.7827,
+      "step": 4509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961831656353489e-06,
+      "loss": 0.958,
+      "step": 4510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618089374564433e-06,
+      "loss": 0.8789,
+      "step": 4511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617862119315656e-06,
+      "loss": 0.999,
+      "step": 4512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617634797790113e-06,
+      "loss": 1.0166,
+      "step": 4513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961740740998938e-06,
+      "loss": 0.9014,
+      "step": 4514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617179955915017e-06,
+      "loss": 1.0967,
+      "step": 4515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616952435568594e-06,
+      "loss": 1.0166,
+      "step": 4516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961672484895168e-06,
+      "loss": 0.8887,
+      "step": 4517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961649719606584e-06,
+      "loss": 1.0703,
+      "step": 4518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961626947691265e-06,
+      "loss": 0.9277,
+      "step": 4519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961604169149367e-06,
+      "loss": 1.0,
+      "step": 4520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615813839810474e-06,
+      "loss": 1.0,
+      "step": 4521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615585921864634e-06,
+      "loss": 1.0752,
+      "step": 4522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615357937657723e-06,
+      "loss": 0.7461,
+      "step": 4523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96151298871913e-06,
+      "loss": 0.7222,
+      "step": 4524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961490177046695e-06,
+      "loss": 1.0684,
+      "step": 4525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614673587486235e-06,
+      "loss": 0.9727,
+      "step": 4526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614445338250737e-06,
+      "loss": 0.9453,
+      "step": 4527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961421702276202e-06,
+      "loss": 0.8232,
+      "step": 4528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613988641021662e-06,
+      "loss": 1.0752,
+      "step": 4529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613760193031234e-06,
+      "loss": 1.0098,
+      "step": 4530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613531678792312e-06,
+      "loss": 0.8506,
+      "step": 4531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961330309830647e-06,
+      "loss": 1.0498,
+      "step": 4532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613074451575286e-06,
+      "loss": 0.8496,
+      "step": 4533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612845738600332e-06,
+      "loss": 0.9551,
+      "step": 4534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612616959383188e-06,
+      "loss": 0.7803,
+      "step": 4535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961238811392543e-06,
+      "loss": 0.7969,
+      "step": 4536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612159202228625e-06,
+      "loss": 0.9014,
+      "step": 4537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611930224294362e-06,
+      "loss": 1.0986,
+      "step": 4538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611701180124215e-06,
+      "loss": 1.0234,
+      "step": 4539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611472069719763e-06,
+      "loss": 1.0361,
+      "step": 4540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611242893082587e-06,
+      "loss": 0.9512,
+      "step": 4541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611013650214256e-06,
+      "loss": 0.9189,
+      "step": 4542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610784341116365e-06,
+      "loss": 0.9688,
+      "step": 4543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610554965790483e-06,
+      "loss": 0.9199,
+      "step": 4544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610325524238196e-06,
+      "loss": 0.8848,
+      "step": 4545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961009601646108e-06,
+      "loss": 0.998,
+      "step": 4546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960986644246072e-06,
+      "loss": 0.4448,
+      "step": 4547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96096368022387e-06,
+      "loss": 1.0,
+      "step": 4548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96094070957966e-06,
+      "loss": 0.834,
+      "step": 4549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609177323136005e-06,
+      "loss": 1.167,
+      "step": 4550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608947484258494e-06,
+      "loss": 0.918,
+      "step": 4551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608717579165655e-06,
+      "loss": 1.0176,
+      "step": 4552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608487607859066e-06,
+      "loss": 0.8828,
+      "step": 4553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960825757034032e-06,
+      "loss": 1.0127,
+      "step": 4554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608027466611e-06,
+      "loss": 0.9287,
+      "step": 4555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960779729667269e-06,
+      "loss": 0.9219,
+      "step": 4556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960756706052697e-06,
+      "loss": 0.9053,
+      "step": 4557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960733675817544e-06,
+      "loss": 0.9766,
+      "step": 4558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607106389619672e-06,
+      "loss": 1.0508,
+      "step": 4559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960687595486127e-06,
+      "loss": 0.7871,
+      "step": 4560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960664545390181e-06,
+      "loss": 0.8428,
+      "step": 4561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606414886742883e-06,
+      "loss": 1.0293,
+      "step": 4562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960618425338608e-06,
+      "loss": 0.9014,
+      "step": 4563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605953553832986e-06,
+      "loss": 0.9375,
+      "step": 4564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605722788085196e-06,
+      "loss": 0.6968,
+      "step": 4565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605491956144296e-06,
+      "loss": 1.0156,
+      "step": 4566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960526105801188e-06,
+      "loss": 0.8848,
+      "step": 4567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605030093689535e-06,
+      "loss": 0.8506,
+      "step": 4568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960479906317886e-06,
+      "loss": 1.0742,
+      "step": 4569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604567966481434e-06,
+      "loss": 0.8174,
+      "step": 4570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960433680359886e-06,
+      "loss": 0.9629,
+      "step": 4571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604105574532736e-06,
+      "loss": 0.9609,
+      "step": 4572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960387427928464e-06,
+      "loss": 1.0059,
+      "step": 4573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603642917856176e-06,
+      "loss": 0.9287,
+      "step": 4574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603411490248936e-06,
+      "loss": 1.2695,
+      "step": 4575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603179996464516e-06,
+      "loss": 0.7305,
+      "step": 4576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960294843650451e-06,
+      "loss": 0.8643,
+      "step": 4577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602716810370515e-06,
+      "loss": 1.0205,
+      "step": 4578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602485118064124e-06,
+      "loss": 0.96,
+      "step": 4579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602253359586935e-06,
+      "loss": 1.0107,
+      "step": 4580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602021534940544e-06,
+      "loss": 0.9839,
+      "step": 4581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960178964412655e-06,
+      "loss": 0.8647,
+      "step": 4582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601557687146556e-06,
+      "loss": 0.8672,
+      "step": 4583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601325664002157e-06,
+      "loss": 1.1592,
+      "step": 4584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601093574694943e-06,
+      "loss": 1.0381,
+      "step": 4585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600861419226528e-06,
+      "loss": 1.2871,
+      "step": 4586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96006291975985e-06,
+      "loss": 0.917,
+      "step": 4587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960039690981246e-06,
+      "loss": 0.9746,
+      "step": 4588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960016455587002e-06,
+      "loss": 0.9868,
+      "step": 4589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959993213577277e-06,
+      "loss": 0.8584,
+      "step": 4590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599699649522312e-06,
+      "loss": 1.0684,
+      "step": 4591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599467097120256e-06,
+      "loss": 0.9443,
+      "step": 4592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599234478568197e-06,
+      "loss": 0.7471,
+      "step": 4593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599001793867743e-06,
+      "loss": 0.9893,
+      "step": 4594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959876904302049e-06,
+      "loss": 0.7896,
+      "step": 4595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959853622602805e-06,
+      "loss": 0.835,
+      "step": 4596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959830334289203e-06,
+      "loss": 0.8828,
+      "step": 4597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959807039361402e-06,
+      "loss": 0.9775,
+      "step": 4598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959783737819564e-06,
+      "loss": 0.8535,
+      "step": 4599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959760429663849e-06,
+      "loss": 0.9766,
+      "step": 4600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597371148944175e-06,
+      "loss": 0.8906,
+      "step": 4601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597137935114305e-06,
+      "loss": 0.9326,
+      "step": 4602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596904655150482e-06,
+      "loss": 0.9297,
+      "step": 4603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596671309054317e-06,
+      "loss": 0.9971,
+      "step": 4604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596437896827418e-06,
+      "loss": 1.1973,
+      "step": 4605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596204418471394e-06,
+      "loss": 0.959,
+      "step": 4606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959597087398785e-06,
+      "loss": 1.2051,
+      "step": 4607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595737263378404e-06,
+      "loss": 1.1426,
+      "step": 4608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959550358664465e-06,
+      "loss": 0.8188,
+      "step": 4609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595269843788216e-06,
+      "loss": 1.0508,
+      "step": 4610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595036034810708e-06,
+      "loss": 1.0742,
+      "step": 4611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594802159713727e-06,
+      "loss": 1.0391,
+      "step": 4612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959456821849889e-06,
+      "loss": 0.9619,
+      "step": 4613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959433421116782e-06,
+      "loss": 1.001,
+      "step": 4614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959410013772211e-06,
+      "loss": 0.8486,
+      "step": 4615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593865998163395e-06,
+      "loss": 0.9424,
+      "step": 4616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959363179249327e-06,
+      "loss": 0.9287,
+      "step": 4617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593397520713357e-06,
+      "loss": 0.8184,
+      "step": 4618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959316318282527e-06,
+      "loss": 0.9648,
+      "step": 4619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592928778830623e-06,
+      "loss": 0.998,
+      "step": 4620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592694308731032e-06,
+      "loss": 0.6064,
+      "step": 4621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959245977252811e-06,
+      "loss": 0.9189,
+      "step": 4622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592225170223473e-06,
+      "loss": 0.6179,
+      "step": 4623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591990501818745e-06,
+      "loss": 0.8438,
+      "step": 4624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591755767315537e-06,
+      "loss": 1.1123,
+      "step": 4625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959152096671547e-06,
+      "loss": 0.9131,
+      "step": 4626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591286100020153e-06,
+      "loss": 0.8779,
+      "step": 4627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591051167231213e-06,
+      "loss": 0.9531,
+      "step": 4628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959081616835027e-06,
+      "loss": 0.8369,
+      "step": 4629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590581103378936e-06,
+      "loss": 0.8218,
+      "step": 4630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959034597231884e-06,
+      "loss": 0.7534,
+      "step": 4631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590110775171594e-06,
+      "loss": 0.9355,
+      "step": 4632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958987551193882e-06,
+      "loss": 0.9912,
+      "step": 4633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589640182622146e-06,
+      "loss": 0.9355,
+      "step": 4634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958940478722319e-06,
+      "loss": 1.002,
+      "step": 4635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958916932574357e-06,
+      "loss": 0.9697,
+      "step": 4636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958893379818491e-06,
+      "loss": 0.9014,
+      "step": 4637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958869820454884e-06,
+      "loss": 1.0049,
+      "step": 4638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588462544836974e-06,
+      "loss": 1.082,
+      "step": 4639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588226819050943e-06,
+      "loss": 1.0547,
+      "step": 4640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587991027192365e-06,
+      "loss": 0.6802,
+      "step": 4641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958775516926287e-06,
+      "loss": 0.8682,
+      "step": 4642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587519245264084e-06,
+      "loss": 1.0996,
+      "step": 4643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958728325519763e-06,
+      "loss": 0.8994,
+      "step": 4644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587047199065136e-06,
+      "loss": 0.9009,
+      "step": 4645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586811076868223e-06,
+      "loss": 0.7886,
+      "step": 4646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586574888608522e-06,
+      "loss": 1.1826,
+      "step": 4647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586338634287663e-06,
+      "loss": 0.999,
+      "step": 4648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586102313907275e-06,
+      "loss": 0.8564,
+      "step": 4649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585865927468982e-06,
+      "loss": 1.0635,
+      "step": 4650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585629474974414e-06,
+      "loss": 0.7344,
+      "step": 4651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585392956425197e-06,
+      "loss": 0.9814,
+      "step": 4652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958515637182297e-06,
+      "loss": 0.8462,
+      "step": 4653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584919721169355e-06,
+      "loss": 0.8906,
+      "step": 4654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584683004465987e-06,
+      "loss": 1.1426,
+      "step": 4655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584446221714497e-06,
+      "loss": 1.0361,
+      "step": 4656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584209372916514e-06,
+      "loss": 0.8989,
+      "step": 4657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958397245807367e-06,
+      "loss": 1.0293,
+      "step": 4658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583735477187604e-06,
+      "loss": 0.6797,
+      "step": 4659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958349843025994e-06,
+      "loss": 0.9639,
+      "step": 4660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958326131729232e-06,
+      "loss": 0.8169,
+      "step": 4661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583024138286372e-06,
+      "loss": 0.9238,
+      "step": 4662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582786893243734e-06,
+      "loss": 0.8262,
+      "step": 4663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582549582166035e-06,
+      "loss": 1.0879,
+      "step": 4664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958231220505492e-06,
+      "loss": 1.0332,
+      "step": 4665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582074761912013e-06,
+      "loss": 0.875,
+      "step": 4666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958183725273896e-06,
+      "loss": 0.583,
+      "step": 4667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581599677537394e-06,
+      "loss": 1.0625,
+      "step": 4668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958136203630895e-06,
+      "loss": 1.1504,
+      "step": 4669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958112432905527e-06,
+      "loss": 0.9893,
+      "step": 4670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580886555777993e-06,
+      "loss": 0.9219,
+      "step": 4671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958064871647875e-06,
+      "loss": 1.0254,
+      "step": 4672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580410811159186e-06,
+      "loss": 0.7803,
+      "step": 4673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580172839820935e-06,
+      "loss": 0.9102,
+      "step": 4674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579934802465647e-06,
+      "loss": 0.9814,
+      "step": 4675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579696699094954e-06,
+      "loss": 0.8037,
+      "step": 4676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579458529710494e-06,
+      "loss": 0.7871,
+      "step": 4677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579220294313914e-06,
+      "loss": 0.8799,
+      "step": 4678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957898199290686e-06,
+      "loss": 0.7114,
+      "step": 4679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957874362549096e-06,
+      "loss": 0.915,
+      "step": 4680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578505192067875e-06,
+      "loss": 0.8535,
+      "step": 4681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578266692639234e-06,
+      "loss": 1.1543,
+      "step": 4682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578028127206685e-06,
+      "loss": 0.8857,
+      "step": 4683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957778949577187e-06,
+      "loss": 0.7671,
+      "step": 4684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957755079833644e-06,
+      "loss": 0.9873,
+      "step": 4685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957731203490203e-06,
+      "loss": 1.0654,
+      "step": 4686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577073205470293e-06,
+      "loss": 0.9287,
+      "step": 4687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576834310042873e-06,
+      "loss": 0.8613,
+      "step": 4688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576595348621413e-06,
+      "loss": 1.1543,
+      "step": 4689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957635632120756e-06,
+      "loss": 0.8032,
+      "step": 4690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957611722780297e-06,
+      "loss": 1.1348,
+      "step": 4691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957587806840928e-06,
+      "loss": 1.1133,
+      "step": 4692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575638843028142e-06,
+      "loss": 0.7556,
+      "step": 4693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575399551661206e-06,
+      "loss": 1.0049,
+      "step": 4694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575160194310115e-06,
+      "loss": 0.7139,
+      "step": 4695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957492077097653e-06,
+      "loss": 0.7769,
+      "step": 4696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574681281662085e-06,
+      "loss": 1.1074,
+      "step": 4697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574441726368445e-06,
+      "loss": 0.9717,
+      "step": 4698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574202105097253e-06,
+      "loss": 0.9502,
+      "step": 4699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957396241785016e-06,
+      "loss": 0.918,
+      "step": 4700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573722664628824e-06,
+      "loss": 0.8145,
+      "step": 4701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957348284543489e-06,
+      "loss": 1.0107,
+      "step": 4702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573242960270012e-06,
+      "loss": 0.8828,
+      "step": 4703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957300300913585e-06,
+      "loss": 1.0674,
+      "step": 4704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572762992034048e-06,
+      "loss": 0.6616,
+      "step": 4705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957252290896626e-06,
+      "loss": 1.1553,
+      "step": 4706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572282759934153e-06,
+      "loss": 0.9678,
+      "step": 4707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572042544939367e-06,
+      "loss": 0.9834,
+      "step": 4708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571802263983568e-06,
+      "loss": 0.9922,
+      "step": 4709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571561917068406e-06,
+      "loss": 1.0732,
+      "step": 4710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571321504195534e-06,
+      "loss": 0.9922,
+      "step": 4711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571081025366617e-06,
+      "loss": 0.9248,
+      "step": 4712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570840480583305e-06,
+      "loss": 1.0625,
+      "step": 4713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570599869847264e-06,
+      "loss": 0.6919,
+      "step": 4714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570359193160145e-06,
+      "loss": 0.8511,
+      "step": 4715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570118450523608e-06,
+      "loss": 1.1348,
+      "step": 4716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569877641939313e-06,
+      "loss": 0.959,
+      "step": 4717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569636767408917e-06,
+      "loss": 0.9609,
+      "step": 4718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569395826934083e-06,
+      "loss": 0.917,
+      "step": 4719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956915482051647e-06,
+      "loss": 1.1621,
+      "step": 4720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956891374815774e-06,
+      "loss": 0.9238,
+      "step": 4721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568672609859553e-06,
+      "loss": 0.9062,
+      "step": 4722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956843140562357e-06,
+      "loss": 1.0615,
+      "step": 4723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956819013545145e-06,
+      "loss": 0.8062,
+      "step": 4724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567948799344868e-06,
+      "loss": 0.8086,
+      "step": 4725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956770739730547e-06,
+      "loss": 0.7227,
+      "step": 4726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567465929334933e-06,
+      "loss": 0.8799,
+      "step": 4727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567224395434915e-06,
+      "loss": 1.0381,
+      "step": 4728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956698279560708e-06,
+      "loss": 1.0557,
+      "step": 4729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566741129853098e-06,
+      "loss": 1.0244,
+      "step": 4730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956649939817463e-06,
+      "loss": 1.0986,
+      "step": 4731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956625760057334e-06,
+      "loss": 0.6719,
+      "step": 4732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566015737050894e-06,
+      "loss": 0.8057,
+      "step": 4733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565773807608966e-06,
+      "loss": 0.8848,
+      "step": 4734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565531812249215e-06,
+      "loss": 1.1465,
+      "step": 4735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565289750973317e-06,
+      "loss": 0.9331,
+      "step": 4736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956504762378293e-06,
+      "loss": 0.8955,
+      "step": 4737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956480543067973e-06,
+      "loss": 0.9248,
+      "step": 4738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564563171665386e-06,
+      "loss": 0.7402,
+      "step": 4739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956432084674156e-06,
+      "loss": 0.9512,
+      "step": 4740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956407845590993e-06,
+      "loss": 1.0938,
+      "step": 4741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563835999172163e-06,
+      "loss": 1.0098,
+      "step": 4742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956359347652993e-06,
+      "loss": 1.0762,
+      "step": 4743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95633508879849e-06,
+      "loss": 1.0986,
+      "step": 4744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563108233538745e-06,
+      "loss": 1.0938,
+      "step": 4745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562865513193143e-06,
+      "loss": 0.917,
+      "step": 4746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956262272694976e-06,
+      "loss": 1.0586,
+      "step": 4747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562379874810273e-06,
+      "loss": 0.9238,
+      "step": 4748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562136956776355e-06,
+      "loss": 0.9365,
+      "step": 4749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561893972849677e-06,
+      "loss": 0.9438,
+      "step": 4750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561650923031916e-06,
+      "loss": 0.9053,
+      "step": 4751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561407807324746e-06,
+      "loss": 0.9023,
+      "step": 4752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561164625729843e-06,
+      "loss": 0.7988,
+      "step": 4753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560921378248885e-06,
+      "loss": 0.9229,
+      "step": 4754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560678064883544e-06,
+      "loss": 0.8467,
+      "step": 4755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560434685635494e-06,
+      "loss": 0.9932,
+      "step": 4756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956019124050642e-06,
+      "loss": 1.1113,
+      "step": 4757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559947729497997e-06,
+      "loss": 0.8516,
+      "step": 4758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95597041526119e-06,
+      "loss": 1.0107,
+      "step": 4759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955946050984981e-06,
+      "loss": 0.7876,
+      "step": 4760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559216801213405e-06,
+      "loss": 1.0537,
+      "step": 4761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558973026704367e-06,
+      "loss": 0.9014,
+      "step": 4762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955872918632437e-06,
+      "loss": 0.7114,
+      "step": 4763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95584852800751e-06,
+      "loss": 0.9658,
+      "step": 4764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558241307958235e-06,
+      "loss": 0.7617,
+      "step": 4765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955799726997546e-06,
+      "loss": 1.0488,
+      "step": 4766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955775316612845e-06,
+      "loss": 1.0801,
+      "step": 4767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955750899641889e-06,
+      "loss": 1.0605,
+      "step": 4768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557264760848465e-06,
+      "loss": 1.332,
+      "step": 4769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557020459418857e-06,
+      "loss": 0.915,
+      "step": 4770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955677609213175e-06,
+      "loss": 0.9707,
+      "step": 4771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556531658988824e-06,
+      "loss": 0.7236,
+      "step": 4772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556287159991766e-06,
+      "loss": 0.6362,
+      "step": 4773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556042595142263e-06,
+      "loss": 0.9648,
+      "step": 4774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555797964442e-06,
+      "loss": 0.9248,
+      "step": 4775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555553267892656e-06,
+      "loss": 0.876,
+      "step": 4776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555308505495923e-06,
+      "loss": 0.9043,
+      "step": 4777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955506367725349e-06,
+      "loss": 1.083,
+      "step": 4778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554818783167038e-06,
+      "loss": 1.0527,
+      "step": 4779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955457382323826e-06,
+      "loss": 1.0234,
+      "step": 4780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554328797468843e-06,
+      "loss": 0.9131,
+      "step": 4781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554083705860473e-06,
+      "loss": 0.8047,
+      "step": 4782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553838548414836e-06,
+      "loss": 0.9053,
+      "step": 4783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955359332513363e-06,
+      "loss": 1.124,
+      "step": 4784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955334803601854e-06,
+      "loss": 0.8135,
+      "step": 4785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553102681071257e-06,
+      "loss": 0.9951,
+      "step": 4786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552857260293466e-06,
+      "loss": 0.8604,
+      "step": 4787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955261177368687e-06,
+      "loss": 1.042,
+      "step": 4788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552366221253147e-06,
+      "loss": 1.0703,
+      "step": 4789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552120602994004e-06,
+      "loss": 0.8887,
+      "step": 4790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955187491891112e-06,
+      "loss": 0.9756,
+      "step": 4791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551629169006197e-06,
+      "loss": 0.9824,
+      "step": 4792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551383353280922e-06,
+      "loss": 1.0107,
+      "step": 4793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551137471736995e-06,
+      "loss": 1.0557,
+      "step": 4794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955089152437611e-06,
+      "loss": 0.7991,
+      "step": 4795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550645511199953e-06,
+      "loss": 1.1338,
+      "step": 4796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550399432210227e-06,
+      "loss": 0.9893,
+      "step": 4797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955015328740863e-06,
+      "loss": 0.8052,
+      "step": 4798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549907076796853e-06,
+      "loss": 0.7949,
+      "step": 4799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954966080037659e-06,
+      "loss": 0.9951,
+      "step": 4800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954941445814955e-06,
+      "loss": 0.9209,
+      "step": 4801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954916805011742e-06,
+      "loss": 1.0576,
+      "step": 4802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548921576281896e-06,
+      "loss": 0.6929,
+      "step": 4803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954867503664469e-06,
+      "loss": 1.0703,
+      "step": 4804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548428431207483e-06,
+      "loss": 0.8975,
+      "step": 4805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954818175997199e-06,
+      "loss": 0.8262,
+      "step": 4806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95479350229399e-06,
+      "loss": 0.8262,
+      "step": 4807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547688220112923e-06,
+      "loss": 1.0352,
+      "step": 4808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954744135149275e-06,
+      "loss": 0.9238,
+      "step": 4809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547194417081088e-06,
+      "loss": 0.7949,
+      "step": 4810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546947416879635e-06,
+      "loss": 0.9805,
+      "step": 4811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546700350890095e-06,
+      "loss": 1.0635,
+      "step": 4812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546453219114175e-06,
+      "loss": 1.0527,
+      "step": 4813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954620602155357e-06,
+      "loss": 1.0352,
+      "step": 4814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545958758209992e-06,
+      "loss": 0.8496,
+      "step": 4815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545711429085136e-06,
+      "loss": 0.6333,
+      "step": 4816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954546403418071e-06,
+      "loss": 0.9424,
+      "step": 4817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954521657349842e-06,
+      "loss": 0.8101,
+      "step": 4818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544969047039973e-06,
+      "loss": 1.1484,
+      "step": 4819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954472145480707e-06,
+      "loss": 1.0176,
+      "step": 4820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954447379680142e-06,
+      "loss": 1.0684,
+      "step": 4821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544226073024727e-06,
+      "loss": 0.9951,
+      "step": 4822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543978283478705e-06,
+      "loss": 0.8633,
+      "step": 4823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954373042816505e-06,
+      "loss": 0.916,
+      "step": 4824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954348250708548e-06,
+      "loss": 0.9805,
+      "step": 4825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95432345202417e-06,
+      "loss": 1.0361,
+      "step": 4826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954298646763542e-06,
+      "loss": 1.0293,
+      "step": 4827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542738349268343e-06,
+      "loss": 1.0889,
+      "step": 4828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954249016514219e-06,
+      "loss": 0.9961,
+      "step": 4829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954224191525866e-06,
+      "loss": 0.8555,
+      "step": 4830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954199359961947e-06,
+      "loss": 1.0273,
+      "step": 4831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541745218226334e-06,
+      "loss": 0.873,
+      "step": 4832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954149677108096e-06,
+      "loss": 0.8945,
+      "step": 4833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541248258185055e-06,
+      "loss": 0.8555,
+      "step": 4834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954099967954034e-06,
+      "loss": 1.0723,
+      "step": 4835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954075103514852e-06,
+      "loss": 0.9971,
+      "step": 4836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540502325011317e-06,
+      "loss": 0.8037,
+      "step": 4837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540253549130436e-06,
+      "loss": 0.8809,
+      "step": 4838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540004707507597e-06,
+      "loss": 0.9277,
+      "step": 4839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953975580014452e-06,
+      "loss": 0.8271,
+      "step": 4840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539506827042904e-06,
+      "loss": 0.6279,
+      "step": 4841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953925778820448e-06,
+      "loss": 0.9736,
+      "step": 4842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539008683630957e-06,
+      "loss": 1.0596,
+      "step": 4843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953875951332405e-06,
+      "loss": 1.0557,
+      "step": 4844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538510277285487e-06,
+      "loss": 0.9365,
+      "step": 4845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953826097551697e-06,
+      "loss": 0.7783,
+      "step": 4846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538011608020227e-06,
+      "loss": 1.1475,
+      "step": 4847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537762174796975e-06,
+      "loss": 0.7998,
+      "step": 4848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953751267584893e-06,
+      "loss": 0.8745,
+      "step": 4849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537263111177815e-06,
+      "loss": 1.0049,
+      "step": 4850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537013480785346e-06,
+      "loss": 0.7881,
+      "step": 4851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536763784673247e-06,
+      "loss": 0.6265,
+      "step": 4852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536514022843237e-06,
+      "loss": 0.9121,
+      "step": 4853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536264195297035e-06,
+      "loss": 0.916,
+      "step": 4854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536014302036366e-06,
+      "loss": 0.832,
+      "step": 4855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953576434306295e-06,
+      "loss": 1.0371,
+      "step": 4856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535514318378506e-06,
+      "loss": 0.7466,
+      "step": 4857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535264227984767e-06,
+      "loss": 0.9502,
+      "step": 4858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535014071883447e-06,
+      "loss": 0.8115,
+      "step": 4859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534763850076273e-06,
+      "loss": 0.9243,
+      "step": 4860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953451356256497e-06,
+      "loss": 0.9844,
+      "step": 4861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534263209351265e-06,
+      "loss": 0.9971,
+      "step": 4862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534012790436877e-06,
+      "loss": 0.9541,
+      "step": 4863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953376230582353e-06,
+      "loss": 0.7859,
+      "step": 4864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533511755512963e-06,
+      "loss": 1.0,
+      "step": 4865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533261139506888e-06,
+      "loss": 1.1523,
+      "step": 4866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533010457807046e-06,
+      "loss": 1.0869,
+      "step": 4867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532759710415154e-06,
+      "loss": 1.0195,
+      "step": 4868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953250889733294e-06,
+      "loss": 0.8398,
+      "step": 4869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532258018562136e-06,
+      "loss": 0.812,
+      "step": 4870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532007074104467e-06,
+      "loss": 0.9531,
+      "step": 4871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953175606396167e-06,
+      "loss": 0.8853,
+      "step": 4872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953150498813547e-06,
+      "loss": 0.9746,
+      "step": 4873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531253846627594e-06,
+      "loss": 1.0469,
+      "step": 4874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953100263943978e-06,
+      "loss": 0.7832,
+      "step": 4875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953075136657375e-06,
+      "loss": 0.7559,
+      "step": 4876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9530500028031245e-06,
+      "loss": 1.0391,
+      "step": 4877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953024862381399e-06,
+      "loss": 1.0439,
+      "step": 4878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952999715392372e-06,
+      "loss": 0.8232,
+      "step": 4879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952974561836217e-06,
+      "loss": 0.8379,
+      "step": 4880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529494017131064e-06,
+      "loss": 0.8291,
+      "step": 4881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952924235023215e-06,
+      "loss": 1.0752,
+      "step": 4882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952899061766715e-06,
+      "loss": 1.1074,
+      "step": 4883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952873881943781e-06,
+      "loss": 0.9385,
+      "step": 4884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528486955545854e-06,
+      "loss": 0.6804,
+      "step": 4885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952823502599302e-06,
+      "loss": 0.9824,
+      "step": 4886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527983030781054e-06,
+      "loss": 0.8164,
+      "step": 4887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952773096991168e-06,
+      "loss": 0.8877,
+      "step": 4888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527478843386637e-06,
+      "loss": 1.0732,
+      "step": 4889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952722665120767e-06,
+      "loss": 0.8857,
+      "step": 4890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952697439337651e-06,
+      "loss": 0.79,
+      "step": 4891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95267220698949e-06,
+      "loss": 0.8584,
+      "step": 4892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526469680764573e-06,
+      "loss": 0.917,
+      "step": 4893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526217225987273e-06,
+      "loss": 1.0332,
+      "step": 4894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952596470556474e-06,
+      "loss": 0.9053,
+      "step": 4895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952571211949871e-06,
+      "loss": 0.8145,
+      "step": 4896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525459467790924e-06,
+      "loss": 0.9531,
+      "step": 4897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952520675044313e-06,
+      "loss": 1.0576,
+      "step": 4898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952495396745706e-06,
+      "loss": 0.999,
+      "step": 4899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524701118834465e-06,
+      "loss": 1.0303,
+      "step": 4900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952444820457708e-06,
+      "loss": 0.9131,
+      "step": 4901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524195224686646e-06,
+      "loss": 0.9795,
+      "step": 4902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523942179164916e-06,
+      "loss": 1.0859,
+      "step": 4903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952368906801363e-06,
+      "loss": 0.9082,
+      "step": 4904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523435891234525e-06,
+      "loss": 0.8135,
+      "step": 4905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523182648829354e-06,
+      "loss": 0.9873,
+      "step": 4906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952292934079986e-06,
+      "loss": 0.873,
+      "step": 4907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522675967147786e-06,
+      "loss": 0.804,
+      "step": 4908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522422527874883e-06,
+      "loss": 0.8818,
+      "step": 4909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522169022982892e-06,
+      "loss": 0.6587,
+      "step": 4910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521915452473563e-06,
+      "loss": 1.0029,
+      "step": 4911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521661816348645e-06,
+      "loss": 1.2744,
+      "step": 4912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952140811460988e-06,
+      "loss": 0.8184,
+      "step": 4913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952115434725902e-06,
+      "loss": 1.0098,
+      "step": 4914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952090051429782e-06,
+      "loss": 0.9834,
+      "step": 4915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520646615728013e-06,
+      "loss": 0.9058,
+      "step": 4916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520392651551363e-06,
+      "loss": 0.9385,
+      "step": 4917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520138621769615e-06,
+      "loss": 0.999,
+      "step": 4918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951988452638452e-06,
+      "loss": 0.8896,
+      "step": 4919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519630365397826e-06,
+      "loss": 0.8584,
+      "step": 4920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519376138811293e-06,
+      "loss": 0.9258,
+      "step": 4921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519121846626665e-06,
+      "loss": 0.8184,
+      "step": 4922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518867488845693e-06,
+      "loss": 0.9492,
+      "step": 4923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518613065470134e-06,
+      "loss": 1.1582,
+      "step": 4924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518358576501745e-06,
+      "loss": 0.9111,
+      "step": 4925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951810402194227e-06,
+      "loss": 1.166,
+      "step": 4926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951784940179347e-06,
+      "loss": 1.0293,
+      "step": 4927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95175947160571e-06,
+      "loss": 0.958,
+      "step": 4928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951733996473491e-06,
+      "loss": 1.1025,
+      "step": 4929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517085147828665e-06,
+      "loss": 0.856,
+      "step": 4930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516830265340106e-06,
+      "loss": 0.8232,
+      "step": 4931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516575317271e-06,
+      "loss": 0.918,
+      "step": 4932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516320303623103e-06,
+      "loss": 0.7419,
+      "step": 4933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951606522439817e-06,
+      "loss": 0.8706,
+      "step": 4934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515810079597962e-06,
+      "loss": 1.04,
+      "step": 4935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515554869224233e-06,
+      "loss": 0.6313,
+      "step": 4936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515299593278744e-06,
+      "loss": 0.9355,
+      "step": 4937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951504425176325e-06,
+      "loss": 1.0703,
+      "step": 4938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514788844679517e-06,
+      "loss": 0.9072,
+      "step": 4939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514533372029304e-06,
+      "loss": 1.0391,
+      "step": 4940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951427783381437e-06,
+      "loss": 0.9121,
+      "step": 4941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514022230036475e-06,
+      "loss": 1.0957,
+      "step": 4942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513766560697376e-06,
+      "loss": 0.9062,
+      "step": 4943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513510825798845e-06,
+      "loss": 0.9355,
+      "step": 4944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513255025342637e-06,
+      "loss": 0.9243,
+      "step": 4945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512999159330518e-06,
+      "loss": 1.0723,
+      "step": 4946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951274322776425e-06,
+      "loss": 1.0498,
+      "step": 4947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512487230645595e-06,
+      "loss": 0.9453,
+      "step": 4948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512231167976315e-06,
+      "loss": 0.916,
+      "step": 4949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511975039758184e-06,
+      "loss": 0.8035,
+      "step": 4950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951171884599296e-06,
+      "loss": 0.8203,
+      "step": 4951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951146258668241e-06,
+      "loss": 0.9717,
+      "step": 4952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95112062618283e-06,
+      "loss": 1.0596,
+      "step": 4953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510949871432396e-06,
+      "loss": 1.0479,
+      "step": 4954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951069341549646e-06,
+      "loss": 0.9463,
+      "step": 4955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510436894022267e-06,
+      "loss": 0.9912,
+      "step": 4956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951018030701158e-06,
+      "loss": 1.0049,
+      "step": 4957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509923654466174e-06,
+      "loss": 0.9619,
+      "step": 4958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509666936387804e-06,
+      "loss": 1.0312,
+      "step": 4959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509410152778256e-06,
+      "loss": 0.748,
+      "step": 4960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509153303639283e-06,
+      "loss": 0.8862,
+      "step": 4961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950889638897267e-06,
+      "loss": 0.8525,
+      "step": 4962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508639408780173e-06,
+      "loss": 0.7832,
+      "step": 4963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508382363063572e-06,
+      "loss": 1.0703,
+      "step": 4964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950812525182464e-06,
+      "loss": 1.0908,
+      "step": 4965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950786807506514e-06,
+      "loss": 0.9717,
+      "step": 4966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507610832786853e-06,
+      "loss": 0.9424,
+      "step": 4967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507353524991545e-06,
+      "loss": 1.0547,
+      "step": 4968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507096151680997e-06,
+      "loss": 0.8984,
+      "step": 4969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506838712856973e-06,
+      "loss": 0.9072,
+      "step": 4970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506581208521252e-06,
+      "loss": 1.0518,
+      "step": 4971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950632363867561e-06,
+      "loss": 0.6567,
+      "step": 4972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950606600332182e-06,
+      "loss": 0.7812,
+      "step": 4973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505808302461653e-06,
+      "loss": 1.04,
+      "step": 4974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505550536096896e-06,
+      "loss": 0.7524,
+      "step": 4975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505292704229315e-06,
+      "loss": 0.9727,
+      "step": 4976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950503480686069e-06,
+      "loss": 1.0908,
+      "step": 4977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95047768439928e-06,
+      "loss": 1.0566,
+      "step": 4978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950451881562742e-06,
+      "loss": 0.9873,
+      "step": 4979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504260721766328e-06,
+      "loss": 0.9414,
+      "step": 4980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504002562411305e-06,
+      "loss": 1.1328,
+      "step": 4981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503744337564126e-06,
+      "loss": 1.1367,
+      "step": 4982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503486047226575e-06,
+      "loss": 0.9619,
+      "step": 4983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950322769140043e-06,
+      "loss": 1.0234,
+      "step": 4984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950296927008747e-06,
+      "loss": 0.7251,
+      "step": 4985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950271078328948e-06,
+      "loss": 0.9648,
+      "step": 4986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502452231008234e-06,
+      "loss": 1.0918,
+      "step": 4987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950219361324552e-06,
+      "loss": 0.9961,
+      "step": 4988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950193493000312e-06,
+      "loss": 0.9492,
+      "step": 4989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501676181282815e-06,
+      "loss": 1.042,
+      "step": 4990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501417367086383e-06,
+      "loss": 0.8633,
+      "step": 4991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501158487415613e-06,
+      "loss": 0.8525,
+      "step": 4992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500899542272292e-06,
+      "loss": 1.0811,
+      "step": 4993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500640531658195e-06,
+      "loss": 0.7583,
+      "step": 4994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500381455575117e-06,
+      "loss": 0.9961,
+      "step": 4995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500122314024838e-06,
+      "loss": 1.0859,
+      "step": 4996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499863107009143e-06,
+      "loss": 0.7959,
+      "step": 4997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949960383452982e-06,
+      "loss": 0.792,
+      "step": 4998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499344496588655e-06,
+      "loss": 0.9727,
+      "step": 4999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499085093187436e-06,
+      "loss": 0.5552,
+      "step": 5000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 39016,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "total_flos": 129762904227840.0,
+  "trial_name": null,
+  "trial_params": null
+}