diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,68872 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.999956425116563,
+  "eval_steps": 500,
+  "global_step": 11474,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.355400696864112e-08,
+      "loss": 0.5441,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.710801393728224e-08,
+      "loss": 0.5458,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3066202090592336e-07,
+      "loss": 0.5258,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7421602787456448e-07,
+      "loss": 0.5308,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1777003484320556e-07,
+      "loss": 0.5349,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6132404181184673e-07,
+      "loss": 0.5188,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0487804878048784e-07,
+      "loss": 0.5624,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4843205574912896e-07,
+      "loss": 0.5625,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9198606271777007e-07,
+      "loss": 0.5185,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3554006968641113e-07,
+      "loss": 0.5375,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.790940766550523e-07,
+      "loss": 0.5388,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.226480836236935e-07,
+      "loss": 0.5418,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.662020905923345e-07,
+      "loss": 0.5538,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.097560975609757e-07,
+      "loss": 0.5591,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.533101045296167e-07,
+      "loss": 0.5338,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.968641114982579e-07,
+      "loss": 0.5412,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.40418118466899e-07,
+      "loss": 0.5299,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.839721254355401e-07,
+      "loss": 0.5205,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.275261324041812e-07,
+      "loss": 0.4959,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.710801393728223e-07,
+      "loss": 0.5375,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.146341463414634e-07,
+      "loss": 0.5251,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.581881533101046e-07,
+      "loss": 0.5087,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0017421602787455e-06,
+      "loss": 0.535,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.045296167247387e-06,
+      "loss": 0.5114,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.088850174216028e-06,
+      "loss": 0.5108,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.132404181184669e-06,
+      "loss": 0.513,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1759581881533102e-06,
+      "loss": 0.5057,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 0.4949,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2630662020905923e-06,
+      "loss": 0.5337,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3066202090592335e-06,
+      "loss": 0.506,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3501742160278747e-06,
+      "loss": 0.4958,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3937282229965158e-06,
+      "loss": 0.4912,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4372822299651568e-06,
+      "loss": 0.5144,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.480836236933798e-06,
+      "loss": 0.4904,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5243902439024391e-06,
+      "loss": 0.4992,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5679442508710803e-06,
+      "loss": 0.4863,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6114982578397212e-06,
+      "loss": 0.4608,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6550522648083624e-06,
+      "loss": 0.4821,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6986062717770036e-06,
+      "loss": 0.4726,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7421602787456445e-06,
+      "loss": 0.4796,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7857142857142857e-06,
+      "loss": 0.4706,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8292682926829268e-06,
+      "loss": 0.4653,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.872822299651568e-06,
+      "loss": 0.4731,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.916376306620209e-06,
+      "loss": 0.4585,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.95993031358885e-06,
+      "loss": 0.459,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.003484320557491e-06,
+      "loss": 0.4659,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0470383275261322e-06,
+      "loss": 0.4701,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.090592334494774e-06,
+      "loss": 0.4818,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.134146341463415e-06,
+      "loss": 0.4446,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.177700348432056e-06,
+      "loss": 0.4621,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.221254355400697e-06,
+      "loss": 0.4607,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.264808362369338e-06,
+      "loss": 0.4565,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3083623693379792e-06,
+      "loss": 0.4688,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3519163763066204e-06,
+      "loss": 0.4635,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3954703832752616e-06,
+      "loss": 0.4489,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 0.4442,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.482578397212544e-06,
+      "loss": 0.4349,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5261324041811846e-06,
+      "loss": 0.4386,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.569686411149826e-06,
+      "loss": 0.4479,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.613240418118467e-06,
+      "loss": 0.4435,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.656794425087108e-06,
+      "loss": 0.4394,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7003484320557493e-06,
+      "loss": 0.4426,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7439024390243905e-06,
+      "loss": 0.4456,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7874564459930316e-06,
+      "loss": 0.4296,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8310104529616724e-06,
+      "loss": 0.4277,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8745644599303136e-06,
+      "loss": 0.4419,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9181184668989547e-06,
+      "loss": 0.4242,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.961672473867596e-06,
+      "loss": 0.4494,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.005226480836237e-06,
+      "loss": 0.4631,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0487804878048782e-06,
+      "loss": 0.4337,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0923344947735194e-06,
+      "loss": 0.4428,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1358885017421605e-06,
+      "loss": 0.4245,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1794425087108017e-06,
+      "loss": 0.4262,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2229965156794425e-06,
+      "loss": 0.4338,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.266550522648084e-06,
+      "loss": 0.431,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3101045296167248e-06,
+      "loss": 0.4253,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3536585365853664e-06,
+      "loss": 0.4321,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.397212543554007e-06,
+      "loss": 0.4523,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4407665505226483e-06,
+      "loss": 0.4286,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.484320557491289e-06,
+      "loss": 0.4262,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5278745644599306e-06,
+      "loss": 0.4551,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 0.4301,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.614982578397213e-06,
+      "loss": 0.4211,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 0.4404,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7020905923344953e-06,
+      "loss": 0.4393,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.745644599303136e-06,
+      "loss": 0.4467,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.789198606271777e-06,
+      "loss": 0.4488,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.832752613240418e-06,
+      "loss": 0.4325,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.876306620209059e-06,
+      "loss": 0.4049,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9198606271777e-06,
+      "loss": 0.4256,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9634146341463414e-06,
+      "loss": 0.4133,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.006968641114982e-06,
+      "loss": 0.4082,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.050522648083624e-06,
+      "loss": 0.4153,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0940766550522645e-06,
+      "loss": 0.4232,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.137630662020906e-06,
+      "loss": 0.4099,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.181184668989548e-06,
+      "loss": 0.4667,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.224738675958188e-06,
+      "loss": 0.4195,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.26829268292683e-06,
+      "loss": 0.4211,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.311846689895471e-06,
+      "loss": 0.4375,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.355400696864112e-06,
+      "loss": 0.4332,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.398954703832753e-06,
+      "loss": 0.4143,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.442508710801394e-06,
+      "loss": 0.4133,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4860627177700346e-06,
+      "loss": 0.4007,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.529616724738676e-06,
+      "loss": 0.4106,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.573170731707317e-06,
+      "loss": 0.4332,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6167247386759585e-06,
+      "loss": 0.4429,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.660278745644599e-06,
+      "loss": 0.4266,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.703832752613241e-06,
+      "loss": 0.4442,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7473867595818816e-06,
+      "loss": 0.4459,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.790940766550523e-06,
+      "loss": 0.4297,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.834494773519164e-06,
+      "loss": 0.4287,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.4174,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921602787456446e-06,
+      "loss": 0.4152,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965156794425088e-06,
+      "loss": 0.431,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0087108013937286e-06,
+      "loss": 0.3814,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.052264808362369e-06,
+      "loss": 0.418,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.09581881533101e-06,
+      "loss": 0.4167,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.139372822299652e-06,
+      "loss": 0.4062,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.182926829268292e-06,
+      "loss": 0.422,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.226480836236934e-06,
+      "loss": 0.4266,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.270034843205575e-06,
+      "loss": 0.393,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.313588850174216e-06,
+      "loss": 0.4157,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.3965,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.400696864111499e-06,
+      "loss": 0.4095,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.444250871080139e-06,
+      "loss": 0.435,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.487804878048781e-06,
+      "loss": 0.4245,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.531358885017422e-06,
+      "loss": 0.4305,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.574912891986063e-06,
+      "loss": 0.4052,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.618466898954704e-06,
+      "loss": 0.422,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.662020905923345e-06,
+      "loss": 0.4153,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.705574912891986e-06,
+      "loss": 0.4294,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.749128919860627e-06,
+      "loss": 0.4165,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.792682926829269e-06,
+      "loss": 0.4225,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8362369337979094e-06,
+      "loss": 0.4223,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.879790940766551e-06,
+      "loss": 0.4139,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.923344947735192e-06,
+      "loss": 0.4011,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.966898954703833e-06,
+      "loss": 0.4009,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.010452961672474e-06,
+      "loss": 0.407,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.054006968641116e-06,
+      "loss": 0.4075,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 0.4342,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.141114982578398e-06,
+      "loss": 0.4144,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.184668989547039e-06,
+      "loss": 0.4126,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2282229965156795e-06,
+      "loss": 0.4084,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.271777003484321e-06,
+      "loss": 0.4095,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.315331010452962e-06,
+      "loss": 0.3912,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.3588850174216034e-06,
+      "loss": 0.4443,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.402439024390244e-06,
+      "loss": 0.4199,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.445993031358885e-06,
+      "loss": 0.4141,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.489547038327526e-06,
+      "loss": 0.4093,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.533101045296168e-06,
+      "loss": 0.3987,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.576655052264809e-06,
+      "loss": 0.4035,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.6202090592334496e-06,
+      "loss": 0.4097,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.66376306620209e-06,
+      "loss": 0.4236,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.707317073170733e-06,
+      "loss": 0.4109,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.7508710801393735e-06,
+      "loss": 0.4084,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.794425087108014e-06,
+      "loss": 0.3981,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.837979094076655e-06,
+      "loss": 0.4036,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.8815331010452966e-06,
+      "loss": 0.4242,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.925087108013937e-06,
+      "loss": 0.4062,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.968641114982578e-06,
+      "loss": 0.418,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.0121951219512205e-06,
+      "loss": 0.4001,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.055749128919861e-06,
+      "loss": 0.4025,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.099303135888502e-06,
+      "loss": 0.4269,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.3955,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.186411149825785e-06,
+      "loss": 0.415,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.229965156794426e-06,
+      "loss": 0.4077,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.273519163763067e-06,
+      "loss": 0.4006,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 0.413,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.360627177700349e-06,
+      "loss": 0.4065,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4041811846689906e-06,
+      "loss": 0.4101,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.447735191637631e-06,
+      "loss": 0.4137,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.491289198606272e-06,
+      "loss": 0.4166,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.534843205574914e-06,
+      "loss": 0.403,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.578397212543554e-06,
+      "loss": 0.4097,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.621951219512195e-06,
+      "loss": 0.4099,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.665505226480837e-06,
+      "loss": 0.4098,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.709059233449478e-06,
+      "loss": 0.4222,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.752613240418118e-06,
+      "loss": 0.4188,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.79616724738676e-06,
+      "loss": 0.4141,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.8397212543554e-06,
+      "loss": 0.4082,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.883275261324043e-06,
+      "loss": 0.4111,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.926829268292683e-06,
+      "loss": 0.4055,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.970383275261324e-06,
+      "loss": 0.4041,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.013937282229964e-06,
+      "loss": 0.402,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.057491289198608e-06,
+      "loss": 0.3943,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.101045296167248e-06,
+      "loss": 0.3966,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.144599303135889e-06,
+      "loss": 0.3943,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.188153310104529e-06,
+      "loss": 0.4093,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.231707317073172e-06,
+      "loss": 0.4199,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.275261324041812e-06,
+      "loss": 0.4122,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.318815331010454e-06,
+      "loss": 0.3999,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.362369337979095e-06,
+      "loss": 0.4112,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.405923344947735e-06,
+      "loss": 0.4032,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.449477351916377e-06,
+      "loss": 0.4134,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.493031358885017e-06,
+      "loss": 0.4125,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.53658536585366e-06,
+      "loss": 0.4406,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.5801393728223e-06,
+      "loss": 0.4052,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.623693379790942e-06,
+      "loss": 0.4315,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.667247386759581e-06,
+      "loss": 0.4195,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.710801393728225e-06,
+      "loss": 0.4015,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.754355400696865e-06,
+      "loss": 0.4075,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.797909407665506e-06,
+      "loss": 0.3965,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.841463414634146e-06,
+      "loss": 0.4104,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.885017421602788e-06,
+      "loss": 0.4046,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.4035,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.972125435540069e-06,
+      "loss": 0.4061,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.01567944250871e-06,
+      "loss": 0.4113,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.059233449477352e-06,
+      "loss": 0.4011,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.102787456445994e-06,
+      "loss": 0.3915,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.146341463414634e-06,
+      "loss": 0.4166,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.189895470383275e-06,
+      "loss": 0.3791,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.233449477351917e-06,
+      "loss": 0.3966,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.277003484320559e-06,
+      "loss": 0.3881,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.320557491289198e-06,
+      "loss": 0.4033,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.36411149825784e-06,
+      "loss": 0.3986,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.407665505226482e-06,
+      "loss": 0.4045,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.451219512195123e-06,
+      "loss": 0.4118,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.494773519163763e-06,
+      "loss": 0.4035,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.538327526132405e-06,
+      "loss": 0.4031,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.581881533101046e-06,
+      "loss": 0.4114,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.625435540069686e-06,
+      "loss": 0.3994,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.668989547038328e-06,
+      "loss": 0.4006,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.712543554006968e-06,
+      "loss": 0.4233,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.4121,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.799651567944251e-06,
+      "loss": 0.4033,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.843205574912892e-06,
+      "loss": 0.4109,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.886759581881534e-06,
+      "loss": 0.3958,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.930313588850176e-06,
+      "loss": 0.3873,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973867595818816e-06,
+      "loss": 0.4078,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0017421602787457e-05,
+      "loss": 0.3917,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0060975609756099e-05,
+      "loss": 0.4104,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0104529616724739e-05,
+      "loss": 0.4013,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.014808362369338e-05,
+      "loss": 0.401,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.019163763066202e-05,
+      "loss": 0.407,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0235191637630663e-05,
+      "loss": 0.391,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0278745644599303e-05,
+      "loss": 0.4227,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0322299651567945e-05,
+      "loss": 0.3906,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0365853658536585e-05,
+      "loss": 0.3972,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0409407665505228e-05,
+      "loss": 0.3849,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0452961672473868e-05,
+      "loss": 0.4133,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.049651567944251e-05,
+      "loss": 0.403,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.054006968641115e-05,
+      "loss": 0.4142,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0583623693379793e-05,
+      "loss": 0.4058,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0627177700348433e-05,
+      "loss": 0.4075,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0670731707317074e-05,
+      "loss": 0.3937,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.3965,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0757839721254356e-05,
+      "loss": 0.39,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0801393728222997e-05,
+      "loss": 0.3896,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0844947735191637e-05,
+      "loss": 0.4078,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0888501742160279e-05,
+      "loss": 0.3965,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.093205574912892e-05,
+      "loss": 0.4128,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0975609756097562e-05,
+      "loss": 0.4011,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1019163763066202e-05,
+      "loss": 0.403,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1062717770034843e-05,
+      "loss": 0.3901,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1106271777003485e-05,
+      "loss": 0.4349,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1149825783972127e-05,
+      "loss": 0.3898,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1193379790940766e-05,
+      "loss": 0.4006,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1236933797909408e-05,
+      "loss": 0.3952,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.128048780487805e-05,
+      "loss": 0.3777,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.132404181184669e-05,
+      "loss": 0.4199,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1367595818815331e-05,
+      "loss": 0.4184,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1411149825783973e-05,
+      "loss": 0.4334,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1454703832752614e-05,
+      "loss": 0.3909,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1498257839721254e-05,
+      "loss": 0.3767,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1541811846689896e-05,
+      "loss": 0.4144,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1585365853658537e-05,
+      "loss": 0.404,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1628919860627179e-05,
+      "loss": 0.4099,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1672473867595819e-05,
+      "loss": 0.4046,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.171602787456446e-05,
+      "loss": 0.4135,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1759581881533102e-05,
+      "loss": 0.3983,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1803135888501744e-05,
+      "loss": 0.3915,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1846689895470384e-05,
+      "loss": 0.4019,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1890243902439025e-05,
+      "loss": 0.3933,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1933797909407667e-05,
+      "loss": 0.3981,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1977351916376307e-05,
+      "loss": 0.389,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2020905923344948e-05,
+      "loss": 0.4099,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2064459930313588e-05,
+      "loss": 0.3931,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2108013937282231e-05,
+      "loss": 0.3945,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2151567944250871e-05,
+      "loss": 0.3892,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 0.417,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2238675958188153e-05,
+      "loss": 0.3938,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2282229965156796e-05,
+      "loss": 0.409,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2325783972125436e-05,
+      "loss": 0.3882,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2369337979094078e-05,
+      "loss": 0.4043,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2412891986062717e-05,
+      "loss": 0.4018,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2456445993031359e-05,
+      "loss": 0.3805,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3902,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2543554006968642e-05,
+      "loss": 0.3975,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2587108013937282e-05,
+      "loss": 0.3986,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2630662020905924e-05,
+      "loss": 0.4043,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2674216027874567e-05,
+      "loss": 0.3872,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2717770034843207e-05,
+      "loss": 0.3734,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2761324041811848e-05,
+      "loss": 0.3934,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2804878048780488e-05,
+      "loss": 0.3857,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.284843205574913e-05,
+      "loss": 0.3788,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.289198606271777e-05,
+      "loss": 0.4005,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2935540069686411e-05,
+      "loss": 0.4242,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2979094076655051e-05,
+      "loss": 0.3848,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3022648083623695e-05,
+      "loss": 0.3857,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3066202090592336e-05,
+      "loss": 0.3938,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3109756097560976e-05,
+      "loss": 0.395,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3153310104529618e-05,
+      "loss": 0.3954,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3196864111498258e-05,
+      "loss": 0.4035,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3240418118466899e-05,
+      "loss": 0.3822,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3283972125435539e-05,
+      "loss": 0.4212,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.332752613240418e-05,
+      "loss": 0.4044,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3371080139372824e-05,
+      "loss": 0.388,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3414634146341466e-05,
+      "loss": 0.4011,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3458188153310105e-05,
+      "loss": 0.3887,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3501742160278747e-05,
+      "loss": 0.4011,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3545296167247387e-05,
+      "loss": 0.3963,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3588850174216028e-05,
+      "loss": 0.3837,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3632404181184668e-05,
+      "loss": 0.384,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.367595818815331e-05,
+      "loss": 0.4,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3719512195121953e-05,
+      "loss": 0.409,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3763066202090593e-05,
+      "loss": 0.3816,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3806620209059235e-05,
+      "loss": 0.42,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3850174216027875e-05,
+      "loss": 0.3984,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3893728222996516e-05,
+      "loss": 0.3948,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3937282229965156e-05,
+      "loss": 0.4055,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3980836236933798e-05,
+      "loss": 0.3822,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4024390243902441e-05,
+      "loss": 0.3961,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4067944250871083e-05,
+      "loss": 0.4008,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4111498257839722e-05,
+      "loss": 0.3963,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4155052264808364e-05,
+      "loss": 0.4166,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4198606271777004e-05,
+      "loss": 0.3798,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4242160278745646e-05,
+      "loss": 0.367,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.3967,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4329268292682927e-05,
+      "loss": 0.4007,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.437282229965157e-05,
+      "loss": 0.3851,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.441637630662021e-05,
+      "loss": 0.3685,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4459930313588852e-05,
+      "loss": 0.3976,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4503484320557492e-05,
+      "loss": 0.3938,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4547038327526133e-05,
+      "loss": 0.3926,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4590592334494773e-05,
+      "loss": 0.3803,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.3901,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4677700348432055e-05,
+      "loss": 0.3828,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4721254355400698e-05,
+      "loss": 0.3916,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.476480836236934e-05,
+      "loss": 0.3943,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4808362369337981e-05,
+      "loss": 0.3939,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4851916376306621e-05,
+      "loss": 0.3874,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4895470383275263e-05,
+      "loss": 0.3984,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4939024390243902e-05,
+      "loss": 0.414,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4982578397212544e-05,
+      "loss": 0.3972,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5026132404181184e-05,
+      "loss": 0.3933,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5069686411149827e-05,
+      "loss": 0.393,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5113240418118469e-05,
+      "loss": 0.426,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5156794425087109e-05,
+      "loss": 0.3955,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.520034843205575e-05,
+      "loss": 0.4021,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.524390243902439e-05,
+      "loss": 0.396,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.528745644599303e-05,
+      "loss": 0.4146,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5331010452961673e-05,
+      "loss": 0.3963,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5374564459930317e-05,
+      "loss": 0.4109,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5418118466898957e-05,
+      "loss": 0.3962,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5461672473867596e-05,
+      "loss": 0.3905,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5505226480836236e-05,
+      "loss": 0.3918,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.554878048780488e-05,
+      "loss": 0.3988,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.559233449477352e-05,
+      "loss": 0.3897,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.563588850174216e-05,
+      "loss": 0.4035,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.56794425087108e-05,
+      "loss": 0.3877,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5722996515679446e-05,
+      "loss": 0.3933,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5766550522648086e-05,
+      "loss": 0.4046,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5810104529616726e-05,
+      "loss": 0.4023,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5853658536585366e-05,
+      "loss": 0.3771,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.589721254355401e-05,
+      "loss": 0.3903,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.594076655052265e-05,
+      "loss": 0.3957,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.598432055749129e-05,
+      "loss": 0.3859,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.602787456445993e-05,
+      "loss": 0.3964,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.3986,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6114982578397215e-05,
+      "loss": 0.3952,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6158536585365855e-05,
+      "loss": 0.3886,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6202090592334495e-05,
+      "loss": 0.3743,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6245644599303135e-05,
+      "loss": 0.3837,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6289198606271778e-05,
+      "loss": 0.3854,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6332752613240418e-05,
+      "loss": 0.3873,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6376306620209058e-05,
+      "loss": 0.3773,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.64198606271777e-05,
+      "loss": 0.3997,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6463414634146345e-05,
+      "loss": 0.385,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6506968641114984e-05,
+      "loss": 0.3969,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6550522648083624e-05,
+      "loss": 0.3838,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6594076655052264e-05,
+      "loss": 0.3608,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6637630662020908e-05,
+      "loss": 0.4044,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6681184668989547e-05,
+      "loss": 0.3978,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.672473867595819e-05,
+      "loss": 0.3991,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.676829268292683e-05,
+      "loss": 0.3733,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.681184668989547e-05,
+      "loss": 0.3847,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6855400696864114e-05,
+      "loss": 0.4059,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6898954703832754e-05,
+      "loss": 0.3888,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6942508710801394e-05,
+      "loss": 0.3737,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6986062717770033e-05,
+      "loss": 0.3598,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7029616724738677e-05,
+      "loss": 0.4085,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.707317073170732e-05,
+      "loss": 0.3914,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.711672473867596e-05,
+      "loss": 0.3668,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.71602787456446e-05,
+      "loss": 0.3935,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7203832752613243e-05,
+      "loss": 0.3847,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7247386759581883e-05,
+      "loss": 0.4153,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7290940766550523e-05,
+      "loss": 0.3894,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7334494773519163e-05,
+      "loss": 0.3834,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7378048780487806e-05,
+      "loss": 0.402,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.742160278745645e-05,
+      "loss": 0.3736,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.746515679442509e-05,
+      "loss": 0.3796,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.750871080139373e-05,
+      "loss": 0.3872,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.755226480836237e-05,
+      "loss": 0.3973,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7595818815331012e-05,
+      "loss": 0.3875,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7639372822299652e-05,
+      "loss": 0.3798,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7682926829268292e-05,
+      "loss": 0.4135,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7726480836236932e-05,
+      "loss": 0.395,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7770034843205575e-05,
+      "loss": 0.3995,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.781358885017422e-05,
+      "loss": 0.411,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.3881,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.79006968641115e-05,
+      "loss": 0.3898,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7944250871080138e-05,
+      "loss": 0.4011,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.798780487804878e-05,
+      "loss": 0.3876,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.803135888501742e-05,
+      "loss": 0.3924,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8074912891986065e-05,
+      "loss": 0.3833,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8118466898954705e-05,
+      "loss": 0.3903,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8162020905923348e-05,
+      "loss": 0.3808,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8205574912891988e-05,
+      "loss": 0.3801,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8249128919860628e-05,
+      "loss": 0.3756,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8292682926829268e-05,
+      "loss": 0.3692,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.833623693379791e-05,
+      "loss": 0.3999,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.837979094076655e-05,
+      "loss": 0.395,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8423344947735194e-05,
+      "loss": 0.4013,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8466898954703834e-05,
+      "loss": 0.3898,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8510452961672474e-05,
+      "loss": 0.3976,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8554006968641117e-05,
+      "loss": 0.3771,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8597560975609757e-05,
+      "loss": 0.3642,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8641114982578397e-05,
+      "loss": 0.3805,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8684668989547037e-05,
+      "loss": 0.3733,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.872822299651568e-05,
+      "loss": 0.3681,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8771777003484323e-05,
+      "loss": 0.3851,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8815331010452963e-05,
+      "loss": 0.3857,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8858885017421603e-05,
+      "loss": 0.3975,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8902439024390246e-05,
+      "loss": 0.3856,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8945993031358886e-05,
+      "loss": 0.4208,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8989547038327526e-05,
+      "loss": 0.4008,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9033101045296166e-05,
+      "loss": 0.3697,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.907665505226481e-05,
+      "loss": 0.3751,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9120209059233453e-05,
+      "loss": 0.393,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9163763066202093e-05,
+      "loss": 0.3869,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9207317073170733e-05,
+      "loss": 0.3668,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9250871080139372e-05,
+      "loss": 0.3955,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9294425087108016e-05,
+      "loss": 0.3942,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9337979094076656e-05,
+      "loss": 0.3808,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9381533101045295e-05,
+      "loss": 0.391,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9425087108013935e-05,
+      "loss": 0.3919,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9468641114982582e-05,
+      "loss": 0.3778,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.3827,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9555749128919862e-05,
+      "loss": 0.3678,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9599303135888502e-05,
+      "loss": 0.4031,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.3758,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9686411149825785e-05,
+      "loss": 0.4008,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9729965156794425e-05,
+      "loss": 0.3857,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9773519163763068e-05,
+      "loss": 0.3712,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9817073170731708e-05,
+      "loss": 0.3588,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986062717770035e-05,
+      "loss": 0.3892,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.990418118466899e-05,
+      "loss": 0.3907,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994773519163763e-05,
+      "loss": 0.3989,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999128919860627e-05,
+      "loss": 0.4027,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0034843205574914e-05,
+      "loss": 0.3859,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0078397212543554e-05,
+      "loss": 0.3845,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0121951219512197e-05,
+      "loss": 0.3909,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0165505226480837e-05,
+      "loss": 0.4278,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0209059233449477e-05,
+      "loss": 0.3893,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.025261324041812e-05,
+      "loss": 0.4036,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.029616724738676e-05,
+      "loss": 0.393,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.03397212543554e-05,
+      "loss": 0.3729,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.038327526132404e-05,
+      "loss": 0.3935,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0426829268292683e-05,
+      "loss": 0.3834,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0470383275261327e-05,
+      "loss": 0.3891,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0513937282229967e-05,
+      "loss": 0.3663,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0557491289198607e-05,
+      "loss": 0.389,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.060104529616725e-05,
+      "loss": 0.3947,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.064459930313589e-05,
+      "loss": 0.3806,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.068815331010453e-05,
+      "loss": 0.3829,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.073170731707317e-05,
+      "loss": 0.3719,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0775261324041813e-05,
+      "loss": 0.3795,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0818815331010456e-05,
+      "loss": 0.3706,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0862369337979096e-05,
+      "loss": 0.3673,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0905923344947736e-05,
+      "loss": 0.3863,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0949477351916376e-05,
+      "loss": 0.3912,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.099303135888502e-05,
+      "loss": 0.3609,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.103658536585366e-05,
+      "loss": 0.3691,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.10801393728223e-05,
+      "loss": 0.3803,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1123693379790942e-05,
+      "loss": 0.3807,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1167247386759585e-05,
+      "loss": 0.3835,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1210801393728225e-05,
+      "loss": 0.3966,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1254355400696865e-05,
+      "loss": 0.379,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1297909407665505e-05,
+      "loss": 0.3951,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.134146341463415e-05,
+      "loss": 0.3866,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1385017421602788e-05,
+      "loss": 0.3704,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.387,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.147212543554007e-05,
+      "loss": 0.3984,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.151567944250871e-05,
+      "loss": 0.3812,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1559233449477355e-05,
+      "loss": 0.377,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1602787456445995e-05,
+      "loss": 0.3839,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1646341463414634e-05,
+      "loss": 0.3818,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1689895470383274e-05,
+      "loss": 0.3881,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1733449477351918e-05,
+      "loss": 0.3693,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1777003484320557e-05,
+      "loss": 0.3902,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18205574912892e-05,
+      "loss": 0.3877,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.186411149825784e-05,
+      "loss": 0.3852,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1907665505226484e-05,
+      "loss": 0.3919,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1951219512195124e-05,
+      "loss": 0.3659,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1994773519163764e-05,
+      "loss": 0.3736,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2038327526132404e-05,
+      "loss": 0.3751,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2081881533101047e-05,
+      "loss": 0.3855,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2125435540069687e-05,
+      "loss": 0.3688,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.216898954703833e-05,
+      "loss": 0.3946,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.221254355400697e-05,
+      "loss": 0.3619,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.225609756097561e-05,
+      "loss": 0.3806,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2299651567944253e-05,
+      "loss": 0.3858,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2343205574912893e-05,
+      "loss": 0.3869,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2386759581881533e-05,
+      "loss": 0.3755,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2430313588850173e-05,
+      "loss": 0.3794,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2473867595818816e-05,
+      "loss": 0.4136,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.251742160278746e-05,
+      "loss": 0.381,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.25609756097561e-05,
+      "loss": 0.3849,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.260452961672474e-05,
+      "loss": 0.3831,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.264808362369338e-05,
+      "loss": 0.3739,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2691637630662022e-05,
+      "loss": 0.3756,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2735191637630662e-05,
+      "loss": 0.3763,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2778745644599302e-05,
+      "loss": 0.372,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2822299651567945e-05,
+      "loss": 0.3915,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.286585365853659e-05,
+      "loss": 0.3915,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.290940766550523e-05,
+      "loss": 0.377,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.295296167247387e-05,
+      "loss": 0.3926,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.299651567944251e-05,
+      "loss": 0.3831,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3040069686411152e-05,
+      "loss": 0.3739,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.308362369337979e-05,
+      "loss": 0.3887,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.312717770034843e-05,
+      "loss": 0.3824,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3170731707317075e-05,
+      "loss": 0.3772,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3214285714285715e-05,
+      "loss": 0.3772,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3257839721254358e-05,
+      "loss": 0.3834,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3301393728222998e-05,
+      "loss": 0.4079,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3344947735191638e-05,
+      "loss": 0.3669,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3388501742160278e-05,
+      "loss": 0.3831,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.343205574912892e-05,
+      "loss": 0.3824,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.347560975609756e-05,
+      "loss": 0.3663,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3519163763066204e-05,
+      "loss": 0.3814,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3562717770034844e-05,
+      "loss": 0.3683,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3606271777003487e-05,
+      "loss": 0.404,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3649825783972127e-05,
+      "loss": 0.3878,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3693379790940767e-05,
+      "loss": 0.3922,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3736933797909407e-05,
+      "loss": 0.3874,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.378048780487805e-05,
+      "loss": 0.357,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3824041811846694e-05,
+      "loss": 0.3743,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3867595818815333e-05,
+      "loss": 0.3721,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3911149825783973e-05,
+      "loss": 0.3936,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3954703832752613e-05,
+      "loss": 0.3838,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3998257839721257e-05,
+      "loss": 0.3721,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4041811846689896e-05,
+      "loss": 0.3943,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4085365853658536e-05,
+      "loss": 0.3832,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4128919860627176e-05,
+      "loss": 0.3913,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4172473867595823e-05,
+      "loss": 0.3874,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4216027874564463e-05,
+      "loss": 0.3686,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4259581881533103e-05,
+      "loss": 0.364,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4303135888501743e-05,
+      "loss": 0.358,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4346689895470386e-05,
+      "loss": 0.3996,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.3929,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4433797909407666e-05,
+      "loss": 0.3783,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4477351916376306e-05,
+      "loss": 0.3616,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.452090592334495e-05,
+      "loss": 0.3751,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4564459930313592e-05,
+      "loss": 0.3876,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4608013937282232e-05,
+      "loss": 0.3951,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4651567944250872e-05,
+      "loss": 0.384,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4695121951219512e-05,
+      "loss": 0.3773,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4738675958188155e-05,
+      "loss": 0.3976,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4782229965156795e-05,
+      "loss": 0.364,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4825783972125435e-05,
+      "loss": 0.3945,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4869337979094078e-05,
+      "loss": 0.3647,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4912891986062718e-05,
+      "loss": 0.3905,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.495644599303136e-05,
+      "loss": 0.3822,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3782,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5043554006968644e-05,
+      "loss": 0.3733,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5087108013937284e-05,
+      "loss": 0.3745,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5130662020905928e-05,
+      "loss": 0.3786,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5174216027874564e-05,
+      "loss": 0.3649,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5217770034843207e-05,
+      "loss": 0.365,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5261324041811847e-05,
+      "loss": 0.3644,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.530487804878049e-05,
+      "loss": 0.3719,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5348432055749134e-05,
+      "loss": 0.3769,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.539198606271777e-05,
+      "loss": 0.3847,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5435540069686414e-05,
+      "loss": 0.389,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5479094076655054e-05,
+      "loss": 0.3985,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5522648083623697e-05,
+      "loss": 0.3939,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5566202090592333e-05,
+      "loss": 0.3842,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5609756097560977e-05,
+      "loss": 0.3776,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5653310104529617e-05,
+      "loss": 0.3752,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.569686411149826e-05,
+      "loss": 0.3905,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5740418118466903e-05,
+      "loss": 0.3717,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.578397212543554e-05,
+      "loss": 0.3815,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5827526132404183e-05,
+      "loss": 0.3716,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5871080139372823e-05,
+      "loss": 0.4058,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5914634146341466e-05,
+      "loss": 0.3653,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5958188153310103e-05,
+      "loss": 0.3909,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6001742160278746e-05,
+      "loss": 0.364,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.604529616724739e-05,
+      "loss": 0.3836,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.608885017421603e-05,
+      "loss": 0.3907,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6132404181184672e-05,
+      "loss": 0.4024,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.617595818815331e-05,
+      "loss": 0.3685,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6219512195121952e-05,
+      "loss": 0.3738,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6263066202090592e-05,
+      "loss": 0.3521,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6306620209059235e-05,
+      "loss": 0.3859,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6350174216027872e-05,
+      "loss": 0.3762,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6393728222996515e-05,
+      "loss": 0.375,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6437282229965162e-05,
+      "loss": 0.3813,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6480836236933798e-05,
+      "loss": 0.3696,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.652439024390244e-05,
+      "loss": 0.3828,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6567944250871078e-05,
+      "loss": 0.3748,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6611498257839725e-05,
+      "loss": 0.385,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.665505226480836e-05,
+      "loss": 0.3981,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6698606271777005e-05,
+      "loss": 0.3705,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6742160278745648e-05,
+      "loss": 0.352,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 0.4066,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.682926829268293e-05,
+      "loss": 0.365,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6872822299651568e-05,
+      "loss": 0.3915,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.691637630662021e-05,
+      "loss": 0.3941,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.695993031358885e-05,
+      "loss": 0.3856,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7003484320557494e-05,
+      "loss": 0.3639,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7047038327526137e-05,
+      "loss": 0.3583,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7090592334494774e-05,
+      "loss": 0.3558,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7134146341463417e-05,
+      "loss": 0.3825,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7177700348432057e-05,
+      "loss": 0.3757,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.72212543554007e-05,
+      "loss": 0.362,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7264808362369337e-05,
+      "loss": 0.3722,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.730836236933798e-05,
+      "loss": 0.363,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.735191637630662e-05,
+      "loss": 0.3718,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7395470383275263e-05,
+      "loss": 0.3667,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7439024390243906e-05,
+      "loss": 0.3811,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7482578397212543e-05,
+      "loss": 0.3801,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7526132404181186e-05,
+      "loss": 0.3698,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7569686411149826e-05,
+      "loss": 0.3679,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.761324041811847e-05,
+      "loss": 0.3725,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7656794425087106e-05,
+      "loss": 0.3663,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.770034843205575e-05,
+      "loss": 0.3629,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7743902439024393e-05,
+      "loss": 0.3818,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7787456445993032e-05,
+      "loss": 0.3732,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7831010452961676e-05,
+      "loss": 0.3914,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7874564459930312e-05,
+      "loss": 0.3621,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7918118466898955e-05,
+      "loss": 0.3688,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7961672473867595e-05,
+      "loss": 0.3897,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.800522648083624e-05,
+      "loss": 0.3713,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8048780487804882e-05,
+      "loss": 0.3825,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.809233449477352e-05,
+      "loss": 0.362,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8135888501742165e-05,
+      "loss": 0.3787,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.81794425087108e-05,
+      "loss": 0.3941,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8222996515679445e-05,
+      "loss": 0.4076,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.826655052264808e-05,
+      "loss": 0.3776,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8310104529616728e-05,
+      "loss": 0.3597,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8353658536585365e-05,
+      "loss": 0.3994,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8397212543554008e-05,
+      "loss": 0.3747,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.844076655052265e-05,
+      "loss": 0.3872,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.848432055749129e-05,
+      "loss": 0.3856,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8527874564459934e-05,
+      "loss": 0.3991,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.391,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8614982578397214e-05,
+      "loss": 0.3778,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8658536585365854e-05,
+      "loss": 0.3756,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8702090592334497e-05,
+      "loss": 0.3468,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.874564459930314e-05,
+      "loss": 0.3662,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8789198606271777e-05,
+      "loss": 0.3764,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.883275261324042e-05,
+      "loss": 0.3612,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.887630662020906e-05,
+      "loss": 0.3438,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8919860627177704e-05,
+      "loss": 0.382,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.896341463414634e-05,
+      "loss": 0.3745,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9006968641114983e-05,
+      "loss": 0.378,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9050522648083623e-05,
+      "loss": 0.3681,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9094076655052267e-05,
+      "loss": 0.3961,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.913763066202091e-05,
+      "loss": 0.3997,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9181184668989546e-05,
+      "loss": 0.3634,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.922473867595819e-05,
+      "loss": 0.3918,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.3657,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9311846689895473e-05,
+      "loss": 0.3691,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.935540069686411e-05,
+      "loss": 0.3775,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9398954703832753e-05,
+      "loss": 0.3873,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9442508710801396e-05,
+      "loss": 0.3791,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9486062717770036e-05,
+      "loss": 0.3845,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.952961672473868e-05,
+      "loss": 0.3681,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9573170731707316e-05,
+      "loss": 0.381,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9616724738675962e-05,
+      "loss": 0.3761,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.96602787456446e-05,
+      "loss": 0.3865,
+      "step": 681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9703832752613242e-05,
+      "loss": 0.3811,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9747386759581885e-05,
+      "loss": 0.3815,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9790940766550525e-05,
+      "loss": 0.392,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.983449477351917e-05,
+      "loss": 0.3917,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9878048780487805e-05,
+      "loss": 0.3812,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9921602787456448e-05,
+      "loss": 0.3973,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9965156794425088e-05,
+      "loss": 0.4061,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.000871080139373e-05,
+      "loss": 0.3921,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0052264808362368e-05,
+      "loss": 0.3841,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.009581881533101e-05,
+      "loss": 0.3739,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0139372822299655e-05,
+      "loss": 0.3943,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0182926829268294e-05,
+      "loss": 0.3854,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0226480836236938e-05,
+      "loss": 0.385,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0270034843205574e-05,
+      "loss": 0.3903,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0313588850174217e-05,
+      "loss": 0.3798,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.3634,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.04006968641115e-05,
+      "loss": 0.388,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0444250871080144e-05,
+      "loss": 0.3719,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.048780487804878e-05,
+      "loss": 0.3711,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0531358885017424e-05,
+      "loss": 0.3773,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.057491289198606e-05,
+      "loss": 0.3689,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0618466898954704e-05,
+      "loss": 0.3785,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.066202090592335e-05,
+      "loss": 0.3784,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.070557491289199e-05,
+      "loss": 0.3738,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.074912891986063e-05,
+      "loss": 0.3622,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.079268292682927e-05,
+      "loss": 0.3638,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.083623693379791e-05,
+      "loss": 0.3777,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.087979094076655e-05,
+      "loss": 0.3672,
+      "step": 709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.092334494773519e-05,
+      "loss": 0.3806,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.096689895470383e-05,
+      "loss": 0.3815,
+      "step": 711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.101045296167247e-05,
+      "loss": 0.3805,
+      "step": 712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1054006968641116e-05,
+      "loss": 0.3931,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.109756097560976e-05,
+      "loss": 0.3818,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.11411149825784e-05,
+      "loss": 0.3738,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.118466898954704e-05,
+      "loss": 0.3614,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.122822299651568e-05,
+      "loss": 0.3553,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.127177700348432e-05,
+      "loss": 0.3728,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.131533101045296e-05,
+      "loss": 0.3589,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.13588850174216e-05,
+      "loss": 0.3707,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.140243902439025e-05,
+      "loss": 0.3783,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.144599303135889e-05,
+      "loss": 0.383,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.148954703832753e-05,
+      "loss": 0.376,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.153310104529617e-05,
+      "loss": 0.3598,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.157665505226481e-05,
+      "loss": 0.3611,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.162020905923345e-05,
+      "loss": 0.3985,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.166376306620209e-05,
+      "loss": 0.3575,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.170731707317073e-05,
+      "loss": 0.3889,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.1750871080139375e-05,
+      "loss": 0.3884,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.179442508710802e-05,
+      "loss": 0.3929,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.183797909407666e-05,
+      "loss": 0.378,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.18815331010453e-05,
+      "loss": 0.378,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.192508710801394e-05,
+      "loss": 0.3693,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.196864111498258e-05,
+      "loss": 0.378,
+      "step": 734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.201219512195122e-05,
+      "loss": 0.3922,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.205574912891986e-05,
+      "loss": 0.3625,
+      "step": 736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.20993031358885e-05,
+      "loss": 0.386,
+      "step": 737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.3746,
+      "step": 738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.218641114982579e-05,
+      "loss": 0.3769,
+      "step": 739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.222996515679443e-05,
+      "loss": 0.369,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.227351916376307e-05,
+      "loss": 0.358,
+      "step": 741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.231707317073171e-05,
+      "loss": 0.3983,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.236062717770035e-05,
+      "loss": 0.366,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.240418118466899e-05,
+      "loss": 0.3638,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.244773519163763e-05,
+      "loss": 0.3851,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.249128919860627e-05,
+      "loss": 0.3754,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.253484320557491e-05,
+      "loss": 0.361,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.2578397212543556e-05,
+      "loss": 0.3954,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.26219512195122e-05,
+      "loss": 0.3915,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.2665505226480836e-05,
+      "loss": 0.3616,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.270905923344948e-05,
+      "loss": 0.3637,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.2752613240418116e-05,
+      "loss": 0.3838,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.279616724738676e-05,
+      "loss": 0.3913,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.28397212543554e-05,
+      "loss": 0.3779,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.288327526132404e-05,
+      "loss": 0.371,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.292682926829269e-05,
+      "loss": 0.388,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.2970383275261326e-05,
+      "loss": 0.3649,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.301393728222997e-05,
+      "loss": 0.352,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3057491289198605e-05,
+      "loss": 0.3759,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.310104529616725e-05,
+      "loss": 0.3852,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.314459930313589e-05,
+      "loss": 0.383,
+      "step": 761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.318815331010453e-05,
+      "loss": 0.3582,
+      "step": 762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.323170731707317e-05,
+      "loss": 0.3686,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3275261324041815e-05,
+      "loss": 0.3808,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.331881533101046e-05,
+      "loss": 0.3753,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3362369337979095e-05,
+      "loss": 0.3818,
+      "step": 766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.340592334494774e-05,
+      "loss": 0.3822,
+      "step": 767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.344947735191638e-05,
+      "loss": 0.3982,
+      "step": 768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.349303135888502e-05,
+      "loss": 0.3596,
+      "step": 769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.353658536585366e-05,
+      "loss": 0.3623,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.35801393728223e-05,
+      "loss": 0.386,
+      "step": 771
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.362369337979094e-05,
+      "loss": 0.3763,
+      "step": 772
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3667247386759584e-05,
+      "loss": 0.369,
+      "step": 773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.371080139372823e-05,
+      "loss": 0.3645,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.3754355400696864e-05,
+      "loss": 0.3614,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.379790940766551e-05,
+      "loss": 0.38,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.384146341463415e-05,
+      "loss": 0.369,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.388501742160279e-05,
+      "loss": 0.3783,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.3676,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.397212543554007e-05,
+      "loss": 0.3567,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.401567944250871e-05,
+      "loss": 0.3554,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4059233449477354e-05,
+      "loss": 0.3733,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4102787456446e-05,
+      "loss": 0.365,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.414634146341464e-05,
+      "loss": 0.3642,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4189895470383277e-05,
+      "loss": 0.3814,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.423344947735192e-05,
+      "loss": 0.4286,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4277003484320556e-05,
+      "loss": 0.3552,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.43205574912892e-05,
+      "loss": 0.3701,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4364111498257836e-05,
+      "loss": 0.3674,
+      "step": 789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4407665505226486e-05,
+      "loss": 0.3614,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.445121951219512e-05,
+      "loss": 0.3583,
+      "step": 791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4494773519163766e-05,
+      "loss": 0.3516,
+      "step": 792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.453832752613241e-05,
+      "loss": 0.3657,
+      "step": 793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4581881533101046e-05,
+      "loss": 0.3535,
+      "step": 794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.462543554006969e-05,
+      "loss": 0.3599,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4668989547038326e-05,
+      "loss": 0.3782,
+      "step": 796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.471254355400697e-05,
+      "loss": 0.3703,
+      "step": 797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.475609756097561e-05,
+      "loss": 0.3783,
+      "step": 798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4799651567944255e-05,
+      "loss": 0.3571,
+      "step": 799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.48432055749129e-05,
+      "loss": 0.3778,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4886759581881535e-05,
+      "loss": 0.3862,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.493031358885018e-05,
+      "loss": 0.3812,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.4973867595818815e-05,
+      "loss": 0.3688,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.501742160278746e-05,
+      "loss": 0.3825,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5060975609756095e-05,
+      "loss": 0.386,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.510452961672474e-05,
+      "loss": 0.3746,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.514808362369338e-05,
+      "loss": 0.3789,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5191637630662025e-05,
+      "loss": 0.3804,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.523519163763067e-05,
+      "loss": 0.3714,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5278745644599304e-05,
+      "loss": 0.3723,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.532229965156795e-05,
+      "loss": 0.356,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5365853658536584e-05,
+      "loss": 0.3765,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.540940766550523e-05,
+      "loss": 0.367,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5452961672473864e-05,
+      "loss": 0.3611,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.549651567944251e-05,
+      "loss": 0.3605,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.554006968641115e-05,
+      "loss": 0.3907,
+      "step": 816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5583623693379794e-05,
+      "loss": 0.3706,
+      "step": 817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.562717770034844e-05,
+      "loss": 0.3586,
+      "step": 818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5670731707317074e-05,
+      "loss": 0.3689,
+      "step": 819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.3619,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5757839721254353e-05,
+      "loss": 0.3768,
+      "step": 821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5801393728223e-05,
+      "loss": 0.3755,
+      "step": 822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.584494773519164e-05,
+      "loss": 0.3495,
+      "step": 823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5888501742160277e-05,
+      "loss": 0.3652,
+      "step": 824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.5932055749128927e-05,
+      "loss": 0.3701,
+      "step": 825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.597560975609756e-05,
+      "loss": 0.3696,
+      "step": 826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6019163763066206e-05,
+      "loss": 0.356,
+      "step": 827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.606271777003484e-05,
+      "loss": 0.3662,
+      "step": 828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6106271777003486e-05,
+      "loss": 0.3708,
+      "step": 829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.614982578397213e-05,
+      "loss": 0.3796,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6193379790940766e-05,
+      "loss": 0.3988,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.623693379790941e-05,
+      "loss": 0.3908,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.628048780487805e-05,
+      "loss": 0.3928,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6324041811846696e-05,
+      "loss": 0.3679,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.636759581881533e-05,
+      "loss": 0.359,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6411149825783976e-05,
+      "loss": 0.3873,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.645470383275261e-05,
+      "loss": 0.3626,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6498257839721255e-05,
+      "loss": 0.3705,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.65418118466899e-05,
+      "loss": 0.3714,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6585365853658535e-05,
+      "loss": 0.3809,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.662891986062718e-05,
+      "loss": 0.3652,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.667247386759582e-05,
+      "loss": 0.3811,
+      "step": 842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6716027874564465e-05,
+      "loss": 0.377,
+      "step": 843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.67595818815331e-05,
+      "loss": 0.3798,
+      "step": 844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6803135888501745e-05,
+      "loss": 0.3501,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.684668989547039e-05,
+      "loss": 0.3646,
+      "step": 846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6890243902439025e-05,
+      "loss": 0.3588,
+      "step": 847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.693379790940767e-05,
+      "loss": 0.3459,
+      "step": 848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6977351916376304e-05,
+      "loss": 0.3584,
+      "step": 849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.702090592334495e-05,
+      "loss": 0.3765,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.706445993031359e-05,
+      "loss": 0.3701,
+      "step": 851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.7108013937282234e-05,
+      "loss": 0.3814,
+      "step": 852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.715156794425087e-05,
+      "loss": 0.3442,
+      "step": 853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.7195121951219514e-05,
+      "loss": 0.3621,
+      "step": 854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.723867595818816e-05,
+      "loss": 0.3616,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.7282229965156794e-05,
+      "loss": 0.3622,
+      "step": 856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.732578397212544e-05,
+      "loss": 0.3484,
+      "step": 857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.7369337979094074e-05,
+      "loss": 0.3786,
+      "step": 858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.741289198606272e-05,
+      "loss": 0.3722,
+      "step": 859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.745644599303136e-05,
+      "loss": 0.3774,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.3762,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.754355400696865e-05,
+      "loss": 0.3582,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.758710801393728e-05,
+      "loss": 0.3884,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.7630662020905927e-05,
+      "loss": 0.3829,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.767421602787456e-05,
+      "loss": 0.3805,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.7717770034843206e-05,
+      "loss": 0.3802,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.776132404181185e-05,
+      "loss": 0.3893,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.780487804878049e-05,
+      "loss": 0.3603,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.7848432055749136e-05,
+      "loss": 0.3772,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.789198606271777e-05,
+      "loss": 0.3828,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.7935540069686416e-05,
+      "loss": 0.3673,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.797909407665505e-05,
+      "loss": 0.3563,
+      "step": 872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8022648083623696e-05,
+      "loss": 0.3725,
+      "step": 873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.806620209059233e-05,
+      "loss": 0.3842,
+      "step": 874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8109756097560976e-05,
+      "loss": 0.3749,
+      "step": 875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.815331010452962e-05,
+      "loss": 0.3878,
+      "step": 876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.819686411149826e-05,
+      "loss": 0.3543,
+      "step": 877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8240418118466905e-05,
+      "loss": 0.3582,
+      "step": 878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.828397212543554e-05,
+      "loss": 0.3716,
+      "step": 879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8327526132404185e-05,
+      "loss": 0.385,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.837108013937282e-05,
+      "loss": 0.3606,
+      "step": 881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8414634146341465e-05,
+      "loss": 0.3537,
+      "step": 882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.84581881533101e-05,
+      "loss": 0.3748,
+      "step": 883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8501742160278745e-05,
+      "loss": 0.3677,
+      "step": 884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.854529616724739e-05,
+      "loss": 0.3669,
+      "step": 885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.858885017421603e-05,
+      "loss": 0.3602,
+      "step": 886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8632404181184675e-05,
+      "loss": 0.3731,
+      "step": 887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.867595818815331e-05,
+      "loss": 0.3735,
+      "step": 888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.8719512195121954e-05,
+      "loss": 0.3561,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.876306620209059e-05,
+      "loss": 0.3723,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8806620209059234e-05,
+      "loss": 0.3678,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.885017421602787e-05,
+      "loss": 0.377,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8893728222996514e-05,
+      "loss": 0.3876,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.8937282229965164e-05,
+      "loss": 0.369,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.89808362369338e-05,
+      "loss": 0.3521,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.3572,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.906794425087108e-05,
+      "loss": 0.374,
+      "step": 897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9111498257839724e-05,
+      "loss": 0.3692,
+      "step": 898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.915505226480836e-05,
+      "loss": 0.3703,
+      "step": 899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9198606271777003e-05,
+      "loss": 0.3775,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.924216027874565e-05,
+      "loss": 0.3683,
+      "step": 901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.3535,
+      "step": 902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.932926829268293e-05,
+      "loss": 0.354,
+      "step": 903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.937282229965157e-05,
+      "loss": 0.3602,
+      "step": 904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.941637630662021e-05,
+      "loss": 0.3656,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.945993031358885e-05,
+      "loss": 0.3642,
+      "step": 906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.950348432055749e-05,
+      "loss": 0.3793,
+      "step": 907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9547038327526136e-05,
+      "loss": 0.3592,
+      "step": 908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.959059233449477e-05,
+      "loss": 0.3869,
+      "step": 909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9634146341463416e-05,
+      "loss": 0.3734,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.967770034843206e-05,
+      "loss": 0.3719,
+      "step": 911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.97212543554007e-05,
+      "loss": 0.3728,
+      "step": 912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.976480836236934e-05,
+      "loss": 0.3817,
+      "step": 913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.980836236933798e-05,
+      "loss": 0.3855,
+      "step": 914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.985191637630662e-05,
+      "loss": 0.3739,
+      "step": 915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.989547038327526e-05,
+      "loss": 0.3758,
+      "step": 916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9939024390243905e-05,
+      "loss": 0.3673,
+      "step": 917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.998257839721254e-05,
+      "loss": 0.3656,
+      "step": 918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.0026132404181185e-05,
+      "loss": 0.3807,
+      "step": 919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.006968641114983e-05,
+      "loss": 0.3953,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.011324041811847e-05,
+      "loss": 0.3618,
+      "step": 921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.015679442508711e-05,
+      "loss": 0.3475,
+      "step": 922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.020034843205575e-05,
+      "loss": 0.3776,
+      "step": 923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.0243902439024395e-05,
+      "loss": 0.3934,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.028745644599303e-05,
+      "loss": 0.3556,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.0331010452961675e-05,
+      "loss": 0.3832,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.037456445993031e-05,
+      "loss": 0.3717,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.0418118466898954e-05,
+      "loss": 0.3616,
+      "step": 928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.04616724738676e-05,
+      "loss": 0.3761,
+      "step": 929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.050522648083624e-05,
+      "loss": 0.3718,
+      "step": 930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.0548780487804884e-05,
+      "loss": 0.3707,
+      "step": 931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.059233449477352e-05,
+      "loss": 0.3679,
+      "step": 932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.0635888501742164e-05,
+      "loss": 0.3751,
+      "step": 933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.06794425087108e-05,
+      "loss": 0.3949,
+      "step": 934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.0722996515679444e-05,
+      "loss": 0.3684,
+      "step": 935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.076655052264808e-05,
+      "loss": 0.3612,
+      "step": 936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.081010452961673e-05,
+      "loss": 0.3538,
+      "step": 937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.085365853658537e-05,
+      "loss": 0.3578,
+      "step": 938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.089721254355401e-05,
+      "loss": 0.3868,
+      "step": 939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.0940766550522653e-05,
+      "loss": 0.4069,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.098432055749129e-05,
+      "loss": 0.362,
+      "step": 941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.102787456445993e-05,
+      "loss": 0.3864,
+      "step": 942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 0.3782,
+      "step": 943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.111498257839721e-05,
+      "loss": 0.3738,
+      "step": 944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1158536585365856e-05,
+      "loss": 0.362,
+      "step": 945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.12020905923345e-05,
+      "loss": 0.3863,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.124564459930314e-05,
+      "loss": 0.3472,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.128919860627178e-05,
+      "loss": 0.3566,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.133275261324042e-05,
+      "loss": 0.384,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.137630662020906e-05,
+      "loss": 0.3862,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.14198606271777e-05,
+      "loss": 0.3731,
+      "step": 951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.146341463414634e-05,
+      "loss": 0.3672,
+      "step": 952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.150696864111498e-05,
+      "loss": 0.3643,
+      "step": 953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1550522648083626e-05,
+      "loss": 0.3739,
+      "step": 954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.159407665505227e-05,
+      "loss": 0.383,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.163763066202091e-05,
+      "loss": 0.378,
+      "step": 956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.168118466898955e-05,
+      "loss": 0.3712,
+      "step": 957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.172473867595819e-05,
+      "loss": 0.3575,
+      "step": 958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.176829268292683e-05,
+      "loss": 0.3811,
+      "step": 959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.181184668989547e-05,
+      "loss": 0.3777,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.185540069686411e-05,
+      "loss": 0.358,
+      "step": 961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.189895470383275e-05,
+      "loss": 0.378,
+      "step": 962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1942508710801395e-05,
+      "loss": 0.3685,
+      "step": 963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.198606271777004e-05,
+      "loss": 0.3734,
+      "step": 964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.202961672473868e-05,
+      "loss": 0.3714,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.207317073170732e-05,
+      "loss": 0.3569,
+      "step": 966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.211672473867596e-05,
+      "loss": 0.3741,
+      "step": 967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.21602787456446e-05,
+      "loss": 0.3521,
+      "step": 968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.220383275261324e-05,
+      "loss": 0.3606,
+      "step": 969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.2247386759581884e-05,
+      "loss": 0.3704,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.229094076655053e-05,
+      "loss": 0.3603,
+      "step": 971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.233449477351917e-05,
+      "loss": 0.3764,
+      "step": 972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.237804878048781e-05,
+      "loss": 0.3755,
+      "step": 973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.242160278745645e-05,
+      "loss": 0.3531,
+      "step": 974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.246515679442509e-05,
+      "loss": 0.3731,
+      "step": 975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.250871080139373e-05,
+      "loss": 0.374,
+      "step": 976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.255226480836237e-05,
+      "loss": 0.3592,
+      "step": 977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.259581881533101e-05,
+      "loss": 0.3793,
+      "step": 978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.2639372822299653e-05,
+      "loss": 0.3868,
+      "step": 979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.26829268292683e-05,
+      "loss": 0.3636,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.272648083623694e-05,
+      "loss": 0.3502,
+      "step": 981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.2770034843205577e-05,
+      "loss": 0.3805,
+      "step": 982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.281358885017422e-05,
+      "loss": 0.3619,
+      "step": 983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.3765,
+      "step": 984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.29006968641115e-05,
+      "loss": 0.3741,
+      "step": 985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.294425087108014e-05,
+      "loss": 0.3592,
+      "step": 986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.298780487804878e-05,
+      "loss": 0.3453,
+      "step": 987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.303135888501742e-05,
+      "loss": 0.3843,
+      "step": 988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.3074912891986066e-05,
+      "loss": 0.3961,
+      "step": 989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.311846689895471e-05,
+      "loss": 0.3787,
+      "step": 990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.3162020905923346e-05,
+      "loss": 0.3921,
+      "step": 991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.320557491289199e-05,
+      "loss": 0.3918,
+      "step": 992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.324912891986063e-05,
+      "loss": 0.3725,
+      "step": 993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.329268292682927e-05,
+      "loss": 0.3702,
+      "step": 994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.333623693379791e-05,
+      "loss": 0.3789,
+      "step": 995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.337979094076655e-05,
+      "loss": 0.3872,
+      "step": 996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.342334494773519e-05,
+      "loss": 0.3472,
+      "step": 997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.3466898954703835e-05,
+      "loss": 0.3734,
+      "step": 998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.351045296167248e-05,
+      "loss": 0.3732,
+      "step": 999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.3554006968641115e-05,
+      "loss": 0.3678,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.359756097560976e-05,
+      "loss": 0.3517,
+      "step": 1001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.36411149825784e-05,
+      "loss": 0.373,
+      "step": 1002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.368466898954704e-05,
+      "loss": 0.3834,
+      "step": 1003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.372822299651568e-05,
+      "loss": 0.3875,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.377177700348432e-05,
+      "loss": 0.3801,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.381533101045297e-05,
+      "loss": 0.3568,
+      "step": 1006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.3858885017421604e-05,
+      "loss": 0.3742,
+      "step": 1007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.390243902439025e-05,
+      "loss": 0.3657,
+      "step": 1008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.394599303135889e-05,
+      "loss": 0.3778,
+      "step": 1009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.398954703832753e-05,
+      "loss": 0.3523,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.403310104529617e-05,
+      "loss": 0.3917,
+      "step": 1011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.407665505226481e-05,
+      "loss": 0.384,
+      "step": 1012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.412020905923345e-05,
+      "loss": 0.3726,
+      "step": 1013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.4163763066202094e-05,
+      "loss": 0.3781,
+      "step": 1014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.420731707317074e-05,
+      "loss": 0.3719,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.4250871080139374e-05,
+      "loss": 0.3636,
+      "step": 1016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.429442508710802e-05,
+      "loss": 0.3663,
+      "step": 1017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.433797909407666e-05,
+      "loss": 0.3575,
+      "step": 1018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.43815331010453e-05,
+      "loss": 0.3582,
+      "step": 1019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.442508710801394e-05,
+      "loss": 0.3688,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.4468641114982576e-05,
+      "loss": 0.3774,
+      "step": 1021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.451219512195122e-05,
+      "loss": 0.3548,
+      "step": 1022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.455574912891986e-05,
+      "loss": 0.38,
+      "step": 1023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.4599303135888506e-05,
+      "loss": 0.3526,
+      "step": 1024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 0.3833,
+      "step": 1025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.4686411149825786e-05,
+      "loss": 0.3587,
+      "step": 1026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.472996515679443e-05,
+      "loss": 0.371,
+      "step": 1027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.4773519163763066e-05,
+      "loss": 0.3657,
+      "step": 1028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.481707317073171e-05,
+      "loss": 0.368,
+      "step": 1029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.4860627177700346e-05,
+      "loss": 0.3656,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.490418118466899e-05,
+      "loss": 0.3855,
+      "step": 1031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.494773519163763e-05,
+      "loss": 0.3777,
+      "step": 1032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.4991289198606276e-05,
+      "loss": 0.3729,
+      "step": 1033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.503484320557492e-05,
+      "loss": 0.3742,
+      "step": 1034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5078397212543555e-05,
+      "loss": 0.3631,
+      "step": 1035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.51219512195122e-05,
+      "loss": 0.3595,
+      "step": 1036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5165505226480835e-05,
+      "loss": 0.3769,
+      "step": 1037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.520905923344948e-05,
+      "loss": 0.383,
+      "step": 1038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5252613240418115e-05,
+      "loss": 0.3783,
+      "step": 1039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.529616724738676e-05,
+      "loss": 0.3838,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.533972125435541e-05,
+      "loss": 0.3487,
+      "step": 1041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5383275261324045e-05,
+      "loss": 0.3683,
+      "step": 1042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.542682926829269e-05,
+      "loss": 0.3719,
+      "step": 1043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5470383275261325e-05,
+      "loss": 0.3616,
+      "step": 1044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.551393728222997e-05,
+      "loss": 0.3729,
+      "step": 1045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5557491289198604e-05,
+      "loss": 0.3742,
+      "step": 1046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.560104529616725e-05,
+      "loss": 0.3611,
+      "step": 1047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.564459930313589e-05,
+      "loss": 0.3649,
+      "step": 1048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5688153310104534e-05,
+      "loss": 0.3551,
+      "step": 1049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.573170731707318e-05,
+      "loss": 0.3788,
+      "step": 1050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5775261324041814e-05,
+      "loss": 0.3798,
+      "step": 1051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.581881533101046e-05,
+      "loss": 0.3685,
+      "step": 1052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5862369337979094e-05,
+      "loss": 0.3835,
+      "step": 1053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.590592334494774e-05,
+      "loss": 0.3827,
+      "step": 1054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.594947735191638e-05,
+      "loss": 0.3597,
+      "step": 1055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.599303135888502e-05,
+      "loss": 0.3831,
+      "step": 1056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.603658536585366e-05,
+      "loss": 0.3641,
+      "step": 1057
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6080139372822303e-05,
+      "loss": 0.3585,
+      "step": 1058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.612369337979095e-05,
+      "loss": 0.372,
+      "step": 1059
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.616724738675958e-05,
+      "loss": 0.3825,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6210801393728226e-05,
+      "loss": 0.369,
+      "step": 1061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.625435540069686e-05,
+      "loss": 0.3752,
+      "step": 1062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6297909407665506e-05,
+      "loss": 0.394,
+      "step": 1063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.634146341463415e-05,
+      "loss": 0.3656,
+      "step": 1064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6385017421602786e-05,
+      "loss": 0.3877,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 0.3324,
+      "step": 1066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.647212543554007e-05,
+      "loss": 0.3702,
+      "step": 1067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6515679442508716e-05,
+      "loss": 0.3734,
+      "step": 1068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.655923344947735e-05,
+      "loss": 0.3695,
+      "step": 1069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6602787456445996e-05,
+      "loss": 0.3683,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.664634146341464e-05,
+      "loss": 0.3519,
+      "step": 1071
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6689895470383275e-05,
+      "loss": 0.3514,
+      "step": 1072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.673344947735192e-05,
+      "loss": 0.3609,
+      "step": 1073
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6777003484320555e-05,
+      "loss": 0.3449,
+      "step": 1074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6820557491289205e-05,
+      "loss": 0.3626,
+      "step": 1075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686411149825784e-05,
+      "loss": 0.37,
+      "step": 1076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6907665505226485e-05,
+      "loss": 0.3505,
+      "step": 1077
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.695121951219512e-05,
+      "loss": 0.3745,
+      "step": 1078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6994773519163765e-05,
+      "loss": 0.3607,
+      "step": 1079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.703832752613241e-05,
+      "loss": 0.3621,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7081881533101045e-05,
+      "loss": 0.3779,
+      "step": 1081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.712543554006969e-05,
+      "loss": 0.3702,
+      "step": 1082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.716898954703833e-05,
+      "loss": 0.362,
+      "step": 1083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7212543554006975e-05,
+      "loss": 0.3417,
+      "step": 1084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.725609756097561e-05,
+      "loss": 0.3845,
+      "step": 1085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7299651567944254e-05,
+      "loss": 0.3799,
+      "step": 1086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.73432055749129e-05,
+      "loss": 0.3834,
+      "step": 1087
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7386759581881534e-05,
+      "loss": 0.3696,
+      "step": 1088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.743031358885018e-05,
+      "loss": 0.3481,
+      "step": 1089
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7473867595818814e-05,
+      "loss": 0.3725,
+      "step": 1090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.751742160278746e-05,
+      "loss": 0.3663,
+      "step": 1091
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.75609756097561e-05,
+      "loss": 0.3735,
+      "step": 1092
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7604529616724744e-05,
+      "loss": 0.3725,
+      "step": 1093
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.764808362369339e-05,
+      "loss": 0.4315,
+      "step": 1094
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7691637630662024e-05,
+      "loss": 0.3658,
+      "step": 1095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.773519163763067e-05,
+      "loss": 0.4336,
+      "step": 1096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.77787456445993e-05,
+      "loss": 0.3628,
+      "step": 1097
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.782229965156795e-05,
+      "loss": 0.362,
+      "step": 1098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.786585365853658e-05,
+      "loss": 0.3682,
+      "step": 1099
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7909407665505226e-05,
+      "loss": 0.3684,
+      "step": 1100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.795296167247387e-05,
+      "loss": 0.3672,
+      "step": 1101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.799651567944251e-05,
+      "loss": 0.3733,
+      "step": 1102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8040069686411156e-05,
+      "loss": 0.3703,
+      "step": 1103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.808362369337979e-05,
+      "loss": 0.3664,
+      "step": 1104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8127177700348436e-05,
+      "loss": 0.3672,
+      "step": 1105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.817073170731707e-05,
+      "loss": 0.3676,
+      "step": 1106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 0.3556,
+      "step": 1107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.825783972125435e-05,
+      "loss": 0.3527,
+      "step": 1108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8301393728222996e-05,
+      "loss": 0.3656,
+      "step": 1109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8344947735191646e-05,
+      "loss": 0.3695,
+      "step": 1110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.838850174216028e-05,
+      "loss": 0.3788,
+      "step": 1111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8432055749128926e-05,
+      "loss": 0.3831,
+      "step": 1112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.847560975609756e-05,
+      "loss": 0.3792,
+      "step": 1113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8519163763066205e-05,
+      "loss": 0.365,
+      "step": 1114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.856271777003484e-05,
+      "loss": 0.3778,
+      "step": 1115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8606271777003485e-05,
+      "loss": 0.3719,
+      "step": 1116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.864982578397213e-05,
+      "loss": 0.3438,
+      "step": 1117
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.869337979094077e-05,
+      "loss": 0.3735,
+      "step": 1118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8736933797909415e-05,
+      "loss": 0.3855,
+      "step": 1119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.878048780487805e-05,
+      "loss": 0.376,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8824041811846695e-05,
+      "loss": 0.3819,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.886759581881533e-05,
+      "loss": 0.3627,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8911149825783975e-05,
+      "loss": 0.4004,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.895470383275261e-05,
+      "loss": 0.3469,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8998257839721254e-05,
+      "loss": 0.3598,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.90418118466899e-05,
+      "loss": 0.3695,
+      "step": 1126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.908536585365854e-05,
+      "loss": 0.3694,
+      "step": 1127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9128919860627184e-05,
+      "loss": 0.3702,
+      "step": 1128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.917247386759582e-05,
+      "loss": 0.3725,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9216027874564464e-05,
+      "loss": 0.3829,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.92595818815331e-05,
+      "loss": 0.3847,
+      "step": 1131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9303135888501744e-05,
+      "loss": 0.3702,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.934668989547039e-05,
+      "loss": 0.3702,
+      "step": 1133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9390243902439024e-05,
+      "loss": 0.3579,
+      "step": 1134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.943379790940767e-05,
+      "loss": 0.362,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.947735191637631e-05,
+      "loss": 0.3769,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.952090592334495e-05,
+      "loss": 0.3707,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.956445993031359e-05,
+      "loss": 0.3529,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.960801393728223e-05,
+      "loss": 0.3562,
+      "step": 1139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.965156794425087e-05,
+      "loss": 0.3704,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.969512195121951e-05,
+      "loss": 0.3664,
+      "step": 1141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9738675958188156e-05,
+      "loss": 0.3709,
+      "step": 1142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.978222996515679e-05,
+      "loss": 0.3443,
+      "step": 1143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9825783972125436e-05,
+      "loss": 0.3747,
+      "step": 1144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.986933797909408e-05,
+      "loss": 0.3722,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.991289198606272e-05,
+      "loss": 0.3654,
+      "step": 1146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.995644599303136e-05,
+      "loss": 0.3729,
+      "step": 1147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5e-05,
+      "loss": 0.381,
+      "step": 1148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999998958670857e-05,
+      "loss": 0.3722,
+      "step": 1149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999999583468353e-05,
+      "loss": 0.3716,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999999062803829e-05,
+      "loss": 0.3558,
+      "step": 1151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999983338735634e-05,
+      "loss": 0.3672,
+      "step": 1152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999973966776235e-05,
+      "loss": 0.3709,
+      "step": 1153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999996251216096e-05,
+      "loss": 0.3703,
+      "step": 1154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999994897489086e-05,
+      "loss": 0.3678,
+      "step": 1155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999993335496721e-05,
+      "loss": 0.3598,
+      "step": 1156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999915652391433e-05,
+      "loss": 0.3621,
+      "step": 1157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999895867165174e-05,
+      "loss": 0.3586,
+      "step": 1158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999987399929027e-05,
+      "loss": 0.3571,
+      "step": 1159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999985004876875e-05,
+      "loss": 0.3635,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999982401560282e-05,
+      "loss": 0.3705,
+      "step": 1161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999979589979489e-05,
+      "loss": 0.3726,
+      "step": 1162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999765701347556e-05,
+      "loss": 0.3967,
+      "step": 1163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999733420263635e-05,
+      "loss": 0.3546,
+      "step": 1164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999699056546096e-05,
+      "loss": 0.3864,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999966261019813e-05,
+      "loss": 0.375,
+      "step": 1166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.99996240812231e-05,
+      "loss": 0.3675,
+      "step": 1167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999583469624585e-05,
+      "loss": 0.3683,
+      "step": 1168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999954077540634e-05,
+      "loss": 0.3658,
+      "step": 1169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999949599857231e-05,
+      "loss": 0.3609,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999944913912665e-05,
+      "loss": 0.3541,
+      "step": 1171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999940019707369e-05,
+      "loss": 0.3645,
+      "step": 1172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999349172417956e-05,
+      "loss": 0.3755,
+      "step": 1173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999296065164186e-05,
+      "loss": 0.3823,
+      "step": 1174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9999240875317285e-05,
+      "loss": 0.3763,
+      "step": 1175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.999918360288237e-05,
+      "loss": 0.3837,
+      "step": 1176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999912424786472e-05,
+      "loss": 0.3602,
+      "step": 1177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999906281026985e-05,
+      "loss": 0.3539,
+      "step": 1178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999899929010345e-05,
+      "loss": 0.3668,
+      "step": 1179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999893368737139e-05,
+      "loss": 0.3385,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999886600207974e-05,
+      "loss": 0.3396,
+      "step": 1181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999879623423476e-05,
+      "loss": 0.3681,
+      "step": 1182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999872438384293e-05,
+      "loss": 0.3463,
+      "step": 1183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999865045091089e-05,
+      "loss": 0.3692,
+      "step": 1184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999857443544547e-05,
+      "loss": 0.3737,
+      "step": 1185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9998496337453726e-05,
+      "loss": 0.3519,
+      "step": 1186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999841615694287e-05,
+      "loss": 0.4049,
+      "step": 1187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9998333893920336e-05,
+      "loss": 0.368,
+      "step": 1188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999824954839374e-05,
+      "loss": 0.3756,
+      "step": 1189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9998163120370874e-05,
+      "loss": 0.3577,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999807460985975e-05,
+      "loss": 0.365,
+      "step": 1191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999798401686856e-05,
+      "loss": 0.3609,
+      "step": 1192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999789134140569e-05,
+      "loss": 0.3789,
+      "step": 1193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.99977965834797e-05,
+      "loss": 0.3667,
+      "step": 1194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.99976997430994e-05,
+      "loss": 0.3762,
+      "step": 1195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999760082027372e-05,
+      "loss": 0.3573,
+      "step": 1196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999749981501183e-05,
+      "loss": 0.3551,
+      "step": 1197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999739672732307e-05,
+      "loss": 0.3446,
+      "step": 1198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999729155721699e-05,
+      "loss": 0.3538,
+      "step": 1199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999718430470333e-05,
+      "loss": 0.3528,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999707496979201e-05,
+      "loss": 0.3654,
+      "step": 1201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9996963552493145e-05,
+      "loss": 0.3735,
+      "step": 1202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9996850052817064e-05,
+      "loss": 0.3777,
+      "step": 1203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999673447077425e-05,
+      "loss": 0.369,
+      "step": 1204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999661680637543e-05,
+      "loss": 0.3688,
+      "step": 1205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999649705963148e-05,
+      "loss": 0.3669,
+      "step": 1206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999637523055348e-05,
+      "loss": 0.3653,
+      "step": 1207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9996251319152704e-05,
+      "loss": 0.3637,
+      "step": 1208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999612532544063e-05,
+      "loss": 0.3749,
+      "step": 1209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999599724942893e-05,
+      "loss": 0.3624,
+      "step": 1210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999586709112944e-05,
+      "loss": 0.3779,
+      "step": 1211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999573485055422e-05,
+      "loss": 0.3582,
+      "step": 1212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.99956005277155e-05,
+      "loss": 0.352,
+      "step": 1213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9995464122625726e-05,
+      "loss": 0.3598,
+      "step": 1214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9995325635297506e-05,
+      "loss": 0.3637,
+      "step": 1215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999518506574368e-05,
+      "loss": 0.3688,
+      "step": 1216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999504241397725e-05,
+      "loss": 0.3791,
+      "step": 1217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9994897680011416e-05,
+      "loss": 0.3525,
+      "step": 1218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999475086385958e-05,
+      "loss": 0.3723,
+      "step": 1219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9994601965535344e-05,
+      "loss": 0.377,
+      "step": 1220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999445098505247e-05,
+      "loss": 0.3551,
+      "step": 1221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9994297922424926e-05,
+      "loss": 0.355,
+      "step": 1222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999414277766691e-05,
+      "loss": 0.3692,
+      "step": 1223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999398555079276e-05,
+      "loss": 0.3547,
+      "step": 1224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999382624181704e-05,
+      "loss": 0.3606,
+      "step": 1225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999366485075448e-05,
+      "loss": 0.3683,
+      "step": 1226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999350137762005e-05,
+      "loss": 0.3827,
+      "step": 1227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9993335822428846e-05,
+      "loss": 0.3605,
+      "step": 1228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999316818519621e-05,
+      "loss": 0.3667,
+      "step": 1229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9992998465937674e-05,
+      "loss": 0.3709,
+      "step": 1230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999282666466892e-05,
+      "loss": 0.3553,
+      "step": 1231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.999265278140586e-05,
+      "loss": 0.3616,
+      "step": 1232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9992476816164594e-05,
+      "loss": 0.3877,
+      "step": 1233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9992298768961406e-05,
+      "loss": 0.3648,
+      "step": 1234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9992118639812775e-05,
+      "loss": 0.3699,
+      "step": 1235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.999193642873537e-05,
+      "loss": 0.3488,
+      "step": 1236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9991752135746076e-05,
+      "loss": 0.3677,
+      "step": 1237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9991565760861936e-05,
+      "loss": 0.3965,
+      "step": 1238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9991377304100204e-05,
+      "loss": 0.3662,
+      "step": 1239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.999118676547832e-05,
+      "loss": 0.3528,
+      "step": 1240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9990994145013924e-05,
+      "loss": 0.3525,
+      "step": 1241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9990799442724856e-05,
+      "loss": 0.3637,
+      "step": 1242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.999060265862912e-05,
+      "loss": 0.3654,
+      "step": 1243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.999040379274494e-05,
+      "loss": 0.3687,
+      "step": 1244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.999020284509072e-05,
+      "loss": 0.3729,
+      "step": 1245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9989999815685076e-05,
+      "loss": 0.3638,
+      "step": 1246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998979470454677e-05,
+      "loss": 0.3496,
+      "step": 1247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998958751169482e-05,
+      "loss": 0.3537,
+      "step": 1248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998937823714838e-05,
+      "loss": 0.3567,
+      "step": 1249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9989166880926834e-05,
+      "loss": 0.3397,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9988953443049755e-05,
+      "loss": 0.3777,
+      "step": 1251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9988737923536874e-05,
+      "loss": 0.3628,
+      "step": 1252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9988520322408155e-05,
+      "loss": 0.3625,
+      "step": 1253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998830063968374e-05,
+      "loss": 0.3723,
+      "step": 1254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9988078875383956e-05,
+      "loss": 0.3605,
+      "step": 1255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998785502952934e-05,
+      "loss": 0.355,
+      "step": 1256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998762910214061e-05,
+      "loss": 0.3746,
+      "step": 1257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998740109323867e-05,
+      "loss": 0.356,
+      "step": 1258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998717100284463e-05,
+      "loss": 0.3593,
+      "step": 1259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998693883097979e-05,
+      "loss": 0.3741,
+      "step": 1260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9986704577665646e-05,
+      "loss": 0.3745,
+      "step": 1261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9986468242923864e-05,
+      "loss": 0.3741,
+      "step": 1262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998622982677633e-05,
+      "loss": 0.3658,
+      "step": 1263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998598932924511e-05,
+      "loss": 0.3653,
+      "step": 1264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998574675035248e-05,
+      "loss": 0.3753,
+      "step": 1265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9985502090120867e-05,
+      "loss": 0.3529,
+      "step": 1266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998525534857294e-05,
+      "loss": 0.3729,
+      "step": 1267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9985006525731524e-05,
+      "loss": 0.3762,
+      "step": 1268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998475562161966e-05,
+      "loss": 0.3696,
+      "step": 1269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998450263626056e-05,
+      "loss": 0.3678,
+      "step": 1270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998424756967766e-05,
+      "loss": 0.3648,
+      "step": 1271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9983990421894555e-05,
+      "loss": 0.3608,
+      "step": 1272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998373119293505e-05,
+      "loss": 0.3738,
+      "step": 1273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998346988282314e-05,
+      "loss": 0.3585,
+      "step": 1274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9983206491583014e-05,
+      "loss": 0.3683,
+      "step": 1275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998294101923906e-05,
+      "loss": 0.3538,
+      "step": 1276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9982673465815826e-05,
+      "loss": 0.3552,
+      "step": 1277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9982403831338104e-05,
+      "loss": 0.3545,
+      "step": 1278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998213211583084e-05,
+      "loss": 0.367,
+      "step": 1279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998185831931919e-05,
+      "loss": 0.3804,
+      "step": 1280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998158244182849e-05,
+      "loss": 0.3428,
+      "step": 1281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998130448338428e-05,
+      "loss": 0.3597,
+      "step": 1282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.99810244440123e-05,
+      "loss": 0.3609,
+      "step": 1283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998074232373845e-05,
+      "loss": 0.3665,
+      "step": 1284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9980458122588855e-05,
+      "loss": 0.3673,
+      "step": 1285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.998017184058982e-05,
+      "loss": 0.3664,
+      "step": 1286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9979883477767835e-05,
+      "loss": 0.3475,
+      "step": 1287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9979593034149616e-05,
+      "loss": 0.3543,
+      "step": 1288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9979300509762017e-05,
+      "loss": 0.3627,
+      "step": 1289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9979005904632134e-05,
+      "loss": 0.361,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9978709218787236e-05,
+      "loss": 0.3574,
+      "step": 1291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997841045225478e-05,
+      "loss": 0.361,
+      "step": 1292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997810960506242e-05,
+      "loss": 0.3672,
+      "step": 1293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997780667723801e-05,
+      "loss": 0.3343,
+      "step": 1294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997750166880958e-05,
+      "loss": 0.3675,
+      "step": 1295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997719457980537e-05,
+      "loss": 0.3736,
+      "step": 1296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.99768854102538e-05,
+      "loss": 0.3778,
+      "step": 1297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9976574160183486e-05,
+      "loss": 0.3676,
+      "step": 1298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9976260829623244e-05,
+      "loss": 0.3756,
+      "step": 1299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997594541860207e-05,
+      "loss": 0.3604,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9975627927149176e-05,
+      "loss": 0.3827,
+      "step": 1301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9975308355293924e-05,
+      "loss": 0.3752,
+      "step": 1302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997498670306591e-05,
+      "loss": 0.3706,
+      "step": 1303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9974662970494915e-05,
+      "loss": 0.3668,
+      "step": 1304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9974337157610885e-05,
+      "loss": 0.38,
+      "step": 1305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997400926444399e-05,
+      "loss": 0.379,
+      "step": 1306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997367929102458e-05,
+      "loss": 0.3666,
+      "step": 1307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997334723738319e-05,
+      "loss": 0.3569,
+      "step": 1308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9973013103550564e-05,
+      "loss": 0.3584,
+      "step": 1309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997267688955763e-05,
+      "loss": 0.3702,
+      "step": 1310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9972338595435505e-05,
+      "loss": 0.3666,
+      "step": 1311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997199822121551e-05,
+      "loss": 0.3783,
+      "step": 1312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9971655766929134e-05,
+      "loss": 0.3616,
+      "step": 1313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.99713112326081e-05,
+      "loss": 0.3578,
+      "step": 1314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9970964618284274e-05,
+      "loss": 0.3623,
+      "step": 1315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997061592398975e-05,
+      "loss": 0.3424,
+      "step": 1316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997026514975682e-05,
+      "loss": 0.3549,
+      "step": 1317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9969912295617924e-05,
+      "loss": 0.3544,
+      "step": 1318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9969557361605744e-05,
+      "loss": 0.3776,
+      "step": 1319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9969200347753126e-05,
+      "loss": 0.3521,
+      "step": 1320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996884125409312e-05,
+      "loss": 0.3715,
+      "step": 1321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9968480080658956e-05,
+      "loss": 0.3745,
+      "step": 1322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996811682748407e-05,
+      "loss": 0.3694,
+      "step": 1323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996775149460209e-05,
+      "loss": 0.3502,
+      "step": 1324
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9967384082046825e-05,
+      "loss": 0.3566,
+      "step": 1325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996701458985229e-05,
+      "loss": 0.367,
+      "step": 1326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9966643018052686e-05,
+      "loss": 0.3606,
+      "step": 1327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.99662693666824e-05,
+      "loss": 0.3846,
+      "step": 1328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9965893635776015e-05,
+      "loss": 0.3618,
+      "step": 1329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996551582536832e-05,
+      "loss": 0.3575,
+      "step": 1330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9965135935494286e-05,
+      "loss": 0.3603,
+      "step": 1331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9964753966189075e-05,
+      "loss": 0.368,
+      "step": 1332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996436991748804e-05,
+      "loss": 0.371,
+      "step": 1333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996398378942673e-05,
+      "loss": 0.3549,
+      "step": 1334
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996359558204088e-05,
+      "loss": 0.3678,
+      "step": 1335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996320529536644e-05,
+      "loss": 0.3492,
+      "step": 1336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996281292943953e-05,
+      "loss": 0.3734,
+      "step": 1337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996241848429645e-05,
+      "loss": 0.3449,
+      "step": 1338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9962021959973736e-05,
+      "loss": 0.344,
+      "step": 1339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996162335650808e-05,
+      "loss": 0.3709,
+      "step": 1340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9961222673936376e-05,
+      "loss": 0.3715,
+      "step": 1341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996081991229572e-05,
+      "loss": 0.3574,
+      "step": 1342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996041507162338e-05,
+      "loss": 0.3535,
+      "step": 1343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996000815195684e-05,
+      "loss": 0.3549,
+      "step": 1344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.995959915333376e-05,
+      "loss": 0.3655,
+      "step": 1345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9959188075792e-05,
+      "loss": 0.3672,
+      "step": 1346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9958774919369624e-05,
+      "loss": 0.3903,
+      "step": 1347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9958359684104846e-05,
+      "loss": 0.353,
+      "step": 1348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9957942370036114e-05,
+      "loss": 0.3776,
+      "step": 1349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995752297720207e-05,
+      "loss": 0.3527,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995710150564152e-05,
+      "loss": 0.3854,
+      "step": 1351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995667795539347e-05,
+      "loss": 0.3404,
+      "step": 1352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995625232649714e-05,
+      "loss": 0.3627,
+      "step": 1353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9955824618991924e-05,
+      "loss": 0.3829,
+      "step": 1354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9955394832917404e-05,
+      "loss": 0.3723,
+      "step": 1355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995496296831337e-05,
+      "loss": 0.3496,
+      "step": 1356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995452902521979e-05,
+      "loss": 0.3774,
+      "step": 1357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995409300367684e-05,
+      "loss": 0.3506,
+      "step": 1358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9953654903724866e-05,
+      "loss": 0.3669,
+      "step": 1359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995321472540444e-05,
+      "loss": 0.3774,
+      "step": 1360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995277246875628e-05,
+      "loss": 0.3619,
+      "step": 1361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995232813382134e-05,
+      "loss": 0.3794,
+      "step": 1362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995188172064075e-05,
+      "loss": 0.3448,
+      "step": 1363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9951433229255824e-05,
+      "loss": 0.3585,
+      "step": 1364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995098265970807e-05,
+      "loss": 0.3921,
+      "step": 1365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.99505300120392e-05,
+      "loss": 0.3628,
+      "step": 1366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9950075286291125e-05,
+      "loss": 0.3585,
+      "step": 1367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994961848250592e-05,
+      "loss": 0.3676,
+      "step": 1368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9949159600725863e-05,
+      "loss": 0.3821,
+      "step": 1369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994869864099345e-05,
+      "loss": 0.3535,
+      "step": 1370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9948235603351326e-05,
+      "loss": 0.3763,
+      "step": 1371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9947770487842366e-05,
+      "loss": 0.3806,
+      "step": 1372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9947303294509626e-05,
+      "loss": 0.3573,
+      "step": 1373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994683402339633e-05,
+      "loss": 0.3659,
+      "step": 1374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994636267454593e-05,
+      "loss": 0.3464,
+      "step": 1375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994588924800206e-05,
+      "loss": 0.3621,
+      "step": 1376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994541374380853e-05,
+      "loss": 0.3641,
+      "step": 1377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9944936162009365e-05,
+      "loss": 0.4407,
+      "step": 1378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9944456502648754e-05,
+      "loss": 0.3385,
+      "step": 1379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994397476577111e-05,
+      "loss": 0.3488,
+      "step": 1380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994349095142102e-05,
+      "loss": 0.3614,
+      "step": 1381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994300505964327e-05,
+      "loss": 0.3355,
+      "step": 1382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994251709048283e-05,
+      "loss": 0.3695,
+      "step": 1383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9942027043984874e-05,
+      "loss": 0.3648,
+      "step": 1384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9941534920194757e-05,
+      "loss": 0.347,
+      "step": 1385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.994104071915803e-05,
+      "loss": 0.3627,
+      "step": 1386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9940544440920434e-05,
+      "loss": 0.3709,
+      "step": 1387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9940046085527924e-05,
+      "loss": 0.3743,
+      "step": 1388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9939545653026605e-05,
+      "loss": 0.3568,
+      "step": 1389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993904314346282e-05,
+      "loss": 0.36,
+      "step": 1390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9938538556883065e-05,
+      "loss": 0.3701,
+      "step": 1391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993803189333406e-05,
+      "loss": 0.3359,
+      "step": 1392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993752315286269e-05,
+      "loss": 0.3631,
+      "step": 1393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993701233551606e-05,
+      "loss": 0.3816,
+      "step": 1394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993649944134144e-05,
+      "loss": 0.3914,
+      "step": 1395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9935984470386304e-05,
+      "loss": 0.3624,
+      "step": 1396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993546742269833e-05,
+      "loss": 0.3624,
+      "step": 1397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993494829832537e-05,
+      "loss": 0.3742,
+      "step": 1398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993442709731548e-05,
+      "loss": 0.3504,
+      "step": 1399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.99339038197169e-05,
+      "loss": 0.3575,
+      "step": 1400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993337846557806e-05,
+      "loss": 0.3561,
+      "step": 1401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.99328510349476e-05,
+      "loss": 0.3733,
+      "step": 1402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993232152787434e-05,
+      "loss": 0.362,
+      "step": 1403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9931789944407284e-05,
+      "loss": 0.3687,
+      "step": 1404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.993125628459564e-05,
+      "loss": 0.3601,
+      "step": 1405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99307205484888e-05,
+      "loss": 0.3633,
+      "step": 1406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.993018273613637e-05,
+      "loss": 0.3641,
+      "step": 1407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992964284758812e-05,
+      "loss": 0.3585,
+      "step": 1408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992910088289401e-05,
+      "loss": 0.3894,
+      "step": 1409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9928556842104226e-05,
+      "loss": 0.3829,
+      "step": 1410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992801072526912e-05,
+      "loss": 0.3696,
+      "step": 1411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992746253243925e-05,
+      "loss": 0.3706,
+      "step": 1412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9926912263665335e-05,
+      "loss": 0.3514,
+      "step": 1413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992635991899832e-05,
+      "loss": 0.3604,
+      "step": 1414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992580549848935e-05,
+      "loss": 0.3649,
+      "step": 1415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992524900218972e-05,
+      "loss": 0.3586,
+      "step": 1416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992469043015094e-05,
+      "loss": 0.3654,
+      "step": 1417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992412978242474e-05,
+      "loss": 0.3466,
+      "step": 1418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9923567059062985e-05,
+      "loss": 0.3586,
+      "step": 1419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992300226011778e-05,
+      "loss": 0.3761,
+      "step": 1420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9922435385641406e-05,
+      "loss": 0.3788,
+      "step": 1421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992186643568631e-05,
+      "loss": 0.3549,
+      "step": 1422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992129541030518e-05,
+      "loss": 0.3392,
+      "step": 1423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.992072230955086e-05,
+      "loss": 0.3596,
+      "step": 1424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9920147133476406e-05,
+      "loss": 0.3404,
+      "step": 1425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9919569882135054e-05,
+      "loss": 0.3756,
+      "step": 1426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9918990555580234e-05,
+      "loss": 0.3495,
+      "step": 1427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.991840915386557e-05,
+      "loss": 0.3473,
+      "step": 1428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.991782567704488e-05,
+      "loss": 0.363,
+      "step": 1429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.991724012517217e-05,
+      "loss": 0.3687,
+      "step": 1430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.991665249830164e-05,
+      "loss": 0.3623,
+      "step": 1431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.991606279648769e-05,
+      "loss": 0.381,
+      "step": 1432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9915471019784895e-05,
+      "loss": 0.3678,
+      "step": 1433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9914877168248034e-05,
+      "loss": 0.3584,
+      "step": 1434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.991428124193207e-05,
+      "loss": 0.347,
+      "step": 1435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9913683240892175e-05,
+      "loss": 0.3638,
+      "step": 1436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99130831651837e-05,
+      "loss": 0.372,
+      "step": 1437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9912481014862184e-05,
+      "loss": 0.364,
+      "step": 1438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.991187678998336e-05,
+      "loss": 0.3611,
+      "step": 1439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9911270490603166e-05,
+      "loss": 0.375,
+      "step": 1440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.991066211677772e-05,
+      "loss": 0.3652,
+      "step": 1441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.991005166856332e-05,
+      "loss": 0.3589,
+      "step": 1442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.990943914601649e-05,
+      "loss": 0.3492,
+      "step": 1443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9908824549193935e-05,
+      "loss": 0.3713,
+      "step": 1444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9908207878152514e-05,
+      "loss": 0.3528,
+      "step": 1445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.990758913294933e-05,
+      "loss": 0.3451,
+      "step": 1446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9906968313641644e-05,
+      "loss": 0.3848,
+      "step": 1447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.990634542028693e-05,
+      "loss": 0.3561,
+      "step": 1448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.990572045294283e-05,
+      "loss": 0.3522,
+      "step": 1449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9905093411667206e-05,
+      "loss": 0.3643,
+      "step": 1450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99044642965181e-05,
+      "loss": 0.3764,
+      "step": 1451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9903833107553735e-05,
+      "loss": 0.3533,
+      "step": 1452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9903199844832536e-05,
+      "loss": 0.3445,
+      "step": 1453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9902564508413136e-05,
+      "loss": 0.3564,
+      "step": 1454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.990192709835432e-05,
+      "loss": 0.3573,
+      "step": 1455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99012876147151e-05,
+      "loss": 0.3897,
+      "step": 1456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9900646057554676e-05,
+      "loss": 0.3553,
+      "step": 1457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9900002426932414e-05,
+      "loss": 0.3653,
+      "step": 1458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.98993567229079e-05,
+      "loss": 0.3571,
+      "step": 1459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.98987089455409e-05,
+      "loss": 0.3583,
+      "step": 1460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9898059094891384e-05,
+      "loss": 0.3519,
+      "step": 1461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.989740717101949e-05,
+      "loss": 0.356,
+      "step": 1462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.989675317398556e-05,
+      "loss": 0.3641,
+      "step": 1463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9896097103850146e-05,
+      "loss": 0.3644,
+      "step": 1464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9895438960673965e-05,
+      "loss": 0.3506,
+      "step": 1465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.989477874451794e-05,
+      "loss": 0.3619,
+      "step": 1466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9894116455443176e-05,
+      "loss": 0.3854,
+      "step": 1467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9893452093510975e-05,
+      "loss": 0.3616,
+      "step": 1468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9892785658782846e-05,
+      "loss": 0.3668,
+      "step": 1469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.989211715132047e-05,
+      "loss": 0.351,
+      "step": 1470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.989144657118572e-05,
+      "loss": 0.3685,
+      "step": 1471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9890773918440664e-05,
+      "loss": 0.3808,
+      "step": 1472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.989009919314758e-05,
+      "loss": 0.3688,
+      "step": 1473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.98894223953689e-05,
+      "loss": 0.3387,
+      "step": 1474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.988874352516729e-05,
+      "loss": 0.3588,
+      "step": 1475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.988806258260558e-05,
+      "loss": 0.3554,
+      "step": 1476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.98873795677468e-05,
+      "loss": 0.3453,
+      "step": 1477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9886694480654174e-05,
+      "loss": 0.3354,
+      "step": 1478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.988600732139111e-05,
+      "loss": 0.3457,
+      "step": 1479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.988531809002122e-05,
+      "loss": 0.3426,
+      "step": 1480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.98846267866083e-05,
+      "loss": 0.3733,
+      "step": 1481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9883933411216324e-05,
+      "loss": 0.3384,
+      "step": 1482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.98832379639095e-05,
+      "loss": 0.3775,
+      "step": 1483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.988254044475218e-05,
+      "loss": 0.3539,
+      "step": 1484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.988184085380893e-05,
+      "loss": 0.3661,
+      "step": 1485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.988113919114451e-05,
+      "loss": 0.3698,
+      "step": 1486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9880435456823876e-05,
+      "loss": 0.3656,
+      "step": 1487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9879729650912154e-05,
+      "loss": 0.3387,
+      "step": 1488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.987902177347468e-05,
+      "loss": 0.3376,
+      "step": 1489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.987831182457698e-05,
+      "loss": 0.3887,
+      "step": 1490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9877599804284755e-05,
+      "loss": 0.3699,
+      "step": 1491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.987688571266393e-05,
+      "loss": 0.3679,
+      "step": 1492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.98761695497806e-05,
+      "loss": 0.3572,
+      "step": 1493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.987545131570105e-05,
+      "loss": 0.3531,
+      "step": 1494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9874731010491753e-05,
+      "loss": 0.3291,
+      "step": 1495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.98740086342194e-05,
+      "loss": 0.3553,
+      "step": 1496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.987328418695084e-05,
+      "loss": 0.362,
+      "step": 1497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.987255766875315e-05,
+      "loss": 0.3641,
+      "step": 1498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9871829079693554e-05,
+      "loss": 0.3531,
+      "step": 1499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.987109841983951e-05,
+      "loss": 0.359,
+      "step": 1500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9870365689258635e-05,
+      "loss": 0.359,
+      "step": 1501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.986963088801876e-05,
+      "loss": 0.3593,
+      "step": 1502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.986889401618791e-05,
+      "loss": 0.3692,
+      "step": 1503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9868155073834285e-05,
+      "loss": 0.3416,
+      "step": 1504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9867414061026265e-05,
+      "loss": 0.3567,
+      "step": 1505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.986667097783247e-05,
+      "loss": 0.3419,
+      "step": 1506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.986592582432166e-05,
+      "loss": 0.3675,
+      "step": 1507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.986517860056282e-05,
+      "loss": 0.3498,
+      "step": 1508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.986442930662511e-05,
+      "loss": 0.3522,
+      "step": 1509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.986367794257789e-05,
+      "loss": 0.3578,
+      "step": 1510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.98629245084907e-05,
+      "loss": 0.3608,
+      "step": 1511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9862169004433296e-05,
+      "loss": 0.3431,
+      "step": 1512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.986141143047559e-05,
+      "loss": 0.372,
+      "step": 1513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.986065178668772e-05,
+      "loss": 0.3659,
+      "step": 1514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.985989007313999e-05,
+      "loss": 0.3768,
+      "step": 1515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.985912628990291e-05,
+      "loss": 0.3599,
+      "step": 1516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.985836043704718e-05,
+      "loss": 0.3525,
+      "step": 1517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.985759251464369e-05,
+      "loss": 0.3614,
+      "step": 1518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.985682252276352e-05,
+      "loss": 0.3636,
+      "step": 1519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.985605046147794e-05,
+      "loss": 0.3554,
+      "step": 1520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.985527633085841e-05,
+      "loss": 0.3647,
+      "step": 1521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.98545001309766e-05,
+      "loss": 0.3659,
+      "step": 1522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.985372186190434e-05,
+      "loss": 0.3366,
+      "step": 1523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9852941523713685e-05,
+      "loss": 0.3633,
+      "step": 1524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.985215911647685e-05,
+      "loss": 0.3513,
+      "step": 1525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9851374640266266e-05,
+      "loss": 0.3545,
+      "step": 1526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9850588095154543e-05,
+      "loss": 0.331,
+      "step": 1527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.984979948121448e-05,
+      "loss": 0.3605,
+      "step": 1528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9849008798519086e-05,
+      "loss": 0.3696,
+      "step": 1529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.984821604714154e-05,
+      "loss": 0.3646,
+      "step": 1530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9847421227155226e-05,
+      "loss": 0.3548,
+      "step": 1531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9846624338633707e-05,
+      "loss": 0.3493,
+      "step": 1532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.984582538165075e-05,
+      "loss": 0.3555,
+      "step": 1533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9845024356280306e-05,
+      "loss": 0.3533,
+      "step": 1534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9844221262596535e-05,
+      "loss": 0.3498,
+      "step": 1535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.984341610067375e-05,
+      "loss": 0.337,
+      "step": 1536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9842608870586485e-05,
+      "loss": 0.3525,
+      "step": 1537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.984179957240947e-05,
+      "loss": 0.3604,
+      "step": 1538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.984098820621762e-05,
+      "loss": 0.3681,
+      "step": 1539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9840174772086015e-05,
+      "loss": 0.3782,
+      "step": 1540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.983935927008995e-05,
+      "loss": 0.3549,
+      "step": 1541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.983854170030494e-05,
+      "loss": 0.3599,
+      "step": 1542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.983772206280663e-05,
+      "loss": 0.3576,
+      "step": 1543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.983690035767091e-05,
+      "loss": 0.3807,
+      "step": 1544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.983607658497381e-05,
+      "loss": 0.3578,
+      "step": 1545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9835250744791615e-05,
+      "loss": 0.3692,
+      "step": 1546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.983442283720074e-05,
+      "loss": 0.3527,
+      "step": 1547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.983359286227784e-05,
+      "loss": 0.3439,
+      "step": 1548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9832760820099724e-05,
+      "loss": 0.3667,
+      "step": 1549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.983192671074341e-05,
+      "loss": 0.361,
+      "step": 1550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9831090534286114e-05,
+      "loss": 0.3445,
+      "step": 1551
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.983025229080523e-05,
+      "loss": 0.3694,
+      "step": 1552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.982941198037834e-05,
+      "loss": 0.3608,
+      "step": 1553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.982856960308324e-05,
+      "loss": 0.3557,
+      "step": 1554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.982772515899789e-05,
+      "loss": 0.3668,
+      "step": 1555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.982687864820045e-05,
+      "loss": 0.3653,
+      "step": 1556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.98260300707693e-05,
+      "loss": 0.3538,
+      "step": 1557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.982517942678296e-05,
+      "loss": 0.3694,
+      "step": 1558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.982432671632017e-05,
+      "loss": 0.351,
+      "step": 1559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9823471939459884e-05,
+      "loss": 0.3564,
+      "step": 1560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9822615096281206e-05,
+      "loss": 0.3606,
+      "step": 1561
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9821756186863435e-05,
+      "loss": 0.3726,
+      "step": 1562
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9820895211286096e-05,
+      "loss": 0.3459,
+      "step": 1563
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9820032169628866e-05,
+      "loss": 0.3505,
+      "step": 1564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.981916706197164e-05,
+      "loss": 0.3339,
+      "step": 1565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9818299888394495e-05,
+      "loss": 0.3438,
+      "step": 1566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.98174306489777e-05,
+      "loss": 0.3534,
+      "step": 1567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.98165593438017e-05,
+      "loss": 0.3589,
+      "step": 1568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.981568597294716e-05,
+      "loss": 0.3512,
+      "step": 1569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9814810536494924e-05,
+      "loss": 0.3819,
+      "step": 1570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9813933034526e-05,
+      "loss": 0.3732,
+      "step": 1571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9813053467121646e-05,
+      "loss": 0.3736,
+      "step": 1572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.981217183436326e-05,
+      "loss": 0.352,
+      "step": 1573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.981128813633244e-05,
+      "loss": 0.3477,
+      "step": 1574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9810402373111e-05,
+      "loss": 0.3545,
+      "step": 1575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.980951454478091e-05,
+      "loss": 0.3516,
+      "step": 1576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.980862465142437e-05,
+      "loss": 0.3476,
+      "step": 1577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9807732693123735e-05,
+      "loss": 0.3586,
+      "step": 1578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.980683866996158e-05,
+      "loss": 0.3621,
+      "step": 1579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.980594258202064e-05,
+      "loss": 0.3753,
+      "step": 1580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9805044429383886e-05,
+      "loss": 0.367,
+      "step": 1581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.980414421213442e-05,
+      "loss": 0.3379,
+      "step": 1582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.98032419303556e-05,
+      "loss": 0.3469,
+      "step": 1583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9802337584130926e-05,
+      "loss": 0.3684,
+      "step": 1584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9801431173544114e-05,
+      "loss": 0.3621,
+      "step": 1585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9800522698679056e-05,
+      "loss": 0.3596,
+      "step": 1586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.979961215961984e-05,
+      "loss": 0.363,
+      "step": 1587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9798699556450765e-05,
+      "loss": 0.3771,
+      "step": 1588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.979778488925629e-05,
+      "loss": 0.3452,
+      "step": 1589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.979686815812108e-05,
+      "loss": 0.3501,
+      "step": 1590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.979594936312999e-05,
+      "loss": 0.3515,
+      "step": 1591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.979502850436808e-05,
+      "loss": 0.3752,
+      "step": 1592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.979410558192057e-05,
+      "loss": 0.3659,
+      "step": 1593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.979318059587289e-05,
+      "loss": 0.3563,
+      "step": 1594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.979225354631067e-05,
+      "loss": 0.3707,
+      "step": 1595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.979132443331971e-05,
+      "loss": 0.3263,
+      "step": 1596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.979039325698601e-05,
+      "loss": 0.3471,
+      "step": 1597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.978946001739577e-05,
+      "loss": 0.3733,
+      "step": 1598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9788524714635374e-05,
+      "loss": 0.3756,
+      "step": 1599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.978758734879138e-05,
+      "loss": 0.3507,
+      "step": 1600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9786647919950575e-05,
+      "loss": 0.3636,
+      "step": 1601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9785706428199905e-05,
+      "loss": 0.3648,
+      "step": 1602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.97847628736265e-05,
+      "loss": 0.3809,
+      "step": 1603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.978381725631773e-05,
+      "loss": 0.3632,
+      "step": 1604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9782869576361105e-05,
+      "loss": 0.3428,
+      "step": 1605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.978191983384435e-05,
+      "loss": 0.3587,
+      "step": 1606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9780968028855366e-05,
+      "loss": 0.3474,
+      "step": 1607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.978001416148227e-05,
+      "loss": 0.3635,
+      "step": 1608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.977905823181334e-05,
+      "loss": 0.3677,
+      "step": 1609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.977810023993706e-05,
+      "loss": 0.3498,
+      "step": 1610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.977714018594212e-05,
+      "loss": 0.3508,
+      "step": 1611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9776178069917365e-05,
+      "loss": 0.3485,
+      "step": 1612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.977521389195187e-05,
+      "loss": 0.3472,
+      "step": 1613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.977424765213487e-05,
+      "loss": 0.3501,
+      "step": 1614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.977327935055581e-05,
+      "loss": 0.3696,
+      "step": 1615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.977230898730431e-05,
+      "loss": 0.3537,
+      "step": 1616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.977133656247018e-05,
+      "loss": 0.3555,
+      "step": 1617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.977036207614346e-05,
+      "loss": 0.3448,
+      "step": 1618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.976938552841433e-05,
+      "loss": 0.3628,
+      "step": 1619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.976840691937319e-05,
+      "loss": 0.3575,
+      "step": 1620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.976742624911061e-05,
+      "loss": 0.3535,
+      "step": 1621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9766443517717375e-05,
+      "loss": 0.3461,
+      "step": 1622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9765458725284445e-05,
+      "loss": 0.3543,
+      "step": 1623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.976447187190298e-05,
+      "loss": 0.3575,
+      "step": 1624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.976348295766432e-05,
+      "loss": 0.3416,
+      "step": 1625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9762491982660004e-05,
+      "loss": 0.3562,
+      "step": 1626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.976149894698175e-05,
+      "loss": 0.3738,
+      "step": 1627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.97605038507215e-05,
+      "loss": 0.3533,
+      "step": 1628
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9759506693971336e-05,
+      "loss": 0.3765,
+      "step": 1629
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.975850747682357e-05,
+      "loss": 0.3661,
+      "step": 1630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9757506199370694e-05,
+      "loss": 0.3487,
+      "step": 1631
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.975650286170539e-05,
+      "loss": 0.3614,
+      "step": 1632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9755497463920514e-05,
+      "loss": 0.3614,
+      "step": 1633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9754490006109146e-05,
+      "loss": 0.3601,
+      "step": 1634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9753480488364526e-05,
+      "loss": 0.3749,
+      "step": 1635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9752468910780106e-05,
+      "loss": 0.362,
+      "step": 1636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.975145527344952e-05,
+      "loss": 0.369,
+      "step": 1637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.975043957646659e-05,
+      "loss": 0.3613,
+      "step": 1638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.974942181992534e-05,
+      "loss": 0.3681,
+      "step": 1639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.974840200391996e-05,
+      "loss": 0.3552,
+      "step": 1640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9747380128544855e-05,
+      "loss": 0.3543,
+      "step": 1641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.974635619389462e-05,
+      "loss": 0.3478,
+      "step": 1642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9745330200064014e-05,
+      "loss": 0.3626,
+      "step": 1643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9744302147148023e-05,
+      "loss": 0.338,
+      "step": 1644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.974327203524181e-05,
+      "loss": 0.3435,
+      "step": 1645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.97422398644407e-05,
+      "loss": 0.3518,
+      "step": 1646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.974120563484026e-05,
+      "loss": 0.3435,
+      "step": 1647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9740169346536206e-05,
+      "loss": 0.357,
+      "step": 1648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9739130999624464e-05,
+      "loss": 0.3488,
+      "step": 1649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9738090594201144e-05,
+      "loss": 0.3631,
+      "step": 1650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.973704813036255e-05,
+      "loss": 0.3555,
+      "step": 1651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.973600360820518e-05,
+      "loss": 0.3569,
+      "step": 1652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.973495702782571e-05,
+      "loss": 0.3642,
+      "step": 1653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.973390838932102e-05,
+      "loss": 0.3411,
+      "step": 1654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.973285769278817e-05,
+      "loss": 0.348,
+      "step": 1655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.973180493832441e-05,
+      "loss": 0.3474,
+      "step": 1656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9730750126027205e-05,
+      "loss": 0.3577,
+      "step": 1657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.972969325599417e-05,
+      "loss": 0.3333,
+      "step": 1658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.972863432832315e-05,
+      "loss": 0.361,
+      "step": 1659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9727573343112136e-05,
+      "loss": 0.3645,
+      "step": 1660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9726510300459364e-05,
+      "loss": 0.3557,
+      "step": 1661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.972544520046322e-05,
+      "loss": 0.3628,
+      "step": 1662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.972437804322229e-05,
+      "loss": 0.3409,
+      "step": 1663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.972330882883536e-05,
+      "loss": 0.3388,
+      "step": 1664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.972223755740139e-05,
+      "loss": 0.3469,
+      "step": 1665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9721164229019544e-05,
+      "loss": 0.3447,
+      "step": 1666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.972008884378918e-05,
+      "loss": 0.3707,
+      "step": 1667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.971901140180983e-05,
+      "loss": 0.3527,
+      "step": 1668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9717931903181215e-05,
+      "loss": 0.3489,
+      "step": 1669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.971685034800327e-05,
+      "loss": 0.3605,
+      "step": 1670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9715766736376105e-05,
+      "loss": 0.3273,
+      "step": 1671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9714681068400026e-05,
+      "loss": 0.366,
+      "step": 1672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.971359334417551e-05,
+      "loss": 0.3306,
+      "step": 1673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.971250356380326e-05,
+      "loss": 0.3589,
+      "step": 1674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9711411727384126e-05,
+      "loss": 0.3432,
+      "step": 1675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.971031783501919e-05,
+      "loss": 0.3621,
+      "step": 1676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9709221886809695e-05,
+      "loss": 0.3558,
+      "step": 1677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.97081238828571e-05,
+      "loss": 0.359,
+      "step": 1678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9707023823263014e-05,
+      "loss": 0.3557,
+      "step": 1679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.970592170812928e-05,
+      "loss": 0.3381,
+      "step": 1680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.970481753755791e-05,
+      "loss": 0.3586,
+      "step": 1681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.970371131165111e-05,
+      "loss": 0.3792,
+      "step": 1682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.970260303051126e-05,
+      "loss": 0.3539,
+      "step": 1683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9701492694240967e-05,
+      "loss": 0.3394,
+      "step": 1684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.970038030294299e-05,
+      "loss": 0.3518,
+      "step": 1685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.96992658567203e-05,
+      "loss": 0.3539,
+      "step": 1686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.969814935567606e-05,
+      "loss": 0.377,
+      "step": 1687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9697030799913606e-05,
+      "loss": 0.3693,
+      "step": 1688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9695910189536484e-05,
+      "loss": 0.3569,
+      "step": 1689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9694787524648415e-05,
+      "loss": 0.3675,
+      "step": 1690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.969366280535331e-05,
+      "loss": 0.3607,
+      "step": 1691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9692536031755297e-05,
+      "loss": 0.3563,
+      "step": 1692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.969140720395865e-05,
+      "loss": 0.3664,
+      "step": 1693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.969027632206786e-05,
+      "loss": 0.3568,
+      "step": 1694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.968914338618761e-05,
+      "loss": 0.3503,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9688008396422774e-05,
+      "loss": 0.3465,
+      "step": 1696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.96868713528784e-05,
+      "loss": 0.3519,
+      "step": 1697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9685732255659736e-05,
+      "loss": 0.3521,
+      "step": 1698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.968459110487222e-05,
+      "loss": 0.3587,
+      "step": 1699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.968344790062149e-05,
+      "loss": 0.3339,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.968230264301334e-05,
+      "loss": 0.3461,
+      "step": 1701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.96811553321538e-05,
+      "loss": 0.3462,
+      "step": 1702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.968000596814907e-05,
+      "loss": 0.3569,
+      "step": 1703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.967885455110552e-05,
+      "loss": 0.3653,
+      "step": 1704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.967770108112974e-05,
+      "loss": 0.3537,
+      "step": 1705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9676545558328504e-05,
+      "loss": 0.3617,
+      "step": 1706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.967538798280875e-05,
+      "loss": 0.3613,
+      "step": 1707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.967422835467764e-05,
+      "loss": 0.363,
+      "step": 1708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.967306667404252e-05,
+      "loss": 0.3335,
+      "step": 1709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9671902941010894e-05,
+      "loss": 0.3574,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9670737155690503e-05,
+      "loss": 0.3434,
+      "step": 1711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.966956931818925e-05,
+      "loss": 0.3698,
+      "step": 1712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.966839942861523e-05,
+      "loss": 0.3699,
+      "step": 1713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.966722748707672e-05,
+      "loss": 0.3524,
+      "step": 1714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.966605349368222e-05,
+      "loss": 0.3567,
+      "step": 1715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.966487744854039e-05,
+      "loss": 0.3512,
+      "step": 1716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.966369935176007e-05,
+      "loss": 0.3467,
+      "step": 1717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.966251920345034e-05,
+      "loss": 0.3593,
+      "step": 1718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.966133700372041e-05,
+      "loss": 0.3784,
+      "step": 1719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9660152752679716e-05,
+      "loss": 0.3432,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9658966450437885e-05,
+      "loss": 0.3609,
+      "step": 1721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.965777809710471e-05,
+      "loss": 0.3534,
+      "step": 1722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.965658769279019e-05,
+      "loss": 0.351,
+      "step": 1723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.965539523760452e-05,
+      "loss": 0.3754,
+      "step": 1724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.965420073165807e-05,
+      "loss": 0.3503,
+      "step": 1725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.965300417506141e-05,
+      "loss": 0.3523,
+      "step": 1726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.96518055679253e-05,
+      "loss": 0.3411,
+      "step": 1727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.965060491036068e-05,
+      "loss": 0.348,
+      "step": 1728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.964940220247869e-05,
+      "loss": 0.3455,
+      "step": 1729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9648197444390654e-05,
+      "loss": 0.373,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.964699063620807e-05,
+      "loss": 0.3601,
+      "step": 1731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.964578177804268e-05,
+      "loss": 0.3719,
+      "step": 1732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.964457087000636e-05,
+      "loss": 0.3425,
+      "step": 1733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.964335791221118e-05,
+      "loss": 0.3597,
+      "step": 1734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9642142904769435e-05,
+      "loss": 0.3696,
+      "step": 1735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9640925847793586e-05,
+      "loss": 0.3618,
+      "step": 1736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.963970674139628e-05,
+      "loss": 0.3663,
+      "step": 1737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9638485585690365e-05,
+      "loss": 0.3487,
+      "step": 1738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.963726238078888e-05,
+      "loss": 0.353,
+      "step": 1739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.963603712680503e-05,
+      "loss": 0.3683,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.963480982385225e-05,
+      "loss": 0.3396,
+      "step": 1741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.963358047204413e-05,
+      "loss": 0.3644,
+      "step": 1742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.963234907149446e-05,
+      "loss": 0.3652,
+      "step": 1743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.963111562231723e-05,
+      "loss": 0.3562,
+      "step": 1744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9629880124626596e-05,
+      "loss": 0.3458,
+      "step": 1745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.962864257853694e-05,
+      "loss": 0.3365,
+      "step": 1746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9627402984162795e-05,
+      "loss": 0.3583,
+      "step": 1747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9626161341618914e-05,
+      "loss": 0.3573,
+      "step": 1748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9624917651020226e-05,
+      "loss": 0.3706,
+      "step": 1749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.962367191248183e-05,
+      "loss": 0.3708,
+      "step": 1750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9622424126119066e-05,
+      "loss": 0.3336,
+      "step": 1751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.96211742920474e-05,
+      "loss": 0.3632,
+      "step": 1752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.961992241038255e-05,
+      "loss": 0.3443,
+      "step": 1753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.961866848124037e-05,
+      "loss": 0.3296,
+      "step": 1754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9617412504736935e-05,
+      "loss": 0.3565,
+      "step": 1755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9616154480988515e-05,
+      "loss": 0.3565,
+      "step": 1756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9614894410111535e-05,
+      "loss": 0.3545,
+      "step": 1757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.961363229222264e-05,
+      "loss": 0.3482,
+      "step": 1758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.961236812743866e-05,
+      "loss": 0.3551,
+      "step": 1759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.961110191587659e-05,
+      "loss": 0.3505,
+      "step": 1760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.960983365765366e-05,
+      "loss": 0.3544,
+      "step": 1761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9608563352887245e-05,
+      "loss": 0.3596,
+      "step": 1762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.960729100169494e-05,
+      "loss": 0.339,
+      "step": 1763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.96060166041945e-05,
+      "loss": 0.3293,
+      "step": 1764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9604740160503906e-05,
+      "loss": 0.3377,
+      "step": 1765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.960346167074129e-05,
+      "loss": 0.3468,
+      "step": 1766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9602181135025006e-05,
+      "loss": 0.3219,
+      "step": 1767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.960089855347358e-05,
+      "loss": 0.344,
+      "step": 1768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9599613926205734e-05,
+      "loss": 0.3531,
+      "step": 1769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.959832725334037e-05,
+      "loss": 0.344,
+      "step": 1770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.959703853499659e-05,
+      "loss": 0.3346,
+      "step": 1771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9595747771293686e-05,
+      "loss": 0.3845,
+      "step": 1772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.959445496235112e-05,
+      "loss": 0.3556,
+      "step": 1773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.959316010828857e-05,
+      "loss": 0.3709,
+      "step": 1774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.959186320922589e-05,
+      "loss": 0.3342,
+      "step": 1775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.959056426528311e-05,
+      "loss": 0.3726,
+      "step": 1776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9589263276580484e-05,
+      "loss": 0.3435,
+      "step": 1777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.958796024323843e-05,
+      "loss": 0.3702,
+      "step": 1778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.958665516537755e-05,
+      "loss": 0.364,
+      "step": 1779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.958534804311866e-05,
+      "loss": 0.3367,
+      "step": 1780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.958403887658273e-05,
+      "loss": 0.3467,
+      "step": 1781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.958272766589096e-05,
+      "loss": 0.3686,
+      "step": 1782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.958141441116471e-05,
+      "loss": 0.3437,
+      "step": 1783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.958009911252555e-05,
+      "loss": 0.3636,
+      "step": 1784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.95787817700952e-05,
+      "loss": 0.3335,
+      "step": 1785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9577462383995624e-05,
+      "loss": 0.3661,
+      "step": 1786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.957614095434893e-05,
+      "loss": 0.3418,
+      "step": 1787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9574817481277444e-05,
+      "loss": 0.341,
+      "step": 1788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9573491964903665e-05,
+      "loss": 0.3404,
+      "step": 1789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.957216440535029e-05,
+      "loss": 0.3494,
+      "step": 1790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9570834802740195e-05,
+      "loss": 0.3464,
+      "step": 1791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.956950315719646e-05,
+      "loss": 0.342,
+      "step": 1792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9568169468842344e-05,
+      "loss": 0.3522,
+      "step": 1793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.956683373780129e-05,
+      "loss": 0.3666,
+      "step": 1794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.956549596419694e-05,
+      "loss": 0.3598,
+      "step": 1795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9564156148153125e-05,
+      "loss": 0.3708,
+      "step": 1796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9562814289793855e-05,
+      "loss": 0.3551,
+      "step": 1797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.956147038924334e-05,
+      "loss": 0.352,
+      "step": 1798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9560124446625984e-05,
+      "loss": 0.3233,
+      "step": 1799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.955877646206635e-05,
+      "loss": 0.3524,
+      "step": 1800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.955742643568923e-05,
+      "loss": 0.3567,
+      "step": 1801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.955607436761958e-05,
+      "loss": 0.3535,
+      "step": 1802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.955472025798254e-05,
+      "loss": 0.3568,
+      "step": 1803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9553364106903466e-05,
+      "loss": 0.3408,
+      "step": 1804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.955200591450788e-05,
+      "loss": 0.3347,
+      "step": 1805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.95506456809215e-05,
+      "loss": 0.3646,
+      "step": 1806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.954928340627022e-05,
+      "loss": 0.3755,
+      "step": 1807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9547919090680175e-05,
+      "loss": 0.3415,
+      "step": 1808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.95465527342776e-05,
+      "loss": 0.3484,
+      "step": 1809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.954518433718901e-05,
+      "loss": 0.3645,
+      "step": 1810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.954381389954104e-05,
+      "loss": 0.359,
+      "step": 1811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.954244142146055e-05,
+      "loss": 0.3394,
+      "step": 1812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9541066903074584e-05,
+      "loss": 0.3624,
+      "step": 1813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.953969034451036e-05,
+      "loss": 0.3468,
+      "step": 1814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.953831174589531e-05,
+      "loss": 0.3774,
+      "step": 1815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.953693110735703e-05,
+      "loss": 0.3475,
+      "step": 1816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.953554842902332e-05,
+      "loss": 0.3474,
+      "step": 1817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9534163711022166e-05,
+      "loss": 0.3536,
+      "step": 1818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9532776953481735e-05,
+      "loss": 0.3544,
+      "step": 1819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.95313881565304e-05,
+      "loss": 0.3579,
+      "step": 1820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.95299973202967e-05,
+      "loss": 0.3662,
+      "step": 1821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.952860444490938e-05,
+      "loss": 0.3268,
+      "step": 1822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9527209530497365e-05,
+      "loss": 0.3285,
+      "step": 1823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.952581257718977e-05,
+      "loss": 0.3637,
+      "step": 1824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.952441358511591e-05,
+      "loss": 0.3471,
+      "step": 1825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.952301255440527e-05,
+      "loss": 0.3871,
+      "step": 1826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.952160948518754e-05,
+      "loss": 0.3488,
+      "step": 1827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.952020437759258e-05,
+      "loss": 0.369,
+      "step": 1828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9518797231750474e-05,
+      "loss": 0.3506,
+      "step": 1829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9517388047791444e-05,
+      "loss": 0.3423,
+      "step": 1830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.951597682584594e-05,
+      "loss": 0.3425,
+      "step": 1831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9514563566044594e-05,
+      "loss": 0.332,
+      "step": 1832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9513148268518216e-05,
+      "loss": 0.3529,
+      "step": 1833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9511730933397807e-05,
+      "loss": 0.3687,
+      "step": 1834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.951031156081456e-05,
+      "loss": 0.3633,
+      "step": 1835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950889015089986e-05,
+      "loss": 0.3607,
+      "step": 1836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950746670378526e-05,
+      "loss": 0.3574,
+      "step": 1837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950604121960254e-05,
+      "loss": 0.366,
+      "step": 1838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9504613698483645e-05,
+      "loss": 0.3465,
+      "step": 1839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9503184140560686e-05,
+      "loss": 0.346,
+      "step": 1840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950175254596601e-05,
+      "loss": 0.3483,
+      "step": 1841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.950031891483213e-05,
+      "loss": 0.3688,
+      "step": 1842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9498883247291725e-05,
+      "loss": 0.338,
+      "step": 1843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9497445543477706e-05,
+      "loss": 0.3491,
+      "step": 1844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9496005803523144e-05,
+      "loss": 0.3637,
+      "step": 1845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.94945640275613e-05,
+      "loss": 0.3508,
+      "step": 1846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.949312021572563e-05,
+      "loss": 0.3638,
+      "step": 1847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.949167436814979e-05,
+      "loss": 0.3812,
+      "step": 1848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.949022648496759e-05,
+      "loss": 0.3665,
+      "step": 1849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.948877656631307e-05,
+      "loss": 0.3511,
+      "step": 1850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.948732461232041e-05,
+      "loss": 0.348,
+      "step": 1851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.948587062312404e-05,
+      "loss": 0.3492,
+      "step": 1852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.948441459885853e-05,
+      "loss": 0.3511,
+      "step": 1853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.948295653965865e-05,
+      "loss": 0.3627,
+      "step": 1854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.948149644565936e-05,
+      "loss": 0.3615,
+      "step": 1855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9480034316995826e-05,
+      "loss": 0.3371,
+      "step": 1856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947857015380337e-05,
+      "loss": 0.3447,
+      "step": 1857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947710395621753e-05,
+      "loss": 0.3732,
+      "step": 1858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9475635724374005e-05,
+      "loss": 0.3465,
+      "step": 1859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9474165458408714e-05,
+      "loss": 0.3585,
+      "step": 1860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9472693158457746e-05,
+      "loss": 0.3517,
+      "step": 1861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.947121882465737e-05,
+      "loss": 0.3655,
+      "step": 1862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9469742457144075e-05,
+      "loss": 0.3419,
+      "step": 1863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9468264056054496e-05,
+      "loss": 0.3269,
+      "step": 1864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9466783621525494e-05,
+      "loss": 0.3539,
+      "step": 1865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9465301153694086e-05,
+      "loss": 0.343,
+      "step": 1866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.946381665269751e-05,
+      "loss": 0.3531,
+      "step": 1867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.946233011867315e-05,
+      "loss": 0.3456,
+      "step": 1868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.946084155175864e-05,
+      "loss": 0.3411,
+      "step": 1869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945935095209173e-05,
+      "loss": 0.3538,
+      "step": 1870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9457858319810426e-05,
+      "loss": 0.3656,
+      "step": 1871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9456363655052864e-05,
+      "loss": 0.3541,
+      "step": 1872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.94548669579574e-05,
+      "loss": 0.3677,
+      "step": 1873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945336822866258e-05,
+      "loss": 0.3217,
+      "step": 1874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945186746730713e-05,
+      "loss": 0.3448,
+      "step": 1875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.945036467402995e-05,
+      "loss": 0.3572,
+      "step": 1876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944885984897016e-05,
+      "loss": 0.3564,
+      "step": 1877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9447352992267045e-05,
+      "loss": 0.363,
+      "step": 1878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944584410406008e-05,
+      "loss": 0.3466,
+      "step": 1879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9444333184488926e-05,
+      "loss": 0.3286,
+      "step": 1880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944282023369345e-05,
+      "loss": 0.3372,
+      "step": 1881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944130525181369e-05,
+      "loss": 0.3727,
+      "step": 1882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.943978823898988e-05,
+      "loss": 0.368,
+      "step": 1883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.943826919536244e-05,
+      "loss": 0.364,
+      "step": 1884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.943674812107196e-05,
+      "loss": 0.3621,
+      "step": 1885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9435225016259244e-05,
+      "loss": 0.3383,
+      "step": 1886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9433699881065286e-05,
+      "loss": 0.3346,
+      "step": 1887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.943217271563125e-05,
+      "loss": 0.3507,
+      "step": 1888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.943064352009848e-05,
+      "loss": 0.3625,
+      "step": 1889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9429112294608546e-05,
+      "loss": 0.3508,
+      "step": 1890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.942757903930316e-05,
+      "loss": 0.315,
+      "step": 1891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.942604375432425e-05,
+      "loss": 0.3548,
+      "step": 1892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.942450643981394e-05,
+      "loss": 0.3611,
+      "step": 1893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9422967095914515e-05,
+      "loss": 0.3445,
+      "step": 1894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9421425722768466e-05,
+      "loss": 0.3455,
+      "step": 1895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.941988232051846e-05,
+      "loss": 0.3575,
+      "step": 1896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9418336889307365e-05,
+      "loss": 0.3639,
+      "step": 1897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.941678942927822e-05,
+      "loss": 0.3667,
+      "step": 1898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.941523994057428e-05,
+      "loss": 0.3458,
+      "step": 1899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.941368842333896e-05,
+      "loss": 0.3659,
+      "step": 1900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.941213487771586e-05,
+      "loss": 0.3439,
+      "step": 1901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.94105793038488e-05,
+      "loss": 0.3574,
+      "step": 1902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.940902170188176e-05,
+      "loss": 0.3353,
+      "step": 1903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.940746207195891e-05,
+      "loss": 0.3396,
+      "step": 1904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.940590041422462e-05,
+      "loss": 0.3398,
+      "step": 1905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9404336728823444e-05,
+      "loss": 0.3511,
+      "step": 1906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.940277101590012e-05,
+      "loss": 0.3494,
+      "step": 1907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9401203275599556e-05,
+      "loss": 0.3514,
+      "step": 1908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.939963350806689e-05,
+      "loss": 0.354,
+      "step": 1909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.939806171344742e-05,
+      "loss": 0.3454,
+      "step": 1910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.939648789188662e-05,
+      "loss": 0.3471,
+      "step": 1911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.939491204353019e-05,
+      "loss": 0.3573,
+      "step": 1912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9393334168523974e-05,
+      "loss": 0.373,
+      "step": 1913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.939175426701403e-05,
+      "loss": 0.3469,
+      "step": 1914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.939017233914661e-05,
+      "loss": 0.3582,
+      "step": 1915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.938858838506812e-05,
+      "loss": 0.3562,
+      "step": 1916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.938700240492518e-05,
+      "loss": 0.3339,
+      "step": 1917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.938541439886462e-05,
+      "loss": 0.3549,
+      "step": 1918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9383824367033395e-05,
+      "loss": 0.3427,
+      "step": 1919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9382232309578694e-05,
+      "loss": 0.3427,
+      "step": 1920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9380638226647894e-05,
+      "loss": 0.377,
+      "step": 1921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.937904211838853e-05,
+      "loss": 0.3574,
+      "step": 1922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9377443984948346e-05,
+      "loss": 0.363,
+      "step": 1923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.937584382647527e-05,
+      "loss": 0.3535,
+      "step": 1924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9374241643117434e-05,
+      "loss": 0.3587,
+      "step": 1925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9372637435023115e-05,
+      "loss": 0.366,
+      "step": 1926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.937103120234081e-05,
+      "loss": 0.3259,
+      "step": 1927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.936942294521921e-05,
+      "loss": 0.3414,
+      "step": 1928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.936781266380717e-05,
+      "loss": 0.3619,
+      "step": 1929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.936620035825373e-05,
+      "loss": 0.341,
+      "step": 1930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.936458602870814e-05,
+      "loss": 0.3472,
+      "step": 1931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9362969675319836e-05,
+      "loss": 0.3573,
+      "step": 1932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9361351298238414e-05,
+      "loss": 0.344,
+      "step": 1933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.935973089761369e-05,
+      "loss": 0.3374,
+      "step": 1934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9358108473595636e-05,
+      "loss": 0.3655,
+      "step": 1935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9356484026334455e-05,
+      "loss": 0.3402,
+      "step": 1936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.935485755598047e-05,
+      "loss": 0.351,
+      "step": 1937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9353229062684266e-05,
+      "loss": 0.3684,
+      "step": 1938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.935159854659657e-05,
+      "loss": 0.3536,
+      "step": 1939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9349966007868306e-05,
+      "loss": 0.3639,
+      "step": 1940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.934833144665058e-05,
+      "loss": 0.3785,
+      "step": 1941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.934669486309471e-05,
+      "loss": 0.3599,
+      "step": 1942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.934505625735216e-05,
+      "loss": 0.3578,
+      "step": 1943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.93434156295746e-05,
+      "loss": 0.3434,
+      "step": 1944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.934177297991392e-05,
+      "loss": 0.3639,
+      "step": 1945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9340128308522144e-05,
+      "loss": 0.3583,
+      "step": 1946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9338481615551525e-05,
+      "loss": 0.3502,
+      "step": 1947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9336832901154455e-05,
+      "loss": 0.3433,
+      "step": 1948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.933518216548358e-05,
+      "loss": 0.3547,
+      "step": 1949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.933352940869167e-05,
+      "loss": 0.3615,
+      "step": 1950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.933187463093172e-05,
+      "loss": 0.3449,
+      "step": 1951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.933021783235689e-05,
+      "loss": 0.357,
+      "step": 1952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.932855901312056e-05,
+      "loss": 0.3392,
+      "step": 1953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.932689817337626e-05,
+      "loss": 0.3487,
+      "step": 1954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.932523531327771e-05,
+      "loss": 0.3508,
+      "step": 1955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.932357043297884e-05,
+      "loss": 0.3386,
+      "step": 1956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.932190353263377e-05,
+      "loss": 0.3245,
+      "step": 1957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.932023461239677e-05,
+      "loss": 0.3758,
+      "step": 1958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.931856367242233e-05,
+      "loss": 0.3476,
+      "step": 1959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.931689071286511e-05,
+      "loss": 0.3611,
+      "step": 1960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.931521573387998e-05,
+      "loss": 0.3359,
+      "step": 1961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9313538735621964e-05,
+      "loss": 0.3429,
+      "step": 1962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.931185971824629e-05,
+      "loss": 0.3707,
+      "step": 1963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.931017868190837e-05,
+      "loss": 0.3439,
+      "step": 1964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.930849562676382e-05,
+      "loss": 0.3259,
+      "step": 1965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.930681055296842e-05,
+      "loss": 0.3438,
+      "step": 1966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9305123460678146e-05,
+      "loss": 0.3424,
+      "step": 1967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.930343435004915e-05,
+      "loss": 0.3343,
+      "step": 1968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9301743221237796e-05,
+      "loss": 0.3773,
+      "step": 1969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9300050074400605e-05,
+      "loss": 0.3563,
+      "step": 1970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9298354909694306e-05,
+      "loss": 0.371,
+      "step": 1971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9296657727275805e-05,
+      "loss": 0.3546,
+      "step": 1972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.929495852730221e-05,
+      "loss": 0.3549,
+      "step": 1973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.929325730993079e-05,
+      "loss": 0.3387,
+      "step": 1974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9291554075319005e-05,
+      "loss": 0.3819,
+      "step": 1975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.928984882362454e-05,
+      "loss": 0.3321,
+      "step": 1976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.928814155500521e-05,
+      "loss": 0.3502,
+      "step": 1977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.928643226961906e-05,
+      "loss": 0.3542,
+      "step": 1978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.92847209676243e-05,
+      "loss": 0.3315,
+      "step": 1979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.928300764917933e-05,
+      "loss": 0.3326,
+      "step": 1980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.928129231444275e-05,
+      "loss": 0.3604,
+      "step": 1981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.927957496357332e-05,
+      "loss": 0.3311,
+      "step": 1982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.927785559673002e-05,
+      "loss": 0.3647,
+      "step": 1983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9276134214071986e-05,
+      "loss": 0.343,
+      "step": 1984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.927441081575855e-05,
+      "loss": 0.3605,
+      "step": 1985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9272685401949245e-05,
+      "loss": 0.342,
+      "step": 1986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.927095797280378e-05,
+      "loss": 0.3706,
+      "step": 1987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9269228528482046e-05,
+      "loss": 0.3402,
+      "step": 1988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.926749706914412e-05,
+      "loss": 0.3413,
+      "step": 1989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.926576359495028e-05,
+      "loss": 0.3438,
+      "step": 1990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9264028106060984e-05,
+      "loss": 0.3533,
+      "step": 1991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9262290602636854e-05,
+      "loss": 0.3425,
+      "step": 1992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.926055108483874e-05,
+      "loss": 0.3288,
+      "step": 1993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.925880955282764e-05,
+      "loss": 0.3755,
+      "step": 1994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9257066006764764e-05,
+      "loss": 0.3507,
+      "step": 1995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.925532044681149e-05,
+      "loss": 0.3225,
+      "step": 1996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.925357287312941e-05,
+      "loss": 0.3504,
+      "step": 1997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.925182328588026e-05,
+      "loss": 0.3426,
+      "step": 1998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9250071685226004e-05,
+      "loss": 0.3543,
+      "step": 1999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.924831807132877e-05,
+      "loss": 0.3566,
+      "step": 2000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9246562444350875e-05,
+      "loss": 0.339,
+      "step": 2001
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.924480480445482e-05,
+      "loss": 0.358,
+      "step": 2002
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9243045151803306e-05,
+      "loss": 0.358,
+      "step": 2003
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9241283486559204e-05,
+      "loss": 0.3539,
+      "step": 2004
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.923951980888558e-05,
+      "loss": 0.3328,
+      "step": 2005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.923775411894569e-05,
+      "loss": 0.3527,
+      "step": 2006
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9235986416902965e-05,
+      "loss": 0.3558,
+      "step": 2007
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9234216702921026e-05,
+      "loss": 0.3654,
+      "step": 2008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9232444977163697e-05,
+      "loss": 0.3303,
+      "step": 2009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.923067123979495e-05,
+      "loss": 0.3308,
+      "step": 2010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.922889549097898e-05,
+      "loss": 0.3464,
+      "step": 2011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9227117730880154e-05,
+      "loss": 0.3428,
+      "step": 2012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.922533795966304e-05,
+      "loss": 0.366,
+      "step": 2013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9223556177492346e-05,
+      "loss": 0.3496,
+      "step": 2014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.922177238453303e-05,
+      "loss": 0.359,
+      "step": 2015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9219986580950186e-05,
+      "loss": 0.3469,
+      "step": 2016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9218198766909116e-05,
+      "loss": 0.3762,
+      "step": 2017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.92164089425753e-05,
+      "loss": 0.3578,
+      "step": 2018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.921461710811443e-05,
+      "loss": 0.3547,
+      "step": 2019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.921282326369234e-05,
+      "loss": 0.3165,
+      "step": 2020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.921102740947509e-05,
+      "loss": 0.3523,
+      "step": 2021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9209229545628886e-05,
+      "loss": 0.3633,
+      "step": 2022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.920742967232016e-05,
+      "loss": 0.3461,
+      "step": 2023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.920562778971552e-05,
+      "loss": 0.3241,
+      "step": 2024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9203823897981736e-05,
+      "loss": 0.3531,
+      "step": 2025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9202017997285785e-05,
+      "loss": 0.3467,
+      "step": 2026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.920021008779483e-05,
+      "loss": 0.3374,
+      "step": 2027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.919840016967622e-05,
+      "loss": 0.3478,
+      "step": 2028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9196588243097476e-05,
+      "loss": 0.3638,
+      "step": 2029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.919477430822632e-05,
+      "loss": 0.3546,
+      "step": 2030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9192958365230646e-05,
+      "loss": 0.3298,
+      "step": 2031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.919114041427855e-05,
+      "loss": 0.3507,
+      "step": 2032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.918932045553831e-05,
+      "loss": 0.3555,
+      "step": 2033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9187498489178384e-05,
+      "loss": 0.3637,
+      "step": 2034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.918567451536742e-05,
+      "loss": 0.3399,
+      "step": 2035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.918384853427424e-05,
+      "loss": 0.3647,
+      "step": 2036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9182020546067856e-05,
+      "loss": 0.3353,
+      "step": 2037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.91801905509175e-05,
+      "loss": 0.3714,
+      "step": 2038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9178358548992535e-05,
+      "loss": 0.3437,
+      "step": 2039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.917652454046254e-05,
+      "loss": 0.3668,
+      "step": 2040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.917468852549729e-05,
+      "loss": 0.3406,
+      "step": 2041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9172850504266714e-05,
+      "loss": 0.3274,
+      "step": 2042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.917101047694096e-05,
+      "loss": 0.3462,
+      "step": 2043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.916916844369033e-05,
+      "loss": 0.3462,
+      "step": 2044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.916732440468533e-05,
+      "loss": 0.3594,
+      "step": 2045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.916547836009666e-05,
+      "loss": 0.3339,
+      "step": 2046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.916363031009519e-05,
+      "loss": 0.3432,
+      "step": 2047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9161780254851966e-05,
+      "loss": 0.3493,
+      "step": 2048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.915992819453826e-05,
+      "loss": 0.3503,
+      "step": 2049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.915807412932548e-05,
+      "loss": 0.3391,
+      "step": 2050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.915621805938526e-05,
+      "loss": 0.3297,
+      "step": 2051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9154359984889395e-05,
+      "loss": 0.3592,
+      "step": 2052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.915249990600987e-05,
+      "loss": 0.3568,
+      "step": 2053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.915063782291886e-05,
+      "loss": 0.3464,
+      "step": 2054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9148773735788735e-05,
+      "loss": 0.3636,
+      "step": 2055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.914690764479202e-05,
+      "loss": 0.3475,
+      "step": 2056
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.914503955010146e-05,
+      "loss": 0.3397,
+      "step": 2057
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.914316945188997e-05,
+      "loss": 0.3381,
+      "step": 2058
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9141297350330654e-05,
+      "loss": 0.3625,
+      "step": 2059
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.913942324559679e-05,
+      "loss": 0.3503,
+      "step": 2060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9137547137861846e-05,
+      "loss": 0.3634,
+      "step": 2061
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.913566902729949e-05,
+      "loss": 0.3679,
+      "step": 2062
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9133788914083555e-05,
+      "loss": 0.3503,
+      "step": 2063
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.913190679838808e-05,
+      "loss": 0.3331,
+      "step": 2064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9130022680387284e-05,
+      "loss": 0.3531,
+      "step": 2065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.912813656025554e-05,
+      "loss": 0.3436,
+      "step": 2066
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.912624843816746e-05,
+      "loss": 0.3528,
+      "step": 2067
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9124358314297794e-05,
+      "loss": 0.3503,
+      "step": 2068
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.91224661888215e-05,
+      "loss": 0.3519,
+      "step": 2069
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.912057206191373e-05,
+      "loss": 0.332,
+      "step": 2070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.91186759337498e-05,
+      "loss": 0.362,
+      "step": 2071
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.911677780450522e-05,
+      "loss": 0.3305,
+      "step": 2072
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.911487767435569e-05,
+      "loss": 0.3443,
+      "step": 2073
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.911297554347708e-05,
+      "loss": 0.3423,
+      "step": 2074
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.911107141204547e-05,
+      "loss": 0.3407,
+      "step": 2075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.910916528023711e-05,
+      "loss": 0.3588,
+      "step": 2076
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9107257148228434e-05,
+      "loss": 0.3532,
+      "step": 2077
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.910534701619605e-05,
+      "loss": 0.3405,
+      "step": 2078
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.910343488431679e-05,
+      "loss": 0.3538,
+      "step": 2079
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9101520752767625e-05,
+      "loss": 0.3535,
+      "step": 2080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9099604621725745e-05,
+      "loss": 0.3325,
+      "step": 2081
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.90976864913685e-05,
+      "loss": 0.3594,
+      "step": 2082
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9095766361873446e-05,
+      "loss": 0.3438,
+      "step": 2083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9093844233418314e-05,
+      "loss": 0.3432,
+      "step": 2084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.909192010618102e-05,
+      "loss": 0.3569,
+      "step": 2085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9089993980339663e-05,
+      "loss": 0.346,
+      "step": 2086
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.908806585607254e-05,
+      "loss": 0.3513,
+      "step": 2087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.908613573355812e-05,
+      "loss": 0.3406,
+      "step": 2088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.908420361297505e-05,
+      "loss": 0.3674,
+      "step": 2089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9082269494502176e-05,
+      "loss": 0.3491,
+      "step": 2090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.908033337831855e-05,
+      "loss": 0.3556,
+      "step": 2091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9078395264603335e-05,
+      "loss": 0.3314,
+      "step": 2092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9076455153535964e-05,
+      "loss": 0.3586,
+      "step": 2093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9074513045296e-05,
+      "loss": 0.3645,
+      "step": 2094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.907256894006324e-05,
+      "loss": 0.3491,
+      "step": 2095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.907062283801761e-05,
+      "loss": 0.33,
+      "step": 2096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9068674739339236e-05,
+      "loss": 0.3395,
+      "step": 2097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.906672464420847e-05,
+      "loss": 0.3353,
+      "step": 2098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9064772552805795e-05,
+      "loss": 0.3518,
+      "step": 2099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.906281846531191e-05,
+      "loss": 0.3401,
+      "step": 2100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.906086238190769e-05,
+      "loss": 0.3391,
+      "step": 2101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.905890430277418e-05,
+      "loss": 0.3805,
+      "step": 2102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.905694422809265e-05,
+      "loss": 0.3608,
+      "step": 2103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.905498215804452e-05,
+      "loss": 0.357,
+      "step": 2104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.90530180928114e-05,
+      "loss": 0.3321,
+      "step": 2105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.905105203257509e-05,
+      "loss": 0.3308,
+      "step": 2106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.904908397751758e-05,
+      "loss": 0.3449,
+      "step": 2107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9047113927821034e-05,
+      "loss": 0.3411,
+      "step": 2108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9045141883667794e-05,
+      "loss": 0.3732,
+      "step": 2109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.904316784524041e-05,
+      "loss": 0.3465,
+      "step": 2110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.904119181272162e-05,
+      "loss": 0.3578,
+      "step": 2111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9039213786294293e-05,
+      "loss": 0.3592,
+      "step": 2112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9037233766141554e-05,
+      "loss": 0.35,
+      "step": 2113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9035251752446656e-05,
+      "loss": 0.379,
+      "step": 2114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9033267745393066e-05,
+      "loss": 0.3492,
+      "step": 2115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9031281745164436e-05,
+      "loss": 0.3625,
+      "step": 2116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.902929375194459e-05,
+      "loss": 0.3504,
+      "step": 2117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.902730376591754e-05,
+      "loss": 0.3565,
+      "step": 2118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9025311787267485e-05,
+      "loss": 0.3401,
+      "step": 2119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.902331781617882e-05,
+      "loss": 0.3681,
+      "step": 2120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.902132185283609e-05,
+      "loss": 0.3313,
+      "step": 2121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.901932389742406e-05,
+      "loss": 0.3386,
+      "step": 2122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.901732395012766e-05,
+      "loss": 0.3589,
+      "step": 2123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9015322011132024e-05,
+      "loss": 0.3495,
+      "step": 2124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.901331808062244e-05,
+      "loss": 0.3452,
+      "step": 2125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.90113121587844e-05,
+      "loss": 0.3407,
+      "step": 2126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9009304245803584e-05,
+      "loss": 0.3511,
+      "step": 2127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.900729434186585e-05,
+      "loss": 0.3691,
+      "step": 2128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.900528244715723e-05,
+      "loss": 0.3344,
+      "step": 2129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.900326856186396e-05,
+      "loss": 0.3597,
+      "step": 2130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9001252686172456e-05,
+      "loss": 0.3484,
+      "step": 2131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.899923482026929e-05,
+      "loss": 0.3329,
+      "step": 2132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.899721496434126e-05,
+      "loss": 0.3414,
+      "step": 2133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8995193118575325e-05,
+      "loss": 0.3428,
+      "step": 2134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.899316928315863e-05,
+      "loss": 0.3441,
+      "step": 2135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.899114345827851e-05,
+      "loss": 0.3701,
+      "step": 2136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8989115644122474e-05,
+      "loss": 0.3328,
+      "step": 2137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8987085840878226e-05,
+      "loss": 0.3487,
+      "step": 2138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.898505404873366e-05,
+      "loss": 0.3432,
+      "step": 2139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.898302026787682e-05,
+      "loss": 0.3388,
+      "step": 2140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.898098449849597e-05,
+      "loss": 0.3483,
+      "step": 2141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.897894674077956e-05,
+      "loss": 0.3475,
+      "step": 2142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.897690699491619e-05,
+      "loss": 0.3329,
+      "step": 2143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8974865261094676e-05,
+      "loss": 0.3332,
+      "step": 2144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8972821539504e-05,
+      "loss": 0.3568,
+      "step": 2145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.897077583033334e-05,
+      "loss": 0.3386,
+      "step": 2146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.896872813377204e-05,
+      "loss": 0.347,
+      "step": 2147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.896667845000966e-05,
+      "loss": 0.341,
+      "step": 2148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.89646267792359e-05,
+      "loss": 0.3451,
+      "step": 2149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8962573121640684e-05,
+      "loss": 0.3413,
+      "step": 2150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.896051747741411e-05,
+      "loss": 0.3376,
+      "step": 2151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.895845984674643e-05,
+      "loss": 0.3556,
+      "step": 2152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8956400229828124e-05,
+      "loss": 0.3531,
+      "step": 2153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.895433862684983e-05,
+      "loss": 0.3378,
+      "step": 2154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8952275038002384e-05,
+      "loss": 0.3378,
+      "step": 2155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.895020946347677e-05,
+      "loss": 0.36,
+      "step": 2156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.894814190346421e-05,
+      "loss": 0.3451,
+      "step": 2157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.894607235815608e-05,
+      "loss": 0.3482,
+      "step": 2158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8944000827743924e-05,
+      "loss": 0.3465,
+      "step": 2159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.894192731241951e-05,
+      "loss": 0.3545,
+      "step": 2160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.893985181237475e-05,
+      "loss": 0.3389,
+      "step": 2161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.893777432780177e-05,
+      "loss": 0.3419,
+      "step": 2162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.893569485889287e-05,
+      "loss": 0.3418,
+      "step": 2163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8933613405840506e-05,
+      "loss": 0.3575,
+      "step": 2164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.893152996883737e-05,
+      "loss": 0.3513,
+      "step": 2165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.89294445480763e-05,
+      "loss": 0.3375,
+      "step": 2166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.892735714375034e-05,
+      "loss": 0.3449,
+      "step": 2167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8925267756052685e-05,
+      "loss": 0.3315,
+      "step": 2168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.892317638517674e-05,
+      "loss": 0.3253,
+      "step": 2169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.89210830313161e-05,
+      "loss": 0.3437,
+      "step": 2170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8918987694664515e-05,
+      "loss": 0.3469,
+      "step": 2171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8916890375415944e-05,
+      "loss": 0.3407,
+      "step": 2172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.891479107376452e-05,
+      "loss": 0.3382,
+      "step": 2173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8912689789904566e-05,
+      "loss": 0.334,
+      "step": 2174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.891058652403056e-05,
+      "loss": 0.3432,
+      "step": 2175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.890848127633721e-05,
+      "loss": 0.3364,
+      "step": 2176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8906374047019374e-05,
+      "loss": 0.3612,
+      "step": 2177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.89042648362721e-05,
+      "loss": 0.3475,
+      "step": 2178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.890215364429062e-05,
+      "loss": 0.3438,
+      "step": 2179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.890004047127037e-05,
+      "loss": 0.3484,
+      "step": 2180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.889792531740693e-05,
+      "loss": 0.3604,
+      "step": 2181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8895808182896086e-05,
+      "loss": 0.3579,
+      "step": 2182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.889368906793381e-05,
+      "loss": 0.3542,
+      "step": 2183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.889156797271626e-05,
+      "loss": 0.3546,
+      "step": 2184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8889444897439756e-05,
+      "loss": 0.3344,
+      "step": 2185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.888731984230082e-05,
+      "loss": 0.3354,
+      "step": 2186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.888519280749616e-05,
+      "loss": 0.3607,
+      "step": 2187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.888306379322266e-05,
+      "loss": 0.3381,
+      "step": 2188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.888093279967737e-05,
+      "loss": 0.3498,
+      "step": 2189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.887879982705756e-05,
+      "loss": 0.3447,
+      "step": 2190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.887666487556065e-05,
+      "loss": 0.3336,
+      "step": 2191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.887452794538427e-05,
+      "loss": 0.3442,
+      "step": 2192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.887238903672621e-05,
+      "loss": 0.354,
+      "step": 2193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.887024814978445e-05,
+      "loss": 0.3487,
+      "step": 2194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8868105284757163e-05,
+      "loss": 0.3327,
+      "step": 2195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8865960441842695e-05,
+      "loss": 0.3342,
+      "step": 2196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.886381362123959e-05,
+      "loss": 0.3472,
+      "step": 2197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.886166482314654e-05,
+      "loss": 0.3517,
+      "step": 2198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.885951404776246e-05,
+      "loss": 0.3125,
+      "step": 2199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.885736129528642e-05,
+      "loss": 0.3526,
+      "step": 2200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8855206565917704e-05,
+      "loss": 0.3368,
+      "step": 2201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.885304985985575e-05,
+      "loss": 0.3472,
+      "step": 2202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8850891177300176e-05,
+      "loss": 0.3402,
+      "step": 2203
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.88487305184508e-05,
+      "loss": 0.3405,
+      "step": 2204
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.884656788350763e-05,
+      "loss": 0.3322,
+      "step": 2205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.884440327267084e-05,
+      "loss": 0.351,
+      "step": 2206
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8842236686140784e-05,
+      "loss": 0.3504,
+      "step": 2207
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8840068124118014e-05,
+      "loss": 0.35,
+      "step": 2208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.883789758680326e-05,
+      "loss": 0.3344,
+      "step": 2209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8835725074397424e-05,
+      "loss": 0.339,
+      "step": 2210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.883355058710161e-05,
+      "loss": 0.3499,
+      "step": 2211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8831374125117083e-05,
+      "loss": 0.3486,
+      "step": 2212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.882919568864531e-05,
+      "loss": 0.3551,
+      "step": 2213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.882701527788794e-05,
+      "loss": 0.3476,
+      "step": 2214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8824832893046773e-05,
+      "loss": 0.3405,
+      "step": 2215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8822648534323833e-05,
+      "loss": 0.3567,
+      "step": 2216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.882046220192131e-05,
+      "loss": 0.3232,
+      "step": 2217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8818273896041575e-05,
+      "loss": 0.3558,
+      "step": 2218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.881608361688718e-05,
+      "loss": 0.3482,
+      "step": 2219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8813891364660876e-05,
+      "loss": 0.3483,
+      "step": 2220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.881169713956556e-05,
+      "loss": 0.3423,
+      "step": 2221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8809500941804354e-05,
+      "loss": 0.3389,
+      "step": 2222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8807302771580535e-05,
+      "loss": 0.3421,
+      "step": 2223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8805102629097574e-05,
+      "loss": 0.3431,
+      "step": 2224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.880290051455913e-05,
+      "loss": 0.3701,
+      "step": 2225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.880069642816902e-05,
+      "loss": 0.3407,
+      "step": 2226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.879849037013127e-05,
+      "loss": 0.3384,
+      "step": 2227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.879628234065008e-05,
+      "loss": 0.326,
+      "step": 2228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.879407233992982e-05,
+      "loss": 0.3553,
+      "step": 2229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.879186036817507e-05,
+      "loss": 0.3391,
+      "step": 2230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.878964642559057e-05,
+      "loss": 0.3387,
+      "step": 2231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.878743051238123e-05,
+      "loss": 0.3405,
+      "step": 2232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.878521262875219e-05,
+      "loss": 0.3584,
+      "step": 2233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8782992774908724e-05,
+      "loss": 0.3365,
+      "step": 2234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.878077095105632e-05,
+      "loss": 0.3531,
+      "step": 2235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.877854715740062e-05,
+      "loss": 0.3459,
+      "step": 2236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.877632139414747e-05,
+      "loss": 0.3517,
+      "step": 2237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.877409366150291e-05,
+      "loss": 0.3308,
+      "step": 2238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.877186395967312e-05,
+      "loss": 0.3495,
+      "step": 2239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.876963228886449e-05,
+      "loss": 0.3493,
+      "step": 2240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.87673986492836e-05,
+      "loss": 0.3437,
+      "step": 2241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8765163041137196e-05,
+      "loss": 0.358,
+      "step": 2242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.876292546463222e-05,
+      "loss": 0.3553,
+      "step": 2243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8760685919975773e-05,
+      "loss": 0.3522,
+      "step": 2244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.875844440737517e-05,
+      "loss": 0.3356,
+      "step": 2245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.875620092703788e-05,
+      "loss": 0.3489,
+      "step": 2246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.875395547917156e-05,
+      "loss": 0.3701,
+      "step": 2247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.875170806398406e-05,
+      "loss": 0.3589,
+      "step": 2248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.874945868168341e-05,
+      "loss": 0.3381,
+      "step": 2249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8747207332477825e-05,
+      "loss": 0.3433,
+      "step": 2250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8744954016575685e-05,
+      "loss": 0.3448,
+      "step": 2251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8742698734185564e-05,
+      "loss": 0.3643,
+      "step": 2252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.874044148551621e-05,
+      "loss": 0.3491,
+      "step": 2253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.873818227077658e-05,
+      "loss": 0.3593,
+      "step": 2254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8735921090175784e-05,
+      "loss": 0.3517,
+      "step": 2255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.873365794392311e-05,
+      "loss": 0.3607,
+      "step": 2256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8731392832228055e-05,
+      "loss": 0.3432,
+      "step": 2257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8729125755300284e-05,
+      "loss": 0.3543,
+      "step": 2258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8726856713349636e-05,
+      "loss": 0.3648,
+      "step": 2259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8724585706586126e-05,
+      "loss": 0.3407,
+      "step": 2260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8722312735219996e-05,
+      "loss": 0.3412,
+      "step": 2261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.872003779946162e-05,
+      "loss": 0.3546,
+      "step": 2262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.871776089952157e-05,
+      "loss": 0.3655,
+      "step": 2263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.871548203561061e-05,
+      "loss": 0.328,
+      "step": 2264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.871320120793967e-05,
+      "loss": 0.3501,
+      "step": 2265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.871091841671987e-05,
+      "loss": 0.3246,
+      "step": 2266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8708633662162525e-05,
+      "loss": 0.3257,
+      "step": 2267
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.87063469444791e-05,
+      "loss": 0.3251,
+      "step": 2268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.870405826388126e-05,
+      "loss": 0.3447,
+      "step": 2269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.870176762058086e-05,
+      "loss": 0.3687,
+      "step": 2270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8699475014789936e-05,
+      "loss": 0.3138,
+      "step": 2271
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.869718044672068e-05,
+      "loss": 0.3596,
+      "step": 2272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8694883916585486e-05,
+      "loss": 0.3471,
+      "step": 2273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8692585424596936e-05,
+      "loss": 0.3536,
+      "step": 2274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.869028497096778e-05,
+      "loss": 0.3377,
+      "step": 2275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.868798255591095e-05,
+      "loss": 0.3455,
+      "step": 2276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.868567817963957e-05,
+      "loss": 0.3503,
+      "step": 2277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.868337184236693e-05,
+      "loss": 0.345,
+      "step": 2278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.868106354430652e-05,
+      "loss": 0.3322,
+      "step": 2279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8678753285671996e-05,
+      "loss": 0.3398,
+      "step": 2280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8676441066677206e-05,
+      "loss": 0.3296,
+      "step": 2281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.867412688753617e-05,
+      "loss": 0.3444,
+      "step": 2282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.86718107484631e-05,
+      "loss": 0.3388,
+      "step": 2283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8669492649672384e-05,
+      "loss": 0.3416,
+      "step": 2284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.866717259137858e-05,
+      "loss": 0.3224,
+      "step": 2285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.866485057379645e-05,
+      "loss": 0.3388,
+      "step": 2286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.866252659714092e-05,
+      "loss": 0.3706,
+      "step": 2287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.866020066162711e-05,
+      "loss": 0.3364,
+      "step": 2288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.865787276747031e-05,
+      "loss": 0.3521,
+      "step": 2289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8655542914885995e-05,
+      "loss": 0.3459,
+      "step": 2290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.865321110408983e-05,
+      "loss": 0.3421,
+      "step": 2291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.865087733529764e-05,
+      "loss": 0.3412,
+      "step": 2292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.864854160872546e-05,
+      "loss": 0.3465,
+      "step": 2293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.864620392458947e-05,
+      "loss": 0.3284,
+      "step": 2294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.864386428310607e-05,
+      "loss": 0.3332,
+      "step": 2295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8641522684491816e-05,
+      "loss": 0.3426,
+      "step": 2296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8639179128963466e-05,
+      "loss": 0.3474,
+      "step": 2297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.863683361673793e-05,
+      "loss": 0.3381,
+      "step": 2298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8634486148032314e-05,
+      "loss": 0.3439,
+      "step": 2299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.863213672306391e-05,
+      "loss": 0.341,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.862978534205019e-05,
+      "loss": 0.3555,
+      "step": 2301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.862743200520881e-05,
+      "loss": 0.3396,
+      "step": 2302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8625076712757575e-05,
+      "loss": 0.3325,
+      "step": 2303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.862271946491452e-05,
+      "loss": 0.3601,
+      "step": 2304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.862036026189784e-05,
+      "loss": 0.3522,
+      "step": 2305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.86179991039259e-05,
+      "loss": 0.3602,
+      "step": 2306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.861563599121725e-05,
+      "loss": 0.3357,
+      "step": 2307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8613270923990636e-05,
+      "loss": 0.3436,
+      "step": 2308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8610903902464966e-05,
+      "loss": 0.371,
+      "step": 2309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.860853492685935e-05,
+      "loss": 0.3325,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.860616399739305e-05,
+      "loss": 0.3368,
+      "step": 2311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.860379111428555e-05,
+      "loss": 0.3607,
+      "step": 2312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.860141627775646e-05,
+      "loss": 0.3364,
+      "step": 2313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8599039488025615e-05,
+      "loss": 0.3515,
+      "step": 2314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8596660745313024e-05,
+      "loss": 0.3642,
+      "step": 2315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8594280049838865e-05,
+      "loss": 0.3296,
+      "step": 2316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.859189740182349e-05,
+      "loss": 0.3368,
+      "step": 2317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.858951280148745e-05,
+      "loss": 0.3449,
+      "step": 2318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8587126249051484e-05,
+      "loss": 0.352,
+      "step": 2319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.858473774473648e-05,
+      "loss": 0.3436,
+      "step": 2320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.858234728876353e-05,
+      "loss": 0.3396,
+      "step": 2321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.85799548813539e-05,
+      "loss": 0.3469,
+      "step": 2322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.857756052272904e-05,
+      "loss": 0.3433,
+      "step": 2323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.857516421311058e-05,
+      "loss": 0.3472,
+      "step": 2324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.857276595272032e-05,
+      "loss": 0.3325,
+      "step": 2325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.857036574178025e-05,
+      "loss": 0.3458,
+      "step": 2326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.856796358051255e-05,
+      "loss": 0.3386,
+      "step": 2327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.856555946913956e-05,
+      "loss": 0.3672,
+      "step": 2328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.856315340788382e-05,
+      "loss": 0.3335,
+      "step": 2329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.856074539696803e-05,
+      "loss": 0.3587,
+      "step": 2330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.855833543661508e-05,
+      "loss": 0.3455,
+      "step": 2331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.855592352704806e-05,
+      "loss": 0.3415,
+      "step": 2332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.855350966849021e-05,
+      "loss": 0.3177,
+      "step": 2333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.855109386116497e-05,
+      "loss": 0.349,
+      "step": 2334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8548676105295946e-05,
+      "loss": 0.3588,
+      "step": 2335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8546256401106936e-05,
+      "loss": 0.3253,
+      "step": 2336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8543834748821904e-05,
+      "loss": 0.3389,
+      "step": 2337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.854141114866502e-05,
+      "loss": 0.3436,
+      "step": 2338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.853898560086061e-05,
+      "loss": 0.3446,
+      "step": 2339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.85365581056332e-05,
+      "loss": 0.3513,
+      "step": 2340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.853412866320746e-05,
+      "loss": 0.3556,
+      "step": 2341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.853169727380829e-05,
+      "loss": 0.3679,
+      "step": 2342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.852926393766074e-05,
+      "loss": 0.353,
+      "step": 2343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8526828654990033e-05,
+      "loss": 0.345,
+      "step": 2344
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8524391426021605e-05,
+      "loss": 0.3429,
+      "step": 2345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.852195225098104e-05,
+      "loss": 0.3472,
+      "step": 2346
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8519511130094115e-05,
+      "loss": 0.3481,
+      "step": 2347
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8517068063586786e-05,
+      "loss": 0.356,
+      "step": 2348
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.851462305168519e-05,
+      "loss": 0.3573,
+      "step": 2349
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.851217609461566e-05,
+      "loss": 0.3324,
+      "step": 2350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.850972719260466e-05,
+      "loss": 0.368,
+      "step": 2351
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8507276345878894e-05,
+      "loss": 0.3427,
+      "step": 2352
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.850482355466521e-05,
+      "loss": 0.3389,
+      "step": 2353
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.850236881919064e-05,
+      "loss": 0.326,
+      "step": 2354
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.849991213968241e-05,
+      "loss": 0.3374,
+      "step": 2355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8497453516367906e-05,
+      "loss": 0.355,
+      "step": 2356
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.849499294947471e-05,
+      "loss": 0.3439,
+      "step": 2357
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.849253043923058e-05,
+      "loss": 0.3348,
+      "step": 2358
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.849006598586345e-05,
+      "loss": 0.359,
+      "step": 2359
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.848759958960144e-05,
+      "loss": 0.3278,
+      "step": 2360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.848513125067285e-05,
+      "loss": 0.3692,
+      "step": 2361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.848266096930614e-05,
+      "loss": 0.3609,
+      "step": 2362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.848018874572997e-05,
+      "loss": 0.3437,
+      "step": 2363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.847771458017318e-05,
+      "loss": 0.36,
+      "step": 2364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.84752384728648e-05,
+      "loss": 0.3341,
+      "step": 2365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8472760424033996e-05,
+      "loss": 0.3499,
+      "step": 2366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.847028043391016e-05,
+      "loss": 0.3307,
+      "step": 2367
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.846779850272284e-05,
+      "loss": 0.3516,
+      "step": 2368
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.846531463070178e-05,
+      "loss": 0.332,
+      "step": 2369
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8462828818076885e-05,
+      "loss": 0.3659,
+      "step": 2370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8460341065078246e-05,
+      "loss": 0.3282,
+      "step": 2371
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.845785137193614e-05,
+      "loss": 0.3466,
+      "step": 2372
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.845535973888102e-05,
+      "loss": 0.3717,
+      "step": 2373
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8452866166143526e-05,
+      "loss": 0.3565,
+      "step": 2374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.845037065395445e-05,
+      "loss": 0.3654,
+      "step": 2375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.844787320254479e-05,
+      "loss": 0.3542,
+      "step": 2376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.844537381214573e-05,
+      "loss": 0.3572,
+      "step": 2377
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.84428724829886e-05,
+      "loss": 0.3345,
+      "step": 2378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.844036921530495e-05,
+      "loss": 0.3329,
+      "step": 2379
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.843786400932648e-05,
+      "loss": 0.362,
+      "step": 2380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.843535686528507e-05,
+      "loss": 0.3274,
+      "step": 2381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.84328477834128e-05,
+      "loss": 0.3638,
+      "step": 2382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.843033676394191e-05,
+      "loss": 0.3418,
+      "step": 2383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8427823807104824e-05,
+      "loss": 0.3431,
+      "step": 2384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.842530891313416e-05,
+      "loss": 0.3307,
+      "step": 2385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8422792082262686e-05,
+      "loss": 0.3477,
+      "step": 2386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8420273314723386e-05,
+      "loss": 0.3691,
+      "step": 2387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8417752610749384e-05,
+      "loss": 0.3318,
+      "step": 2388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.841522997057402e-05,
+      "loss": 0.3435,
+      "step": 2389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8412705394430784e-05,
+      "loss": 0.3522,
+      "step": 2390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.841017888255337e-05,
+      "loss": 0.3398,
+      "step": 2391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8407650435175615e-05,
+      "loss": 0.3609,
+      "step": 2392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8405120052531584e-05,
+      "loss": 0.3493,
+      "step": 2393
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.840258773485549e-05,
+      "loss": 0.3529,
+      "step": 2394
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8400053482381715e-05,
+      "loss": 0.3466,
+      "step": 2395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.839751729534485e-05,
+      "loss": 0.3275,
+      "step": 2396
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8394979173979655e-05,
+      "loss": 0.361,
+      "step": 2397
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.839243911852105e-05,
+      "loss": 0.3494,
+      "step": 2398
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.838989712920417e-05,
+      "loss": 0.3247,
+      "step": 2399
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.838735320626429e-05,
+      "loss": 0.351,
+      "step": 2400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.838480734993689e-05,
+      "loss": 0.3651,
+      "step": 2401
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.838225956045761e-05,
+      "loss": 0.3301,
+      "step": 2402
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.837970983806229e-05,
+      "loss": 0.3673,
+      "step": 2403
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.837715818298695e-05,
+      "loss": 0.3405,
+      "step": 2404
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8374604595467754e-05,
+      "loss": 0.3375,
+      "step": 2405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.837204907574109e-05,
+      "loss": 0.3355,
+      "step": 2406
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.836949162404348e-05,
+      "loss": 0.3455,
+      "step": 2407
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8366932240611676e-05,
+      "loss": 0.3469,
+      "step": 2408
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.836437092568257e-05,
+      "loss": 0.336,
+      "step": 2409
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8361807679493226e-05,
+      "loss": 0.3599,
+      "step": 2410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8359242502280927e-05,
+      "loss": 0.3477,
+      "step": 2411
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8356675394283106e-05,
+      "loss": 0.3537,
+      "step": 2412
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8354106355737387e-05,
+      "loss": 0.3368,
+      "step": 2413
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.835153538688155e-05,
+      "loss": 0.3295,
+      "step": 2414
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.834896248795359e-05,
+      "loss": 0.3628,
+      "step": 2415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.834638765919165e-05,
+      "loss": 0.3594,
+      "step": 2416
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.834381090083406e-05,
+      "loss": 0.3441,
+      "step": 2417
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8341232213119336e-05,
+      "loss": 0.3482,
+      "step": 2418
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8338651596286175e-05,
+      "loss": 0.3467,
+      "step": 2419
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.833606905057343e-05,
+      "loss": 0.3531,
+      "step": 2420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.833348457622017e-05,
+      "loss": 0.3366,
+      "step": 2421
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.83308981734656e-05,
+      "loss": 0.3385,
+      "step": 2422
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8328309842549135e-05,
+      "loss": 0.3568,
+      "step": 2423
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.832571958371035e-05,
+      "loss": 0.33,
+      "step": 2424
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8323127397189015e-05,
+      "loss": 0.3426,
+      "step": 2425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8320533283225066e-05,
+      "loss": 0.3482,
+      "step": 2426
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.831793724205861e-05,
+      "loss": 0.3478,
+      "step": 2427
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.831533927392997e-05,
+      "loss": 0.3409,
+      "step": 2428
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8312739379079596e-05,
+      "loss": 0.3483,
+      "step": 2429
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.831013755774814e-05,
+      "loss": 0.3376,
+      "step": 2430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.830753381017646e-05,
+      "loss": 0.3775,
+      "step": 2431
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.830492813660553e-05,
+      "loss": 0.3301,
+      "step": 2432
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.830232053727657e-05,
+      "loss": 0.3622,
+      "step": 2433
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.829971101243093e-05,
+      "loss": 0.3417,
+      "step": 2434
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.829709956231015e-05,
+      "loss": 0.3533,
+      "step": 2435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8294486187155967e-05,
+      "loss": 0.334,
+      "step": 2436
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.829187088721026e-05,
+      "loss": 0.3454,
+      "step": 2437
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.828925366271514e-05,
+      "loss": 0.3398,
+      "step": 2438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.828663451391284e-05,
+      "loss": 0.3563,
+      "step": 2439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.82840134410458e-05,
+      "loss": 0.3422,
+      "step": 2440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.828139044435663e-05,
+      "loss": 0.3503,
+      "step": 2441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.827876552408812e-05,
+      "loss": 0.3432,
+      "step": 2442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.827613868048326e-05,
+      "loss": 0.3334,
+      "step": 2443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.827350991378518e-05,
+      "loss": 0.3327,
+      "step": 2444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8270879224237195e-05,
+      "loss": 0.3387,
+      "step": 2445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.826824661208283e-05,
+      "loss": 0.353,
+      "step": 2446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8265612077565755e-05,
+      "loss": 0.3476,
+      "step": 2447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.826297562092984e-05,
+      "loss": 0.3438,
+      "step": 2448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8260337242419106e-05,
+      "loss": 0.3386,
+      "step": 2449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.825769694227778e-05,
+      "loss": 0.3417,
+      "step": 2450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8255054720750254e-05,
+      "loss": 0.3451,
+      "step": 2451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.82524105780811e-05,
+      "loss": 0.3523,
+      "step": 2452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8249764514515054e-05,
+      "loss": 0.3415,
+      "step": 2453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.824711653029706e-05,
+      "loss": 0.3468,
+      "step": 2454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8244466625672205e-05,
+      "loss": 0.354,
+      "step": 2455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.824181480088578e-05,
+      "loss": 0.3371,
+      "step": 2456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.823916105618326e-05,
+      "loss": 0.3347,
+      "step": 2457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.823650539181025e-05,
+      "loss": 0.3512,
+      "step": 2458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8233847808012586e-05,
+      "loss": 0.3212,
+      "step": 2459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.823118830503626e-05,
+      "loss": 0.3437,
+      "step": 2460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.822852688312744e-05,
+      "loss": 0.3551,
+      "step": 2461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.822586354253247e-05,
+      "loss": 0.339,
+      "step": 2462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8223198283497886e-05,
+      "loss": 0.3313,
+      "step": 2463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.822053110627038e-05,
+      "loss": 0.3426,
+      "step": 2464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8217862011096844e-05,
+      "loss": 0.3264,
+      "step": 2465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.821519099822432e-05,
+      "loss": 0.3345,
+      "step": 2466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.821251806790007e-05,
+      "loss": 0.3379,
+      "step": 2467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8209843220371476e-05,
+      "loss": 0.3556,
+      "step": 2468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.820716645588615e-05,
+      "loss": 0.3552,
+      "step": 2469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8204487774691855e-05,
+      "loss": 0.3569,
+      "step": 2470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.820180717703653e-05,
+      "loss": 0.3605,
+      "step": 2471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.819912466316831e-05,
+      "loss": 0.3565,
+      "step": 2472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.819644023333549e-05,
+      "loss": 0.3355,
+      "step": 2473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.819375388778654e-05,
+      "loss": 0.3526,
+      "step": 2474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.819106562677013e-05,
+      "loss": 0.3316,
+      "step": 2475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.818837545053508e-05,
+      "loss": 0.3733,
+      "step": 2476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8185683359330406e-05,
+      "loss": 0.3393,
+      "step": 2477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.818298935340529e-05,
+      "loss": 0.3246,
+      "step": 2478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.818029343300911e-05,
+      "loss": 0.3219,
+      "step": 2479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.817759559839139e-05,
+      "loss": 0.3313,
+      "step": 2480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8174895849801855e-05,
+      "loss": 0.3505,
+      "step": 2481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.817219418749041e-05,
+      "loss": 0.3346,
+      "step": 2482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.816949061170711e-05,
+      "loss": 0.3623,
+      "step": 2483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.816678512270221e-05,
+      "loss": 0.3564,
+      "step": 2484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8164077720726144e-05,
+      "loss": 0.3571,
+      "step": 2485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8161368406029514e-05,
+      "loss": 0.3322,
+      "step": 2486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8158657178863115e-05,
+      "loss": 0.3481,
+      "step": 2487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.815594403947787e-05,
+      "loss": 0.3376,
+      "step": 2488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.815322898812494e-05,
+      "loss": 0.3504,
+      "step": 2489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8150512025055647e-05,
+      "loss": 0.3585,
+      "step": 2490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.814779315052145e-05,
+      "loss": 0.3286,
+      "step": 2491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.814507236477404e-05,
+      "loss": 0.3098,
+      "step": 2492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8142349668065235e-05,
+      "loss": 0.3466,
+      "step": 2493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.813962506064708e-05,
+      "loss": 0.3548,
+      "step": 2494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.813689854277176e-05,
+      "loss": 0.3504,
+      "step": 2495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.813417011469165e-05,
+      "loss": 0.3451,
+      "step": 2496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.813143977665931e-05,
+      "loss": 0.3494,
+      "step": 2497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8128707528927445e-05,
+      "loss": 0.3429,
+      "step": 2498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.812597337174898e-05,
+      "loss": 0.3468,
+      "step": 2499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8123237305376975e-05,
+      "loss": 0.3386,
+      "step": 2500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.812049933006471e-05,
+      "loss": 0.3599,
+      "step": 2501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.811775944606561e-05,
+      "loss": 0.3292,
+      "step": 2502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8115017653633285e-05,
+      "loss": 0.3564,
+      "step": 2503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.811227395302152e-05,
+      "loss": 0.3264,
+      "step": 2504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.810952834448429e-05,
+      "loss": 0.3448,
+      "step": 2505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.810678082827571e-05,
+      "loss": 0.3428,
+      "step": 2506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8104031404650126e-05,
+      "loss": 0.3429,
+      "step": 2507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.810128007386202e-05,
+      "loss": 0.335,
+      "step": 2508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8098526836166055e-05,
+      "loss": 0.3297,
+      "step": 2509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.809577169181709e-05,
+      "loss": 0.3371,
+      "step": 2510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.809301464107014e-05,
+      "loss": 0.3635,
+      "step": 2511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.809025568418041e-05,
+      "loss": 0.3423,
+      "step": 2512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8087494821403275e-05,
+      "loss": 0.3525,
+      "step": 2513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.808473205299428e-05,
+      "loss": 0.3241,
+      "step": 2514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.808196737920917e-05,
+      "loss": 0.339,
+      "step": 2515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.807920080030384e-05,
+      "loss": 0.3437,
+      "step": 2516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.807643231653437e-05,
+      "loss": 0.3321,
+      "step": 2517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8073661928157027e-05,
+      "loss": 0.3322,
+      "step": 2518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.807088963542824e-05,
+      "loss": 0.3534,
+      "step": 2519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.80681154386046e-05,
+      "loss": 0.3661,
+      "step": 2520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.806533933794293e-05,
+      "loss": 0.3496,
+      "step": 2521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8062561333700166e-05,
+      "loss": 0.3481,
+      "step": 2522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8059781426133465e-05,
+      "loss": 0.3603,
+      "step": 2523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.805699961550013e-05,
+      "loss": 0.3302,
+      "step": 2524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8054215902057664e-05,
+      "loss": 0.3449,
+      "step": 2525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.805143028606373e-05,
+      "loss": 0.3513,
+      "step": 2526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.804864276777615e-05,
+      "loss": 0.3322,
+      "step": 2527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.804585334745298e-05,
+      "loss": 0.3367,
+      "step": 2528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.80430620253524e-05,
+      "loss": 0.3306,
+      "step": 2529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.804026880173277e-05,
+      "loss": 0.3457,
+      "step": 2530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.803747367685266e-05,
+      "loss": 0.356,
+      "step": 2531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8034676650970784e-05,
+      "loss": 0.3419,
+      "step": 2532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.803187772434604e-05,
+      "loss": 0.3387,
+      "step": 2533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8029076897237496e-05,
+      "loss": 0.3643,
+      "step": 2534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.802627416990442e-05,
+      "loss": 0.3562,
+      "step": 2535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.802346954260624e-05,
+      "loss": 0.3395,
+      "step": 2536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.802066301560254e-05,
+      "loss": 0.341,
+      "step": 2537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.801785458915312e-05,
+      "loss": 0.335,
+      "step": 2538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.801504426351793e-05,
+      "loss": 0.3495,
+      "step": 2539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.801223203895709e-05,
+      "loss": 0.3448,
+      "step": 2540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.800941791573092e-05,
+      "loss": 0.3581,
+      "step": 2541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8006601894099896e-05,
+      "loss": 0.3509,
+      "step": 2542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.800378397432468e-05,
+      "loss": 0.3195,
+      "step": 2543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8000964156666094e-05,
+      "loss": 0.3599,
+      "step": 2544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.799814244138516e-05,
+      "loss": 0.3367,
+      "step": 2545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7995318828743074e-05,
+      "loss": 0.35,
+      "step": 2546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.799249331900117e-05,
+      "loss": 0.3211,
+      "step": 2547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7989665912421e-05,
+      "loss": 0.3501,
+      "step": 2548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7986836609264285e-05,
+      "loss": 0.3455,
+      "step": 2549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.798400540979289e-05,
+      "loss": 0.3556,
+      "step": 2550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.798117231426889e-05,
+      "loss": 0.3429,
+      "step": 2551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7978337322954536e-05,
+      "loss": 0.3428,
+      "step": 2552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7975500436112214e-05,
+      "loss": 0.3627,
+      "step": 2553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7972661654004536e-05,
+      "loss": 0.3471,
+      "step": 2554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.796982097689426e-05,
+      "loss": 0.3519,
+      "step": 2555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.796697840504433e-05,
+      "loss": 0.3488,
+      "step": 2556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.796413393871785e-05,
+      "loss": 0.3337,
+      "step": 2557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7961287578178126e-05,
+      "loss": 0.3313,
+      "step": 2558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.795843932368862e-05,
+      "loss": 0.3402,
+      "step": 2559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7955589175512974e-05,
+      "loss": 0.3431,
+      "step": 2560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7952737133914995e-05,
+      "loss": 0.334,
+      "step": 2561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.794988319915868e-05,
+      "loss": 0.321,
+      "step": 2562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.794702737150821e-05,
+      "loss": 0.3318,
+      "step": 2563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7944169651227915e-05,
+      "loss": 0.3563,
+      "step": 2564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.794131003858232e-05,
+      "loss": 0.3512,
+      "step": 2565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7938448533836104e-05,
+      "loss": 0.3335,
+      "step": 2566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7935585137254155e-05,
+      "loss": 0.3397,
+      "step": 2567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.79327198491015e-05,
+      "loss": 0.3595,
+      "step": 2568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.792985266964337e-05,
+      "loss": 0.3347,
+      "step": 2569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.792698359914515e-05,
+      "loss": 0.3386,
+      "step": 2570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.792411263787241e-05,
+      "loss": 0.3484,
+      "step": 2571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7921239786090896e-05,
+      "loss": 0.3329,
+      "step": 2572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7918365044066526e-05,
+      "loss": 0.341,
+      "step": 2573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7915488412065394e-05,
+      "loss": 0.319,
+      "step": 2574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.791260989035376e-05,
+      "loss": 0.3329,
+      "step": 2575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.790972947919809e-05,
+      "loss": 0.351,
+      "step": 2576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7906847178864975e-05,
+      "loss": 0.3381,
+      "step": 2577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7903962989621224e-05,
+      "loss": 0.3477,
+      "step": 2578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.79010769117338e-05,
+      "loss": 0.3632,
+      "step": 2579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.789818894546986e-05,
+      "loss": 0.334,
+      "step": 2580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.78952990910967e-05,
+      "loss": 0.3367,
+      "step": 2581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.789240734888182e-05,
+      "loss": 0.3493,
+      "step": 2582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.788951371909289e-05,
+      "loss": 0.3449,
+      "step": 2583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7886618201997747e-05,
+      "loss": 0.3551,
+      "step": 2584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7883720797864404e-05,
+      "loss": 0.3293,
+      "step": 2585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.788082150696108e-05,
+      "loss": 0.3396,
+      "step": 2586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.787792032955611e-05,
+      "loss": 0.3469,
+      "step": 2587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7875017265918036e-05,
+      "loss": 0.3376,
+      "step": 2588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7872112316315596e-05,
+      "loss": 0.3547,
+      "step": 2589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.786920548101766e-05,
+      "loss": 0.3394,
+      "step": 2590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.786629676029329e-05,
+      "loss": 0.3227,
+      "step": 2591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.786338615441174e-05,
+      "loss": 0.3305,
+      "step": 2592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.786047366364242e-05,
+      "loss": 0.3333,
+      "step": 2593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.785755928825491e-05,
+      "loss": 0.3169,
+      "step": 2594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7854643028518974e-05,
+      "loss": 0.3443,
+      "step": 2595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.785172488470455e-05,
+      "loss": 0.354,
+      "step": 2596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7848804857081746e-05,
+      "loss": 0.3308,
+      "step": 2597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.784588294592085e-05,
+      "loss": 0.3239,
+      "step": 2598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.784295915149233e-05,
+      "loss": 0.3292,
+      "step": 2599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.78400334740668e-05,
+      "loss": 0.3384,
+      "step": 2600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7837105913915086e-05,
+      "loss": 0.3437,
+      "step": 2601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.783417647130817e-05,
+      "loss": 0.3313,
+      "step": 2602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.78312451465172e-05,
+      "loss": 0.316,
+      "step": 2603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7828311939813515e-05,
+      "loss": 0.3472,
+      "step": 2604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.782537685146861e-05,
+      "loss": 0.31,
+      "step": 2605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.782243988175417e-05,
+      "loss": 0.3399,
+      "step": 2606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781950103094205e-05,
+      "loss": 0.3445,
+      "step": 2607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781656029930427e-05,
+      "loss": 0.3416,
+      "step": 2608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7813617687113044e-05,
+      "loss": 0.3253,
+      "step": 2609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.781067319464075e-05,
+      "loss": 0.3711,
+      "step": 2610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.780772682215992e-05,
+      "loss": 0.3361,
+      "step": 2611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.780477856994329e-05,
+      "loss": 0.3285,
+      "step": 2612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7801828438263755e-05,
+      "loss": 0.3744,
+      "step": 2613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.779887642739438e-05,
+      "loss": 0.3502,
+      "step": 2614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.779592253760843e-05,
+      "loss": 0.3332,
+      "step": 2615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.779296676917931e-05,
+      "loss": 0.3671,
+      "step": 2616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7790009122380614e-05,
+      "loss": 0.3616,
+      "step": 2617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.778704959748611e-05,
+      "loss": 0.3354,
+      "step": 2618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.778408819476974e-05,
+      "loss": 0.3445,
+      "step": 2619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.778112491450562e-05,
+      "loss": 0.3397,
+      "step": 2620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7778159756968046e-05,
+      "loss": 0.3496,
+      "step": 2621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7775192722431474e-05,
+      "loss": 0.3047,
+      "step": 2622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.777222381117053e-05,
+      "loss": 0.3103,
+      "step": 2623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776925302346004e-05,
+      "loss": 0.328,
+      "step": 2624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776628035957498e-05,
+      "loss": 0.3335,
+      "step": 2625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7763305819790515e-05,
+      "loss": 0.3284,
+      "step": 2626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.776032940438197e-05,
+      "loss": 0.3586,
+      "step": 2627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7757351113624846e-05,
+      "loss": 0.3539,
+      "step": 2628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.775437094779483e-05,
+      "loss": 0.3478,
+      "step": 2629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7751388907167765e-05,
+      "loss": 0.3403,
+      "step": 2630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774840499201969e-05,
+      "loss": 0.3433,
+      "step": 2631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.774541920262679e-05,
+      "loss": 0.325,
+      "step": 2632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7742431539265434e-05,
+      "loss": 0.3326,
+      "step": 2633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773944200221218e-05,
+      "loss": 0.362,
+      "step": 2634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7736450591743745e-05,
+      "loss": 0.351,
+      "step": 2635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773345730813702e-05,
+      "loss": 0.3588,
+      "step": 2636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.773046215166908e-05,
+      "loss": 0.3386,
+      "step": 2637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7727465122617144e-05,
+      "loss": 0.3447,
+      "step": 2638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772446622125864e-05,
+      "loss": 0.3558,
+      "step": 2639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7721465447871156e-05,
+      "loss": 0.3444,
+      "step": 2640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7718462802732436e-05,
+      "loss": 0.3249,
+      "step": 2641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.771545828612044e-05,
+      "loss": 0.3421,
+      "step": 2642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.771245189831323e-05,
+      "loss": 0.3459,
+      "step": 2643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.770944363958912e-05,
+      "loss": 0.347,
+      "step": 2644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7706433510226564e-05,
+      "loss": 0.3312,
+      "step": 2645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.770342151050417e-05,
+      "loss": 0.3319,
+      "step": 2646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.770040764070074e-05,
+      "loss": 0.3348,
+      "step": 2647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.769739190109526e-05,
+      "loss": 0.3325,
+      "step": 2648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.769437429196687e-05,
+      "loss": 0.3409,
+      "step": 2649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7691354813594854e-05,
+      "loss": 0.3612,
+      "step": 2650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.768833346625875e-05,
+      "loss": 0.3499,
+      "step": 2651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7685310250238194e-05,
+      "loss": 0.3461,
+      "step": 2652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.768228516581304e-05,
+      "loss": 0.3252,
+      "step": 2653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.767925821326328e-05,
+      "loss": 0.3588,
+      "step": 2654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.76762293928691e-05,
+      "loss": 0.3328,
+      "step": 2655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.767319870491087e-05,
+      "loss": 0.3303,
+      "step": 2656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.767016614966911e-05,
+      "loss": 0.3408,
+      "step": 2657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.766713172742451e-05,
+      "loss": 0.358,
+      "step": 2658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.766409543845796e-05,
+      "loss": 0.3268,
+      "step": 2659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7661057283050505e-05,
+      "loss": 0.3505,
+      "step": 2660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.765801726148336e-05,
+      "loss": 0.3232,
+      "step": 2661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7654975374037907e-05,
+      "loss": 0.3261,
+      "step": 2662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.765193162099573e-05,
+      "loss": 0.3518,
+      "step": 2663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7648886002638554e-05,
+      "loss": 0.3274,
+      "step": 2664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7645838519248294e-05,
+      "loss": 0.3313,
+      "step": 2665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7642789171107026e-05,
+      "loss": 0.3408,
+      "step": 2666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.763973795849702e-05,
+      "loss": 0.3398,
+      "step": 2667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.763668488170069e-05,
+      "loss": 0.3604,
+      "step": 2668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.763362994100066e-05,
+      "loss": 0.3472,
+      "step": 2669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7630573136679666e-05,
+      "loss": 0.3308,
+      "step": 2670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.762751446902068e-05,
+      "loss": 0.3335,
+      "step": 2671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7624453938306815e-05,
+      "loss": 0.3459,
+      "step": 2672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7621391544821356e-05,
+      "loss": 0.3359,
+      "step": 2673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.761832728884778e-05,
+      "loss": 0.3841,
+      "step": 2674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.76152611706697e-05,
+      "loss": 0.3356,
+      "step": 2675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.761219319057095e-05,
+      "loss": 0.3304,
+      "step": 2676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.760912334883549e-05,
+      "loss": 0.3548,
+      "step": 2677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7606051645747484e-05,
+      "loss": 0.3408,
+      "step": 2678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7602978081591246e-05,
+      "loss": 0.3527,
+      "step": 2679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7599902656651285e-05,
+      "loss": 0.3474,
+      "step": 2680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.759682537121227e-05,
+      "loss": 0.3424,
+      "step": 2681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.759374622555903e-05,
+      "loss": 0.3483,
+      "step": 2682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.759066521997658e-05,
+      "loss": 0.343,
+      "step": 2683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.758758235475011e-05,
+      "loss": 0.3596,
+      "step": 2684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7584497630165e-05,
+      "loss": 0.3518,
+      "step": 2685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7581411046506745e-05,
+      "loss": 0.3527,
+      "step": 2686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.757832260406106e-05,
+      "loss": 0.3448,
+      "step": 2687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.757523230311383e-05,
+      "loss": 0.3289,
+      "step": 2688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.757214014395109e-05,
+      "loss": 0.3352,
+      "step": 2689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.756904612685906e-05,
+      "loss": 0.344,
+      "step": 2690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.756595025212412e-05,
+      "loss": 0.3301,
+      "step": 2691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7562852520032856e-05,
+      "loss": 0.3564,
+      "step": 2692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7559752930871984e-05,
+      "loss": 0.3229,
+      "step": 2693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.755665148492842e-05,
+      "loss": 0.3404,
+      "step": 2694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.755354818248924e-05,
+      "loss": 0.3603,
+      "step": 2695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7550443023841676e-05,
+      "loss": 0.333,
+      "step": 2696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7547336009273184e-05,
+      "loss": 0.3251,
+      "step": 2697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.754422713907132e-05,
+      "loss": 0.3261,
+      "step": 2698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.754111641352388e-05,
+      "loss": 0.3437,
+      "step": 2699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.753800383291878e-05,
+      "loss": 0.3414,
+      "step": 2700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.753488939754415e-05,
+      "loss": 0.325,
+      "step": 2701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.753177310768824e-05,
+      "loss": 0.3239,
+      "step": 2702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7528654963639527e-05,
+      "loss": 0.3525,
+      "step": 2703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7525534965686614e-05,
+      "loss": 0.3611,
+      "step": 2704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7522413114118316e-05,
+      "loss": 0.3381,
+      "step": 2705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.751928940922359e-05,
+      "loss": 0.3567,
+      "step": 2706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.751616385129158e-05,
+      "loss": 0.3219,
+      "step": 2707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.75130364406116e-05,
+      "loss": 0.3338,
+      "step": 2708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7509907177473104e-05,
+      "loss": 0.338,
+      "step": 2709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.750677606216577e-05,
+      "loss": 0.3459,
+      "step": 2710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7503643094979425e-05,
+      "loss": 0.3434,
+      "step": 2711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7500508276204046e-05,
+      "loss": 0.354,
+      "step": 2712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.749737160612981e-05,
+      "loss": 0.3458,
+      "step": 2713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7494233085047055e-05,
+      "loss": 0.3405,
+      "step": 2714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7491092713246283e-05,
+      "loss": 0.3381,
+      "step": 2715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.748795049101819e-05,
+      "loss": 0.3374,
+      "step": 2716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.748480641865362e-05,
+      "loss": 0.3427,
+      "step": 2717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.74816604964436e-05,
+      "loss": 0.3459,
+      "step": 2718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.747851272467931e-05,
+      "loss": 0.3506,
+      "step": 2719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7475363103652135e-05,
+      "loss": 0.3449,
+      "step": 2720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7472211633653604e-05,
+      "loss": 0.3651,
+      "step": 2721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.746905831497543e-05,
+      "loss": 0.3452,
+      "step": 2722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.746590314790948e-05,
+      "loss": 0.3374,
+      "step": 2723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.746274613274782e-05,
+      "loss": 0.3335,
+      "step": 2724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.745958726978266e-05,
+      "loss": 0.3433,
+      "step": 2725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7456426559306385e-05,
+      "loss": 0.3207,
+      "step": 2726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.745326400161159e-05,
+      "loss": 0.3354,
+      "step": 2727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7450099596990975e-05,
+      "loss": 0.3264,
+      "step": 2728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.744693334573747e-05,
+      "loss": 0.3363,
+      "step": 2729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.744376524814413e-05,
+      "loss": 0.3461,
+      "step": 2730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.744059530450422e-05,
+      "loss": 0.3275,
+      "step": 2731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.743742351511115e-05,
+      "loss": 0.3292,
+      "step": 2732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.743424988025852e-05,
+      "loss": 0.3394,
+      "step": 2733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.743107440024007e-05,
+      "loss": 0.3548,
+      "step": 2734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7427897075349746e-05,
+      "loss": 0.3246,
+      "step": 2735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7424717905881645e-05,
+      "loss": 0.3646,
+      "step": 2736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.742153689213004e-05,
+      "loss": 0.3239,
+      "step": 2737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7418354034389376e-05,
+      "loss": 0.3242,
+      "step": 2738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7415169332954254e-05,
+      "loss": 0.3463,
+      "step": 2739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.741198278811947e-05,
+      "loss": 0.3544,
+      "step": 2740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.740879440017999e-05,
+      "loss": 0.3384,
+      "step": 2741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.740560416943091e-05,
+      "loss": 0.3501,
+      "step": 2742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7402412096167545e-05,
+      "loss": 0.3574,
+      "step": 2743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.739921818068536e-05,
+      "loss": 0.3578,
+      "step": 2744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.739602242327999e-05,
+      "loss": 0.3252,
+      "step": 2745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.739282482424724e-05,
+      "loss": 0.3331,
+      "step": 2746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7389625383883104e-05,
+      "loss": 0.3635,
+      "step": 2747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.738642410248372e-05,
+      "loss": 0.343,
+      "step": 2748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.738322098034539e-05,
+      "loss": 0.3566,
+      "step": 2749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.738001601776462e-05,
+      "loss": 0.3388,
+      "step": 2750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.737680921503807e-05,
+      "loss": 0.3495,
+      "step": 2751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7373600572462565e-05,
+      "loss": 0.3368,
+      "step": 2752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.737039009033512e-05,
+      "loss": 0.3407,
+      "step": 2753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7367177768952874e-05,
+      "loss": 0.3424,
+      "step": 2754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7363963608613205e-05,
+      "loss": 0.3514,
+      "step": 2755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.73607476096136e-05,
+      "loss": 0.3407,
+      "step": 2756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7357529772251743e-05,
+      "loss": 0.3474,
+      "step": 2757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.735431009682549e-05,
+      "loss": 0.3355,
+      "step": 2758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.735108858363286e-05,
+      "loss": 0.3342,
+      "step": 2759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.734786523297205e-05,
+      "loss": 0.3356,
+      "step": 2760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.734464004514141e-05,
+      "loss": 0.3405,
+      "step": 2761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.734141302043949e-05,
+      "loss": 0.3384,
+      "step": 2762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7338184159164975e-05,
+      "loss": 0.3313,
+      "step": 2763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.733495346161675e-05,
+      "loss": 0.3413,
+      "step": 2764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7331720928093845e-05,
+      "loss": 0.3318,
+      "step": 2765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7328486558895474e-05,
+      "loss": 0.335,
+      "step": 2766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.732525035432101e-05,
+      "loss": 0.3528,
+      "step": 2767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.732201231467004e-05,
+      "loss": 0.3462,
+      "step": 2768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7318772440242245e-05,
+      "loss": 0.3164,
+      "step": 2769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.731553073133754e-05,
+      "loss": 0.3288,
+      "step": 2770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.731228718825597e-05,
+      "loss": 0.3276,
+      "step": 2771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7309041811297775e-05,
+      "loss": 0.3396,
+      "step": 2772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7305794600763356e-05,
+      "loss": 0.3308,
+      "step": 2773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.730254555695327e-05,
+      "loss": 0.3294,
+      "step": 2774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.729929468016829e-05,
+      "loss": 0.3319,
+      "step": 2775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.729604197070929e-05,
+      "loss": 0.3511,
+      "step": 2776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7292787428877366e-05,
+      "loss": 0.3175,
+      "step": 2777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.728953105497376e-05,
+      "loss": 0.3245,
+      "step": 2778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7286272849299904e-05,
+      "loss": 0.3246,
+      "step": 2779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.728301281215736e-05,
+      "loss": 0.3329,
+      "step": 2780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7279750943847916e-05,
+      "loss": 0.3311,
+      "step": 2781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.727648724467348e-05,
+      "loss": 0.3608,
+      "step": 2782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.727322171493616e-05,
+      "loss": 0.3422,
+      "step": 2783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.72699543549382e-05,
+      "loss": 0.3222,
+      "step": 2784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7266685164982053e-05,
+      "loss": 0.3391,
+      "step": 2785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.726341414537033e-05,
+      "loss": 0.3395,
+      "step": 2786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7260141296405794e-05,
+      "loss": 0.3348,
+      "step": 2787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.725686661839139e-05,
+      "loss": 0.3497,
+      "step": 2788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.725359011163023e-05,
+      "loss": 0.3551,
+      "step": 2789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7250311776425596e-05,
+      "loss": 0.3386,
+      "step": 2790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.724703161308094e-05,
+      "loss": 0.3595,
+      "step": 2791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.724374962189989e-05,
+      "loss": 0.3609,
+      "step": 2792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.724046580318622e-05,
+      "loss": 0.3456,
+      "step": 2793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7237180157243895e-05,
+      "loss": 0.3366,
+      "step": 2794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7233892684377055e-05,
+      "loss": 0.3432,
+      "step": 2795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.723060338488997e-05,
+      "loss": 0.3697,
+      "step": 2796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.722731225908714e-05,
+      "loss": 0.3359,
+      "step": 2797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.722401930727317e-05,
+      "loss": 0.3305,
+      "step": 2798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.722072452975289e-05,
+      "loss": 0.3423,
+      "step": 2799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.721742792683125e-05,
+      "loss": 0.3443,
+      "step": 2800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.721412949881341e-05,
+      "loss": 0.333,
+      "step": 2801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.721082924600467e-05,
+      "loss": 0.329,
+      "step": 2802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.720752716871052e-05,
+      "loss": 0.3503,
+      "step": 2803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.720422326723659e-05,
+      "loss": 0.3595,
+      "step": 2804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.720091754188872e-05,
+      "loss": 0.3363,
+      "step": 2805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.719760999297288e-05,
+      "loss": 0.3296,
+      "step": 2806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7194300620795234e-05,
+      "loss": 0.3323,
+      "step": 2807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.719098942566211e-05,
+      "loss": 0.3402,
+      "step": 2808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.718767640787999e-05,
+      "loss": 0.355,
+      "step": 2809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7184361567755534e-05,
+      "loss": 0.3313,
+      "step": 2810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.718104490559559e-05,
+      "loss": 0.3392,
+      "step": 2811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7177726421707135e-05,
+      "loss": 0.3548,
+      "step": 2812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.717440611639736e-05,
+      "loss": 0.3388,
+      "step": 2813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.717108398997357e-05,
+      "loss": 0.3499,
+      "step": 2814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.71677600427433e-05,
+      "loss": 0.3454,
+      "step": 2815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.716443427501421e-05,
+      "loss": 0.3537,
+      "step": 2816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7161106687094134e-05,
+      "loss": 0.3214,
+      "step": 2817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.715777727929109e-05,
+      "loss": 0.3164,
+      "step": 2818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.715444605191327e-05,
+      "loss": 0.3451,
+      "step": 2819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.715111300526899e-05,
+      "loss": 0.3158,
+      "step": 2820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.71477781396668e-05,
+      "loss": 0.3177,
+      "step": 2821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.714444145541536e-05,
+      "loss": 0.3249,
+      "step": 2822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7141102952823534e-05,
+      "loss": 0.3474,
+      "step": 2823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.713776263220033e-05,
+      "loss": 0.355,
+      "step": 2824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.713442049385495e-05,
+      "loss": 0.3462,
+      "step": 2825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.713107653809675e-05,
+      "loss": 0.3173,
+      "step": 2826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.712773076523524e-05,
+      "loss": 0.3633,
+      "step": 2827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.712438317558013e-05,
+      "loss": 0.347,
+      "step": 2828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.712103376944128e-05,
+      "loss": 0.3596,
+      "step": 2829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.71176825471287e-05,
+      "loss": 0.3375,
+      "step": 2830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7114329508952624e-05,
+      "loss": 0.3592,
+      "step": 2831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7110974655223386e-05,
+      "loss": 0.3313,
+      "step": 2832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.710761798625153e-05,
+      "loss": 0.3508,
+      "step": 2833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.710425950234777e-05,
+      "loss": 0.3376,
+      "step": 2834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.710089920382295e-05,
+      "loss": 0.3423,
+      "step": 2835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7097537090988126e-05,
+      "loss": 0.3402,
+      "step": 2836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7094173164154504e-05,
+      "loss": 0.3379,
+      "step": 2837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.709080742363345e-05,
+      "loss": 0.4023,
+      "step": 2838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.708743986973652e-05,
+      "loss": 0.3578,
+      "step": 2839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.708407050277541e-05,
+      "loss": 0.3367,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7080699323062e-05,
+      "loss": 0.3329,
+      "step": 2841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.707732633090834e-05,
+      "loss": 0.3438,
+      "step": 2842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.707395152662663e-05,
+      "loss": 0.3529,
+      "step": 2843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7070574910529266e-05,
+      "loss": 0.342,
+      "step": 2844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.706719648292879e-05,
+      "loss": 0.3556,
+      "step": 2845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.706381624413792e-05,
+      "loss": 0.3355,
+      "step": 2846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.706043419446953e-05,
+      "loss": 0.3279,
+      "step": 2847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.70570503342367e-05,
+      "loss": 0.3459,
+      "step": 2848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.70536646637526e-05,
+      "loss": 0.3314,
+      "step": 2849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.705027718333066e-05,
+      "loss": 0.3318,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.704688789328442e-05,
+      "loss": 0.3339,
+      "step": 2851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.704349679392759e-05,
+      "loss": 0.3377,
+      "step": 2852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.704010388557408e-05,
+      "loss": 0.3439,
+      "step": 2853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.703670916853792e-05,
+      "loss": 0.34,
+      "step": 2854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.703331264313336e-05,
+      "loss": 0.3471,
+      "step": 2855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7029914309674785e-05,
+      "loss": 0.333,
+      "step": 2856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7026514168476736e-05,
+      "loss": 0.3471,
+      "step": 2857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.702311221985397e-05,
+      "loss": 0.3305,
+      "step": 2858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7019708464121344e-05,
+      "loss": 0.3292,
+      "step": 2859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7016302901593935e-05,
+      "loss": 0.3268,
+      "step": 2860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.701289553258698e-05,
+      "loss": 0.3433,
+      "step": 2861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.700948635741586e-05,
+      "loss": 0.3335,
+      "step": 2862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7006075376396155e-05,
+      "loss": 0.3392,
+      "step": 2863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.700266258984357e-05,
+      "loss": 0.3411,
+      "step": 2864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6999247998074023e-05,
+      "loss": 0.3344,
+      "step": 2865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6995831601403563e-05,
+      "loss": 0.3406,
+      "step": 2866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6992413400148423e-05,
+      "loss": 0.3339,
+      "step": 2867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6988993394625e-05,
+      "loss": 0.3382,
+      "step": 2868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6985571585149876e-05,
+      "loss": 0.3359,
+      "step": 2869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.698214797203976e-05,
+      "loss": 0.3276,
+      "step": 2870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.697872255561155e-05,
+      "loss": 0.3524,
+      "step": 2871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.697529533618233e-05,
+      "loss": 0.3142,
+      "step": 2872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6971866314069324e-05,
+      "loss": 0.3291,
+      "step": 2873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.696843548958992e-05,
+      "loss": 0.3353,
+      "step": 2874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6965002863061705e-05,
+      "loss": 0.349,
+      "step": 2875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.696156843480239e-05,
+      "loss": 0.3363,
+      "step": 2876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.695813220512989e-05,
+      "loss": 0.3356,
+      "step": 2877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6954694174362264e-05,
+      "loss": 0.324,
+      "step": 2878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.695125434281775e-05,
+      "loss": 0.3395,
+      "step": 2879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.694781271081474e-05,
+      "loss": 0.334,
+      "step": 2880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.694436927867182e-05,
+      "loss": 0.3402,
+      "step": 2881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.694092404670769e-05,
+      "loss": 0.3333,
+      "step": 2882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.693747701524128e-05,
+      "loss": 0.3337,
+      "step": 2883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.693402818459164e-05,
+      "loss": 0.3507,
+      "step": 2884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.693057755507801e-05,
+      "loss": 0.3312,
+      "step": 2885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6927125127019775e-05,
+      "loss": 0.3319,
+      "step": 2886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.692367090073652e-05,
+      "loss": 0.3546,
+      "step": 2887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6920214876547976e-05,
+      "loss": 0.3469,
+      "step": 2888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.691675705477402e-05,
+      "loss": 0.3394,
+      "step": 2889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6913297435734736e-05,
+      "loss": 0.3412,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.690983601975035e-05,
+      "loss": 0.3294,
+      "step": 2891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6906372807141265e-05,
+      "loss": 0.3423,
+      "step": 2892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.690290779822804e-05,
+      "loss": 0.3313,
+      "step": 2893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6899440993331404e-05,
+      "loss": 0.3279,
+      "step": 2894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.689597239277226e-05,
+      "loss": 0.3516,
+      "step": 2895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6892501996871654e-05,
+      "loss": 0.3465,
+      "step": 2896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.688902980595084e-05,
+      "loss": 0.3566,
+      "step": 2897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.688555582033118e-05,
+      "loss": 0.3391,
+      "step": 2898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.688208004033427e-05,
+      "loss": 0.3319,
+      "step": 2899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.687860246628182e-05,
+      "loss": 0.3296,
+      "step": 2900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.687512309849572e-05,
+      "loss": 0.3292,
+      "step": 2901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6871641937298036e-05,
+      "loss": 0.3395,
+      "step": 2902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.686815898301099e-05,
+      "loss": 0.3465,
+      "step": 2903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.686467423595698e-05,
+      "loss": 0.334,
+      "step": 2904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.686118769645856e-05,
+      "loss": 0.3281,
+      "step": 2905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.685769936483845e-05,
+      "loss": 0.3307,
+      "step": 2906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.685420924141953e-05,
+      "loss": 0.3359,
+      "step": 2907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.685071732652487e-05,
+      "loss": 0.3625,
+      "step": 2908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6847223620477686e-05,
+      "loss": 0.3202,
+      "step": 2909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.684372812360137e-05,
+      "loss": 0.3252,
+      "step": 2910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.684023083621946e-05,
+      "loss": 0.3371,
+      "step": 2911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.683673175865569e-05,
+      "loss": 0.3276,
+      "step": 2912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.683323089123393e-05,
+      "loss": 0.3389,
+      "step": 2913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6829728234278234e-05,
+      "loss": 0.3404,
+      "step": 2914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.682622378811282e-05,
+      "loss": 0.346,
+      "step": 2915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.682271755306207e-05,
+      "loss": 0.3177,
+      "step": 2916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6819209529450526e-05,
+      "loss": 0.348,
+      "step": 2917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.68156997176029e-05,
+      "loss": 0.3333,
+      "step": 2918
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6812188117844075e-05,
+      "loss": 0.3232,
+      "step": 2919
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.680867473049908e-05,
+      "loss": 0.3337,
+      "step": 2920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.680515955589314e-05,
+      "loss": 0.3551,
+      "step": 2921
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.680164259435162e-05,
+      "loss": 0.3449,
+      "step": 2922
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.679812384620006e-05,
+      "loss": 0.3272,
+      "step": 2923
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.679460331176416e-05,
+      "loss": 0.3411,
+      "step": 2924
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.679108099136979e-05,
+      "loss": 0.3207,
+      "step": 2925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6787556885343006e-05,
+      "loss": 0.3294,
+      "step": 2926
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.678403099400999e-05,
+      "loss": 0.3668,
+      "step": 2927
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.678050331769709e-05,
+      "loss": 0.3342,
+      "step": 2928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6776973856730874e-05,
+      "loss": 0.335,
+      "step": 2929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.677344261143802e-05,
+      "loss": 0.3641,
+      "step": 2930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6769909582145374e-05,
+      "loss": 0.339,
+      "step": 2931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.676637476917999e-05,
+      "loss": 0.3703,
+      "step": 2932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6762838172869046e-05,
+      "loss": 0.3458,
+      "step": 2933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.67592997935399e-05,
+      "loss": 0.3239,
+      "step": 2934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6755759631520076e-05,
+      "loss": 0.3541,
+      "step": 2935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6752217687137246e-05,
+      "loss": 0.3658,
+      "step": 2936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6748673960719285e-05,
+      "loss": 0.326,
+      "step": 2937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.67451284525942e-05,
+      "loss": 0.3419,
+      "step": 2938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.674158116309016e-05,
+      "loss": 0.3303,
+      "step": 2939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6738032092535523e-05,
+      "loss": 0.3334,
+      "step": 2940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.67344812412588e-05,
+      "loss": 0.3575,
+      "step": 2941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.673092860958867e-05,
+      "loss": 0.3483,
+      "step": 2942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6727374197853965e-05,
+      "loss": 0.3367,
+      "step": 2943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.67238180063837e-05,
+      "loss": 0.3227,
+      "step": 2944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.672026003550704e-05,
+      "loss": 0.3315,
+      "step": 2945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.671670028555332e-05,
+      "loss": 0.3391,
+      "step": 2946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6713138756852035e-05,
+      "loss": 0.33,
+      "step": 2947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6709575449732864e-05,
+      "loss": 0.3312,
+      "step": 2948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.670601036452561e-05,
+      "loss": 0.3357,
+      "step": 2949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.67024435015603e-05,
+      "loss": 0.3238,
+      "step": 2950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.669887486116708e-05,
+      "loss": 0.33,
+      "step": 2951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.669530444367625e-05,
+      "loss": 0.3566,
+      "step": 2952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.669173224941832e-05,
+      "loss": 0.342,
+      "step": 2953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.668815827872394e-05,
+      "loss": 0.34,
+      "step": 2954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.668458253192391e-05,
+      "loss": 0.3392,
+      "step": 2955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.668100500934924e-05,
+      "loss": 0.343,
+      "step": 2956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.667742571133105e-05,
+      "loss": 0.321,
+      "step": 2957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.667384463820065e-05,
+      "loss": 0.3238,
+      "step": 2958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.667026179028953e-05,
+      "loss": 0.3212,
+      "step": 2959
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.666667716792931e-05,
+      "loss": 0.3384,
+      "step": 2960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.666309077145181e-05,
+      "loss": 0.3361,
+      "step": 2961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6659502601188975e-05,
+      "loss": 0.3476,
+      "step": 2962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6655912657472945e-05,
+      "loss": 0.3347,
+      "step": 2963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.665232094063603e-05,
+      "loss": 0.3416,
+      "step": 2964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.664872745101067e-05,
+      "loss": 0.3271,
+      "step": 2965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.664513218892948e-05,
+      "loss": 0.3397,
+      "step": 2966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6641535154725265e-05,
+      "loss": 0.3478,
+      "step": 2967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.663793634873097e-05,
+      "loss": 0.341,
+      "step": 2968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6634335771279705e-05,
+      "loss": 0.3141,
+      "step": 2969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6630733422704756e-05,
+      "loss": 0.3384,
+      "step": 2970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.662712930333956e-05,
+      "loss": 0.3399,
+      "step": 2971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6623523413517725e-05,
+      "loss": 0.3252,
+      "step": 2972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.661991575357304e-05,
+      "loss": 0.3461,
+      "step": 2973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.66163063238394e-05,
+      "loss": 0.3314,
+      "step": 2974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.661269512465094e-05,
+      "loss": 0.3404,
+      "step": 2975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.66090821563419e-05,
+      "loss": 0.3381,
+      "step": 2976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.660546741924672e-05,
+      "loss": 0.3444,
+      "step": 2977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.660185091369997e-05,
+      "loss": 0.3401,
+      "step": 2978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.659823264003642e-05,
+      "loss": 0.3358,
+      "step": 2979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.659461259859099e-05,
+      "loss": 0.3442,
+      "step": 2980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.659099078969875e-05,
+      "loss": 0.3386,
+      "step": 2981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.658736721369495e-05,
+      "loss": 0.3378,
+      "step": 2982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.658374187091498e-05,
+      "loss": 0.3292,
+      "step": 2983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6580114761694435e-05,
+      "loss": 0.3405,
+      "step": 2984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6576485886369045e-05,
+      "loss": 0.3482,
+      "step": 2985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.65728552452747e-05,
+      "loss": 0.318,
+      "step": 2986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6569222838747455e-05,
+      "loss": 0.336,
+      "step": 2987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6565588667123557e-05,
+      "loss": 0.3516,
+      "step": 2988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.656195273073938e-05,
+      "loss": 0.3397,
+      "step": 2989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6558315029931467e-05,
+      "loss": 0.3343,
+      "step": 2990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.655467556503655e-05,
+      "loss": 0.327,
+      "step": 2991
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.655103433639151e-05,
+      "loss": 0.3396,
+      "step": 2992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.654739134433338e-05,
+      "loss": 0.3423,
+      "step": 2993
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.654374658919935e-05,
+      "loss": 0.325,
+      "step": 2994
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.65401000713268e-05,
+      "loss": 0.3547,
+      "step": 2995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.653645179105328e-05,
+      "loss": 0.3293,
+      "step": 2996
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6532801748716456e-05,
+      "loss": 0.3325,
+      "step": 2997
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.65291499446542e-05,
+      "loss": 0.3377,
+      "step": 2998
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.652549637920453e-05,
+      "loss": 0.3245,
+      "step": 2999
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6521841052705634e-05,
+      "loss": 0.3343,
+      "step": 3000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.355400696864112e-08,
+      "loss": 0.3482,
+      "step": 3001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.710801393728224e-08,
+      "loss": 0.3206,
+      "step": 3002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3066202090592336e-07,
+      "loss": 0.3134,
+      "step": 3003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7421602787456448e-07,
+      "loss": 0.3305,
+      "step": 3004
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.1777003484320556e-07,
+      "loss": 0.3503,
+      "step": 3005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6132404181184673e-07,
+      "loss": 0.3406,
+      "step": 3006
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.0487804878048784e-07,
+      "loss": 0.346,
+      "step": 3007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4843205574912896e-07,
+      "loss": 0.3519,
+      "step": 3008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9198606271777007e-07,
+      "loss": 0.3293,
+      "step": 3009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.3554006968641113e-07,
+      "loss": 0.343,
+      "step": 3010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.790940766550523e-07,
+      "loss": 0.314,
+      "step": 3011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.226480836236935e-07,
+      "loss": 0.3114,
+      "step": 3012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.662020905923345e-07,
+      "loss": 0.3273,
+      "step": 3013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.097560975609757e-07,
+      "loss": 0.3297,
+      "step": 3014
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.533101045296167e-07,
+      "loss": 0.333,
+      "step": 3015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.968641114982579e-07,
+      "loss": 0.3603,
+      "step": 3016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.40418118466899e-07,
+      "loss": 0.3446,
+      "step": 3017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.839721254355401e-07,
+      "loss": 0.3285,
+      "step": 3018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.275261324041812e-07,
+      "loss": 0.3535,
+      "step": 3019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.710801393728223e-07,
+      "loss": 0.3451,
+      "step": 3020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.146341463414634e-07,
+      "loss": 0.3374,
+      "step": 3021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581881533101046e-07,
+      "loss": 0.3278,
+      "step": 3022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.0017421602787455e-06,
+      "loss": 0.3545,
+      "step": 3023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.045296167247387e-06,
+      "loss": 0.3391,
+      "step": 3024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.088850174216028e-06,
+      "loss": 0.3557,
+      "step": 3025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.132404181184669e-06,
+      "loss": 0.3419,
+      "step": 3026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1759581881533102e-06,
+      "loss": 0.3361,
+      "step": 3027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 0.3357,
+      "step": 3028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.2630662020905923e-06,
+      "loss": 0.344,
+      "step": 3029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.3066202090592335e-06,
+      "loss": 0.3295,
+      "step": 3030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.3501742160278747e-06,
+      "loss": 0.3276,
+      "step": 3031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.3937282229965158e-06,
+      "loss": 0.3326,
+      "step": 3032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4372822299651568e-06,
+      "loss": 0.3198,
+      "step": 3033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.480836236933798e-06,
+      "loss": 0.3484,
+      "step": 3034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5243902439024391e-06,
+      "loss": 0.3472,
+      "step": 3035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.5679442508710803e-06,
+      "loss": 0.315,
+      "step": 3036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6114982578397212e-06,
+      "loss": 0.3438,
+      "step": 3037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6550522648083624e-06,
+      "loss": 0.3287,
+      "step": 3038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6986062717770036e-06,
+      "loss": 0.3359,
+      "step": 3039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7421602787456445e-06,
+      "loss": 0.3433,
+      "step": 3040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7857142857142857e-06,
+      "loss": 0.3344,
+      "step": 3041
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8292682926829268e-06,
+      "loss": 0.3575,
+      "step": 3042
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.872822299651568e-06,
+      "loss": 0.3393,
+      "step": 3043
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.916376306620209e-06,
+      "loss": 0.3388,
+      "step": 3044
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.95993031358885e-06,
+      "loss": 0.3324,
+      "step": 3045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.003484320557491e-06,
+      "loss": 0.3333,
+      "step": 3046
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.0470383275261322e-06,
+      "loss": 0.3334,
+      "step": 3047
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.090592334494774e-06,
+      "loss": 0.3324,
+      "step": 3048
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.134146341463415e-06,
+      "loss": 0.333,
+      "step": 3049
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.177700348432056e-06,
+      "loss": 0.3365,
+      "step": 3050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.221254355400697e-06,
+      "loss": 0.3233,
+      "step": 3051
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.264808362369338e-06,
+      "loss": 0.3486,
+      "step": 3052
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3083623693379792e-06,
+      "loss": 0.3422,
+      "step": 3053
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3519163763066204e-06,
+      "loss": 0.3179,
+      "step": 3054
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3954703832752616e-06,
+      "loss": 0.3233,
+      "step": 3055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 0.3286,
+      "step": 3056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.482578397212544e-06,
+      "loss": 0.3348,
+      "step": 3057
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.5261324041811846e-06,
+      "loss": 0.3525,
+      "step": 3058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.569686411149826e-06,
+      "loss": 0.3599,
+      "step": 3059
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.613240418118467e-06,
+      "loss": 0.3275,
+      "step": 3060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.656794425087108e-06,
+      "loss": 0.3543,
+      "step": 3061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7003484320557493e-06,
+      "loss": 0.3278,
+      "step": 3062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7439024390243905e-06,
+      "loss": 0.3235,
+      "step": 3063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7874564459930316e-06,
+      "loss": 0.3444,
+      "step": 3064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8310104529616724e-06,
+      "loss": 0.3507,
+      "step": 3065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.8745644599303136e-06,
+      "loss": 0.3407,
+      "step": 3066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9181184668989547e-06,
+      "loss": 0.3265,
+      "step": 3067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.961672473867596e-06,
+      "loss": 0.3293,
+      "step": 3068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.005226480836237e-06,
+      "loss": 0.3341,
+      "step": 3069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.0487804878048782e-06,
+      "loss": 0.3526,
+      "step": 3070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.0923344947735194e-06,
+      "loss": 0.3339,
+      "step": 3071
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.1358885017421605e-06,
+      "loss": 0.33,
+      "step": 3072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.1794425087108017e-06,
+      "loss": 0.3097,
+      "step": 3073
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.2229965156794425e-06,
+      "loss": 0.3433,
+      "step": 3074
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.266550522648084e-06,
+      "loss": 0.3263,
+      "step": 3075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.3101045296167248e-06,
+      "loss": 0.3345,
+      "step": 3076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.3536585365853664e-06,
+      "loss": 0.3387,
+      "step": 3077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.397212543554007e-06,
+      "loss": 0.3344,
+      "step": 3078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4407665505226483e-06,
+      "loss": 0.3512,
+      "step": 3079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.484320557491289e-06,
+      "loss": 0.339,
+      "step": 3080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.5278745644599306e-06,
+      "loss": 0.3503,
+      "step": 3081
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 0.3263,
+      "step": 3082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.614982578397213e-06,
+      "loss": 0.3427,
+      "step": 3083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 0.3446,
+      "step": 3084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.7020905923344953e-06,
+      "loss": 0.3361,
+      "step": 3085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.745644599303136e-06,
+      "loss": 0.3276,
+      "step": 3086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.789198606271777e-06,
+      "loss": 0.3107,
+      "step": 3087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.832752613240418e-06,
+      "loss": 0.3279,
+      "step": 3088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.876306620209059e-06,
+      "loss": 0.3253,
+      "step": 3089
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9198606271777e-06,
+      "loss": 0.3357,
+      "step": 3090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9634146341463414e-06,
+      "loss": 0.3067,
+      "step": 3091
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.006968641114982e-06,
+      "loss": 0.3216,
+      "step": 3092
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.050522648083624e-06,
+      "loss": 0.3486,
+      "step": 3093
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0940766550522645e-06,
+      "loss": 0.3168,
+      "step": 3094
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.137630662020906e-06,
+      "loss": 0.3319,
+      "step": 3095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.181184668989548e-06,
+      "loss": 0.3208,
+      "step": 3096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.224738675958188e-06,
+      "loss": 0.324,
+      "step": 3097
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.26829268292683e-06,
+      "loss": 0.3433,
+      "step": 3098
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.311846689895471e-06,
+      "loss": 0.3499,
+      "step": 3099
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.355400696864112e-06,
+      "loss": 0.321,
+      "step": 3100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.398954703832753e-06,
+      "loss": 0.3398,
+      "step": 3101
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.442508710801394e-06,
+      "loss": 0.3376,
+      "step": 3102
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4860627177700346e-06,
+      "loss": 0.3398,
+      "step": 3103
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.529616724738676e-06,
+      "loss": 0.4027,
+      "step": 3104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.573170731707317e-06,
+      "loss": 0.3361,
+      "step": 3105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6167247386759585e-06,
+      "loss": 0.3417,
+      "step": 3106
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660278745644599e-06,
+      "loss": 0.3356,
+      "step": 3107
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.703832752613241e-06,
+      "loss": 0.3274,
+      "step": 3108
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7473867595818816e-06,
+      "loss": 0.3358,
+      "step": 3109
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.790940766550523e-06,
+      "loss": 0.3297,
+      "step": 3110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.834494773519164e-06,
+      "loss": 0.3483,
+      "step": 3111
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.3427,
+      "step": 3112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921602787456446e-06,
+      "loss": 0.3472,
+      "step": 3113
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.965156794425088e-06,
+      "loss": 0.3418,
+      "step": 3114
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.0087108013937286e-06,
+      "loss": 0.3396,
+      "step": 3115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.052264808362369e-06,
+      "loss": 0.3388,
+      "step": 3116
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.09581881533101e-06,
+      "loss": 0.3394,
+      "step": 3117
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.139372822299652e-06,
+      "loss": 0.3431,
+      "step": 3118
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.182926829268292e-06,
+      "loss": 0.3549,
+      "step": 3119
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.226480836236934e-06,
+      "loss": 0.3301,
+      "step": 3120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.270034843205575e-06,
+      "loss": 0.3262,
+      "step": 3121
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.313588850174216e-06,
+      "loss": 0.3402,
+      "step": 3122
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.3386,
+      "step": 3123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.400696864111499e-06,
+      "loss": 0.3337,
+      "step": 3124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.444250871080139e-06,
+      "loss": 0.3203,
+      "step": 3125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.487804878048781e-06,
+      "loss": 0.3464,
+      "step": 3126
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.531358885017422e-06,
+      "loss": 0.3413,
+      "step": 3127
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.574912891986063e-06,
+      "loss": 0.3264,
+      "step": 3128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.618466898954704e-06,
+      "loss": 0.3503,
+      "step": 3129
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.662020905923345e-06,
+      "loss": 0.3441,
+      "step": 3130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.705574912891986e-06,
+      "loss": 0.3387,
+      "step": 3131
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.749128919860627e-06,
+      "loss": 0.3395,
+      "step": 3132
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.792682926829269e-06,
+      "loss": 0.3386,
+      "step": 3133
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.8362369337979094e-06,
+      "loss": 0.3417,
+      "step": 3134
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.879790940766551e-06,
+      "loss": 0.3232,
+      "step": 3135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.923344947735192e-06,
+      "loss": 0.359,
+      "step": 3136
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.966898954703833e-06,
+      "loss": 0.3326,
+      "step": 3137
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.010452961672474e-06,
+      "loss": 0.3455,
+      "step": 3138
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.054006968641116e-06,
+      "loss": 0.3475,
+      "step": 3139
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 0.3295,
+      "step": 3140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.141114982578398e-06,
+      "loss": 0.3357,
+      "step": 3141
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.184668989547039e-06,
+      "loss": 0.3178,
+      "step": 3142
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.2282229965156795e-06,
+      "loss": 0.333,
+      "step": 3143
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.271777003484321e-06,
+      "loss": 0.3315,
+      "step": 3144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.315331010452962e-06,
+      "loss": 0.345,
+      "step": 3145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.3588850174216034e-06,
+      "loss": 0.3449,
+      "step": 3146
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.402439024390244e-06,
+      "loss": 0.3315,
+      "step": 3147
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.445993031358885e-06,
+      "loss": 0.3143,
+      "step": 3148
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.489547038327526e-06,
+      "loss": 0.326,
+      "step": 3149
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.533101045296168e-06,
+      "loss": 0.3389,
+      "step": 3150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.576655052264809e-06,
+      "loss": 0.3285,
+      "step": 3151
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.6202090592334496e-06,
+      "loss": 0.348,
+      "step": 3152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.66376306620209e-06,
+      "loss": 0.3374,
+      "step": 3153
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.707317073170733e-06,
+      "loss": 0.3348,
+      "step": 3154
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.7508710801393735e-06,
+      "loss": 0.3621,
+      "step": 3155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.794425087108014e-06,
+      "loss": 0.3052,
+      "step": 3156
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.837979094076655e-06,
+      "loss": 0.3153,
+      "step": 3157
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.8815331010452966e-06,
+      "loss": 0.341,
+      "step": 3158
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.925087108013937e-06,
+      "loss": 0.3433,
+      "step": 3159
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.968641114982578e-06,
+      "loss": 0.3334,
+      "step": 3160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.0121951219512205e-06,
+      "loss": 0.3335,
+      "step": 3161
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.055749128919861e-06,
+      "loss": 0.3305,
+      "step": 3162
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.099303135888502e-06,
+      "loss": 0.3469,
+      "step": 3163
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.3285,
+      "step": 3164
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.186411149825785e-06,
+      "loss": 0.3374,
+      "step": 3165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.229965156794426e-06,
+      "loss": 0.3176,
+      "step": 3166
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.273519163763067e-06,
+      "loss": 0.342,
+      "step": 3167
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 0.3239,
+      "step": 3168
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.360627177700349e-06,
+      "loss": 0.3335,
+      "step": 3169
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.4041811846689906e-06,
+      "loss": 0.3223,
+      "step": 3170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.447735191637631e-06,
+      "loss": 0.3347,
+      "step": 3171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.491289198606272e-06,
+      "loss": 0.3521,
+      "step": 3172
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.534843205574914e-06,
+      "loss": 0.3261,
+      "step": 3173
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.578397212543554e-06,
+      "loss": 0.3369,
+      "step": 3174
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.621951219512195e-06,
+      "loss": 0.3274,
+      "step": 3175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.665505226480837e-06,
+      "loss": 0.3396,
+      "step": 3176
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.709059233449478e-06,
+      "loss": 0.3462,
+      "step": 3177
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.752613240418118e-06,
+      "loss": 0.3328,
+      "step": 3178
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.79616724738676e-06,
+      "loss": 0.324,
+      "step": 3179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.8397212543554e-06,
+      "loss": 0.3076,
+      "step": 3180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.883275261324043e-06,
+      "loss": 0.3355,
+      "step": 3181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.926829268292683e-06,
+      "loss": 0.341,
+      "step": 3182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.970383275261324e-06,
+      "loss": 0.316,
+      "step": 3183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.013937282229964e-06,
+      "loss": 0.3534,
+      "step": 3184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.057491289198608e-06,
+      "loss": 0.3446,
+      "step": 3185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.101045296167248e-06,
+      "loss": 0.3335,
+      "step": 3186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.144599303135889e-06,
+      "loss": 0.3297,
+      "step": 3187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.188153310104529e-06,
+      "loss": 0.3323,
+      "step": 3188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.231707317073172e-06,
+      "loss": 0.343,
+      "step": 3189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.275261324041812e-06,
+      "loss": 0.3326,
+      "step": 3190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.318815331010454e-06,
+      "loss": 0.3442,
+      "step": 3191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.362369337979095e-06,
+      "loss": 0.3533,
+      "step": 3192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.405923344947735e-06,
+      "loss": 0.3378,
+      "step": 3193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449477351916377e-06,
+      "loss": 0.3347,
+      "step": 3194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.493031358885017e-06,
+      "loss": 0.3315,
+      "step": 3195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53658536585366e-06,
+      "loss": 0.3523,
+      "step": 3196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5801393728223e-06,
+      "loss": 0.3161,
+      "step": 3197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623693379790942e-06,
+      "loss": 0.3362,
+      "step": 3198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667247386759581e-06,
+      "loss": 0.3262,
+      "step": 3199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.710801393728225e-06,
+      "loss": 0.3335,
+      "step": 3200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.754355400696865e-06,
+      "loss": 0.3435,
+      "step": 3201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.797909407665506e-06,
+      "loss": 0.3523,
+      "step": 3202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.841463414634146e-06,
+      "loss": 0.339,
+      "step": 3203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.885017421602788e-06,
+      "loss": 0.3413,
+      "step": 3204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.3453,
+      "step": 3205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.972125435540069e-06,
+      "loss": 0.3308,
+      "step": 3206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.01567944250871e-06,
+      "loss": 0.3542,
+      "step": 3207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.059233449477352e-06,
+      "loss": 0.3294,
+      "step": 3208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.102787456445994e-06,
+      "loss": 0.3305,
+      "step": 3209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.146341463414634e-06,
+      "loss": 0.3416,
+      "step": 3210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.189895470383275e-06,
+      "loss": 0.3062,
+      "step": 3211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.233449477351917e-06,
+      "loss": 0.4035,
+      "step": 3212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.277003484320559e-06,
+      "loss": 0.3559,
+      "step": 3213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.320557491289198e-06,
+      "loss": 0.3359,
+      "step": 3214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.36411149825784e-06,
+      "loss": 0.3402,
+      "step": 3215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.407665505226482e-06,
+      "loss": 0.3513,
+      "step": 3216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.451219512195123e-06,
+      "loss": 0.3345,
+      "step": 3217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.494773519163763e-06,
+      "loss": 0.3237,
+      "step": 3218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.538327526132405e-06,
+      "loss": 0.3338,
+      "step": 3219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.581881533101046e-06,
+      "loss": 0.3442,
+      "step": 3220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.625435540069686e-06,
+      "loss": 0.3285,
+      "step": 3221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.668989547038328e-06,
+      "loss": 0.3255,
+      "step": 3222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.712543554006968e-06,
+      "loss": 0.3405,
+      "step": 3223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.3451,
+      "step": 3224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.799651567944251e-06,
+      "loss": 0.3398,
+      "step": 3225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.843205574912892e-06,
+      "loss": 0.3193,
+      "step": 3226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.886759581881534e-06,
+      "loss": 0.3278,
+      "step": 3227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.930313588850176e-06,
+      "loss": 0.3493,
+      "step": 3228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.973867595818816e-06,
+      "loss": 0.3235,
+      "step": 3229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0017421602787457e-05,
+      "loss": 0.3158,
+      "step": 3230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0060975609756099e-05,
+      "loss": 0.3274,
+      "step": 3231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0104529616724739e-05,
+      "loss": 0.3378,
+      "step": 3232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.014808362369338e-05,
+      "loss": 0.3328,
+      "step": 3233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.019163763066202e-05,
+      "loss": 0.3593,
+      "step": 3234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0235191637630663e-05,
+      "loss": 0.3429,
+      "step": 3235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0278745644599303e-05,
+      "loss": 0.3361,
+      "step": 3236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0322299651567945e-05,
+      "loss": 0.3404,
+      "step": 3237
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0365853658536585e-05,
+      "loss": 0.3201,
+      "step": 3238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0409407665505228e-05,
+      "loss": 0.3283,
+      "step": 3239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0452961672473868e-05,
+      "loss": 0.3261,
+      "step": 3240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.049651567944251e-05,
+      "loss": 0.3283,
+      "step": 3241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.054006968641115e-05,
+      "loss": 0.3275,
+      "step": 3242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0583623693379793e-05,
+      "loss": 0.3401,
+      "step": 3243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0627177700348433e-05,
+      "loss": 0.3388,
+      "step": 3244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0670731707317074e-05,
+      "loss": 0.3198,
+      "step": 3245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.3513,
+      "step": 3246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0757839721254356e-05,
+      "loss": 0.3207,
+      "step": 3247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0801393728222997e-05,
+      "loss": 0.3351,
+      "step": 3248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0844947735191637e-05,
+      "loss": 0.3377,
+      "step": 3249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0888501742160279e-05,
+      "loss": 0.3266,
+      "step": 3250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.093205574912892e-05,
+      "loss": 0.3363,
+      "step": 3251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0975609756097562e-05,
+      "loss": 0.3279,
+      "step": 3252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1019163763066202e-05,
+      "loss": 0.3278,
+      "step": 3253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1062717770034843e-05,
+      "loss": 0.3352,
+      "step": 3254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1106271777003485e-05,
+      "loss": 0.343,
+      "step": 3255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1149825783972127e-05,
+      "loss": 0.347,
+      "step": 3256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1193379790940766e-05,
+      "loss": 0.3755,
+      "step": 3257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1236933797909408e-05,
+      "loss": 0.3432,
+      "step": 3258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.128048780487805e-05,
+      "loss": 0.3321,
+      "step": 3259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.132404181184669e-05,
+      "loss": 0.3288,
+      "step": 3260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1367595818815331e-05,
+      "loss": 0.3633,
+      "step": 3261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1411149825783973e-05,
+      "loss": 0.3281,
+      "step": 3262
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1454703832752614e-05,
+      "loss": 0.3545,
+      "step": 3263
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1498257839721254e-05,
+      "loss": 0.3378,
+      "step": 3264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1541811846689896e-05,
+      "loss": 0.3356,
+      "step": 3265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1585365853658537e-05,
+      "loss": 0.3373,
+      "step": 3266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1628919860627179e-05,
+      "loss": 0.3311,
+      "step": 3267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1672473867595819e-05,
+      "loss": 0.3284,
+      "step": 3268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.171602787456446e-05,
+      "loss": 0.3396,
+      "step": 3269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1759581881533102e-05,
+      "loss": 0.3353,
+      "step": 3270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1803135888501744e-05,
+      "loss": 0.3368,
+      "step": 3271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1846689895470384e-05,
+      "loss": 0.3305,
+      "step": 3272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1890243902439025e-05,
+      "loss": 0.3266,
+      "step": 3273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1933797909407667e-05,
+      "loss": 0.3312,
+      "step": 3274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.1977351916376307e-05,
+      "loss": 0.3363,
+      "step": 3275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2020905923344948e-05,
+      "loss": 0.3312,
+      "step": 3276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2064459930313588e-05,
+      "loss": 0.3449,
+      "step": 3277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2108013937282231e-05,
+      "loss": 0.3408,
+      "step": 3278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2151567944250871e-05,
+      "loss": 0.3422,
+      "step": 3279
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 0.3283,
+      "step": 3280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2238675958188153e-05,
+      "loss": 0.3222,
+      "step": 3281
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2282229965156796e-05,
+      "loss": 0.3387,
+      "step": 3282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2325783972125436e-05,
+      "loss": 0.3215,
+      "step": 3283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2369337979094078e-05,
+      "loss": 0.3388,
+      "step": 3284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2412891986062717e-05,
+      "loss": 0.326,
+      "step": 3285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2456445993031359e-05,
+      "loss": 0.3337,
+      "step": 3286
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3407,
+      "step": 3287
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2543554006968642e-05,
+      "loss": 0.3459,
+      "step": 3288
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2587108013937282e-05,
+      "loss": 0.3396,
+      "step": 3289
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2630662020905924e-05,
+      "loss": 0.3396,
+      "step": 3290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2674216027874567e-05,
+      "loss": 0.3292,
+      "step": 3291
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2717770034843207e-05,
+      "loss": 0.3437,
+      "step": 3292
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2761324041811848e-05,
+      "loss": 0.3199,
+      "step": 3293
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2804878048780488e-05,
+      "loss": 0.3432,
+      "step": 3294
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.284843205574913e-05,
+      "loss": 0.3343,
+      "step": 3295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.289198606271777e-05,
+      "loss": 0.3293,
+      "step": 3296
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2935540069686411e-05,
+      "loss": 0.3377,
+      "step": 3297
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2979094076655051e-05,
+      "loss": 0.3231,
+      "step": 3298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3022648083623695e-05,
+      "loss": 0.3304,
+      "step": 3299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3066202090592336e-05,
+      "loss": 0.3529,
+      "step": 3300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3109756097560976e-05,
+      "loss": 0.3607,
+      "step": 3301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3153310104529618e-05,
+      "loss": 0.346,
+      "step": 3302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3196864111498258e-05,
+      "loss": 0.3529,
+      "step": 3303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3240418118466899e-05,
+      "loss": 0.3449,
+      "step": 3304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3283972125435539e-05,
+      "loss": 0.3404,
+      "step": 3305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.332752613240418e-05,
+      "loss": 0.3446,
+      "step": 3306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3371080139372824e-05,
+      "loss": 0.3035,
+      "step": 3307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3414634146341466e-05,
+      "loss": 0.313,
+      "step": 3308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3458188153310105e-05,
+      "loss": 0.3414,
+      "step": 3309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3501742160278747e-05,
+      "loss": 0.3421,
+      "step": 3310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3545296167247387e-05,
+      "loss": 0.3419,
+      "step": 3311
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3588850174216028e-05,
+      "loss": 0.3223,
+      "step": 3312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3632404181184668e-05,
+      "loss": 0.3299,
+      "step": 3313
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.367595818815331e-05,
+      "loss": 0.3206,
+      "step": 3314
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3719512195121953e-05,
+      "loss": 0.3407,
+      "step": 3315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3763066202090593e-05,
+      "loss": 0.348,
+      "step": 3316
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3806620209059235e-05,
+      "loss": 0.3564,
+      "step": 3317
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3850174216027875e-05,
+      "loss": 0.3195,
+      "step": 3318
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3893728222996516e-05,
+      "loss": 0.3293,
+      "step": 3319
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3937282229965156e-05,
+      "loss": 0.332,
+      "step": 3320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.3980836236933798e-05,
+      "loss": 0.3221,
+      "step": 3321
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4024390243902441e-05,
+      "loss": 0.3262,
+      "step": 3322
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4067944250871083e-05,
+      "loss": 0.3409,
+      "step": 3323
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4111498257839722e-05,
+      "loss": 0.359,
+      "step": 3324
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4155052264808364e-05,
+      "loss": 0.3346,
+      "step": 3325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4198606271777004e-05,
+      "loss": 0.3249,
+      "step": 3326
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4242160278745646e-05,
+      "loss": 0.3501,
+      "step": 3327
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.3465,
+      "step": 3328
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4329268292682927e-05,
+      "loss": 0.3587,
+      "step": 3329
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.437282229965157e-05,
+      "loss": 0.3718,
+      "step": 3330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.441637630662021e-05,
+      "loss": 0.366,
+      "step": 3331
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4459930313588852e-05,
+      "loss": 0.34,
+      "step": 3332
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4503484320557492e-05,
+      "loss": 0.3478,
+      "step": 3333
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4547038327526133e-05,
+      "loss": 0.3357,
+      "step": 3334
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4590592334494773e-05,
+      "loss": 0.3355,
+      "step": 3335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.3373,
+      "step": 3336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4677700348432055e-05,
+      "loss": 0.3441,
+      "step": 3337
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4721254355400698e-05,
+      "loss": 0.3296,
+      "step": 3338
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.476480836236934e-05,
+      "loss": 0.3394,
+      "step": 3339
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4808362369337981e-05,
+      "loss": 0.3223,
+      "step": 3340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4851916376306621e-05,
+      "loss": 0.3123,
+      "step": 3341
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4895470383275263e-05,
+      "loss": 0.3357,
+      "step": 3342
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4939024390243902e-05,
+      "loss": 0.339,
+      "step": 3343
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4982578397212544e-05,
+      "loss": 0.3326,
+      "step": 3344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5026132404181184e-05,
+      "loss": 0.3287,
+      "step": 3345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5069686411149827e-05,
+      "loss": 0.3379,
+      "step": 3346
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5113240418118469e-05,
+      "loss": 0.3312,
+      "step": 3347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5156794425087109e-05,
+      "loss": 0.3385,
+      "step": 3348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.520034843205575e-05,
+      "loss": 0.3319,
+      "step": 3349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.524390243902439e-05,
+      "loss": 0.3214,
+      "step": 3350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.528745644599303e-05,
+      "loss": 0.3402,
+      "step": 3351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5331010452961673e-05,
+      "loss": 0.326,
+      "step": 3352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5374564459930317e-05,
+      "loss": 0.3276,
+      "step": 3353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5418118466898957e-05,
+      "loss": 0.3375,
+      "step": 3354
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5461672473867596e-05,
+      "loss": 0.3217,
+      "step": 3355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5505226480836236e-05,
+      "loss": 0.3457,
+      "step": 3356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.554878048780488e-05,
+      "loss": 0.3356,
+      "step": 3357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.559233449477352e-05,
+      "loss": 0.3291,
+      "step": 3358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.563588850174216e-05,
+      "loss": 0.374,
+      "step": 3359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.56794425087108e-05,
+      "loss": 0.326,
+      "step": 3360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5722996515679446e-05,
+      "loss": 0.3193,
+      "step": 3361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5766550522648086e-05,
+      "loss": 0.34,
+      "step": 3362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5810104529616726e-05,
+      "loss": 0.3274,
+      "step": 3363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5853658536585366e-05,
+      "loss": 0.3153,
+      "step": 3364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.589721254355401e-05,
+      "loss": 0.342,
+      "step": 3365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.594076655052265e-05,
+      "loss": 0.3438,
+      "step": 3366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.598432055749129e-05,
+      "loss": 0.359,
+      "step": 3367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.602787456445993e-05,
+      "loss": 0.3267,
+      "step": 3368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.3593,
+      "step": 3369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6114982578397215e-05,
+      "loss": 0.3443,
+      "step": 3370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6158536585365855e-05,
+      "loss": 0.3407,
+      "step": 3371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6202090592334495e-05,
+      "loss": 0.3338,
+      "step": 3372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6245644599303135e-05,
+      "loss": 0.3474,
+      "step": 3373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6289198606271778e-05,
+      "loss": 0.3404,
+      "step": 3374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6332752613240418e-05,
+      "loss": 0.3303,
+      "step": 3375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6376306620209058e-05,
+      "loss": 0.3258,
+      "step": 3376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.64198606271777e-05,
+      "loss": 0.3361,
+      "step": 3377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6463414634146345e-05,
+      "loss": 0.3223,
+      "step": 3378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6506968641114984e-05,
+      "loss": 0.3364,
+      "step": 3379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6550522648083624e-05,
+      "loss": 0.3364,
+      "step": 3380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6594076655052264e-05,
+      "loss": 0.3194,
+      "step": 3381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6637630662020908e-05,
+      "loss": 0.3528,
+      "step": 3382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6681184668989547e-05,
+      "loss": 0.3269,
+      "step": 3383
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.672473867595819e-05,
+      "loss": 0.3271,
+      "step": 3384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.676829268292683e-05,
+      "loss": 0.3365,
+      "step": 3385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.681184668989547e-05,
+      "loss": 0.3499,
+      "step": 3386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6855400696864114e-05,
+      "loss": 0.3357,
+      "step": 3387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6898954703832754e-05,
+      "loss": 0.3487,
+      "step": 3388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6942508710801394e-05,
+      "loss": 0.314,
+      "step": 3389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6986062717770033e-05,
+      "loss": 0.3341,
+      "step": 3390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7029616724738677e-05,
+      "loss": 0.3335,
+      "step": 3391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.707317073170732e-05,
+      "loss": 0.3317,
+      "step": 3392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.711672473867596e-05,
+      "loss": 0.3501,
+      "step": 3393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.71602787456446e-05,
+      "loss": 0.3403,
+      "step": 3394
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7203832752613243e-05,
+      "loss": 0.3342,
+      "step": 3395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7247386759581883e-05,
+      "loss": 0.3308,
+      "step": 3396
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7290940766550523e-05,
+      "loss": 0.3435,
+      "step": 3397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7334494773519163e-05,
+      "loss": 0.3293,
+      "step": 3398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7378048780487806e-05,
+      "loss": 0.3209,
+      "step": 3399
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.742160278745645e-05,
+      "loss": 0.3323,
+      "step": 3400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.746515679442509e-05,
+      "loss": 0.344,
+      "step": 3401
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.750871080139373e-05,
+      "loss": 0.3058,
+      "step": 3402
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.755226480836237e-05,
+      "loss": 0.3357,
+      "step": 3403
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7595818815331012e-05,
+      "loss": 0.3388,
+      "step": 3404
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7639372822299652e-05,
+      "loss": 0.3292,
+      "step": 3405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7682926829268292e-05,
+      "loss": 0.3371,
+      "step": 3406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7726480836236932e-05,
+      "loss": 0.3288,
+      "step": 3407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7770034843205575e-05,
+      "loss": 0.3234,
+      "step": 3408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.781358885017422e-05,
+      "loss": 0.3381,
+      "step": 3409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.3296,
+      "step": 3410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.79006968641115e-05,
+      "loss": 0.3441,
+      "step": 3411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7944250871080138e-05,
+      "loss": 0.3482,
+      "step": 3412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.798780487804878e-05,
+      "loss": 0.3136,
+      "step": 3413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.803135888501742e-05,
+      "loss": 0.3361,
+      "step": 3414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8074912891986065e-05,
+      "loss": 0.3476,
+      "step": 3415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8118466898954705e-05,
+      "loss": 0.3346,
+      "step": 3416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8162020905923348e-05,
+      "loss": 0.3186,
+      "step": 3417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8205574912891988e-05,
+      "loss": 0.3305,
+      "step": 3418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8249128919860628e-05,
+      "loss": 0.3325,
+      "step": 3419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8292682926829268e-05,
+      "loss": 0.3366,
+      "step": 3420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.833623693379791e-05,
+      "loss": 0.3239,
+      "step": 3421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.837979094076655e-05,
+      "loss": 0.33,
+      "step": 3422
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8423344947735194e-05,
+      "loss": 0.3441,
+      "step": 3423
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8466898954703834e-05,
+      "loss": 0.3307,
+      "step": 3424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8510452961672474e-05,
+      "loss": 0.355,
+      "step": 3425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8554006968641117e-05,
+      "loss": 0.3268,
+      "step": 3426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8597560975609757e-05,
+      "loss": 0.311,
+      "step": 3427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8641114982578397e-05,
+      "loss": 0.3743,
+      "step": 3428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8684668989547037e-05,
+      "loss": 0.3271,
+      "step": 3429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.872822299651568e-05,
+      "loss": 0.3281,
+      "step": 3430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8771777003484323e-05,
+      "loss": 0.3413,
+      "step": 3431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8815331010452963e-05,
+      "loss": 0.351,
+      "step": 3432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8858885017421603e-05,
+      "loss": 0.3433,
+      "step": 3433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8902439024390246e-05,
+      "loss": 0.3425,
+      "step": 3434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8945993031358886e-05,
+      "loss": 0.3374,
+      "step": 3435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8989547038327526e-05,
+      "loss": 0.3207,
+      "step": 3436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9033101045296166e-05,
+      "loss": 0.3265,
+      "step": 3437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.907665505226481e-05,
+      "loss": 0.373,
+      "step": 3438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9120209059233453e-05,
+      "loss": 0.3371,
+      "step": 3439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9163763066202093e-05,
+      "loss": 0.3434,
+      "step": 3440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9207317073170733e-05,
+      "loss": 0.3172,
+      "step": 3441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9250871080139372e-05,
+      "loss": 0.332,
+      "step": 3442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9294425087108016e-05,
+      "loss": 0.3413,
+      "step": 3443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9337979094076656e-05,
+      "loss": 0.3253,
+      "step": 3444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9381533101045295e-05,
+      "loss": 0.3369,
+      "step": 3445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9425087108013935e-05,
+      "loss": 0.3501,
+      "step": 3446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9468641114982582e-05,
+      "loss": 0.3225,
+      "step": 3447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.3344,
+      "step": 3448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9555749128919862e-05,
+      "loss": 0.3426,
+      "step": 3449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9599303135888502e-05,
+      "loss": 0.3142,
+      "step": 3450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.3283,
+      "step": 3451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9686411149825785e-05,
+      "loss": 0.3577,
+      "step": 3452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9729965156794425e-05,
+      "loss": 0.3263,
+      "step": 3453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9773519163763068e-05,
+      "loss": 0.338,
+      "step": 3454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9817073170731708e-05,
+      "loss": 0.3188,
+      "step": 3455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.986062717770035e-05,
+      "loss": 0.3142,
+      "step": 3456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.990418118466899e-05,
+      "loss": 0.3368,
+      "step": 3457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.994773519163763e-05,
+      "loss": 0.3485,
+      "step": 3458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999128919860627e-05,
+      "loss": 0.3408,
+      "step": 3459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0034843205574914e-05,
+      "loss": 0.3172,
+      "step": 3460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0078397212543554e-05,
+      "loss": 0.3325,
+      "step": 3461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0121951219512197e-05,
+      "loss": 0.3406,
+      "step": 3462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0165505226480837e-05,
+      "loss": 0.3333,
+      "step": 3463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0209059233449477e-05,
+      "loss": 0.3431,
+      "step": 3464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.025261324041812e-05,
+      "loss": 0.3506,
+      "step": 3465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.029616724738676e-05,
+      "loss": 0.3152,
+      "step": 3466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.03397212543554e-05,
+      "loss": 0.3352,
+      "step": 3467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.038327526132404e-05,
+      "loss": 0.3387,
+      "step": 3468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0426829268292683e-05,
+      "loss": 0.3369,
+      "step": 3469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0470383275261327e-05,
+      "loss": 0.3244,
+      "step": 3470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0513937282229967e-05,
+      "loss": 0.335,
+      "step": 3471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0557491289198607e-05,
+      "loss": 0.343,
+      "step": 3472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.060104529616725e-05,
+      "loss": 0.3564,
+      "step": 3473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.064459930313589e-05,
+      "loss": 0.3435,
+      "step": 3474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.068815331010453e-05,
+      "loss": 0.3241,
+      "step": 3475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.073170731707317e-05,
+      "loss": 0.3126,
+      "step": 3476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0775261324041813e-05,
+      "loss": 0.3286,
+      "step": 3477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0818815331010456e-05,
+      "loss": 0.3334,
+      "step": 3478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0862369337979096e-05,
+      "loss": 0.3479,
+      "step": 3479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0905923344947736e-05,
+      "loss": 0.3433,
+      "step": 3480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0949477351916376e-05,
+      "loss": 0.3151,
+      "step": 3481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.099303135888502e-05,
+      "loss": 0.3359,
+      "step": 3482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.103658536585366e-05,
+      "loss": 0.3445,
+      "step": 3483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.10801393728223e-05,
+      "loss": 0.3377,
+      "step": 3484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1123693379790942e-05,
+      "loss": 0.328,
+      "step": 3485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1167247386759585e-05,
+      "loss": 0.3412,
+      "step": 3486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1210801393728225e-05,
+      "loss": 0.3366,
+      "step": 3487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1254355400696865e-05,
+      "loss": 0.3252,
+      "step": 3488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1297909407665505e-05,
+      "loss": 0.3305,
+      "step": 3489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.134146341463415e-05,
+      "loss": 0.3386,
+      "step": 3490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1385017421602788e-05,
+      "loss": 0.3235,
+      "step": 3491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.3568,
+      "step": 3492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.147212543554007e-05,
+      "loss": 0.3338,
+      "step": 3493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.151567944250871e-05,
+      "loss": 0.3258,
+      "step": 3494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1559233449477355e-05,
+      "loss": 0.3248,
+      "step": 3495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1602787456445995e-05,
+      "loss": 0.3266,
+      "step": 3496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1646341463414634e-05,
+      "loss": 0.3368,
+      "step": 3497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1689895470383274e-05,
+      "loss": 0.326,
+      "step": 3498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1733449477351918e-05,
+      "loss": 0.3194,
+      "step": 3499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1777003484320557e-05,
+      "loss": 0.318,
+      "step": 3500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.18205574912892e-05,
+      "loss": 0.3204,
+      "step": 3501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.186411149825784e-05,
+      "loss": 0.3258,
+      "step": 3502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1907665505226484e-05,
+      "loss": 0.3203,
+      "step": 3503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1951219512195124e-05,
+      "loss": 0.3365,
+      "step": 3504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1994773519163764e-05,
+      "loss": 0.3315,
+      "step": 3505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2038327526132404e-05,
+      "loss": 0.3216,
+      "step": 3506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2081881533101047e-05,
+      "loss": 0.3303,
+      "step": 3507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2125435540069687e-05,
+      "loss": 0.3458,
+      "step": 3508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.216898954703833e-05,
+      "loss": 0.3394,
+      "step": 3509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.221254355400697e-05,
+      "loss": 0.3232,
+      "step": 3510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.225609756097561e-05,
+      "loss": 0.3439,
+      "step": 3511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2299651567944253e-05,
+      "loss": 0.3392,
+      "step": 3512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2343205574912893e-05,
+      "loss": 0.352,
+      "step": 3513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2386759581881533e-05,
+      "loss": 0.3609,
+      "step": 3514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2430313588850173e-05,
+      "loss": 0.3233,
+      "step": 3515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2473867595818816e-05,
+      "loss": 0.343,
+      "step": 3516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.251742160278746e-05,
+      "loss": 0.3291,
+      "step": 3517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.25609756097561e-05,
+      "loss": 0.3395,
+      "step": 3518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.260452961672474e-05,
+      "loss": 0.3424,
+      "step": 3519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.264808362369338e-05,
+      "loss": 0.3366,
+      "step": 3520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2691637630662022e-05,
+      "loss": 0.3266,
+      "step": 3521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2735191637630662e-05,
+      "loss": 0.3354,
+      "step": 3522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2778745644599302e-05,
+      "loss": 0.3597,
+      "step": 3523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.2822299651567945e-05,
+      "loss": 0.3412,
+      "step": 3524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.286585365853659e-05,
+      "loss": 0.3407,
+      "step": 3525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.290940766550523e-05,
+      "loss": 0.33,
+      "step": 3526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.295296167247387e-05,
+      "loss": 0.3456,
+      "step": 3527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.299651567944251e-05,
+      "loss": 0.3289,
+      "step": 3528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3040069686411152e-05,
+      "loss": 0.3209,
+      "step": 3529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.308362369337979e-05,
+      "loss": 0.3523,
+      "step": 3530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.312717770034843e-05,
+      "loss": 0.3298,
+      "step": 3531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3170731707317075e-05,
+      "loss": 0.3418,
+      "step": 3532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3214285714285715e-05,
+      "loss": 0.3509,
+      "step": 3533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3257839721254358e-05,
+      "loss": 0.3368,
+      "step": 3534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3301393728222998e-05,
+      "loss": 0.3333,
+      "step": 3535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3344947735191638e-05,
+      "loss": 0.3342,
+      "step": 3536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3388501742160278e-05,
+      "loss": 0.3422,
+      "step": 3537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.343205574912892e-05,
+      "loss": 0.3485,
+      "step": 3538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.347560975609756e-05,
+      "loss": 0.3218,
+      "step": 3539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3519163763066204e-05,
+      "loss": 0.3472,
+      "step": 3540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3562717770034844e-05,
+      "loss": 0.3405,
+      "step": 3541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3606271777003487e-05,
+      "loss": 0.3451,
+      "step": 3542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3649825783972127e-05,
+      "loss": 0.3393,
+      "step": 3543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3693379790940767e-05,
+      "loss": 0.3328,
+      "step": 3544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3736933797909407e-05,
+      "loss": 0.3314,
+      "step": 3545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.378048780487805e-05,
+      "loss": 0.3211,
+      "step": 3546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3824041811846694e-05,
+      "loss": 0.332,
+      "step": 3547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3867595818815333e-05,
+      "loss": 0.3102,
+      "step": 3548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3911149825783973e-05,
+      "loss": 0.3121,
+      "step": 3549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3954703832752613e-05,
+      "loss": 0.3398,
+      "step": 3550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3998257839721257e-05,
+      "loss": 0.3332,
+      "step": 3551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4041811846689896e-05,
+      "loss": 0.3393,
+      "step": 3552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4085365853658536e-05,
+      "loss": 0.3266,
+      "step": 3553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4128919860627176e-05,
+      "loss": 0.3188,
+      "step": 3554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4172473867595823e-05,
+      "loss": 0.3297,
+      "step": 3555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4216027874564463e-05,
+      "loss": 0.3292,
+      "step": 3556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4259581881533103e-05,
+      "loss": 0.3384,
+      "step": 3557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4303135888501743e-05,
+      "loss": 0.3343,
+      "step": 3558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4346689895470386e-05,
+      "loss": 0.3145,
+      "step": 3559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.3318,
+      "step": 3560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4433797909407666e-05,
+      "loss": 0.3405,
+      "step": 3561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4477351916376306e-05,
+      "loss": 0.3274,
+      "step": 3562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.452090592334495e-05,
+      "loss": 0.3223,
+      "step": 3563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4564459930313592e-05,
+      "loss": 0.3193,
+      "step": 3564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4608013937282232e-05,
+      "loss": 0.3341,
+      "step": 3565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4651567944250872e-05,
+      "loss": 0.3421,
+      "step": 3566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4695121951219512e-05,
+      "loss": 0.357,
+      "step": 3567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4738675958188155e-05,
+      "loss": 0.3274,
+      "step": 3568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4782229965156795e-05,
+      "loss": 0.3346,
+      "step": 3569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4825783972125435e-05,
+      "loss": 0.3393,
+      "step": 3570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4869337979094078e-05,
+      "loss": 0.3491,
+      "step": 3571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.4912891986062718e-05,
+      "loss": 0.3349,
+      "step": 3572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.495644599303136e-05,
+      "loss": 0.3284,
+      "step": 3573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3362,
+      "step": 3574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.5043554006968644e-05,
+      "loss": 0.3177,
+      "step": 3575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.5087108013937284e-05,
+      "loss": 0.3416,
+      "step": 3576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.5130662020905928e-05,
+      "loss": 0.331,
+      "step": 3577
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.5174216027874564e-05,
+      "loss": 0.318,
+      "step": 3578
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.5217770034843207e-05,
+      "loss": 0.307,
+      "step": 3579
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.5261324041811847e-05,
+      "loss": 0.323,
+      "step": 3580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.530487804878049e-05,
+      "loss": 0.3285,
+      "step": 3581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.5348432055749134e-05,
+      "loss": 0.3305,
+      "step": 3582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.539198606271777e-05,
+      "loss": 0.3395,
+      "step": 3583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.5435540069686414e-05,
+      "loss": 0.327,
+      "step": 3584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.5479094076655054e-05,
+      "loss": 0.3441,
+      "step": 3585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.5522648083623697e-05,
+      "loss": 0.3545,
+      "step": 3586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.5566202090592333e-05,
+      "loss": 0.3239,
+      "step": 3587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.5609756097560977e-05,
+      "loss": 0.3267,
+      "step": 3588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.5653310104529617e-05,
+      "loss": 0.3423,
+      "step": 3589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.569686411149826e-05,
+      "loss": 0.3212,
+      "step": 3590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.5740418118466903e-05,
+      "loss": 0.3334,
+      "step": 3591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.578397212543554e-05,
+      "loss": 0.3359,
+      "step": 3592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.5827526132404183e-05,
+      "loss": 0.3294,
+      "step": 3593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.5871080139372823e-05,
+      "loss": 0.3251,
+      "step": 3594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.5914634146341466e-05,
+      "loss": 0.3509,
+      "step": 3595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.5958188153310103e-05,
+      "loss": 0.3182,
+      "step": 3596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6001742160278746e-05,
+      "loss": 0.3449,
+      "step": 3597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.604529616724739e-05,
+      "loss": 0.3743,
+      "step": 3598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.608885017421603e-05,
+      "loss": 0.3182,
+      "step": 3599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6132404181184672e-05,
+      "loss": 0.2976,
+      "step": 3600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.617595818815331e-05,
+      "loss": 0.3394,
+      "step": 3601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6219512195121952e-05,
+      "loss": 0.3238,
+      "step": 3602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6263066202090592e-05,
+      "loss": 0.3453,
+      "step": 3603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6306620209059235e-05,
+      "loss": 0.3293,
+      "step": 3604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6350174216027872e-05,
+      "loss": 0.3291,
+      "step": 3605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6393728222996515e-05,
+      "loss": 0.3404,
+      "step": 3606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6437282229965162e-05,
+      "loss": 0.3105,
+      "step": 3607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6480836236933798e-05,
+      "loss": 0.3299,
+      "step": 3608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.652439024390244e-05,
+      "loss": 0.3322,
+      "step": 3609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6567944250871078e-05,
+      "loss": 0.3277,
+      "step": 3610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6611498257839725e-05,
+      "loss": 0.3379,
+      "step": 3611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.665505226480836e-05,
+      "loss": 0.373,
+      "step": 3612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6698606271777005e-05,
+      "loss": 0.326,
+      "step": 3613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6742160278745648e-05,
+      "loss": 0.3257,
+      "step": 3614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 0.3297,
+      "step": 3615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.682926829268293e-05,
+      "loss": 0.326,
+      "step": 3616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6872822299651568e-05,
+      "loss": 0.3494,
+      "step": 3617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.691637630662021e-05,
+      "loss": 0.3364,
+      "step": 3618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.695993031358885e-05,
+      "loss": 0.3619,
+      "step": 3619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7003484320557494e-05,
+      "loss": 0.3318,
+      "step": 3620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7047038327526137e-05,
+      "loss": 0.3192,
+      "step": 3621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7090592334494774e-05,
+      "loss": 0.3304,
+      "step": 3622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7134146341463417e-05,
+      "loss": 0.3408,
+      "step": 3623
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7177700348432057e-05,
+      "loss": 0.3425,
+      "step": 3624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.72212543554007e-05,
+      "loss": 0.3109,
+      "step": 3625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7264808362369337e-05,
+      "loss": 0.3542,
+      "step": 3626
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.730836236933798e-05,
+      "loss": 0.3199,
+      "step": 3627
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.735191637630662e-05,
+      "loss": 0.3396,
+      "step": 3628
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7395470383275263e-05,
+      "loss": 0.3321,
+      "step": 3629
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7439024390243906e-05,
+      "loss": 0.3234,
+      "step": 3630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7482578397212543e-05,
+      "loss": 0.3326,
+      "step": 3631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7526132404181186e-05,
+      "loss": 0.3538,
+      "step": 3632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7569686411149826e-05,
+      "loss": 0.325,
+      "step": 3633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.761324041811847e-05,
+      "loss": 0.3422,
+      "step": 3634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7656794425087106e-05,
+      "loss": 0.3623,
+      "step": 3635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.770034843205575e-05,
+      "loss": 0.3447,
+      "step": 3636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7743902439024393e-05,
+      "loss": 0.3261,
+      "step": 3637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7787456445993032e-05,
+      "loss": 0.3413,
+      "step": 3638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7831010452961676e-05,
+      "loss": 0.3274,
+      "step": 3639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7874564459930312e-05,
+      "loss": 0.3187,
+      "step": 3640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7918118466898955e-05,
+      "loss": 0.3265,
+      "step": 3641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7961672473867595e-05,
+      "loss": 0.336,
+      "step": 3642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.800522648083624e-05,
+      "loss": 0.3347,
+      "step": 3643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8048780487804882e-05,
+      "loss": 0.3333,
+      "step": 3644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.809233449477352e-05,
+      "loss": 0.3311,
+      "step": 3645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8135888501742165e-05,
+      "loss": 0.3278,
+      "step": 3646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.81794425087108e-05,
+      "loss": 0.3123,
+      "step": 3647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8222996515679445e-05,
+      "loss": 0.324,
+      "step": 3648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.826655052264808e-05,
+      "loss": 0.3244,
+      "step": 3649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8310104529616728e-05,
+      "loss": 0.3434,
+      "step": 3650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8353658536585365e-05,
+      "loss": 0.3505,
+      "step": 3651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8397212543554008e-05,
+      "loss": 0.3179,
+      "step": 3652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.844076655052265e-05,
+      "loss": 0.3536,
+      "step": 3653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.848432055749129e-05,
+      "loss": 0.3217,
+      "step": 3654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8527874564459934e-05,
+      "loss": 0.3245,
+      "step": 3655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.3587,
+      "step": 3656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8614982578397214e-05,
+      "loss": 0.3312,
+      "step": 3657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8658536585365854e-05,
+      "loss": 0.3181,
+      "step": 3658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8702090592334497e-05,
+      "loss": 0.3466,
+      "step": 3659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.874564459930314e-05,
+      "loss": 0.345,
+      "step": 3660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8789198606271777e-05,
+      "loss": 0.3071,
+      "step": 3661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.883275261324042e-05,
+      "loss": 0.34,
+      "step": 3662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.887630662020906e-05,
+      "loss": 0.3347,
+      "step": 3663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8919860627177704e-05,
+      "loss": 0.3343,
+      "step": 3664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.896341463414634e-05,
+      "loss": 0.3172,
+      "step": 3665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9006968641114983e-05,
+      "loss": 0.3364,
+      "step": 3666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9050522648083623e-05,
+      "loss": 0.3174,
+      "step": 3667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9094076655052267e-05,
+      "loss": 0.3414,
+      "step": 3668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.913763066202091e-05,
+      "loss": 0.3117,
+      "step": 3669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9181184668989546e-05,
+      "loss": 0.334,
+      "step": 3670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.922473867595819e-05,
+      "loss": 0.3424,
+      "step": 3671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.3129,
+      "step": 3672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9311846689895473e-05,
+      "loss": 0.3408,
+      "step": 3673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.935540069686411e-05,
+      "loss": 0.3276,
+      "step": 3674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9398954703832753e-05,
+      "loss": 0.3349,
+      "step": 3675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9442508710801396e-05,
+      "loss": 0.3438,
+      "step": 3676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9486062717770036e-05,
+      "loss": 0.3263,
+      "step": 3677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.952961672473868e-05,
+      "loss": 0.3458,
+      "step": 3678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9573170731707316e-05,
+      "loss": 0.3235,
+      "step": 3679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9616724738675962e-05,
+      "loss": 0.3241,
+      "step": 3680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.96602787456446e-05,
+      "loss": 0.3407,
+      "step": 3681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9703832752613242e-05,
+      "loss": 0.3551,
+      "step": 3682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9747386759581885e-05,
+      "loss": 0.3354,
+      "step": 3683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9790940766550525e-05,
+      "loss": 0.3397,
+      "step": 3684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.983449477351917e-05,
+      "loss": 0.3189,
+      "step": 3685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9878048780487805e-05,
+      "loss": 0.3478,
+      "step": 3686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9921602787456448e-05,
+      "loss": 0.3511,
+      "step": 3687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9965156794425088e-05,
+      "loss": 0.3377,
+      "step": 3688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.000871080139373e-05,
+      "loss": 0.3436,
+      "step": 3689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0052264808362368e-05,
+      "loss": 0.3356,
+      "step": 3690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.009581881533101e-05,
+      "loss": 0.3463,
+      "step": 3691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0139372822299655e-05,
+      "loss": 0.3352,
+      "step": 3692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0182926829268294e-05,
+      "loss": 0.3151,
+      "step": 3693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0226480836236938e-05,
+      "loss": 0.3287,
+      "step": 3694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0270034843205574e-05,
+      "loss": 0.3411,
+      "step": 3695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0313588850174217e-05,
+      "loss": 0.3546,
+      "step": 3696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.3242,
+      "step": 3697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.04006968641115e-05,
+      "loss": 0.3528,
+      "step": 3698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0444250871080144e-05,
+      "loss": 0.3498,
+      "step": 3699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.048780487804878e-05,
+      "loss": 0.3295,
+      "step": 3700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0531358885017424e-05,
+      "loss": 0.3353,
+      "step": 3701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.057491289198606e-05,
+      "loss": 0.3387,
+      "step": 3702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0618466898954704e-05,
+      "loss": 0.3448,
+      "step": 3703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.066202090592335e-05,
+      "loss": 0.3463,
+      "step": 3704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.070557491289199e-05,
+      "loss": 0.3243,
+      "step": 3705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.074912891986063e-05,
+      "loss": 0.3388,
+      "step": 3706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.079268292682927e-05,
+      "loss": 0.3281,
+      "step": 3707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.083623693379791e-05,
+      "loss": 0.3372,
+      "step": 3708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.087979094076655e-05,
+      "loss": 0.3195,
+      "step": 3709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.092334494773519e-05,
+      "loss": 0.3446,
+      "step": 3710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.096689895470383e-05,
+      "loss": 0.3394,
+      "step": 3711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.101045296167247e-05,
+      "loss": 0.3443,
+      "step": 3712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1054006968641116e-05,
+      "loss": 0.3167,
+      "step": 3713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.109756097560976e-05,
+      "loss": 0.3274,
+      "step": 3714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.11411149825784e-05,
+      "loss": 0.3348,
+      "step": 3715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.118466898954704e-05,
+      "loss": 0.3427,
+      "step": 3716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.122822299651568e-05,
+      "loss": 0.3625,
+      "step": 3717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.127177700348432e-05,
+      "loss": 0.3319,
+      "step": 3718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.131533101045296e-05,
+      "loss": 0.3256,
+      "step": 3719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.13588850174216e-05,
+      "loss": 0.3432,
+      "step": 3720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.140243902439025e-05,
+      "loss": 0.3362,
+      "step": 3721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.144599303135889e-05,
+      "loss": 0.3299,
+      "step": 3722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.148954703832753e-05,
+      "loss": 0.37,
+      "step": 3723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.153310104529617e-05,
+      "loss": 0.3393,
+      "step": 3724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.157665505226481e-05,
+      "loss": 0.3196,
+      "step": 3725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.162020905923345e-05,
+      "loss": 0.3094,
+      "step": 3726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.166376306620209e-05,
+      "loss": 0.3203,
+      "step": 3727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.170731707317073e-05,
+      "loss": 0.3278,
+      "step": 3728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1750871080139375e-05,
+      "loss": 0.3405,
+      "step": 3729
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.179442508710802e-05,
+      "loss": 0.3378,
+      "step": 3730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.183797909407666e-05,
+      "loss": 0.3267,
+      "step": 3731
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.18815331010453e-05,
+      "loss": 0.328,
+      "step": 3732
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.192508710801394e-05,
+      "loss": 0.3431,
+      "step": 3733
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.196864111498258e-05,
+      "loss": 0.3273,
+      "step": 3734
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.201219512195122e-05,
+      "loss": 0.3268,
+      "step": 3735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.205574912891986e-05,
+      "loss": 0.3336,
+      "step": 3736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.20993031358885e-05,
+      "loss": 0.3297,
+      "step": 3737
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.3275,
+      "step": 3738
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.218641114982579e-05,
+      "loss": 0.356,
+      "step": 3739
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.222996515679443e-05,
+      "loss": 0.3273,
+      "step": 3740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.227351916376307e-05,
+      "loss": 0.3354,
+      "step": 3741
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.231707317073171e-05,
+      "loss": 0.3187,
+      "step": 3742
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.236062717770035e-05,
+      "loss": 0.3327,
+      "step": 3743
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.240418118466899e-05,
+      "loss": 0.325,
+      "step": 3744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.244773519163763e-05,
+      "loss": 0.3499,
+      "step": 3745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.249128919860627e-05,
+      "loss": 0.3364,
+      "step": 3746
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.253484320557491e-05,
+      "loss": 0.3402,
+      "step": 3747
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2578397212543556e-05,
+      "loss": 0.3386,
+      "step": 3748
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.26219512195122e-05,
+      "loss": 0.332,
+      "step": 3749
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2665505226480836e-05,
+      "loss": 0.3221,
+      "step": 3750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.270905923344948e-05,
+      "loss": 0.3275,
+      "step": 3751
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2752613240418116e-05,
+      "loss": 0.329,
+      "step": 3752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.279616724738676e-05,
+      "loss": 0.3301,
+      "step": 3753
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.28397212543554e-05,
+      "loss": 0.3151,
+      "step": 3754
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.288327526132404e-05,
+      "loss": 0.3277,
+      "step": 3755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.292682926829269e-05,
+      "loss": 0.3191,
+      "step": 3756
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2970383275261326e-05,
+      "loss": 0.3424,
+      "step": 3757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.301393728222997e-05,
+      "loss": 0.3153,
+      "step": 3758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3057491289198605e-05,
+      "loss": 0.3363,
+      "step": 3759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.310104529616725e-05,
+      "loss": 0.3149,
+      "step": 3760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.314459930313589e-05,
+      "loss": 0.3233,
+      "step": 3761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.318815331010453e-05,
+      "loss": 0.3422,
+      "step": 3762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.323170731707317e-05,
+      "loss": 0.3389,
+      "step": 3763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3275261324041815e-05,
+      "loss": 0.3338,
+      "step": 3764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.331881533101046e-05,
+      "loss": 0.3225,
+      "step": 3765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3362369337979095e-05,
+      "loss": 0.3367,
+      "step": 3766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.340592334494774e-05,
+      "loss": 0.3275,
+      "step": 3767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.344947735191638e-05,
+      "loss": 0.3479,
+      "step": 3768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.349303135888502e-05,
+      "loss": 0.3124,
+      "step": 3769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.353658536585366e-05,
+      "loss": 0.3485,
+      "step": 3770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.35801393728223e-05,
+      "loss": 0.3433,
+      "step": 3771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.362369337979094e-05,
+      "loss": 0.3537,
+      "step": 3772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3667247386759584e-05,
+      "loss": 0.3414,
+      "step": 3773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.371080139372823e-05,
+      "loss": 0.314,
+      "step": 3774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3754355400696864e-05,
+      "loss": 0.3136,
+      "step": 3775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.379790940766551e-05,
+      "loss": 0.3177,
+      "step": 3776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.384146341463415e-05,
+      "loss": 0.342,
+      "step": 3777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.388501742160279e-05,
+      "loss": 0.3421,
+      "step": 3778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.332,
+      "step": 3779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.397212543554007e-05,
+      "loss": 0.3169,
+      "step": 3780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.401567944250871e-05,
+      "loss": 0.3432,
+      "step": 3781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4059233449477354e-05,
+      "loss": 0.3304,
+      "step": 3782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4102787456446e-05,
+      "loss": 0.3472,
+      "step": 3783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.414634146341464e-05,
+      "loss": 0.3401,
+      "step": 3784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4189895470383277e-05,
+      "loss": 0.3441,
+      "step": 3785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.423344947735192e-05,
+      "loss": 0.3284,
+      "step": 3786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4277003484320556e-05,
+      "loss": 0.3346,
+      "step": 3787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.43205574912892e-05,
+      "loss": 0.3214,
+      "step": 3788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4364111498257836e-05,
+      "loss": 0.3398,
+      "step": 3789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4407665505226486e-05,
+      "loss": 0.3271,
+      "step": 3790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.445121951219512e-05,
+      "loss": 0.3416,
+      "step": 3791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4494773519163766e-05,
+      "loss": 0.3268,
+      "step": 3792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.453832752613241e-05,
+      "loss": 0.3131,
+      "step": 3793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4581881533101046e-05,
+      "loss": 0.3272,
+      "step": 3794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.462543554006969e-05,
+      "loss": 0.3031,
+      "step": 3795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4668989547038326e-05,
+      "loss": 0.3351,
+      "step": 3796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.471254355400697e-05,
+      "loss": 0.3407,
+      "step": 3797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.475609756097561e-05,
+      "loss": 0.3527,
+      "step": 3798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4799651567944255e-05,
+      "loss": 0.3178,
+      "step": 3799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.48432055749129e-05,
+      "loss": 0.3494,
+      "step": 3800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4886759581881535e-05,
+      "loss": 0.3402,
+      "step": 3801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.493031358885018e-05,
+      "loss": 0.3258,
+      "step": 3802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.4973867595818815e-05,
+      "loss": 0.3149,
+      "step": 3803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.501742160278746e-05,
+      "loss": 0.3423,
+      "step": 3804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.5060975609756095e-05,
+      "loss": 0.3277,
+      "step": 3805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.510452961672474e-05,
+      "loss": 0.3046,
+      "step": 3806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.514808362369338e-05,
+      "loss": 0.3404,
+      "step": 3807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.5191637630662025e-05,
+      "loss": 0.3338,
+      "step": 3808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.523519163763067e-05,
+      "loss": 0.3281,
+      "step": 3809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.5278745644599304e-05,
+      "loss": 0.3417,
+      "step": 3810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.532229965156795e-05,
+      "loss": 0.3345,
+      "step": 3811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.5365853658536584e-05,
+      "loss": 0.3215,
+      "step": 3812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.540940766550523e-05,
+      "loss": 0.328,
+      "step": 3813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.5452961672473864e-05,
+      "loss": 0.3223,
+      "step": 3814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.549651567944251e-05,
+      "loss": 0.3126,
+      "step": 3815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.554006968641115e-05,
+      "loss": 0.3339,
+      "step": 3816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.5583623693379794e-05,
+      "loss": 0.336,
+      "step": 3817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.562717770034844e-05,
+      "loss": 0.3046,
+      "step": 3818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.5670731707317074e-05,
+      "loss": 0.3467,
+      "step": 3819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.3494,
+      "step": 3820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.5757839721254353e-05,
+      "loss": 0.3194,
+      "step": 3821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.5801393728223e-05,
+      "loss": 0.3244,
+      "step": 3822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.584494773519164e-05,
+      "loss": 0.331,
+      "step": 3823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.5888501742160277e-05,
+      "loss": 0.3204,
+      "step": 3824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.5932055749128927e-05,
+      "loss": 0.3291,
+      "step": 3825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.597560975609756e-05,
+      "loss": 0.3261,
+      "step": 3826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6019163763066206e-05,
+      "loss": 0.3453,
+      "step": 3827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.606271777003484e-05,
+      "loss": 0.3374,
+      "step": 3828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6106271777003486e-05,
+      "loss": 0.3286,
+      "step": 3829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.614982578397213e-05,
+      "loss": 0.3314,
+      "step": 3830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6193379790940766e-05,
+      "loss": 0.3197,
+      "step": 3831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.623693379790941e-05,
+      "loss": 0.3267,
+      "step": 3832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.628048780487805e-05,
+      "loss": 0.333,
+      "step": 3833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6324041811846696e-05,
+      "loss": 0.3242,
+      "step": 3834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.636759581881533e-05,
+      "loss": 0.3409,
+      "step": 3835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6411149825783976e-05,
+      "loss": 0.3157,
+      "step": 3836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.645470383275261e-05,
+      "loss": 0.3306,
+      "step": 3837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6498257839721255e-05,
+      "loss": 0.3419,
+      "step": 3838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.65418118466899e-05,
+      "loss": 0.3208,
+      "step": 3839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6585365853658535e-05,
+      "loss": 0.3451,
+      "step": 3840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.662891986062718e-05,
+      "loss": 0.3336,
+      "step": 3841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.667247386759582e-05,
+      "loss": 0.3276,
+      "step": 3842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6716027874564465e-05,
+      "loss": 0.3096,
+      "step": 3843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.67595818815331e-05,
+      "loss": 0.3188,
+      "step": 3844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6803135888501745e-05,
+      "loss": 0.3262,
+      "step": 3845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.684668989547039e-05,
+      "loss": 0.331,
+      "step": 3846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6890243902439025e-05,
+      "loss": 0.3246,
+      "step": 3847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.693379790940767e-05,
+      "loss": 0.33,
+      "step": 3848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.6977351916376304e-05,
+      "loss": 0.3323,
+      "step": 3849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.702090592334495e-05,
+      "loss": 0.3419,
+      "step": 3850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.706445993031359e-05,
+      "loss": 0.3324,
+      "step": 3851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7108013937282234e-05,
+      "loss": 0.3249,
+      "step": 3852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.715156794425087e-05,
+      "loss": 0.361,
+      "step": 3853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7195121951219514e-05,
+      "loss": 0.3377,
+      "step": 3854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.723867595818816e-05,
+      "loss": 0.3195,
+      "step": 3855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7282229965156794e-05,
+      "loss": 0.3656,
+      "step": 3856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.732578397212544e-05,
+      "loss": 0.3426,
+      "step": 3857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7369337979094074e-05,
+      "loss": 0.3197,
+      "step": 3858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.741289198606272e-05,
+      "loss": 0.3365,
+      "step": 3859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.745644599303136e-05,
+      "loss": 0.332,
+      "step": 3860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.316,
+      "step": 3861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.754355400696865e-05,
+      "loss": 0.3186,
+      "step": 3862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.758710801393728e-05,
+      "loss": 0.3205,
+      "step": 3863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7630662020905927e-05,
+      "loss": 0.3233,
+      "step": 3864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.767421602787456e-05,
+      "loss": 0.3408,
+      "step": 3865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7717770034843206e-05,
+      "loss": 0.3196,
+      "step": 3866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.776132404181185e-05,
+      "loss": 0.3345,
+      "step": 3867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.780487804878049e-05,
+      "loss": 0.3436,
+      "step": 3868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7848432055749136e-05,
+      "loss": 0.3198,
+      "step": 3869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.789198606271777e-05,
+      "loss": 0.326,
+      "step": 3870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7935540069686416e-05,
+      "loss": 0.3517,
+      "step": 3871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.797909407665505e-05,
+      "loss": 0.3472,
+      "step": 3872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8022648083623696e-05,
+      "loss": 0.3294,
+      "step": 3873
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.806620209059233e-05,
+      "loss": 0.3087,
+      "step": 3874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8109756097560976e-05,
+      "loss": 0.3211,
+      "step": 3875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.815331010452962e-05,
+      "loss": 0.3205,
+      "step": 3876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.819686411149826e-05,
+      "loss": 0.3232,
+      "step": 3877
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8240418118466905e-05,
+      "loss": 0.3369,
+      "step": 3878
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.828397212543554e-05,
+      "loss": 0.345,
+      "step": 3879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8327526132404185e-05,
+      "loss": 0.3328,
+      "step": 3880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.837108013937282e-05,
+      "loss": 0.3155,
+      "step": 3881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8414634146341465e-05,
+      "loss": 0.3261,
+      "step": 3882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.84581881533101e-05,
+      "loss": 0.3348,
+      "step": 3883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8501742160278745e-05,
+      "loss": 0.3268,
+      "step": 3884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.854529616724739e-05,
+      "loss": 0.3167,
+      "step": 3885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.858885017421603e-05,
+      "loss": 0.3102,
+      "step": 3886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8632404181184675e-05,
+      "loss": 0.3291,
+      "step": 3887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867595818815331e-05,
+      "loss": 0.3504,
+      "step": 3888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8719512195121954e-05,
+      "loss": 0.328,
+      "step": 3889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876306620209059e-05,
+      "loss": 0.3247,
+      "step": 3890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8806620209059234e-05,
+      "loss": 0.3185,
+      "step": 3891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.885017421602787e-05,
+      "loss": 0.3297,
+      "step": 3892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8893728222996514e-05,
+      "loss": 0.3317,
+      "step": 3893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8937282229965164e-05,
+      "loss": 0.3195,
+      "step": 3894
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.89808362369338e-05,
+      "loss": 0.3256,
+      "step": 3895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.3448,
+      "step": 3896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.906794425087108e-05,
+      "loss": 0.3335,
+      "step": 3897
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9111498257839724e-05,
+      "loss": 0.3379,
+      "step": 3898
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.915505226480836e-05,
+      "loss": 0.3411,
+      "step": 3899
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9198606271777003e-05,
+      "loss": 0.3334,
+      "step": 3900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.924216027874565e-05,
+      "loss": 0.3333,
+      "step": 3901
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.3223,
+      "step": 3902
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.932926829268293e-05,
+      "loss": 0.3329,
+      "step": 3903
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.937282229965157e-05,
+      "loss": 0.3538,
+      "step": 3904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.941637630662021e-05,
+      "loss": 0.3316,
+      "step": 3905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.945993031358885e-05,
+      "loss": 0.3263,
+      "step": 3906
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.950348432055749e-05,
+      "loss": 0.3521,
+      "step": 3907
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9547038327526136e-05,
+      "loss": 0.3306,
+      "step": 3908
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.959059233449477e-05,
+      "loss": 0.3479,
+      "step": 3909
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9634146341463416e-05,
+      "loss": 0.3182,
+      "step": 3910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.967770034843206e-05,
+      "loss": 0.3341,
+      "step": 3911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.97212543554007e-05,
+      "loss": 0.3232,
+      "step": 3912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.976480836236934e-05,
+      "loss": 0.3404,
+      "step": 3913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.980836236933798e-05,
+      "loss": 0.3365,
+      "step": 3914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.985191637630662e-05,
+      "loss": 0.3289,
+      "step": 3915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.989547038327526e-05,
+      "loss": 0.3331,
+      "step": 3916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9939024390243905e-05,
+      "loss": 0.3371,
+      "step": 3917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.998257839721254e-05,
+      "loss": 0.3432,
+      "step": 3918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.0026132404181185e-05,
+      "loss": 0.3374,
+      "step": 3919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.006968641114983e-05,
+      "loss": 0.3311,
+      "step": 3920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.011324041811847e-05,
+      "loss": 0.3339,
+      "step": 3921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.015679442508711e-05,
+      "loss": 0.3168,
+      "step": 3922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.020034843205575e-05,
+      "loss": 0.341,
+      "step": 3923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.0243902439024395e-05,
+      "loss": 0.3324,
+      "step": 3924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.028745644599303e-05,
+      "loss": 0.3496,
+      "step": 3925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.0331010452961675e-05,
+      "loss": 0.3414,
+      "step": 3926
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.037456445993031e-05,
+      "loss": 0.3444,
+      "step": 3927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.0418118466898954e-05,
+      "loss": 0.3153,
+      "step": 3928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.04616724738676e-05,
+      "loss": 0.3184,
+      "step": 3929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.050522648083624e-05,
+      "loss": 0.3383,
+      "step": 3930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.0548780487804884e-05,
+      "loss": 0.3263,
+      "step": 3931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.059233449477352e-05,
+      "loss": 0.3452,
+      "step": 3932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.0635888501742164e-05,
+      "loss": 0.339,
+      "step": 3933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.06794425087108e-05,
+      "loss": 0.2929,
+      "step": 3934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.0722996515679444e-05,
+      "loss": 0.3523,
+      "step": 3935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.076655052264808e-05,
+      "loss": 0.3373,
+      "step": 3936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.081010452961673e-05,
+      "loss": 0.3363,
+      "step": 3937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.085365853658537e-05,
+      "loss": 0.3326,
+      "step": 3938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.089721254355401e-05,
+      "loss": 0.3223,
+      "step": 3939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.0940766550522653e-05,
+      "loss": 0.319,
+      "step": 3940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.098432055749129e-05,
+      "loss": 0.3125,
+      "step": 3941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.102787456445993e-05,
+      "loss": 0.3237,
+      "step": 3942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 0.3255,
+      "step": 3943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.111498257839721e-05,
+      "loss": 0.3081,
+      "step": 3944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1158536585365856e-05,
+      "loss": 0.3369,
+      "step": 3945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.12020905923345e-05,
+      "loss": 0.331,
+      "step": 3946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.124564459930314e-05,
+      "loss": 0.3197,
+      "step": 3947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.128919860627178e-05,
+      "loss": 0.3376,
+      "step": 3948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.133275261324042e-05,
+      "loss": 0.3209,
+      "step": 3949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.137630662020906e-05,
+      "loss": 0.3222,
+      "step": 3950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.14198606271777e-05,
+      "loss": 0.3203,
+      "step": 3951
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.146341463414634e-05,
+      "loss": 0.3219,
+      "step": 3952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.150696864111498e-05,
+      "loss": 0.3454,
+      "step": 3953
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1550522648083626e-05,
+      "loss": 0.3288,
+      "step": 3954
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.159407665505227e-05,
+      "loss": 0.3271,
+      "step": 3955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.163763066202091e-05,
+      "loss": 0.3362,
+      "step": 3956
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.168118466898955e-05,
+      "loss": 0.3145,
+      "step": 3957
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.172473867595819e-05,
+      "loss": 0.3273,
+      "step": 3958
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.176829268292683e-05,
+      "loss": 0.3306,
+      "step": 3959
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.181184668989547e-05,
+      "loss": 0.3233,
+      "step": 3960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.185540069686411e-05,
+      "loss": 0.3232,
+      "step": 3961
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.189895470383275e-05,
+      "loss": 0.3352,
+      "step": 3962
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1942508710801395e-05,
+      "loss": 0.3288,
+      "step": 3963
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.198606271777004e-05,
+      "loss": 0.3441,
+      "step": 3964
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.202961672473868e-05,
+      "loss": 0.3525,
+      "step": 3965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.207317073170732e-05,
+      "loss": 0.323,
+      "step": 3966
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.211672473867596e-05,
+      "loss": 0.3401,
+      "step": 3967
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.21602787456446e-05,
+      "loss": 0.3198,
+      "step": 3968
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.220383275261324e-05,
+      "loss": 0.331,
+      "step": 3969
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2247386759581884e-05,
+      "loss": 0.3346,
+      "step": 3970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.229094076655053e-05,
+      "loss": 0.324,
+      "step": 3971
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.233449477351917e-05,
+      "loss": 0.3215,
+      "step": 3972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.237804878048781e-05,
+      "loss": 0.3254,
+      "step": 3973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.242160278745645e-05,
+      "loss": 0.3317,
+      "step": 3974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.246515679442509e-05,
+      "loss": 0.3193,
+      "step": 3975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.250871080139373e-05,
+      "loss": 0.3301,
+      "step": 3976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.255226480836237e-05,
+      "loss": 0.336,
+      "step": 3977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.259581881533101e-05,
+      "loss": 0.3314,
+      "step": 3978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2639372822299653e-05,
+      "loss": 0.3314,
+      "step": 3979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.26829268292683e-05,
+      "loss": 0.3284,
+      "step": 3980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.272648083623694e-05,
+      "loss": 0.3259,
+      "step": 3981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2770034843205577e-05,
+      "loss": 0.3168,
+      "step": 3982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.281358885017422e-05,
+      "loss": 0.3313,
+      "step": 3983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.3381,
+      "step": 3984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.29006968641115e-05,
+      "loss": 0.3316,
+      "step": 3985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.294425087108014e-05,
+      "loss": 0.3646,
+      "step": 3986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.298780487804878e-05,
+      "loss": 0.3164,
+      "step": 3987
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.303135888501742e-05,
+      "loss": 0.3225,
+      "step": 3988
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3074912891986066e-05,
+      "loss": 0.3317,
+      "step": 3989
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311846689895471e-05,
+      "loss": 0.3296,
+      "step": 3990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3162020905923346e-05,
+      "loss": 0.3297,
+      "step": 3991
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320557491289199e-05,
+      "loss": 0.334,
+      "step": 3992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324912891986063e-05,
+      "loss": 0.3304,
+      "step": 3993
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.329268292682927e-05,
+      "loss": 0.3237,
+      "step": 3994
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333623693379791e-05,
+      "loss": 0.3407,
+      "step": 3995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337979094076655e-05,
+      "loss": 0.3384,
+      "step": 3996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342334494773519e-05,
+      "loss": 0.3382,
+      "step": 3997
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3466898954703835e-05,
+      "loss": 0.3331,
+      "step": 3998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351045296167248e-05,
+      "loss": 0.3419,
+      "step": 3999
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3554006968641115e-05,
+      "loss": 0.3391,
+      "step": 4000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359756097560976e-05,
+      "loss": 0.3144,
+      "step": 4001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36411149825784e-05,
+      "loss": 0.3256,
+      "step": 4002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368466898954704e-05,
+      "loss": 0.326,
+      "step": 4003
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372822299651568e-05,
+      "loss": 0.3471,
+      "step": 4004
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377177700348432e-05,
+      "loss": 0.3517,
+      "step": 4005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381533101045297e-05,
+      "loss": 0.3148,
+      "step": 4006
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3858885017421604e-05,
+      "loss": 0.3328,
+      "step": 4007
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390243902439025e-05,
+      "loss": 0.3261,
+      "step": 4008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394599303135889e-05,
+      "loss": 0.3314,
+      "step": 4009
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398954703832753e-05,
+      "loss": 0.3258,
+      "step": 4010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403310104529617e-05,
+      "loss": 0.3368,
+      "step": 4011
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.407665505226481e-05,
+      "loss": 0.3205,
+      "step": 4012
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412020905923345e-05,
+      "loss": 0.3298,
+      "step": 4013
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4163763066202094e-05,
+      "loss": 0.3443,
+      "step": 4014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420731707317074e-05,
+      "loss": 0.3274,
+      "step": 4015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4250871080139374e-05,
+      "loss": 0.3083,
+      "step": 4016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429442508710802e-05,
+      "loss": 0.336,
+      "step": 4017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433797909407666e-05,
+      "loss": 0.3229,
+      "step": 4018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43815331010453e-05,
+      "loss": 0.315,
+      "step": 4019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442508710801394e-05,
+      "loss": 0.3278,
+      "step": 4020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4468641114982576e-05,
+      "loss": 0.3413,
+      "step": 4021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.451219512195122e-05,
+      "loss": 0.3398,
+      "step": 4022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.455574912891986e-05,
+      "loss": 0.3236,
+      "step": 4023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4599303135888506e-05,
+      "loss": 0.3475,
+      "step": 4024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 0.3471,
+      "step": 4025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4686411149825786e-05,
+      "loss": 0.3263,
+      "step": 4026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472996515679443e-05,
+      "loss": 0.3356,
+      "step": 4027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4773519163763066e-05,
+      "loss": 0.3381,
+      "step": 4028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481707317073171e-05,
+      "loss": 0.3559,
+      "step": 4029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4860627177700346e-05,
+      "loss": 0.3198,
+      "step": 4030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.490418118466899e-05,
+      "loss": 0.3329,
+      "step": 4031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.494773519163763e-05,
+      "loss": 0.3269,
+      "step": 4032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4991289198606276e-05,
+      "loss": 0.3228,
+      "step": 4033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.503484320557492e-05,
+      "loss": 0.3223,
+      "step": 4034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5078397212543555e-05,
+      "loss": 0.3247,
+      "step": 4035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.51219512195122e-05,
+      "loss": 0.3465,
+      "step": 4036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5165505226480835e-05,
+      "loss": 0.3262,
+      "step": 4037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.520905923344948e-05,
+      "loss": 0.3264,
+      "step": 4038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5252613240418115e-05,
+      "loss": 0.3333,
+      "step": 4039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.529616724738676e-05,
+      "loss": 0.3415,
+      "step": 4040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.533972125435541e-05,
+      "loss": 0.3417,
+      "step": 4041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5383275261324045e-05,
+      "loss": 0.3418,
+      "step": 4042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.542682926829269e-05,
+      "loss": 0.3531,
+      "step": 4043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5470383275261325e-05,
+      "loss": 0.3183,
+      "step": 4044
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.551393728222997e-05,
+      "loss": 0.3309,
+      "step": 4045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5557491289198604e-05,
+      "loss": 0.3382,
+      "step": 4046
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.560104529616725e-05,
+      "loss": 0.3346,
+      "step": 4047
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.564459930313589e-05,
+      "loss": 0.3198,
+      "step": 4048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5688153310104534e-05,
+      "loss": 0.3476,
+      "step": 4049
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.573170731707318e-05,
+      "loss": 0.3425,
+      "step": 4050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5775261324041814e-05,
+      "loss": 0.3234,
+      "step": 4051
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.581881533101046e-05,
+      "loss": 0.3252,
+      "step": 4052
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5862369337979094e-05,
+      "loss": 0.3513,
+      "step": 4053
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.590592334494774e-05,
+      "loss": 0.3247,
+      "step": 4054
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.594947735191638e-05,
+      "loss": 0.3469,
+      "step": 4055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.599303135888502e-05,
+      "loss": 0.3297,
+      "step": 4056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.603658536585366e-05,
+      "loss": 0.3519,
+      "step": 4057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6080139372822303e-05,
+      "loss": 0.3231,
+      "step": 4058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.612369337979095e-05,
+      "loss": 0.3093,
+      "step": 4059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.616724738675958e-05,
+      "loss": 0.3261,
+      "step": 4060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6210801393728226e-05,
+      "loss": 0.3324,
+      "step": 4061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.625435540069686e-05,
+      "loss": 0.3345,
+      "step": 4062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6297909407665506e-05,
+      "loss": 0.3289,
+      "step": 4063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.634146341463415e-05,
+      "loss": 0.332,
+      "step": 4064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6385017421602786e-05,
+      "loss": 0.3233,
+      "step": 4065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 0.328,
+      "step": 4066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.647212543554007e-05,
+      "loss": 0.3477,
+      "step": 4067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6515679442508716e-05,
+      "loss": 0.3257,
+      "step": 4068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.655923344947735e-05,
+      "loss": 0.3282,
+      "step": 4069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6602787456445996e-05,
+      "loss": 0.3342,
+      "step": 4070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.664634146341464e-05,
+      "loss": 0.345,
+      "step": 4071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6689895470383275e-05,
+      "loss": 0.3246,
+      "step": 4072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.673344947735192e-05,
+      "loss": 0.3338,
+      "step": 4073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6777003484320555e-05,
+      "loss": 0.2951,
+      "step": 4074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6820557491289205e-05,
+      "loss": 0.3425,
+      "step": 4075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.686411149825784e-05,
+      "loss": 0.3218,
+      "step": 4076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6907665505226485e-05,
+      "loss": 0.3171,
+      "step": 4077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.695121951219512e-05,
+      "loss": 0.33,
+      "step": 4078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6994773519163765e-05,
+      "loss": 0.3226,
+      "step": 4079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.703832752613241e-05,
+      "loss": 0.3134,
+      "step": 4080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7081881533101045e-05,
+      "loss": 0.3445,
+      "step": 4081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.712543554006969e-05,
+      "loss": 0.3329,
+      "step": 4082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.716898954703833e-05,
+      "loss": 0.3286,
+      "step": 4083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7212543554006975e-05,
+      "loss": 0.318,
+      "step": 4084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.725609756097561e-05,
+      "loss": 0.3289,
+      "step": 4085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7299651567944254e-05,
+      "loss": 0.3279,
+      "step": 4086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.73432055749129e-05,
+      "loss": 0.3314,
+      "step": 4087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7386759581881534e-05,
+      "loss": 0.3235,
+      "step": 4088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.743031358885018e-05,
+      "loss": 0.3353,
+      "step": 4089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7473867595818814e-05,
+      "loss": 0.3317,
+      "step": 4090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.751742160278746e-05,
+      "loss": 0.3314,
+      "step": 4091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.75609756097561e-05,
+      "loss": 0.3296,
+      "step": 4092
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7604529616724744e-05,
+      "loss": 0.3375,
+      "step": 4093
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.764808362369339e-05,
+      "loss": 0.363,
+      "step": 4094
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7691637630662024e-05,
+      "loss": 0.3355,
+      "step": 4095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.773519163763067e-05,
+      "loss": 0.3162,
+      "step": 4096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.77787456445993e-05,
+      "loss": 0.3246,
+      "step": 4097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.782229965156795e-05,
+      "loss": 0.3347,
+      "step": 4098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.786585365853658e-05,
+      "loss": 0.3373,
+      "step": 4099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7909407665505226e-05,
+      "loss": 0.3341,
+      "step": 4100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.795296167247387e-05,
+      "loss": 0.3288,
+      "step": 4101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.799651567944251e-05,
+      "loss": 0.3239,
+      "step": 4102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8040069686411156e-05,
+      "loss": 0.3309,
+      "step": 4103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.808362369337979e-05,
+      "loss": 0.3375,
+      "step": 4104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8127177700348436e-05,
+      "loss": 0.3254,
+      "step": 4105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.817073170731707e-05,
+      "loss": 0.3333,
+      "step": 4106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 0.3092,
+      "step": 4107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.825783972125435e-05,
+      "loss": 0.3324,
+      "step": 4108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8301393728222996e-05,
+      "loss": 0.3297,
+      "step": 4109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8344947735191646e-05,
+      "loss": 0.3456,
+      "step": 4110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.838850174216028e-05,
+      "loss": 0.3394,
+      "step": 4111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8432055749128926e-05,
+      "loss": 0.33,
+      "step": 4112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.847560975609756e-05,
+      "loss": 0.3392,
+      "step": 4113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8519163763066205e-05,
+      "loss": 0.3213,
+      "step": 4114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.856271777003484e-05,
+      "loss": 0.3188,
+      "step": 4115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8606271777003485e-05,
+      "loss": 0.3203,
+      "step": 4116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.864982578397213e-05,
+      "loss": 0.3262,
+      "step": 4117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.869337979094077e-05,
+      "loss": 0.3226,
+      "step": 4118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8736933797909415e-05,
+      "loss": 0.3438,
+      "step": 4119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.878048780487805e-05,
+      "loss": 0.3162,
+      "step": 4120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8824041811846695e-05,
+      "loss": 0.3381,
+      "step": 4121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.886759581881533e-05,
+      "loss": 0.3253,
+      "step": 4122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8911149825783975e-05,
+      "loss": 0.3186,
+      "step": 4123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.895470383275261e-05,
+      "loss": 0.3213,
+      "step": 4124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8998257839721254e-05,
+      "loss": 0.3387,
+      "step": 4125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.90418118466899e-05,
+      "loss": 0.3235,
+      "step": 4126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.908536585365854e-05,
+      "loss": 0.3379,
+      "step": 4127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9128919860627184e-05,
+      "loss": 0.3377,
+      "step": 4128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.917247386759582e-05,
+      "loss": 0.353,
+      "step": 4129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9216027874564464e-05,
+      "loss": 0.3131,
+      "step": 4130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.92595818815331e-05,
+      "loss": 0.3186,
+      "step": 4131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9303135888501744e-05,
+      "loss": 0.3459,
+      "step": 4132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.934668989547039e-05,
+      "loss": 0.3462,
+      "step": 4133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9390243902439024e-05,
+      "loss": 0.3327,
+      "step": 4134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.943379790940767e-05,
+      "loss": 0.3336,
+      "step": 4135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.947735191637631e-05,
+      "loss": 0.3271,
+      "step": 4136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.952090592334495e-05,
+      "loss": 0.3412,
+      "step": 4137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.956445993031359e-05,
+      "loss": 0.3135,
+      "step": 4138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.960801393728223e-05,
+      "loss": 0.3098,
+      "step": 4139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.965156794425087e-05,
+      "loss": 0.3279,
+      "step": 4140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.969512195121951e-05,
+      "loss": 0.3306,
+      "step": 4141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9738675958188156e-05,
+      "loss": 0.3187,
+      "step": 4142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.978222996515679e-05,
+      "loss": 0.3401,
+      "step": 4143
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9825783972125436e-05,
+      "loss": 0.3364,
+      "step": 4144
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.986933797909408e-05,
+      "loss": 0.3256,
+      "step": 4145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.991289198606272e-05,
+      "loss": 0.3541,
+      "step": 4146
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.995644599303136e-05,
+      "loss": 0.3121,
+      "step": 4147
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5e-05,
+      "loss": 0.3335,
+      "step": 4148
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9999998958670857e-05,
+      "loss": 0.3349,
+      "step": 4149
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.999999583468353e-05,
+      "loss": 0.3262,
+      "step": 4150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.999999062803829e-05,
+      "loss": 0.3406,
+      "step": 4151
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9999983338735634e-05,
+      "loss": 0.3149,
+      "step": 4152
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9999973966776235e-05,
+      "loss": 0.3271,
+      "step": 4153
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.999996251216096e-05,
+      "loss": 0.3224,
+      "step": 4154
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.999994897489086e-05,
+      "loss": 0.3314,
+      "step": 4155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.999993335496721e-05,
+      "loss": 0.3327,
+      "step": 4156
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9999915652391433e-05,
+      "loss": 0.3336,
+      "step": 4157
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9999895867165174e-05,
+      "loss": 0.3319,
+      "step": 4158
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.999987399929027e-05,
+      "loss": 0.3306,
+      "step": 4159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999985004876875e-05,
+      "loss": 0.3344,
+      "step": 4160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999982401560282e-05,
+      "loss": 0.3311,
+      "step": 4161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999979589979489e-05,
+      "loss": 0.3326,
+      "step": 4162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9999765701347556e-05,
+      "loss": 0.3315,
+      "step": 4163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9999733420263635e-05,
+      "loss": 0.3384,
+      "step": 4164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9999699056546096e-05,
+      "loss": 0.3371,
+      "step": 4165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999966261019813e-05,
+      "loss": 0.3275,
+      "step": 4166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.99996240812231e-05,
+      "loss": 0.3157,
+      "step": 4167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9999583469624585e-05,
+      "loss": 0.3181,
+      "step": 4168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999954077540634e-05,
+      "loss": 0.3302,
+      "step": 4169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999949599857231e-05,
+      "loss": 0.3266,
+      "step": 4170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999944913912665e-05,
+      "loss": 0.3333,
+      "step": 4171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999940019707369e-05,
+      "loss": 0.3497,
+      "step": 4172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9999349172417956e-05,
+      "loss": 0.3246,
+      "step": 4173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9999296065164186e-05,
+      "loss": 0.3513,
+      "step": 4174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9999240875317285e-05,
+      "loss": 0.3287,
+      "step": 4175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999918360288237e-05,
+      "loss": 0.3269,
+      "step": 4176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999912424786472e-05,
+      "loss": 0.3397,
+      "step": 4177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999906281026985e-05,
+      "loss": 0.3442,
+      "step": 4178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999899929010345e-05,
+      "loss": 0.322,
+      "step": 4179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999893368737139e-05,
+      "loss": 0.3321,
+      "step": 4180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999886600207974e-05,
+      "loss": 0.336,
+      "step": 4181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999879623423476e-05,
+      "loss": 0.3381,
+      "step": 4182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999872438384293e-05,
+      "loss": 0.324,
+      "step": 4183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999865045091089e-05,
+      "loss": 0.339,
+      "step": 4184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999857443544547e-05,
+      "loss": 0.3459,
+      "step": 4185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9998496337453726e-05,
+      "loss": 0.3222,
+      "step": 4186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999841615694287e-05,
+      "loss": 0.3312,
+      "step": 4187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9998333893920336e-05,
+      "loss": 0.3487,
+      "step": 4188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999824954839374e-05,
+      "loss": 0.3267,
+      "step": 4189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9998163120370874e-05,
+      "loss": 0.3528,
+      "step": 4190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999807460985975e-05,
+      "loss": 0.3502,
+      "step": 4191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999798401686856e-05,
+      "loss": 0.3223,
+      "step": 4192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999789134140569e-05,
+      "loss": 0.3432,
+      "step": 4193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.99977965834797e-05,
+      "loss": 0.3273,
+      "step": 4194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.99976997430994e-05,
+      "loss": 0.3236,
+      "step": 4195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999760082027372e-05,
+      "loss": 0.3352,
+      "step": 4196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999749981501183e-05,
+      "loss": 0.329,
+      "step": 4197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999739672732307e-05,
+      "loss": 0.3473,
+      "step": 4198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999729155721699e-05,
+      "loss": 0.3238,
+      "step": 4199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999718430470333e-05,
+      "loss": 0.3176,
+      "step": 4200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999707496979201e-05,
+      "loss": 0.3392,
+      "step": 4201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9996963552493145e-05,
+      "loss": 0.3152,
+      "step": 4202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9996850052817064e-05,
+      "loss": 0.3258,
+      "step": 4203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999673447077425e-05,
+      "loss": 0.3211,
+      "step": 4204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999661680637543e-05,
+      "loss": 0.3339,
+      "step": 4205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999649705963148e-05,
+      "loss": 0.3588,
+      "step": 4206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999637523055348e-05,
+      "loss": 0.3458,
+      "step": 4207
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9996251319152704e-05,
+      "loss": 0.312,
+      "step": 4208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999612532544063e-05,
+      "loss": 0.3088,
+      "step": 4209
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999599724942893e-05,
+      "loss": 0.3308,
+      "step": 4210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999586709112944e-05,
+      "loss": 0.3297,
+      "step": 4211
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999573485055422e-05,
+      "loss": 0.3311,
+      "step": 4212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.99956005277155e-05,
+      "loss": 0.3598,
+      "step": 4213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9995464122625726e-05,
+      "loss": 0.3289,
+      "step": 4214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9995325635297506e-05,
+      "loss": 0.3413,
+      "step": 4215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.999518506574368e-05,
+      "loss": 0.3194,
+      "step": 4216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999504241397725e-05,
+      "loss": 0.3326,
+      "step": 4217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9994897680011416e-05,
+      "loss": 0.341,
+      "step": 4218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999475086385958e-05,
+      "loss": 0.3465,
+      "step": 4219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9994601965535344e-05,
+      "loss": 0.3089,
+      "step": 4220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999445098505247e-05,
+      "loss": 0.3517,
+      "step": 4221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9994297922424926e-05,
+      "loss": 0.3219,
+      "step": 4222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999414277766691e-05,
+      "loss": 0.3418,
+      "step": 4223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999398555079276e-05,
+      "loss": 0.3334,
+      "step": 4224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999382624181704e-05,
+      "loss": 0.3347,
+      "step": 4225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999366485075448e-05,
+      "loss": 0.332,
+      "step": 4226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999350137762005e-05,
+      "loss": 0.3383,
+      "step": 4227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9993335822428846e-05,
+      "loss": 0.3537,
+      "step": 4228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999316818519621e-05,
+      "loss": 0.3317,
+      "step": 4229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9992998465937674e-05,
+      "loss": 0.3236,
+      "step": 4230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999282666466892e-05,
+      "loss": 0.3217,
+      "step": 4231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999265278140586e-05,
+      "loss": 0.3322,
+      "step": 4232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9992476816164594e-05,
+      "loss": 0.3341,
+      "step": 4233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9992298768961406e-05,
+      "loss": 0.3385,
+      "step": 4234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9992118639812775e-05,
+      "loss": 0.3383,
+      "step": 4235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999193642873537e-05,
+      "loss": 0.3242,
+      "step": 4236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9991752135746076e-05,
+      "loss": 0.3398,
+      "step": 4237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9991565760861936e-05,
+      "loss": 0.3211,
+      "step": 4238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9991377304100204e-05,
+      "loss": 0.3039,
+      "step": 4239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999118676547832e-05,
+      "loss": 0.3496,
+      "step": 4240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9990994145013924e-05,
+      "loss": 0.336,
+      "step": 4241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9990799442724856e-05,
+      "loss": 0.3226,
+      "step": 4242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999060265862912e-05,
+      "loss": 0.3347,
+      "step": 4243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999040379274494e-05,
+      "loss": 0.3252,
+      "step": 4244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.999020284509072e-05,
+      "loss": 0.3212,
+      "step": 4245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9989999815685076e-05,
+      "loss": 0.3339,
+      "step": 4246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998979470454677e-05,
+      "loss": 0.3283,
+      "step": 4247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998958751169482e-05,
+      "loss": 0.3292,
+      "step": 4248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998937823714838e-05,
+      "loss": 0.3293,
+      "step": 4249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9989166880926834e-05,
+      "loss": 0.342,
+      "step": 4250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9988953443049755e-05,
+      "loss": 0.3295,
+      "step": 4251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9988737923536874e-05,
+      "loss": 0.3316,
+      "step": 4252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9988520322408155e-05,
+      "loss": 0.3158,
+      "step": 4253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998830063968374e-05,
+      "loss": 0.33,
+      "step": 4254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9988078875383956e-05,
+      "loss": 0.3329,
+      "step": 4255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998785502952934e-05,
+      "loss": 0.3251,
+      "step": 4256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998762910214061e-05,
+      "loss": 0.3396,
+      "step": 4257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998740109323867e-05,
+      "loss": 0.3224,
+      "step": 4258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998717100284463e-05,
+      "loss": 0.3463,
+      "step": 4259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998693883097979e-05,
+      "loss": 0.3317,
+      "step": 4260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9986704577665646e-05,
+      "loss": 0.3201,
+      "step": 4261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9986468242923864e-05,
+      "loss": 0.3309,
+      "step": 4262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998622982677633e-05,
+      "loss": 0.3211,
+      "step": 4263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998598932924511e-05,
+      "loss": 0.3245,
+      "step": 4264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998574675035248e-05,
+      "loss": 0.3179,
+      "step": 4265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9985502090120867e-05,
+      "loss": 0.3251,
+      "step": 4266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998525534857294e-05,
+      "loss": 0.3314,
+      "step": 4267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9985006525731524e-05,
+      "loss": 0.3142,
+      "step": 4268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998475562161966e-05,
+      "loss": 0.3395,
+      "step": 4269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998450263626056e-05,
+      "loss": 0.3316,
+      "step": 4270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998424756967766e-05,
+      "loss": 0.3388,
+      "step": 4271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9983990421894555e-05,
+      "loss": 0.3068,
+      "step": 4272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998373119293505e-05,
+      "loss": 0.3417,
+      "step": 4273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.998346988282314e-05,
+      "loss": 0.3336,
+      "step": 4274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9983206491583014e-05,
+      "loss": 0.3189,
+      "step": 4275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.998294101923906e-05,
+      "loss": 0.3319,
+      "step": 4276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9982673465815826e-05,
+      "loss": 0.3143,
+      "step": 4277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9982403831338104e-05,
+      "loss": 0.32,
+      "step": 4278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.998213211583084e-05,
+      "loss": 0.311,
+      "step": 4279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.998185831931919e-05,
+      "loss": 0.3068,
+      "step": 4280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.998158244182849e-05,
+      "loss": 0.3275,
+      "step": 4281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.998130448338428e-05,
+      "loss": 0.3001,
+      "step": 4282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.99810244440123e-05,
+      "loss": 0.3267,
+      "step": 4283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.998074232373845e-05,
+      "loss": 0.3391,
+      "step": 4284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9980458122588855e-05,
+      "loss": 0.3136,
+      "step": 4285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.998017184058982e-05,
+      "loss": 0.3146,
+      "step": 4286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9979883477767835e-05,
+      "loss": 0.3336,
+      "step": 4287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9979593034149616e-05,
+      "loss": 0.3381,
+      "step": 4288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9979300509762017e-05,
+      "loss": 0.3313,
+      "step": 4289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9979005904632134e-05,
+      "loss": 0.3276,
+      "step": 4290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9978709218787236e-05,
+      "loss": 0.3044,
+      "step": 4291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997841045225478e-05,
+      "loss": 0.3361,
+      "step": 4292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997810960506242e-05,
+      "loss": 0.3425,
+      "step": 4293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997780667723801e-05,
+      "loss": 0.3297,
+      "step": 4294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997750166880958e-05,
+      "loss": 0.3174,
+      "step": 4295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997719457980537e-05,
+      "loss": 0.3075,
+      "step": 4296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.99768854102538e-05,
+      "loss": 0.3219,
+      "step": 4297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9976574160183486e-05,
+      "loss": 0.3324,
+      "step": 4298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9976260829623244e-05,
+      "loss": 0.3304,
+      "step": 4299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997594541860207e-05,
+      "loss": 0.3244,
+      "step": 4300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9975627927149176e-05,
+      "loss": 0.3322,
+      "step": 4301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9975308355293924e-05,
+      "loss": 0.3427,
+      "step": 4302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997498670306591e-05,
+      "loss": 0.3157,
+      "step": 4303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9974662970494915e-05,
+      "loss": 0.3523,
+      "step": 4304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9974337157610885e-05,
+      "loss": 0.3113,
+      "step": 4305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997400926444399e-05,
+      "loss": 0.3396,
+      "step": 4306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997367929102458e-05,
+      "loss": 0.3201,
+      "step": 4307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997334723738319e-05,
+      "loss": 0.324,
+      "step": 4308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9973013103550564e-05,
+      "loss": 0.3281,
+      "step": 4309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997267688955763e-05,
+      "loss": 0.3371,
+      "step": 4310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9972338595435505e-05,
+      "loss": 0.3164,
+      "step": 4311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997199822121551e-05,
+      "loss": 0.3274,
+      "step": 4312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9971655766929134e-05,
+      "loss": 0.3433,
+      "step": 4313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.99713112326081e-05,
+      "loss": 0.3245,
+      "step": 4314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9970964618284274e-05,
+      "loss": 0.3321,
+      "step": 4315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997061592398975e-05,
+      "loss": 0.3265,
+      "step": 4316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997026514975682e-05,
+      "loss": 0.3381,
+      "step": 4317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9969912295617924e-05,
+      "loss": 0.3473,
+      "step": 4318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9969557361605744e-05,
+      "loss": 0.3216,
+      "step": 4319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9969200347753126e-05,
+      "loss": 0.3315,
+      "step": 4320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.996884125409312e-05,
+      "loss": 0.3408,
+      "step": 4321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9968480080658956e-05,
+      "loss": 0.3157,
+      "step": 4322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.996811682748407e-05,
+      "loss": 0.3524,
+      "step": 4323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.996775149460209e-05,
+      "loss": 0.3381,
+      "step": 4324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9967384082046825e-05,
+      "loss": 0.3274,
+      "step": 4325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.996701458985229e-05,
+      "loss": 0.3234,
+      "step": 4326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9966643018052686e-05,
+      "loss": 0.32,
+      "step": 4327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.99662693666824e-05,
+      "loss": 0.341,
+      "step": 4328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9965893635776015e-05,
+      "loss": 0.3304,
+      "step": 4329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.996551582536832e-05,
+      "loss": 0.3354,
+      "step": 4330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9965135935494286e-05,
+      "loss": 0.3145,
+      "step": 4331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9964753966189075e-05,
+      "loss": 0.3087,
+      "step": 4332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.996436991748804e-05,
+      "loss": 0.3129,
+      "step": 4333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.996398378942673e-05,
+      "loss": 0.3388,
+      "step": 4334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.996359558204088e-05,
+      "loss": 0.3294,
+      "step": 4335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.996320529536644e-05,
+      "loss": 0.344,
+      "step": 4336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.996281292943953e-05,
+      "loss": 0.3087,
+      "step": 4337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.996241848429645e-05,
+      "loss": 0.3248,
+      "step": 4338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9962021959973736e-05,
+      "loss": 0.3533,
+      "step": 4339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.996162335650808e-05,
+      "loss": 0.3278,
+      "step": 4340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9961222673936376e-05,
+      "loss": 0.3207,
+      "step": 4341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.996081991229572e-05,
+      "loss": 0.3322,
+      "step": 4342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.996041507162338e-05,
+      "loss": 0.3163,
+      "step": 4343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.996000815195684e-05,
+      "loss": 0.3236,
+      "step": 4344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995959915333376e-05,
+      "loss": 0.3251,
+      "step": 4345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9959188075792e-05,
+      "loss": 0.317,
+      "step": 4346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9958774919369624e-05,
+      "loss": 0.3287,
+      "step": 4347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9958359684104846e-05,
+      "loss": 0.3452,
+      "step": 4348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9957942370036114e-05,
+      "loss": 0.3236,
+      "step": 4349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995752297720207e-05,
+      "loss": 0.3083,
+      "step": 4350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995710150564152e-05,
+      "loss": 0.3359,
+      "step": 4351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995667795539347e-05,
+      "loss": 0.3554,
+      "step": 4352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995625232649714e-05,
+      "loss": 0.344,
+      "step": 4353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9955824618991924e-05,
+      "loss": 0.3264,
+      "step": 4354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9955394832917404e-05,
+      "loss": 0.3194,
+      "step": 4355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995496296831337e-05,
+      "loss": 0.3508,
+      "step": 4356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995452902521979e-05,
+      "loss": 0.3428,
+      "step": 4357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995409300367684e-05,
+      "loss": 0.3134,
+      "step": 4358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9953654903724866e-05,
+      "loss": 0.3426,
+      "step": 4359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995321472540444e-05,
+      "loss": 0.3221,
+      "step": 4360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995277246875628e-05,
+      "loss": 0.3563,
+      "step": 4361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995232813382134e-05,
+      "loss": 0.325,
+      "step": 4362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995188172064075e-05,
+      "loss": 0.327,
+      "step": 4363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9951433229255824e-05,
+      "loss": 0.3123,
+      "step": 4364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995098265970807e-05,
+      "loss": 0.3235,
+      "step": 4365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.99505300120392e-05,
+      "loss": 0.348,
+      "step": 4366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9950075286291125e-05,
+      "loss": 0.3318,
+      "step": 4367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994961848250592e-05,
+      "loss": 0.3256,
+      "step": 4368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9949159600725863e-05,
+      "loss": 0.3206,
+      "step": 4369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994869864099345e-05,
+      "loss": 0.3126,
+      "step": 4370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9948235603351326e-05,
+      "loss": 0.3451,
+      "step": 4371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9947770487842366e-05,
+      "loss": 0.3434,
+      "step": 4372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9947303294509626e-05,
+      "loss": 0.333,
+      "step": 4373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994683402339633e-05,
+      "loss": 0.3397,
+      "step": 4374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994636267454593e-05,
+      "loss": 0.309,
+      "step": 4375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994588924800206e-05,
+      "loss": 0.3426,
+      "step": 4376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994541374380853e-05,
+      "loss": 0.3221,
+      "step": 4377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9944936162009365e-05,
+      "loss": 0.332,
+      "step": 4378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9944456502648754e-05,
+      "loss": 0.3426,
+      "step": 4379
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994397476577111e-05,
+      "loss": 0.3099,
+      "step": 4380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994349095142102e-05,
+      "loss": 0.3451,
+      "step": 4381
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994300505964327e-05,
+      "loss": 0.3259,
+      "step": 4382
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994251709048283e-05,
+      "loss": 0.3359,
+      "step": 4383
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9942027043984874e-05,
+      "loss": 0.3368,
+      "step": 4384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9941534920194757e-05,
+      "loss": 0.3182,
+      "step": 4385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.994104071915803e-05,
+      "loss": 0.338,
+      "step": 4386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9940544440920434e-05,
+      "loss": 0.333,
+      "step": 4387
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9940046085527924e-05,
+      "loss": 0.3233,
+      "step": 4388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9939545653026605e-05,
+      "loss": 0.327,
+      "step": 4389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993904314346282e-05,
+      "loss": 0.3137,
+      "step": 4390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9938538556883065e-05,
+      "loss": 0.3234,
+      "step": 4391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993803189333406e-05,
+      "loss": 0.3272,
+      "step": 4392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993752315286269e-05,
+      "loss": 0.3206,
+      "step": 4393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993701233551606e-05,
+      "loss": 0.337,
+      "step": 4394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993649944134144e-05,
+      "loss": 0.3553,
+      "step": 4395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9935984470386304e-05,
+      "loss": 0.3175,
+      "step": 4396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993546742269833e-05,
+      "loss": 0.3368,
+      "step": 4397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993494829832537e-05,
+      "loss": 0.3167,
+      "step": 4398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993442709731548e-05,
+      "loss": 0.3253,
+      "step": 4399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.99339038197169e-05,
+      "loss": 0.3147,
+      "step": 4400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993337846557806e-05,
+      "loss": 0.332,
+      "step": 4401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.99328510349476e-05,
+      "loss": 0.3297,
+      "step": 4402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993232152787434e-05,
+      "loss": 0.3135,
+      "step": 4403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9931789944407284e-05,
+      "loss": 0.3158,
+      "step": 4404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993125628459564e-05,
+      "loss": 0.3272,
+      "step": 4405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.99307205484888e-05,
+      "loss": 0.3269,
+      "step": 4406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993018273613637e-05,
+      "loss": 0.3284,
+      "step": 4407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992964284758812e-05,
+      "loss": 0.3232,
+      "step": 4408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992910088289401e-05,
+      "loss": 0.338,
+      "step": 4409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9928556842104226e-05,
+      "loss": 0.34,
+      "step": 4410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992801072526912e-05,
+      "loss": 0.3222,
+      "step": 4411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992746253243925e-05,
+      "loss": 0.3425,
+      "step": 4412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9926912263665335e-05,
+      "loss": 0.3409,
+      "step": 4413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992635991899832e-05,
+      "loss": 0.3124,
+      "step": 4414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992580549848935e-05,
+      "loss": 0.3106,
+      "step": 4415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992524900218972e-05,
+      "loss": 0.3316,
+      "step": 4416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992469043015094e-05,
+      "loss": 0.3127,
+      "step": 4417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992412978242474e-05,
+      "loss": 0.3453,
+      "step": 4418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9923567059062985e-05,
+      "loss": 0.3232,
+      "step": 4419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992300226011778e-05,
+      "loss": 0.3233,
+      "step": 4420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9922435385641406e-05,
+      "loss": 0.3543,
+      "step": 4421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992186643568631e-05,
+      "loss": 0.3281,
+      "step": 4422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992129541030518e-05,
+      "loss": 0.3304,
+      "step": 4423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.992072230955086e-05,
+      "loss": 0.3469,
+      "step": 4424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9920147133476406e-05,
+      "loss": 0.3228,
+      "step": 4425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9919569882135054e-05,
+      "loss": 0.3406,
+      "step": 4426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9918990555580234e-05,
+      "loss": 0.3154,
+      "step": 4427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.991840915386557e-05,
+      "loss": 0.3045,
+      "step": 4428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.991782567704488e-05,
+      "loss": 0.3307,
+      "step": 4429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.991724012517217e-05,
+      "loss": 0.3335,
+      "step": 4430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.991665249830164e-05,
+      "loss": 0.3292,
+      "step": 4431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.991606279648769e-05,
+      "loss": 0.3166,
+      "step": 4432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9915471019784895e-05,
+      "loss": 0.356,
+      "step": 4433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9914877168248034e-05,
+      "loss": 0.331,
+      "step": 4434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.991428124193207e-05,
+      "loss": 0.3384,
+      "step": 4435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9913683240892175e-05,
+      "loss": 0.3261,
+      "step": 4436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.99130831651837e-05,
+      "loss": 0.3149,
+      "step": 4437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9912481014862184e-05,
+      "loss": 0.3205,
+      "step": 4438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.991187678998336e-05,
+      "loss": 0.3171,
+      "step": 4439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9911270490603166e-05,
+      "loss": 0.3274,
+      "step": 4440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.991066211677772e-05,
+      "loss": 0.3279,
+      "step": 4441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.991005166856332e-05,
+      "loss": 0.3349,
+      "step": 4442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.990943914601649e-05,
+      "loss": 0.3513,
+      "step": 4443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9908824549193935e-05,
+      "loss": 0.3253,
+      "step": 4444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9908207878152514e-05,
+      "loss": 0.3298,
+      "step": 4445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.990758913294933e-05,
+      "loss": 0.3193,
+      "step": 4446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9906968313641644e-05,
+      "loss": 0.3294,
+      "step": 4447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.990634542028693e-05,
+      "loss": 0.3298,
+      "step": 4448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.990572045294283e-05,
+      "loss": 0.3145,
+      "step": 4449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9905093411667206e-05,
+      "loss": 0.3281,
+      "step": 4450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.99044642965181e-05,
+      "loss": 0.3093,
+      "step": 4451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9903833107553735e-05,
+      "loss": 0.3304,
+      "step": 4452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9903199844832536e-05,
+      "loss": 0.3156,
+      "step": 4453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9902564508413136e-05,
+      "loss": 0.3057,
+      "step": 4454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.990192709835432e-05,
+      "loss": 0.3264,
+      "step": 4455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.99012876147151e-05,
+      "loss": 0.3167,
+      "step": 4456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9900646057554676e-05,
+      "loss": 0.3749,
+      "step": 4457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9900002426932414e-05,
+      "loss": 0.3169,
+      "step": 4458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.98993567229079e-05,
+      "loss": 0.322,
+      "step": 4459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.98987089455409e-05,
+      "loss": 0.3305,
+      "step": 4460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9898059094891384e-05,
+      "loss": 0.312,
+      "step": 4461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.989740717101949e-05,
+      "loss": 0.3264,
+      "step": 4462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.989675317398556e-05,
+      "loss": 0.3262,
+      "step": 4463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9896097103850146e-05,
+      "loss": 0.3297,
+      "step": 4464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9895438960673965e-05,
+      "loss": 0.3442,
+      "step": 4465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.989477874451794e-05,
+      "loss": 0.3239,
+      "step": 4466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9894116455443176e-05,
+      "loss": 0.3382,
+      "step": 4467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9893452093510975e-05,
+      "loss": 0.3262,
+      "step": 4468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9892785658782846e-05,
+      "loss": 0.3354,
+      "step": 4469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.989211715132047e-05,
+      "loss": 0.3431,
+      "step": 4470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.989144657118572e-05,
+      "loss": 0.3331,
+      "step": 4471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9890773918440664e-05,
+      "loss": 0.331,
+      "step": 4472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.989009919314758e-05,
+      "loss": 0.3226,
+      "step": 4473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.98894223953689e-05,
+      "loss": 0.3337,
+      "step": 4474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.988874352516729e-05,
+      "loss": 0.3265,
+      "step": 4475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.988806258260558e-05,
+      "loss": 0.3274,
+      "step": 4476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.98873795677468e-05,
+      "loss": 0.3214,
+      "step": 4477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9886694480654174e-05,
+      "loss": 0.3191,
+      "step": 4478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.988600732139111e-05,
+      "loss": 0.328,
+      "step": 4479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.988531809002122e-05,
+      "loss": 0.3391,
+      "step": 4480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.98846267866083e-05,
+      "loss": 0.3218,
+      "step": 4481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9883933411216324e-05,
+      "loss": 0.327,
+      "step": 4482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.98832379639095e-05,
+      "loss": 0.3308,
+      "step": 4483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.988254044475218e-05,
+      "loss": 0.3368,
+      "step": 4484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.988184085380893e-05,
+      "loss": 0.3348,
+      "step": 4485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.988113919114451e-05,
+      "loss": 0.3423,
+      "step": 4486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9880435456823876e-05,
+      "loss": 0.3256,
+      "step": 4487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9879729650912154e-05,
+      "loss": 0.3213,
+      "step": 4488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.987902177347468e-05,
+      "loss": 0.3163,
+      "step": 4489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.987831182457698e-05,
+      "loss": 0.3431,
+      "step": 4490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9877599804284755e-05,
+      "loss": 0.3407,
+      "step": 4491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.987688571266393e-05,
+      "loss": 0.3338,
+      "step": 4492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.98761695497806e-05,
+      "loss": 0.3194,
+      "step": 4493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.987545131570105e-05,
+      "loss": 0.3239,
+      "step": 4494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9874731010491753e-05,
+      "loss": 0.3392,
+      "step": 4495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.98740086342194e-05,
+      "loss": 0.3155,
+      "step": 4496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.987328418695084e-05,
+      "loss": 0.3176,
+      "step": 4497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.987255766875315e-05,
+      "loss": 0.3183,
+      "step": 4498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9871829079693554e-05,
+      "loss": 0.3301,
+      "step": 4499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.987109841983951e-05,
+      "loss": 0.319,
+      "step": 4500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9870365689258635e-05,
+      "loss": 0.3362,
+      "step": 4501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.986963088801876e-05,
+      "loss": 0.3084,
+      "step": 4502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.986889401618791e-05,
+      "loss": 0.3291,
+      "step": 4503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9868155073834285e-05,
+      "loss": 0.3293,
+      "step": 4504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9867414061026265e-05,
+      "loss": 0.3403,
+      "step": 4505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.986667097783247e-05,
+      "loss": 0.3219,
+      "step": 4506
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.986592582432166e-05,
+      "loss": 0.3437,
+      "step": 4507
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.986517860056282e-05,
+      "loss": 0.3145,
+      "step": 4508
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.986442930662511e-05,
+      "loss": 0.3417,
+      "step": 4509
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.986367794257789e-05,
+      "loss": 0.3257,
+      "step": 4510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.98629245084907e-05,
+      "loss": 0.3252,
+      "step": 4511
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9862169004433296e-05,
+      "loss": 0.3196,
+      "step": 4512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.986141143047559e-05,
+      "loss": 0.3166,
+      "step": 4513
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.986065178668772e-05,
+      "loss": 0.3419,
+      "step": 4514
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.985989007313999e-05,
+      "loss": 0.336,
+      "step": 4515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.985912628990291e-05,
+      "loss": 0.32,
+      "step": 4516
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.985836043704718e-05,
+      "loss": 0.3292,
+      "step": 4517
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.985759251464369e-05,
+      "loss": 0.3142,
+      "step": 4518
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.985682252276352e-05,
+      "loss": 0.3145,
+      "step": 4519
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.985605046147794e-05,
+      "loss": 0.3513,
+      "step": 4520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.985527633085841e-05,
+      "loss": 0.335,
+      "step": 4521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.98545001309766e-05,
+      "loss": 0.3334,
+      "step": 4522
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.985372186190434e-05,
+      "loss": 0.342,
+      "step": 4523
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9852941523713685e-05,
+      "loss": 0.3161,
+      "step": 4524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.985215911647685e-05,
+      "loss": 0.3077,
+      "step": 4525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9851374640266266e-05,
+      "loss": 0.3169,
+      "step": 4526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9850588095154543e-05,
+      "loss": 0.3266,
+      "step": 4527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.984979948121448e-05,
+      "loss": 0.3139,
+      "step": 4528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9849008798519086e-05,
+      "loss": 0.3277,
+      "step": 4529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.984821604714154e-05,
+      "loss": 0.3234,
+      "step": 4530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9847421227155226e-05,
+      "loss": 0.3359,
+      "step": 4531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9846624338633707e-05,
+      "loss": 0.3211,
+      "step": 4532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.984582538165075e-05,
+      "loss": 0.3266,
+      "step": 4533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9845024356280306e-05,
+      "loss": 0.3162,
+      "step": 4534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9844221262596535e-05,
+      "loss": 0.3268,
+      "step": 4535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.984341610067375e-05,
+      "loss": 0.3393,
+      "step": 4536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9842608870586485e-05,
+      "loss": 0.3455,
+      "step": 4537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.984179957240947e-05,
+      "loss": 0.3334,
+      "step": 4538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.984098820621762e-05,
+      "loss": 0.3088,
+      "step": 4539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9840174772086015e-05,
+      "loss": 0.3358,
+      "step": 4540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.983935927008995e-05,
+      "loss": 0.3098,
+      "step": 4541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.983854170030494e-05,
+      "loss": 0.3354,
+      "step": 4542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.983772206280663e-05,
+      "loss": 0.3552,
+      "step": 4543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.983690035767091e-05,
+      "loss": 0.3455,
+      "step": 4544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.983607658497381e-05,
+      "loss": 0.3169,
+      "step": 4545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9835250744791615e-05,
+      "loss": 0.3437,
+      "step": 4546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.983442283720074e-05,
+      "loss": 0.3232,
+      "step": 4547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.983359286227784e-05,
+      "loss": 0.3305,
+      "step": 4548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9832760820099724e-05,
+      "loss": 0.3302,
+      "step": 4549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.983192671074341e-05,
+      "loss": 0.3335,
+      "step": 4550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9831090534286114e-05,
+      "loss": 0.3241,
+      "step": 4551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.983025229080523e-05,
+      "loss": 0.338,
+      "step": 4552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.982941198037834e-05,
+      "loss": 0.3317,
+      "step": 4553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.982856960308324e-05,
+      "loss": 0.3285,
+      "step": 4554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.982772515899789e-05,
+      "loss": 0.3325,
+      "step": 4555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.982687864820045e-05,
+      "loss": 0.3424,
+      "step": 4556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.98260300707693e-05,
+      "loss": 0.3374,
+      "step": 4557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.982517942678296e-05,
+      "loss": 0.3212,
+      "step": 4558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.982432671632017e-05,
+      "loss": 0.3269,
+      "step": 4559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9823471939459884e-05,
+      "loss": 0.3201,
+      "step": 4560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9822615096281206e-05,
+      "loss": 0.3305,
+      "step": 4561
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9821756186863435e-05,
+      "loss": 0.3119,
+      "step": 4562
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9820895211286096e-05,
+      "loss": 0.3346,
+      "step": 4563
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9820032169628866e-05,
+      "loss": 0.3275,
+      "step": 4564
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.981916706197164e-05,
+      "loss": 0.3382,
+      "step": 4565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9818299888394495e-05,
+      "loss": 0.338,
+      "step": 4566
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.98174306489777e-05,
+      "loss": 0.3238,
+      "step": 4567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.98165593438017e-05,
+      "loss": 0.3431,
+      "step": 4568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.981568597294716e-05,
+      "loss": 0.3271,
+      "step": 4569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9814810536494924e-05,
+      "loss": 0.3161,
+      "step": 4570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9813933034526e-05,
+      "loss": 0.3224,
+      "step": 4571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9813053467121646e-05,
+      "loss": 0.3136,
+      "step": 4572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.981217183436326e-05,
+      "loss": 0.3141,
+      "step": 4573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.981128813633244e-05,
+      "loss": 0.3351,
+      "step": 4574
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9810402373111e-05,
+      "loss": 0.3351,
+      "step": 4575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.980951454478091e-05,
+      "loss": 0.3512,
+      "step": 4576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.980862465142437e-05,
+      "loss": 0.3214,
+      "step": 4577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9807732693123735e-05,
+      "loss": 0.3349,
+      "step": 4578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.980683866996158e-05,
+      "loss": 0.3245,
+      "step": 4579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.980594258202064e-05,
+      "loss": 0.3442,
+      "step": 4580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9805044429383886e-05,
+      "loss": 0.3285,
+      "step": 4581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.980414421213442e-05,
+      "loss": 0.3249,
+      "step": 4582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.98032419303556e-05,
+      "loss": 0.3348,
+      "step": 4583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9802337584130926e-05,
+      "loss": 0.3373,
+      "step": 4584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9801431173544114e-05,
+      "loss": 0.3182,
+      "step": 4585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9800522698679056e-05,
+      "loss": 0.3325,
+      "step": 4586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.979961215961984e-05,
+      "loss": 0.3191,
+      "step": 4587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9798699556450765e-05,
+      "loss": 0.3232,
+      "step": 4588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.979778488925629e-05,
+      "loss": 0.3242,
+      "step": 4589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.979686815812108e-05,
+      "loss": 0.3302,
+      "step": 4590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.979594936312999e-05,
+      "loss": 0.3394,
+      "step": 4591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.979502850436808e-05,
+      "loss": 0.3418,
+      "step": 4592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.979410558192057e-05,
+      "loss": 0.3245,
+      "step": 4593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.979318059587289e-05,
+      "loss": 0.3304,
+      "step": 4594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.979225354631067e-05,
+      "loss": 0.3238,
+      "step": 4595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.979132443331971e-05,
+      "loss": 0.3322,
+      "step": 4596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.979039325698601e-05,
+      "loss": 0.3266,
+      "step": 4597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.978946001739577e-05,
+      "loss": 0.3402,
+      "step": 4598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9788524714635374e-05,
+      "loss": 0.3254,
+      "step": 4599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.978758734879138e-05,
+      "loss": 0.3306,
+      "step": 4600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9786647919950575e-05,
+      "loss": 0.3177,
+      "step": 4601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9785706428199905e-05,
+      "loss": 0.3191,
+      "step": 4602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.97847628736265e-05,
+      "loss": 0.3443,
+      "step": 4603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.978381725631773e-05,
+      "loss": 0.3239,
+      "step": 4604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9782869576361105e-05,
+      "loss": 0.3358,
+      "step": 4605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.978191983384435e-05,
+      "loss": 0.3178,
+      "step": 4606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9780968028855366e-05,
+      "loss": 0.3202,
+      "step": 4607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.978001416148227e-05,
+      "loss": 0.3242,
+      "step": 4608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977905823181334e-05,
+      "loss": 0.3358,
+      "step": 4609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977810023993706e-05,
+      "loss": 0.3315,
+      "step": 4610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977714018594212e-05,
+      "loss": 0.3333,
+      "step": 4611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9776178069917365e-05,
+      "loss": 0.3141,
+      "step": 4612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977521389195187e-05,
+      "loss": 0.3231,
+      "step": 4613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977424765213487e-05,
+      "loss": 0.3285,
+      "step": 4614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977327935055581e-05,
+      "loss": 0.3128,
+      "step": 4615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977230898730431e-05,
+      "loss": 0.3303,
+      "step": 4616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977133656247018e-05,
+      "loss": 0.3305,
+      "step": 4617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977036207614346e-05,
+      "loss": 0.3211,
+      "step": 4618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.976938552841433e-05,
+      "loss": 0.3408,
+      "step": 4619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.976840691937319e-05,
+      "loss": 0.3399,
+      "step": 4620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.976742624911061e-05,
+      "loss": 0.3159,
+      "step": 4621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9766443517717375e-05,
+      "loss": 0.2938,
+      "step": 4622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9765458725284445e-05,
+      "loss": 0.3183,
+      "step": 4623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.976447187190298e-05,
+      "loss": 0.3261,
+      "step": 4624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.976348295766432e-05,
+      "loss": 0.3299,
+      "step": 4625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9762491982660004e-05,
+      "loss": 0.3215,
+      "step": 4626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.976149894698175e-05,
+      "loss": 0.3322,
+      "step": 4627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.97605038507215e-05,
+      "loss": 0.329,
+      "step": 4628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9759506693971336e-05,
+      "loss": 0.331,
+      "step": 4629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.975850747682357e-05,
+      "loss": 0.3524,
+      "step": 4630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9757506199370694e-05,
+      "loss": 0.3473,
+      "step": 4631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.975650286170539e-05,
+      "loss": 0.3183,
+      "step": 4632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9755497463920514e-05,
+      "loss": 0.3208,
+      "step": 4633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9754490006109146e-05,
+      "loss": 0.3351,
+      "step": 4634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9753480488364526e-05,
+      "loss": 0.339,
+      "step": 4635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9752468910780106e-05,
+      "loss": 0.3218,
+      "step": 4636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.975145527344952e-05,
+      "loss": 0.3395,
+      "step": 4637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.975043957646659e-05,
+      "loss": 0.3444,
+      "step": 4638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.974942181992534e-05,
+      "loss": 0.3414,
+      "step": 4639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.974840200391996e-05,
+      "loss": 0.333,
+      "step": 4640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9747380128544855e-05,
+      "loss": 0.3257,
+      "step": 4641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.974635619389462e-05,
+      "loss": 0.3175,
+      "step": 4642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9745330200064014e-05,
+      "loss": 0.3296,
+      "step": 4643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9744302147148023e-05,
+      "loss": 0.3327,
+      "step": 4644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.974327203524181e-05,
+      "loss": 0.3193,
+      "step": 4645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.97422398644407e-05,
+      "loss": 0.3449,
+      "step": 4646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.974120563484026e-05,
+      "loss": 0.3201,
+      "step": 4647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9740169346536206e-05,
+      "loss": 0.3374,
+      "step": 4648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9739130999624464e-05,
+      "loss": 0.3224,
+      "step": 4649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9738090594201144e-05,
+      "loss": 0.3388,
+      "step": 4650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.973704813036255e-05,
+      "loss": 0.3134,
+      "step": 4651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.973600360820518e-05,
+      "loss": 0.3204,
+      "step": 4652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.973495702782571e-05,
+      "loss": 0.3408,
+      "step": 4653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.973390838932102e-05,
+      "loss": 0.3406,
+      "step": 4654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.973285769278817e-05,
+      "loss": 0.3316,
+      "step": 4655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.973180493832441e-05,
+      "loss": 0.3333,
+      "step": 4656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9730750126027205e-05,
+      "loss": 0.3362,
+      "step": 4657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.972969325599417e-05,
+      "loss": 0.3332,
+      "step": 4658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.972863432832315e-05,
+      "loss": 0.3469,
+      "step": 4659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9727573343112136e-05,
+      "loss": 0.3195,
+      "step": 4660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9726510300459364e-05,
+      "loss": 0.2842,
+      "step": 4661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.972544520046322e-05,
+      "loss": 0.3424,
+      "step": 4662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.972437804322229e-05,
+      "loss": 0.3441,
+      "step": 4663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.972330882883536e-05,
+      "loss": 0.3309,
+      "step": 4664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.972223755740139e-05,
+      "loss": 0.3323,
+      "step": 4665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9721164229019544e-05,
+      "loss": 0.3465,
+      "step": 4666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.972008884378918e-05,
+      "loss": 0.3297,
+      "step": 4667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.971901140180983e-05,
+      "loss": 0.3292,
+      "step": 4668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9717931903181215e-05,
+      "loss": 0.3057,
+      "step": 4669
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.971685034800327e-05,
+      "loss": 0.3334,
+      "step": 4670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9715766736376105e-05,
+      "loss": 0.3307,
+      "step": 4671
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9714681068400026e-05,
+      "loss": 0.3424,
+      "step": 4672
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.971359334417551e-05,
+      "loss": 0.3417,
+      "step": 4673
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.971250356380326e-05,
+      "loss": 0.3217,
+      "step": 4674
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9711411727384126e-05,
+      "loss": 0.3177,
+      "step": 4675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.971031783501919e-05,
+      "loss": 0.3392,
+      "step": 4676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9709221886809695e-05,
+      "loss": 0.3318,
+      "step": 4677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.97081238828571e-05,
+      "loss": 0.3392,
+      "step": 4678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9707023823263014e-05,
+      "loss": 0.3161,
+      "step": 4679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.970592170812928e-05,
+      "loss": 0.3235,
+      "step": 4680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.970481753755791e-05,
+      "loss": 0.3191,
+      "step": 4681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.970371131165111e-05,
+      "loss": 0.3481,
+      "step": 4682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.970260303051126e-05,
+      "loss": 0.3297,
+      "step": 4683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9701492694240967e-05,
+      "loss": 0.3043,
+      "step": 4684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.970038030294299e-05,
+      "loss": 0.3304,
+      "step": 4685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.96992658567203e-05,
+      "loss": 0.361,
+      "step": 4686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.969814935567606e-05,
+      "loss": 0.3358,
+      "step": 4687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9697030799913606e-05,
+      "loss": 0.3169,
+      "step": 4688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9695910189536484e-05,
+      "loss": 0.3226,
+      "step": 4689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9694787524648415e-05,
+      "loss": 0.336,
+      "step": 4690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.969366280535331e-05,
+      "loss": 0.3151,
+      "step": 4691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9692536031755297e-05,
+      "loss": 0.3457,
+      "step": 4692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.969140720395865e-05,
+      "loss": 0.3067,
+      "step": 4693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.969027632206786e-05,
+      "loss": 0.3333,
+      "step": 4694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.968914338618761e-05,
+      "loss": 0.3844,
+      "step": 4695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9688008396422774e-05,
+      "loss": 0.326,
+      "step": 4696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.96868713528784e-05,
+      "loss": 0.3249,
+      "step": 4697
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9685732255659736e-05,
+      "loss": 0.3159,
+      "step": 4698
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.968459110487222e-05,
+      "loss": 0.3495,
+      "step": 4699
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.968344790062149e-05,
+      "loss": 0.3307,
+      "step": 4700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.968230264301334e-05,
+      "loss": 0.3061,
+      "step": 4701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.96811553321538e-05,
+      "loss": 0.3324,
+      "step": 4702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.968000596814907e-05,
+      "loss": 0.3123,
+      "step": 4703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.967885455110552e-05,
+      "loss": 0.3068,
+      "step": 4704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.967770108112974e-05,
+      "loss": 0.3184,
+      "step": 4705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9676545558328504e-05,
+      "loss": 0.3267,
+      "step": 4706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.967538798280875e-05,
+      "loss": 0.338,
+      "step": 4707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.967422835467764e-05,
+      "loss": 0.3318,
+      "step": 4708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.967306667404252e-05,
+      "loss": 0.325,
+      "step": 4709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9671902941010894e-05,
+      "loss": 0.311,
+      "step": 4710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9670737155690503e-05,
+      "loss": 0.3206,
+      "step": 4711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.966956931818925e-05,
+      "loss": 0.3397,
+      "step": 4712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.966839942861523e-05,
+      "loss": 0.3215,
+      "step": 4713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.966722748707672e-05,
+      "loss": 0.3364,
+      "step": 4714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.966605349368222e-05,
+      "loss": 0.3314,
+      "step": 4715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.966487744854039e-05,
+      "loss": 0.3174,
+      "step": 4716
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.966369935176007e-05,
+      "loss": 0.3243,
+      "step": 4717
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.966251920345034e-05,
+      "loss": 0.3248,
+      "step": 4718
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.966133700372041e-05,
+      "loss": 0.3337,
+      "step": 4719
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9660152752679716e-05,
+      "loss": 0.3216,
+      "step": 4720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9658966450437885e-05,
+      "loss": 0.3277,
+      "step": 4721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.965777809710471e-05,
+      "loss": 0.3184,
+      "step": 4722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.965658769279019e-05,
+      "loss": 0.3231,
+      "step": 4723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.965539523760452e-05,
+      "loss": 0.3294,
+      "step": 4724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.965420073165807e-05,
+      "loss": 0.3045,
+      "step": 4725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.965300417506141e-05,
+      "loss": 0.3207,
+      "step": 4726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.96518055679253e-05,
+      "loss": 0.3276,
+      "step": 4727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.965060491036068e-05,
+      "loss": 0.321,
+      "step": 4728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.964940220247869e-05,
+      "loss": 0.3208,
+      "step": 4729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9648197444390654e-05,
+      "loss": 0.3258,
+      "step": 4730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.964699063620807e-05,
+      "loss": 0.3295,
+      "step": 4731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.964578177804268e-05,
+      "loss": 0.3073,
+      "step": 4732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.964457087000636e-05,
+      "loss": 0.317,
+      "step": 4733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.964335791221118e-05,
+      "loss": 0.3366,
+      "step": 4734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9642142904769435e-05,
+      "loss": 0.3221,
+      "step": 4735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9640925847793586e-05,
+      "loss": 0.3282,
+      "step": 4736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.963970674139628e-05,
+      "loss": 0.3217,
+      "step": 4737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9638485585690365e-05,
+      "loss": 0.354,
+      "step": 4738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.963726238078888e-05,
+      "loss": 0.3408,
+      "step": 4739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.963603712680503e-05,
+      "loss": 0.3185,
+      "step": 4740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.963480982385225e-05,
+      "loss": 0.3383,
+      "step": 4741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.963358047204413e-05,
+      "loss": 0.3423,
+      "step": 4742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.963234907149446e-05,
+      "loss": 0.3274,
+      "step": 4743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.963111562231723e-05,
+      "loss": 0.3288,
+      "step": 4744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9629880124626596e-05,
+      "loss": 0.3229,
+      "step": 4745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.962864257853694e-05,
+      "loss": 0.3297,
+      "step": 4746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9627402984162795e-05,
+      "loss": 0.3264,
+      "step": 4747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9626161341618914e-05,
+      "loss": 0.3428,
+      "step": 4748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9624917651020226e-05,
+      "loss": 0.3424,
+      "step": 4749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.962367191248183e-05,
+      "loss": 0.3175,
+      "step": 4750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9622424126119066e-05,
+      "loss": 0.3214,
+      "step": 4751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.96211742920474e-05,
+      "loss": 0.3049,
+      "step": 4752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.961992241038255e-05,
+      "loss": 0.3177,
+      "step": 4753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.961866848124037e-05,
+      "loss": 0.325,
+      "step": 4754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9617412504736935e-05,
+      "loss": 0.3213,
+      "step": 4755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9616154480988515e-05,
+      "loss": 0.3293,
+      "step": 4756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9614894410111535e-05,
+      "loss": 0.3472,
+      "step": 4757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.961363229222264e-05,
+      "loss": 0.3437,
+      "step": 4758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.961236812743866e-05,
+      "loss": 0.2999,
+      "step": 4759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.961110191587659e-05,
+      "loss": 0.3471,
+      "step": 4760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.960983365765366e-05,
+      "loss": 0.3312,
+      "step": 4761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9608563352887245e-05,
+      "loss": 0.3431,
+      "step": 4762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.960729100169494e-05,
+      "loss": 0.3124,
+      "step": 4763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.96060166041945e-05,
+      "loss": 0.34,
+      "step": 4764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9604740160503906e-05,
+      "loss": 0.3336,
+      "step": 4765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.960346167074129e-05,
+      "loss": 0.3191,
+      "step": 4766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9602181135025006e-05,
+      "loss": 0.3106,
+      "step": 4767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.960089855347358e-05,
+      "loss": 0.3067,
+      "step": 4768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9599613926205734e-05,
+      "loss": 0.3245,
+      "step": 4769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.959832725334037e-05,
+      "loss": 0.3058,
+      "step": 4770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.959703853499659e-05,
+      "loss": 0.3114,
+      "step": 4771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9595747771293686e-05,
+      "loss": 0.319,
+      "step": 4772
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.959445496235112e-05,
+      "loss": 0.3104,
+      "step": 4773
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.959316010828857e-05,
+      "loss": 0.3136,
+      "step": 4774
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.959186320922589e-05,
+      "loss": 0.3192,
+      "step": 4775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.959056426528311e-05,
+      "loss": 0.3173,
+      "step": 4776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9589263276580484e-05,
+      "loss": 0.3069,
+      "step": 4777
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.958796024323843e-05,
+      "loss": 0.3345,
+      "step": 4778
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.958665516537755e-05,
+      "loss": 0.3189,
+      "step": 4779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.958534804311866e-05,
+      "loss": 0.3244,
+      "step": 4780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.958403887658273e-05,
+      "loss": 0.3226,
+      "step": 4781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.958272766589096e-05,
+      "loss": 0.3195,
+      "step": 4782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.958141441116471e-05,
+      "loss": 0.3268,
+      "step": 4783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.958009911252555e-05,
+      "loss": 0.3324,
+      "step": 4784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.95787817700952e-05,
+      "loss": 0.3239,
+      "step": 4785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9577462383995624e-05,
+      "loss": 0.3518,
+      "step": 4786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.957614095434893e-05,
+      "loss": 0.3304,
+      "step": 4787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9574817481277444e-05,
+      "loss": 0.3134,
+      "step": 4788
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9573491964903665e-05,
+      "loss": 0.3378,
+      "step": 4789
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.957216440535029e-05,
+      "loss": 0.3273,
+      "step": 4790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9570834802740195e-05,
+      "loss": 0.3305,
+      "step": 4791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.956950315719646e-05,
+      "loss": 0.335,
+      "step": 4792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9568169468842344e-05,
+      "loss": 0.3208,
+      "step": 4793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.956683373780129e-05,
+      "loss": 0.3405,
+      "step": 4794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.956549596419694e-05,
+      "loss": 0.3406,
+      "step": 4795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9564156148153125e-05,
+      "loss": 0.3359,
+      "step": 4796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9562814289793855e-05,
+      "loss": 0.3255,
+      "step": 4797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.956147038924334e-05,
+      "loss": 0.3195,
+      "step": 4798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9560124446625984e-05,
+      "loss": 0.3126,
+      "step": 4799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.955877646206635e-05,
+      "loss": 0.3425,
+      "step": 4800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.955742643568923e-05,
+      "loss": 0.3266,
+      "step": 4801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.955607436761958e-05,
+      "loss": 0.3179,
+      "step": 4802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.955472025798254e-05,
+      "loss": 0.338,
+      "step": 4803
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9553364106903466e-05,
+      "loss": 0.3172,
+      "step": 4804
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.955200591450788e-05,
+      "loss": 0.3429,
+      "step": 4805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.95506456809215e-05,
+      "loss": 0.3402,
+      "step": 4806
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.954928340627022e-05,
+      "loss": 0.3426,
+      "step": 4807
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9547919090680175e-05,
+      "loss": 0.3247,
+      "step": 4808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.95465527342776e-05,
+      "loss": 0.3437,
+      "step": 4809
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.954518433718901e-05,
+      "loss": 0.3128,
+      "step": 4810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.954381389954104e-05,
+      "loss": 0.3325,
+      "step": 4811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.954244142146055e-05,
+      "loss": 0.3386,
+      "step": 4812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9541066903074584e-05,
+      "loss": 0.3266,
+      "step": 4813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.953969034451036e-05,
+      "loss": 0.323,
+      "step": 4814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.953831174589531e-05,
+      "loss": 0.3187,
+      "step": 4815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.953693110735703e-05,
+      "loss": 0.3289,
+      "step": 4816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.953554842902332e-05,
+      "loss": 0.3252,
+      "step": 4817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9534163711022166e-05,
+      "loss": 0.3381,
+      "step": 4818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9532776953481735e-05,
+      "loss": 0.3122,
+      "step": 4819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.95313881565304e-05,
+      "loss": 0.3438,
+      "step": 4820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.95299973202967e-05,
+      "loss": 0.3234,
+      "step": 4821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.952860444490938e-05,
+      "loss": 0.344,
+      "step": 4822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9527209530497365e-05,
+      "loss": 0.3272,
+      "step": 4823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.952581257718977e-05,
+      "loss": 0.3265,
+      "step": 4824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.952441358511591e-05,
+      "loss": 0.3148,
+      "step": 4825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.952301255440527e-05,
+      "loss": 0.3408,
+      "step": 4826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.952160948518754e-05,
+      "loss": 0.3166,
+      "step": 4827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.952020437759258e-05,
+      "loss": 0.3305,
+      "step": 4828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9518797231750474e-05,
+      "loss": 0.3316,
+      "step": 4829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9517388047791444e-05,
+      "loss": 0.326,
+      "step": 4830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.951597682584594e-05,
+      "loss": 0.3267,
+      "step": 4831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9514563566044594e-05,
+      "loss": 0.3135,
+      "step": 4832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9513148268518216e-05,
+      "loss": 0.3211,
+      "step": 4833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9511730933397807e-05,
+      "loss": 0.3282,
+      "step": 4834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.951031156081456e-05,
+      "loss": 0.3265,
+      "step": 4835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.950889015089986e-05,
+      "loss": 0.3268,
+      "step": 4836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.950746670378526e-05,
+      "loss": 0.3132,
+      "step": 4837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.950604121960254e-05,
+      "loss": 0.3252,
+      "step": 4838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9504613698483645e-05,
+      "loss": 0.333,
+      "step": 4839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9503184140560686e-05,
+      "loss": 0.3405,
+      "step": 4840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.950175254596601e-05,
+      "loss": 0.3206,
+      "step": 4841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.950031891483213e-05,
+      "loss": 0.3221,
+      "step": 4842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9498883247291725e-05,
+      "loss": 0.3286,
+      "step": 4843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9497445543477706e-05,
+      "loss": 0.3156,
+      "step": 4844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9496005803523144e-05,
+      "loss": 0.3275,
+      "step": 4845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.94945640275613e-05,
+      "loss": 0.3152,
+      "step": 4846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.949312021572563e-05,
+      "loss": 0.3154,
+      "step": 4847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.949167436814979e-05,
+      "loss": 0.3514,
+      "step": 4848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.949022648496759e-05,
+      "loss": 0.337,
+      "step": 4849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.948877656631307e-05,
+      "loss": 0.3248,
+      "step": 4850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.948732461232041e-05,
+      "loss": 0.3269,
+      "step": 4851
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.948587062312404e-05,
+      "loss": 0.3137,
+      "step": 4852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.948441459885853e-05,
+      "loss": 0.3163,
+      "step": 4853
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.948295653965865e-05,
+      "loss": 0.3343,
+      "step": 4854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.948149644565936e-05,
+      "loss": 0.3228,
+      "step": 4855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9480034316995826e-05,
+      "loss": 0.3188,
+      "step": 4856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.947857015380337e-05,
+      "loss": 0.3204,
+      "step": 4857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.947710395621753e-05,
+      "loss": 0.3272,
+      "step": 4858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9475635724374005e-05,
+      "loss": 0.3179,
+      "step": 4859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9474165458408714e-05,
+      "loss": 0.3211,
+      "step": 4860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9472693158457746e-05,
+      "loss": 0.3271,
+      "step": 4861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.947121882465737e-05,
+      "loss": 0.3457,
+      "step": 4862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9469742457144075e-05,
+      "loss": 0.3237,
+      "step": 4863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9468264056054496e-05,
+      "loss": 0.3101,
+      "step": 4864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9466783621525494e-05,
+      "loss": 0.3626,
+      "step": 4865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9465301153694086e-05,
+      "loss": 0.3307,
+      "step": 4866
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.946381665269751e-05,
+      "loss": 0.3304,
+      "step": 4867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.946233011867315e-05,
+      "loss": 0.3289,
+      "step": 4868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.946084155175864e-05,
+      "loss": 0.3198,
+      "step": 4869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.945935095209173e-05,
+      "loss": 0.3294,
+      "step": 4870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9457858319810426e-05,
+      "loss": 0.3009,
+      "step": 4871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9456363655052864e-05,
+      "loss": 0.319,
+      "step": 4872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.94548669579574e-05,
+      "loss": 0.3161,
+      "step": 4873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.945336822866258e-05,
+      "loss": 0.3255,
+      "step": 4874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.945186746730713e-05,
+      "loss": 0.2929,
+      "step": 4875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.945036467402995e-05,
+      "loss": 0.3514,
+      "step": 4876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.944885984897016e-05,
+      "loss": 0.3338,
+      "step": 4877
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9447352992267045e-05,
+      "loss": 0.3253,
+      "step": 4878
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.944584410406008e-05,
+      "loss": 0.3324,
+      "step": 4879
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9444333184488926e-05,
+      "loss": 0.3338,
+      "step": 4880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.944282023369345e-05,
+      "loss": 0.3352,
+      "step": 4881
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.944130525181369e-05,
+      "loss": 0.3265,
+      "step": 4882
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.943978823898988e-05,
+      "loss": 0.3291,
+      "step": 4883
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.943826919536244e-05,
+      "loss": 0.3031,
+      "step": 4884
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.943674812107196e-05,
+      "loss": 0.3168,
+      "step": 4885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9435225016259244e-05,
+      "loss": 0.3275,
+      "step": 4886
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9433699881065286e-05,
+      "loss": 0.3147,
+      "step": 4887
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.943217271563125e-05,
+      "loss": 0.3304,
+      "step": 4888
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.943064352009848e-05,
+      "loss": 0.3333,
+      "step": 4889
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9429112294608546e-05,
+      "loss": 0.3226,
+      "step": 4890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.942757903930316e-05,
+      "loss": 0.3278,
+      "step": 4891
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.942604375432425e-05,
+      "loss": 0.297,
+      "step": 4892
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.942450643981394e-05,
+      "loss": 0.3223,
+      "step": 4893
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9422967095914515e-05,
+      "loss": 0.3311,
+      "step": 4894
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9421425722768466e-05,
+      "loss": 0.338,
+      "step": 4895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.941988232051846e-05,
+      "loss": 0.3323,
+      "step": 4896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9418336889307365e-05,
+      "loss": 0.3292,
+      "step": 4897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.941678942927822e-05,
+      "loss": 0.3311,
+      "step": 4898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.941523994057428e-05,
+      "loss": 0.3432,
+      "step": 4899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.941368842333896e-05,
+      "loss": 0.3417,
+      "step": 4900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.941213487771586e-05,
+      "loss": 0.3274,
+      "step": 4901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.94105793038488e-05,
+      "loss": 0.3162,
+      "step": 4902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.940902170188176e-05,
+      "loss": 0.3443,
+      "step": 4903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.940746207195891e-05,
+      "loss": 0.3318,
+      "step": 4904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.940590041422462e-05,
+      "loss": 0.3284,
+      "step": 4905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9404336728823444e-05,
+      "loss": 0.3184,
+      "step": 4906
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.940277101590012e-05,
+      "loss": 0.3374,
+      "step": 4907
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9401203275599556e-05,
+      "loss": 0.3355,
+      "step": 4908
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.939963350806689e-05,
+      "loss": 0.3272,
+      "step": 4909
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.939806171344742e-05,
+      "loss": 0.33,
+      "step": 4910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.939648789188662e-05,
+      "loss": 0.3314,
+      "step": 4911
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.939491204353019e-05,
+      "loss": 0.3339,
+      "step": 4912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9393334168523974e-05,
+      "loss": 0.3325,
+      "step": 4913
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.939175426701403e-05,
+      "loss": 0.3228,
+      "step": 4914
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.939017233914661e-05,
+      "loss": 0.3214,
+      "step": 4915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.938858838506812e-05,
+      "loss": 0.3197,
+      "step": 4916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.938700240492518e-05,
+      "loss": 0.3178,
+      "step": 4917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.938541439886462e-05,
+      "loss": 0.3366,
+      "step": 4918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9383824367033395e-05,
+      "loss": 0.3176,
+      "step": 4919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9382232309578694e-05,
+      "loss": 0.3218,
+      "step": 4920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9380638226647894e-05,
+      "loss": 0.3293,
+      "step": 4921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.937904211838853e-05,
+      "loss": 0.335,
+      "step": 4922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9377443984948346e-05,
+      "loss": 0.326,
+      "step": 4923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.937584382647527e-05,
+      "loss": 0.3167,
+      "step": 4924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9374241643117434e-05,
+      "loss": 0.3202,
+      "step": 4925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9372637435023115e-05,
+      "loss": 0.3413,
+      "step": 4926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.937103120234081e-05,
+      "loss": 0.3256,
+      "step": 4927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.936942294521921e-05,
+      "loss": 0.3147,
+      "step": 4928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.936781266380717e-05,
+      "loss": 0.352,
+      "step": 4929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.936620035825373e-05,
+      "loss": 0.3285,
+      "step": 4930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.936458602870814e-05,
+      "loss": 0.3294,
+      "step": 4931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9362969675319836e-05,
+      "loss": 0.3353,
+      "step": 4932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9361351298238414e-05,
+      "loss": 0.3302,
+      "step": 4933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.935973089761369e-05,
+      "loss": 0.3167,
+      "step": 4934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9358108473595636e-05,
+      "loss": 0.311,
+      "step": 4935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9356484026334455e-05,
+      "loss": 0.3135,
+      "step": 4936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.935485755598047e-05,
+      "loss": 0.335,
+      "step": 4937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9353229062684266e-05,
+      "loss": 0.3357,
+      "step": 4938
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.935159854659657e-05,
+      "loss": 0.3292,
+      "step": 4939
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9349966007868306e-05,
+      "loss": 0.3127,
+      "step": 4940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.934833144665058e-05,
+      "loss": 0.313,
+      "step": 4941
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.934669486309471e-05,
+      "loss": 0.334,
+      "step": 4942
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.934505625735216e-05,
+      "loss": 0.3137,
+      "step": 4943
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.93434156295746e-05,
+      "loss": 0.342,
+      "step": 4944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.934177297991392e-05,
+      "loss": 0.3197,
+      "step": 4945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9340128308522144e-05,
+      "loss": 0.3206,
+      "step": 4946
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9338481615551525e-05,
+      "loss": 0.3412,
+      "step": 4947
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9336832901154455e-05,
+      "loss": 0.3386,
+      "step": 4948
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.933518216548358e-05,
+      "loss": 0.3178,
+      "step": 4949
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.933352940869167e-05,
+      "loss": 0.333,
+      "step": 4950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.933187463093172e-05,
+      "loss": 0.327,
+      "step": 4951
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.933021783235689e-05,
+      "loss": 0.3205,
+      "step": 4952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.932855901312056e-05,
+      "loss": 0.33,
+      "step": 4953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.932689817337626e-05,
+      "loss": 0.3238,
+      "step": 4954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.932523531327771e-05,
+      "loss": 0.3145,
+      "step": 4955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.932357043297884e-05,
+      "loss": 0.3393,
+      "step": 4956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.932190353263377e-05,
+      "loss": 0.3143,
+      "step": 4957
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.932023461239677e-05,
+      "loss": 0.3119,
+      "step": 4958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.931856367242233e-05,
+      "loss": 0.3291,
+      "step": 4959
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.931689071286511e-05,
+      "loss": 0.3208,
+      "step": 4960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.931521573387998e-05,
+      "loss": 0.341,
+      "step": 4961
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9313538735621964e-05,
+      "loss": 0.3494,
+      "step": 4962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.931185971824629e-05,
+      "loss": 0.3059,
+      "step": 4963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.931017868190837e-05,
+      "loss": 0.3239,
+      "step": 4964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.930849562676382e-05,
+      "loss": 0.3274,
+      "step": 4965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.930681055296842e-05,
+      "loss": 0.3053,
+      "step": 4966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9305123460678146e-05,
+      "loss": 0.3174,
+      "step": 4967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.930343435004915e-05,
+      "loss": 0.3177,
+      "step": 4968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9301743221237796e-05,
+      "loss": 0.3204,
+      "step": 4969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9300050074400605e-05,
+      "loss": 0.3208,
+      "step": 4970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9298354909694306e-05,
+      "loss": 0.3359,
+      "step": 4971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9296657727275805e-05,
+      "loss": 0.3189,
+      "step": 4972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.929495852730221e-05,
+      "loss": 0.3228,
+      "step": 4973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.929325730993079e-05,
+      "loss": 0.3352,
+      "step": 4974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9291554075319005e-05,
+      "loss": 0.3131,
+      "step": 4975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.928984882362454e-05,
+      "loss": 0.3223,
+      "step": 4976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.928814155500521e-05,
+      "loss": 0.3377,
+      "step": 4977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.928643226961906e-05,
+      "loss": 0.3152,
+      "step": 4978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.92847209676243e-05,
+      "loss": 0.3317,
+      "step": 4979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.928300764917933e-05,
+      "loss": 0.3369,
+      "step": 4980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.928129231444275e-05,
+      "loss": 0.3301,
+      "step": 4981
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.927957496357332e-05,
+      "loss": 0.3288,
+      "step": 4982
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.927785559673002e-05,
+      "loss": 0.3287,
+      "step": 4983
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9276134214071986e-05,
+      "loss": 0.3243,
+      "step": 4984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.927441081575855e-05,
+      "loss": 0.3297,
+      "step": 4985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9272685401949245e-05,
+      "loss": 0.3165,
+      "step": 4986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.927095797280378e-05,
+      "loss": 0.3269,
+      "step": 4987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9269228528482046e-05,
+      "loss": 0.3096,
+      "step": 4988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.926749706914412e-05,
+      "loss": 0.3372,
+      "step": 4989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.926576359495028e-05,
+      "loss": 0.3334,
+      "step": 4990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9264028106060984e-05,
+      "loss": 0.3124,
+      "step": 4991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9262290602636854e-05,
+      "loss": 0.3456,
+      "step": 4992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.926055108483874e-05,
+      "loss": 0.3095,
+      "step": 4993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.925880955282764e-05,
+      "loss": 0.3397,
+      "step": 4994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9257066006764764e-05,
+      "loss": 0.3125,
+      "step": 4995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.925532044681149e-05,
+      "loss": 0.333,
+      "step": 4996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.925357287312941e-05,
+      "loss": 0.3221,
+      "step": 4997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.925182328588026e-05,
+      "loss": 0.3305,
+      "step": 4998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9250071685226004e-05,
+      "loss": 0.314,
+      "step": 4999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.924831807132877e-05,
+      "loss": 0.3198,
+      "step": 5000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9246562444350875e-05,
+      "loss": 0.3313,
+      "step": 5001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.924480480445482e-05,
+      "loss": 0.3107,
+      "step": 5002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9243045151803306e-05,
+      "loss": 0.3452,
+      "step": 5003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9241283486559204e-05,
+      "loss": 0.33,
+      "step": 5004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.923951980888558e-05,
+      "loss": 0.3257,
+      "step": 5005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.923775411894569e-05,
+      "loss": 0.3278,
+      "step": 5006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9235986416902965e-05,
+      "loss": 0.3296,
+      "step": 5007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9234216702921026e-05,
+      "loss": 0.3154,
+      "step": 5008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9232444977163697e-05,
+      "loss": 0.3291,
+      "step": 5009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.923067123979495e-05,
+      "loss": 0.3278,
+      "step": 5010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.922889549097898e-05,
+      "loss": 0.3173,
+      "step": 5011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9227117730880154e-05,
+      "loss": 0.3358,
+      "step": 5012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.922533795966304e-05,
+      "loss": 0.3442,
+      "step": 5013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9223556177492346e-05,
+      "loss": 0.3244,
+      "step": 5014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.922177238453303e-05,
+      "loss": 0.3185,
+      "step": 5015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9219986580950186e-05,
+      "loss": 0.3428,
+      "step": 5016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9218198766909116e-05,
+      "loss": 0.3216,
+      "step": 5017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.92164089425753e-05,
+      "loss": 0.3185,
+      "step": 5018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.921461710811443e-05,
+      "loss": 0.3297,
+      "step": 5019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.921282326369234e-05,
+      "loss": 0.3285,
+      "step": 5020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.921102740947509e-05,
+      "loss": 0.3288,
+      "step": 5021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9209229545628886e-05,
+      "loss": 0.3106,
+      "step": 5022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.920742967232016e-05,
+      "loss": 0.3339,
+      "step": 5023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.920562778971552e-05,
+      "loss": 0.3182,
+      "step": 5024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9203823897981736e-05,
+      "loss": 0.3182,
+      "step": 5025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9202017997285785e-05,
+      "loss": 0.3461,
+      "step": 5026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.920021008779483e-05,
+      "loss": 0.3322,
+      "step": 5027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.919840016967622e-05,
+      "loss": 0.3197,
+      "step": 5028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9196588243097476e-05,
+      "loss": 0.3376,
+      "step": 5029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.919477430822632e-05,
+      "loss": 0.3049,
+      "step": 5030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9192958365230646e-05,
+      "loss": 0.3318,
+      "step": 5031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.919114041427855e-05,
+      "loss": 0.3137,
+      "step": 5032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.918932045553831e-05,
+      "loss": 0.3313,
+      "step": 5033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9187498489178384e-05,
+      "loss": 0.3359,
+      "step": 5034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.918567451536742e-05,
+      "loss": 0.3329,
+      "step": 5035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.918384853427424e-05,
+      "loss": 0.3084,
+      "step": 5036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9182020546067856e-05,
+      "loss": 0.3329,
+      "step": 5037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.91801905509175e-05,
+      "loss": 0.3195,
+      "step": 5038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9178358548992535e-05,
+      "loss": 0.3079,
+      "step": 5039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.917652454046254e-05,
+      "loss": 0.3153,
+      "step": 5040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.917468852549729e-05,
+      "loss": 0.3347,
+      "step": 5041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9172850504266714e-05,
+      "loss": 0.339,
+      "step": 5042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.917101047694096e-05,
+      "loss": 0.3326,
+      "step": 5043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.916916844369033e-05,
+      "loss": 0.3131,
+      "step": 5044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.916732440468533e-05,
+      "loss": 0.3208,
+      "step": 5045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.916547836009666e-05,
+      "loss": 0.3194,
+      "step": 5046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.916363031009519e-05,
+      "loss": 0.3276,
+      "step": 5047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9161780254851966e-05,
+      "loss": 0.3272,
+      "step": 5048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.915992819453826e-05,
+      "loss": 0.3175,
+      "step": 5049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.915807412932548e-05,
+      "loss": 0.3348,
+      "step": 5050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.915621805938526e-05,
+      "loss": 0.3246,
+      "step": 5051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9154359984889395e-05,
+      "loss": 0.3183,
+      "step": 5052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.915249990600987e-05,
+      "loss": 0.3275,
+      "step": 5053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.915063782291886e-05,
+      "loss": 0.3113,
+      "step": 5054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9148773735788735e-05,
+      "loss": 0.3233,
+      "step": 5055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.914690764479202e-05,
+      "loss": 0.3244,
+      "step": 5056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.914503955010146e-05,
+      "loss": 0.3259,
+      "step": 5057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.914316945188997e-05,
+      "loss": 0.3198,
+      "step": 5058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9141297350330654e-05,
+      "loss": 0.3245,
+      "step": 5059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.913942324559679e-05,
+      "loss": 0.3089,
+      "step": 5060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9137547137861846e-05,
+      "loss": 0.3211,
+      "step": 5061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.913566902729949e-05,
+      "loss": 0.325,
+      "step": 5062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9133788914083555e-05,
+      "loss": 0.3356,
+      "step": 5063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.913190679838808e-05,
+      "loss": 0.3367,
+      "step": 5064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9130022680387284e-05,
+      "loss": 0.3331,
+      "step": 5065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.912813656025554e-05,
+      "loss": 0.3132,
+      "step": 5066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.912624843816746e-05,
+      "loss": 0.3036,
+      "step": 5067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9124358314297794e-05,
+      "loss": 0.3324,
+      "step": 5068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.91224661888215e-05,
+      "loss": 0.3252,
+      "step": 5069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.912057206191373e-05,
+      "loss": 0.328,
+      "step": 5070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.91186759337498e-05,
+      "loss": 0.3223,
+      "step": 5071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.911677780450522e-05,
+      "loss": 0.3298,
+      "step": 5072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.911487767435569e-05,
+      "loss": 0.3285,
+      "step": 5073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.911297554347708e-05,
+      "loss": 0.3196,
+      "step": 5074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.911107141204547e-05,
+      "loss": 0.3289,
+      "step": 5075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.910916528023711e-05,
+      "loss": 0.3443,
+      "step": 5076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9107257148228434e-05,
+      "loss": 0.3152,
+      "step": 5077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.910534701619605e-05,
+      "loss": 0.3308,
+      "step": 5078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.910343488431679e-05,
+      "loss": 0.3264,
+      "step": 5079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9101520752767625e-05,
+      "loss": 0.3257,
+      "step": 5080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9099604621725745e-05,
+      "loss": 0.3348,
+      "step": 5081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.90976864913685e-05,
+      "loss": 0.335,
+      "step": 5082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9095766361873446e-05,
+      "loss": 0.3161,
+      "step": 5083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9093844233418314e-05,
+      "loss": 0.3133,
+      "step": 5084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.909192010618102e-05,
+      "loss": 0.345,
+      "step": 5085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9089993980339663e-05,
+      "loss": 0.3259,
+      "step": 5086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.908806585607254e-05,
+      "loss": 0.3024,
+      "step": 5087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.908613573355812e-05,
+      "loss": 0.3297,
+      "step": 5088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.908420361297505e-05,
+      "loss": 0.3163,
+      "step": 5089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9082269494502176e-05,
+      "loss": 0.3133,
+      "step": 5090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.908033337831855e-05,
+      "loss": 0.3216,
+      "step": 5091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9078395264603335e-05,
+      "loss": 0.3063,
+      "step": 5092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9076455153535964e-05,
+      "loss": 0.3284,
+      "step": 5093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9074513045296e-05,
+      "loss": 0.3369,
+      "step": 5094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.907256894006324e-05,
+      "loss": 0.3156,
+      "step": 5095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.907062283801761e-05,
+      "loss": 0.335,
+      "step": 5096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9068674739339236e-05,
+      "loss": 0.3354,
+      "step": 5097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.906672464420847e-05,
+      "loss": 0.3239,
+      "step": 5098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9064772552805795e-05,
+      "loss": 0.3061,
+      "step": 5099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.906281846531191e-05,
+      "loss": 0.3322,
+      "step": 5100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.906086238190769e-05,
+      "loss": 0.3179,
+      "step": 5101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.905890430277418e-05,
+      "loss": 0.3375,
+      "step": 5102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.905694422809265e-05,
+      "loss": 0.3243,
+      "step": 5103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.905498215804452e-05,
+      "loss": 0.3244,
+      "step": 5104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.90530180928114e-05,
+      "loss": 0.3322,
+      "step": 5105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.905105203257509e-05,
+      "loss": 0.3311,
+      "step": 5106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.904908397751758e-05,
+      "loss": 0.3237,
+      "step": 5107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9047113927821034e-05,
+      "loss": 0.3149,
+      "step": 5108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9045141883667794e-05,
+      "loss": 0.3334,
+      "step": 5109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.904316784524041e-05,
+      "loss": 0.3334,
+      "step": 5110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.904119181272162e-05,
+      "loss": 0.3069,
+      "step": 5111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9039213786294293e-05,
+      "loss": 0.3247,
+      "step": 5112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9037233766141554e-05,
+      "loss": 0.2983,
+      "step": 5113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9035251752446656e-05,
+      "loss": 0.3123,
+      "step": 5114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9033267745393066e-05,
+      "loss": 0.337,
+      "step": 5115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9031281745164436e-05,
+      "loss": 0.3307,
+      "step": 5116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.902929375194459e-05,
+      "loss": 0.3273,
+      "step": 5117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.902730376591754e-05,
+      "loss": 0.3207,
+      "step": 5118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9025311787267485e-05,
+      "loss": 0.3315,
+      "step": 5119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.902331781617882e-05,
+      "loss": 0.3227,
+      "step": 5120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.902132185283609e-05,
+      "loss": 0.3401,
+      "step": 5121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.901932389742406e-05,
+      "loss": 0.3159,
+      "step": 5122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.901732395012766e-05,
+      "loss": 0.3188,
+      "step": 5123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9015322011132024e-05,
+      "loss": 0.3378,
+      "step": 5124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.901331808062244e-05,
+      "loss": 0.3189,
+      "step": 5125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.90113121587844e-05,
+      "loss": 0.3231,
+      "step": 5126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9009304245803584e-05,
+      "loss": 0.3221,
+      "step": 5127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.900729434186585e-05,
+      "loss": 0.3358,
+      "step": 5128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.900528244715723e-05,
+      "loss": 0.3194,
+      "step": 5129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.900326856186396e-05,
+      "loss": 0.3155,
+      "step": 5130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9001252686172456e-05,
+      "loss": 0.3115,
+      "step": 5131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.899923482026929e-05,
+      "loss": 0.3224,
+      "step": 5132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.899721496434126e-05,
+      "loss": 0.3273,
+      "step": 5133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.8995193118575325e-05,
+      "loss": 0.3175,
+      "step": 5134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.899316928315863e-05,
+      "loss": 0.331,
+      "step": 5135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.899114345827851e-05,
+      "loss": 0.3315,
+      "step": 5136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8989115644122474e-05,
+      "loss": 0.3197,
+      "step": 5137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8987085840878226e-05,
+      "loss": 0.3254,
+      "step": 5138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898505404873366e-05,
+      "loss": 0.3282,
+      "step": 5139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898302026787682e-05,
+      "loss": 0.3245,
+      "step": 5140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898098449849597e-05,
+      "loss": 0.3391,
+      "step": 5141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.897894674077956e-05,
+      "loss": 0.3255,
+      "step": 5142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.897690699491619e-05,
+      "loss": 0.3312,
+      "step": 5143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8974865261094676e-05,
+      "loss": 0.3065,
+      "step": 5144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8972821539504e-05,
+      "loss": 0.3288,
+      "step": 5145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.897077583033334e-05,
+      "loss": 0.3358,
+      "step": 5146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896872813377204e-05,
+      "loss": 0.3198,
+      "step": 5147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896667845000966e-05,
+      "loss": 0.3509,
+      "step": 5148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.89646267792359e-05,
+      "loss": 0.3271,
+      "step": 5149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8962573121640684e-05,
+      "loss": 0.3808,
+      "step": 5150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896051747741411e-05,
+      "loss": 0.3309,
+      "step": 5151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.895845984674643e-05,
+      "loss": 0.3192,
+      "step": 5152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8956400229828124e-05,
+      "loss": 0.3174,
+      "step": 5153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.895433862684983e-05,
+      "loss": 0.3312,
+      "step": 5154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8952275038002384e-05,
+      "loss": 0.3169,
+      "step": 5155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.895020946347677e-05,
+      "loss": 0.3015,
+      "step": 5156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.894814190346421e-05,
+      "loss": 0.3321,
+      "step": 5157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.894607235815608e-05,
+      "loss": 0.3333,
+      "step": 5158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8944000827743924e-05,
+      "loss": 0.3321,
+      "step": 5159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.894192731241951e-05,
+      "loss": 0.3296,
+      "step": 5160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.893985181237475e-05,
+      "loss": 0.3304,
+      "step": 5161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.893777432780177e-05,
+      "loss": 0.3348,
+      "step": 5162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.893569485889287e-05,
+      "loss": 0.3224,
+      "step": 5163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8933613405840506e-05,
+      "loss": 0.3175,
+      "step": 5164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.893152996883737e-05,
+      "loss": 0.3271,
+      "step": 5165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.89294445480763e-05,
+      "loss": 0.3382,
+      "step": 5166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.892735714375034e-05,
+      "loss": 0.3261,
+      "step": 5167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8925267756052685e-05,
+      "loss": 0.3294,
+      "step": 5168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.892317638517674e-05,
+      "loss": 0.3112,
+      "step": 5169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.89210830313161e-05,
+      "loss": 0.3311,
+      "step": 5170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8918987694664515e-05,
+      "loss": 0.3228,
+      "step": 5171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8916890375415944e-05,
+      "loss": 0.3152,
+      "step": 5172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.891479107376452e-05,
+      "loss": 0.318,
+      "step": 5173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8912689789904566e-05,
+      "loss": 0.3516,
+      "step": 5174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.891058652403056e-05,
+      "loss": 0.3351,
+      "step": 5175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.890848127633721e-05,
+      "loss": 0.3448,
+      "step": 5176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8906374047019374e-05,
+      "loss": 0.3362,
+      "step": 5177
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.89042648362721e-05,
+      "loss": 0.3191,
+      "step": 5178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.890215364429062e-05,
+      "loss": 0.3284,
+      "step": 5179
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.890004047127037e-05,
+      "loss": 0.3206,
+      "step": 5180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.889792531740693e-05,
+      "loss": 0.3316,
+      "step": 5181
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8895808182896086e-05,
+      "loss": 0.3399,
+      "step": 5182
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.889368906793381e-05,
+      "loss": 0.3307,
+      "step": 5183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.889156797271626e-05,
+      "loss": 0.3103,
+      "step": 5184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8889444897439756e-05,
+      "loss": 0.3275,
+      "step": 5185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888731984230082e-05,
+      "loss": 0.3157,
+      "step": 5186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888519280749616e-05,
+      "loss": 0.3207,
+      "step": 5187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888306379322266e-05,
+      "loss": 0.2987,
+      "step": 5188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888093279967737e-05,
+      "loss": 0.3451,
+      "step": 5189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.887879982705756e-05,
+      "loss": 0.3274,
+      "step": 5190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.887666487556065e-05,
+      "loss": 0.3296,
+      "step": 5191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.887452794538427e-05,
+      "loss": 0.3255,
+      "step": 5192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.887238903672621e-05,
+      "loss": 0.3211,
+      "step": 5193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.887024814978445e-05,
+      "loss": 0.3136,
+      "step": 5194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8868105284757163e-05,
+      "loss": 0.3393,
+      "step": 5195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8865960441842695e-05,
+      "loss": 0.3375,
+      "step": 5196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.886381362123959e-05,
+      "loss": 0.323,
+      "step": 5197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.886166482314654e-05,
+      "loss": 0.3262,
+      "step": 5198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.885951404776246e-05,
+      "loss": 0.332,
+      "step": 5199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.885736129528642e-05,
+      "loss": 0.3301,
+      "step": 5200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8855206565917704e-05,
+      "loss": 0.338,
+      "step": 5201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.885304985985575e-05,
+      "loss": 0.3279,
+      "step": 5202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8850891177300176e-05,
+      "loss": 0.3066,
+      "step": 5203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.88487305184508e-05,
+      "loss": 0.3319,
+      "step": 5204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.884656788350763e-05,
+      "loss": 0.3292,
+      "step": 5205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.884440327267084e-05,
+      "loss": 0.3334,
+      "step": 5206
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8842236686140784e-05,
+      "loss": 0.3334,
+      "step": 5207
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8840068124118014e-05,
+      "loss": 0.3314,
+      "step": 5208
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.883789758680326e-05,
+      "loss": 0.3277,
+      "step": 5209
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8835725074397424e-05,
+      "loss": 0.297,
+      "step": 5210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.883355058710161e-05,
+      "loss": 0.3277,
+      "step": 5211
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8831374125117083e-05,
+      "loss": 0.3095,
+      "step": 5212
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.882919568864531e-05,
+      "loss": 0.3256,
+      "step": 5213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.882701527788794e-05,
+      "loss": 0.3328,
+      "step": 5214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8824832893046773e-05,
+      "loss": 0.3258,
+      "step": 5215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8822648534323833e-05,
+      "loss": 0.3251,
+      "step": 5216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.882046220192131e-05,
+      "loss": 0.3187,
+      "step": 5217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8818273896041575e-05,
+      "loss": 0.2975,
+      "step": 5218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.881608361688718e-05,
+      "loss": 0.3259,
+      "step": 5219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8813891364660876e-05,
+      "loss": 0.3189,
+      "step": 5220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.881169713956556e-05,
+      "loss": 0.3097,
+      "step": 5221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8809500941804354e-05,
+      "loss": 0.3305,
+      "step": 5222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8807302771580535e-05,
+      "loss": 0.3253,
+      "step": 5223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8805102629097574e-05,
+      "loss": 0.3152,
+      "step": 5224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.880290051455913e-05,
+      "loss": 0.3223,
+      "step": 5225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.880069642816902e-05,
+      "loss": 0.3269,
+      "step": 5226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.879849037013127e-05,
+      "loss": 0.3227,
+      "step": 5227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.879628234065008e-05,
+      "loss": 0.32,
+      "step": 5228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.879407233992982e-05,
+      "loss": 0.3412,
+      "step": 5229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.879186036817507e-05,
+      "loss": 0.3361,
+      "step": 5230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.878964642559057e-05,
+      "loss": 0.3139,
+      "step": 5231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.878743051238123e-05,
+      "loss": 0.3345,
+      "step": 5232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.878521262875219e-05,
+      "loss": 0.311,
+      "step": 5233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8782992774908724e-05,
+      "loss": 0.3278,
+      "step": 5234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.878077095105632e-05,
+      "loss": 0.3195,
+      "step": 5235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.877854715740062e-05,
+      "loss": 0.3204,
+      "step": 5236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.877632139414747e-05,
+      "loss": 0.3079,
+      "step": 5237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.877409366150291e-05,
+      "loss": 0.3287,
+      "step": 5238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.877186395967312e-05,
+      "loss": 0.3352,
+      "step": 5239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.876963228886449e-05,
+      "loss": 0.3358,
+      "step": 5240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.87673986492836e-05,
+      "loss": 0.311,
+      "step": 5241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8765163041137196e-05,
+      "loss": 0.3187,
+      "step": 5242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.876292546463222e-05,
+      "loss": 0.3192,
+      "step": 5243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8760685919975773e-05,
+      "loss": 0.3352,
+      "step": 5244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.875844440737517e-05,
+      "loss": 0.3119,
+      "step": 5245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.875620092703788e-05,
+      "loss": 0.3276,
+      "step": 5246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.875395547917156e-05,
+      "loss": 0.3129,
+      "step": 5247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.875170806398406e-05,
+      "loss": 0.312,
+      "step": 5248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.874945868168341e-05,
+      "loss": 0.3225,
+      "step": 5249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8747207332477825e-05,
+      "loss": 0.3254,
+      "step": 5250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8744954016575685e-05,
+      "loss": 0.3181,
+      "step": 5251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8742698734185564e-05,
+      "loss": 0.3342,
+      "step": 5252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.874044148551621e-05,
+      "loss": 0.3071,
+      "step": 5253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.873818227077658e-05,
+      "loss": 0.3272,
+      "step": 5254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8735921090175784e-05,
+      "loss": 0.329,
+      "step": 5255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.873365794392311e-05,
+      "loss": 0.3263,
+      "step": 5256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8731392832228055e-05,
+      "loss": 0.3275,
+      "step": 5257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8729125755300284e-05,
+      "loss": 0.3254,
+      "step": 5258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8726856713349636e-05,
+      "loss": 0.3317,
+      "step": 5259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8724585706586126e-05,
+      "loss": 0.3322,
+      "step": 5260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8722312735219996e-05,
+      "loss": 0.3339,
+      "step": 5261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.872003779946162e-05,
+      "loss": 0.3291,
+      "step": 5262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.871776089952157e-05,
+      "loss": 0.3213,
+      "step": 5263
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.871548203561061e-05,
+      "loss": 0.3469,
+      "step": 5264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.871320120793967e-05,
+      "loss": 0.3078,
+      "step": 5265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.871091841671987e-05,
+      "loss": 0.3217,
+      "step": 5266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8708633662162525e-05,
+      "loss": 0.3271,
+      "step": 5267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.87063469444791e-05,
+      "loss": 0.328,
+      "step": 5268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.870405826388126e-05,
+      "loss": 0.3347,
+      "step": 5269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.870176762058086e-05,
+      "loss": 0.3246,
+      "step": 5270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8699475014789936e-05,
+      "loss": 0.3141,
+      "step": 5271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.869718044672068e-05,
+      "loss": 0.3251,
+      "step": 5272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8694883916585486e-05,
+      "loss": 0.3422,
+      "step": 5273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8692585424596936e-05,
+      "loss": 0.3094,
+      "step": 5274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.869028497096778e-05,
+      "loss": 0.3372,
+      "step": 5275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.868798255591095e-05,
+      "loss": 0.3367,
+      "step": 5276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.868567817963957e-05,
+      "loss": 0.3322,
+      "step": 5277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.868337184236693e-05,
+      "loss": 0.3331,
+      "step": 5278
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.868106354430652e-05,
+      "loss": 0.3412,
+      "step": 5279
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8678753285671996e-05,
+      "loss": 0.338,
+      "step": 5280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8676441066677206e-05,
+      "loss": 0.319,
+      "step": 5281
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.867412688753617e-05,
+      "loss": 0.335,
+      "step": 5282
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.86718107484631e-05,
+      "loss": 0.3209,
+      "step": 5283
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8669492649672384e-05,
+      "loss": 0.3192,
+      "step": 5284
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.866717259137858e-05,
+      "loss": 0.3535,
+      "step": 5285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.866485057379645e-05,
+      "loss": 0.3472,
+      "step": 5286
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.866252659714092e-05,
+      "loss": 0.3273,
+      "step": 5287
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.866020066162711e-05,
+      "loss": 0.3113,
+      "step": 5288
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.865787276747031e-05,
+      "loss": 0.3168,
+      "step": 5289
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8655542914885995e-05,
+      "loss": 0.3144,
+      "step": 5290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.865321110408983e-05,
+      "loss": 0.3399,
+      "step": 5291
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.865087733529764e-05,
+      "loss": 0.323,
+      "step": 5292
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.864854160872546e-05,
+      "loss": 0.3292,
+      "step": 5293
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.864620392458947e-05,
+      "loss": 0.3257,
+      "step": 5294
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.864386428310607e-05,
+      "loss": 0.335,
+      "step": 5295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8641522684491816e-05,
+      "loss": 0.3119,
+      "step": 5296
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8639179128963466e-05,
+      "loss": 0.3196,
+      "step": 5297
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.863683361673793e-05,
+      "loss": 0.3371,
+      "step": 5298
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8634486148032314e-05,
+      "loss": 0.325,
+      "step": 5299
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.863213672306391e-05,
+      "loss": 0.3336,
+      "step": 5300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.862978534205019e-05,
+      "loss": 0.3248,
+      "step": 5301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.862743200520881e-05,
+      "loss": 0.3221,
+      "step": 5302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.8625076712757575e-05,
+      "loss": 0.3248,
+      "step": 5303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.862271946491452e-05,
+      "loss": 0.3549,
+      "step": 5304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.862036026189784e-05,
+      "loss": 0.3099,
+      "step": 5305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.86179991039259e-05,
+      "loss": 0.3291,
+      "step": 5306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.861563599121725e-05,
+      "loss": 0.3202,
+      "step": 5307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8613270923990636e-05,
+      "loss": 0.3087,
+      "step": 5308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8610903902464966e-05,
+      "loss": 0.3228,
+      "step": 5309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.860853492685935e-05,
+      "loss": 0.3184,
+      "step": 5310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.860616399739305e-05,
+      "loss": 0.3286,
+      "step": 5311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.860379111428555e-05,
+      "loss": 0.3294,
+      "step": 5312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.860141627775646e-05,
+      "loss": 0.3184,
+      "step": 5313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8599039488025615e-05,
+      "loss": 0.3177,
+      "step": 5314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8596660745313024e-05,
+      "loss": 0.3305,
+      "step": 5315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8594280049838865e-05,
+      "loss": 0.3101,
+      "step": 5316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.859189740182349e-05,
+      "loss": 0.3282,
+      "step": 5317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.858951280148745e-05,
+      "loss": 0.3389,
+      "step": 5318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8587126249051484e-05,
+      "loss": 0.3096,
+      "step": 5319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.858473774473648e-05,
+      "loss": 0.3191,
+      "step": 5320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.858234728876353e-05,
+      "loss": 0.3343,
+      "step": 5321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.85799548813539e-05,
+      "loss": 0.3272,
+      "step": 5322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.857756052272904e-05,
+      "loss": 0.3241,
+      "step": 5323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.857516421311058e-05,
+      "loss": 0.3235,
+      "step": 5324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.857276595272032e-05,
+      "loss": 0.3226,
+      "step": 5325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.857036574178025e-05,
+      "loss": 0.3094,
+      "step": 5326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.856796358051255e-05,
+      "loss": 0.3352,
+      "step": 5327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.856555946913956e-05,
+      "loss": 0.3409,
+      "step": 5328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.856315340788382e-05,
+      "loss": 0.3273,
+      "step": 5329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.856074539696803e-05,
+      "loss": 0.3228,
+      "step": 5330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.855833543661508e-05,
+      "loss": 0.3347,
+      "step": 5331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.855592352704806e-05,
+      "loss": 0.3181,
+      "step": 5332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.855350966849021e-05,
+      "loss": 0.3187,
+      "step": 5333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.855109386116497e-05,
+      "loss": 0.3201,
+      "step": 5334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8548676105295946e-05,
+      "loss": 0.3312,
+      "step": 5335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8546256401106936e-05,
+      "loss": 0.3199,
+      "step": 5336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8543834748821904e-05,
+      "loss": 0.31,
+      "step": 5337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.854141114866502e-05,
+      "loss": 0.3586,
+      "step": 5338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.853898560086061e-05,
+      "loss": 0.3244,
+      "step": 5339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.85365581056332e-05,
+      "loss": 0.3142,
+      "step": 5340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.853412866320746e-05,
+      "loss": 0.3219,
+      "step": 5341
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.853169727380829e-05,
+      "loss": 0.3151,
+      "step": 5342
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.852926393766074e-05,
+      "loss": 0.3132,
+      "step": 5343
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8526828654990033e-05,
+      "loss": 0.3286,
+      "step": 5344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8524391426021605e-05,
+      "loss": 0.3131,
+      "step": 5345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.852195225098104e-05,
+      "loss": 0.3214,
+      "step": 5346
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8519511130094115e-05,
+      "loss": 0.3395,
+      "step": 5347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8517068063586786e-05,
+      "loss": 0.2999,
+      "step": 5348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.851462305168519e-05,
+      "loss": 0.3176,
+      "step": 5349
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.851217609461566e-05,
+      "loss": 0.3332,
+      "step": 5350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.850972719260466e-05,
+      "loss": 0.3304,
+      "step": 5351
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8507276345878894e-05,
+      "loss": 0.3217,
+      "step": 5352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.850482355466521e-05,
+      "loss": 0.3293,
+      "step": 5353
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.850236881919064e-05,
+      "loss": 0.3224,
+      "step": 5354
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.849991213968241e-05,
+      "loss": 0.3243,
+      "step": 5355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8497453516367906e-05,
+      "loss": 0.3155,
+      "step": 5356
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.849499294947471e-05,
+      "loss": 0.3184,
+      "step": 5357
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.849253043923058e-05,
+      "loss": 0.3191,
+      "step": 5358
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.849006598586345e-05,
+      "loss": 0.3105,
+      "step": 5359
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848759958960144e-05,
+      "loss": 0.2954,
+      "step": 5360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848513125067285e-05,
+      "loss": 0.3231,
+      "step": 5361
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848266096930614e-05,
+      "loss": 0.3053,
+      "step": 5362
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.848018874572997e-05,
+      "loss": 0.3273,
+      "step": 5363
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.847771458017318e-05,
+      "loss": 0.3435,
+      "step": 5364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.84752384728648e-05,
+      "loss": 0.3245,
+      "step": 5365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8472760424033996e-05,
+      "loss": 0.3279,
+      "step": 5366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.847028043391016e-05,
+      "loss": 0.3218,
+      "step": 5367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.846779850272284e-05,
+      "loss": 0.3424,
+      "step": 5368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.846531463070178e-05,
+      "loss": 0.3037,
+      "step": 5369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8462828818076885e-05,
+      "loss": 0.3187,
+      "step": 5370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8460341065078246e-05,
+      "loss": 0.323,
+      "step": 5371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.845785137193614e-05,
+      "loss": 0.3151,
+      "step": 5372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.845535973888102e-05,
+      "loss": 0.3187,
+      "step": 5373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8452866166143526e-05,
+      "loss": 0.3203,
+      "step": 5374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.845037065395445e-05,
+      "loss": 0.3176,
+      "step": 5375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.844787320254479e-05,
+      "loss": 0.3409,
+      "step": 5376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.844537381214573e-05,
+      "loss": 0.3012,
+      "step": 5377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.84428724829886e-05,
+      "loss": 0.3224,
+      "step": 5378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.844036921530495e-05,
+      "loss": 0.3387,
+      "step": 5379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843786400932648e-05,
+      "loss": 0.3278,
+      "step": 5380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843535686528507e-05,
+      "loss": 0.3242,
+      "step": 5381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.84328477834128e-05,
+      "loss": 0.3305,
+      "step": 5382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.843033676394191e-05,
+      "loss": 0.3024,
+      "step": 5383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8427823807104824e-05,
+      "loss": 0.3172,
+      "step": 5384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.842530891313416e-05,
+      "loss": 0.3194,
+      "step": 5385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8422792082262686e-05,
+      "loss": 0.3268,
+      "step": 5386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8420273314723386e-05,
+      "loss": 0.326,
+      "step": 5387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8417752610749384e-05,
+      "loss": 0.3297,
+      "step": 5388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.841522997057402e-05,
+      "loss": 0.3077,
+      "step": 5389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8412705394430784e-05,
+      "loss": 0.334,
+      "step": 5390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.841017888255337e-05,
+      "loss": 0.3381,
+      "step": 5391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8407650435175615e-05,
+      "loss": 0.3318,
+      "step": 5392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8405120052531584e-05,
+      "loss": 0.3325,
+      "step": 5393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.840258773485549e-05,
+      "loss": 0.2975,
+      "step": 5394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8400053482381715e-05,
+      "loss": 0.3393,
+      "step": 5395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.839751729534485e-05,
+      "loss": 0.3267,
+      "step": 5396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8394979173979655e-05,
+      "loss": 0.3219,
+      "step": 5397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.839243911852105e-05,
+      "loss": 0.3112,
+      "step": 5398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.838989712920417e-05,
+      "loss": 0.3161,
+      "step": 5399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.838735320626429e-05,
+      "loss": 0.3218,
+      "step": 5400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.838480734993689e-05,
+      "loss": 0.3311,
+      "step": 5401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.838225956045761e-05,
+      "loss": 0.3117,
+      "step": 5402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.837970983806229e-05,
+      "loss": 0.3289,
+      "step": 5403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.837715818298695e-05,
+      "loss": 0.3289,
+      "step": 5404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8374604595467754e-05,
+      "loss": 0.3326,
+      "step": 5405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.837204907574109e-05,
+      "loss": 0.3164,
+      "step": 5406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.836949162404348e-05,
+      "loss": 0.3381,
+      "step": 5407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8366932240611676e-05,
+      "loss": 0.311,
+      "step": 5408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.836437092568257e-05,
+      "loss": 0.3262,
+      "step": 5409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8361807679493226e-05,
+      "loss": 0.3297,
+      "step": 5410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8359242502280927e-05,
+      "loss": 0.3298,
+      "step": 5411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8356675394283106e-05,
+      "loss": 0.3262,
+      "step": 5412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8354106355737387e-05,
+      "loss": 0.3219,
+      "step": 5413
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.835153538688155e-05,
+      "loss": 0.3157,
+      "step": 5414
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.834896248795359e-05,
+      "loss": 0.3189,
+      "step": 5415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.834638765919165e-05,
+      "loss": 0.3139,
+      "step": 5416
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.834381090083406e-05,
+      "loss": 0.32,
+      "step": 5417
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8341232213119336e-05,
+      "loss": 0.3314,
+      "step": 5418
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.8338651596286175e-05,
+      "loss": 0.2989,
+      "step": 5419
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.833606905057343e-05,
+      "loss": 0.3121,
+      "step": 5420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.833348457622017e-05,
+      "loss": 0.3211,
+      "step": 5421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.83308981734656e-05,
+      "loss": 0.3355,
+      "step": 5422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8328309842549135e-05,
+      "loss": 0.3168,
+      "step": 5423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.832571958371035e-05,
+      "loss": 0.3392,
+      "step": 5424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8323127397189015e-05,
+      "loss": 0.2985,
+      "step": 5425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8320533283225066e-05,
+      "loss": 0.3408,
+      "step": 5426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.831793724205861e-05,
+      "loss": 0.3142,
+      "step": 5427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.831533927392997e-05,
+      "loss": 0.3271,
+      "step": 5428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8312739379079596e-05,
+      "loss": 0.3351,
+      "step": 5429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.831013755774814e-05,
+      "loss": 0.3212,
+      "step": 5430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.830753381017646e-05,
+      "loss": 0.3069,
+      "step": 5431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.830492813660553e-05,
+      "loss": 0.2957,
+      "step": 5432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.830232053727657e-05,
+      "loss": 0.3297,
+      "step": 5433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.829971101243093e-05,
+      "loss": 0.3379,
+      "step": 5434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.829709956231015e-05,
+      "loss": 0.3165,
+      "step": 5435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8294486187155967e-05,
+      "loss": 0.322,
+      "step": 5436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.829187088721026e-05,
+      "loss": 0.2934,
+      "step": 5437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.828925366271514e-05,
+      "loss": 0.308,
+      "step": 5438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.828663451391284e-05,
+      "loss": 0.3423,
+      "step": 5439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.82840134410458e-05,
+      "loss": 0.3067,
+      "step": 5440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.828139044435663e-05,
+      "loss": 0.3182,
+      "step": 5441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.827876552408812e-05,
+      "loss": 0.3178,
+      "step": 5442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.827613868048326e-05,
+      "loss": 0.3202,
+      "step": 5443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.827350991378518e-05,
+      "loss": 0.3311,
+      "step": 5444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8270879224237195e-05,
+      "loss": 0.3341,
+      "step": 5445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.826824661208283e-05,
+      "loss": 0.3365,
+      "step": 5446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8265612077565755e-05,
+      "loss": 0.3316,
+      "step": 5447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.826297562092984e-05,
+      "loss": 0.3064,
+      "step": 5448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8260337242419106e-05,
+      "loss": 0.3216,
+      "step": 5449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.825769694227778e-05,
+      "loss": 0.3359,
+      "step": 5450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8255054720750254e-05,
+      "loss": 0.3318,
+      "step": 5451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.82524105780811e-05,
+      "loss": 0.3241,
+      "step": 5452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8249764514515054e-05,
+      "loss": 0.3161,
+      "step": 5453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.824711653029706e-05,
+      "loss": 0.321,
+      "step": 5454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8244466625672205e-05,
+      "loss": 0.3292,
+      "step": 5455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.824181480088578e-05,
+      "loss": 0.3343,
+      "step": 5456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.823916105618326e-05,
+      "loss": 0.3305,
+      "step": 5457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.823650539181025e-05,
+      "loss": 0.3207,
+      "step": 5458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8233847808012586e-05,
+      "loss": 0.3318,
+      "step": 5459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.823118830503626e-05,
+      "loss": 0.3258,
+      "step": 5460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.822852688312744e-05,
+      "loss": 0.3095,
+      "step": 5461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.822586354253247e-05,
+      "loss": 0.3364,
+      "step": 5462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8223198283497886e-05,
+      "loss": 0.3211,
+      "step": 5463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.822053110627038e-05,
+      "loss": 0.3634,
+      "step": 5464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8217862011096844e-05,
+      "loss": 0.3136,
+      "step": 5465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.821519099822432e-05,
+      "loss": 0.3171,
+      "step": 5466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.821251806790007e-05,
+      "loss": 0.3296,
+      "step": 5467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8209843220371476e-05,
+      "loss": 0.3139,
+      "step": 5468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.820716645588615e-05,
+      "loss": 0.322,
+      "step": 5469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8204487774691855e-05,
+      "loss": 0.3079,
+      "step": 5470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.820180717703653e-05,
+      "loss": 0.2989,
+      "step": 5471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.819912466316831e-05,
+      "loss": 0.3123,
+      "step": 5472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.819644023333549e-05,
+      "loss": 0.3251,
+      "step": 5473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.819375388778654e-05,
+      "loss": 0.3468,
+      "step": 5474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.819106562677013e-05,
+      "loss": 0.3225,
+      "step": 5475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.818837545053508e-05,
+      "loss": 0.3235,
+      "step": 5476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.8185683359330406e-05,
+      "loss": 0.3457,
+      "step": 5477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.818298935340529e-05,
+      "loss": 0.32,
+      "step": 5478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.818029343300911e-05,
+      "loss": 0.3398,
+      "step": 5479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.817759559839139e-05,
+      "loss": 0.3168,
+      "step": 5480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8174895849801855e-05,
+      "loss": 0.3432,
+      "step": 5481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.817219418749041e-05,
+      "loss": 0.3381,
+      "step": 5482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.816949061170711e-05,
+      "loss": 0.3516,
+      "step": 5483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.816678512270221e-05,
+      "loss": 0.3277,
+      "step": 5484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8164077720726144e-05,
+      "loss": 0.3362,
+      "step": 5485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8161368406029514e-05,
+      "loss": 0.326,
+      "step": 5486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8158657178863115e-05,
+      "loss": 0.3263,
+      "step": 5487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.815594403947787e-05,
+      "loss": 0.3111,
+      "step": 5488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.815322898812494e-05,
+      "loss": 0.3296,
+      "step": 5489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8150512025055647e-05,
+      "loss": 0.3089,
+      "step": 5490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.814779315052145e-05,
+      "loss": 0.3309,
+      "step": 5491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.814507236477404e-05,
+      "loss": 0.3459,
+      "step": 5492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8142349668065235e-05,
+      "loss": 0.3308,
+      "step": 5493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.813962506064708e-05,
+      "loss": 0.323,
+      "step": 5494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.813689854277176e-05,
+      "loss": 0.3284,
+      "step": 5495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.813417011469165e-05,
+      "loss": 0.2979,
+      "step": 5496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.813143977665931e-05,
+      "loss": 0.3235,
+      "step": 5497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8128707528927445e-05,
+      "loss": 0.3289,
+      "step": 5498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.812597337174898e-05,
+      "loss": 0.3192,
+      "step": 5499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8123237305376975e-05,
+      "loss": 0.3098,
+      "step": 5500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.812049933006471e-05,
+      "loss": 0.3248,
+      "step": 5501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.811775944606561e-05,
+      "loss": 0.3074,
+      "step": 5502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8115017653633285e-05,
+      "loss": 0.3126,
+      "step": 5503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.811227395302152e-05,
+      "loss": 0.3284,
+      "step": 5504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.810952834448429e-05,
+      "loss": 0.3247,
+      "step": 5505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.810678082827571e-05,
+      "loss": 0.3164,
+      "step": 5506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8104031404650126e-05,
+      "loss": 0.3139,
+      "step": 5507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.810128007386202e-05,
+      "loss": 0.3306,
+      "step": 5508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8098526836166055e-05,
+      "loss": 0.3211,
+      "step": 5509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.809577169181709e-05,
+      "loss": 0.3182,
+      "step": 5510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.809301464107014e-05,
+      "loss": 0.3014,
+      "step": 5511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.809025568418041e-05,
+      "loss": 0.3262,
+      "step": 5512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8087494821403275e-05,
+      "loss": 0.333,
+      "step": 5513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.808473205299428e-05,
+      "loss": 0.3337,
+      "step": 5514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.808196737920917e-05,
+      "loss": 0.3165,
+      "step": 5515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.807920080030384e-05,
+      "loss": 0.3135,
+      "step": 5516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.807643231653437e-05,
+      "loss": 0.352,
+      "step": 5517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8073661928157027e-05,
+      "loss": 0.32,
+      "step": 5518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.807088963542824e-05,
+      "loss": 0.3344,
+      "step": 5519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.80681154386046e-05,
+      "loss": 0.3217,
+      "step": 5520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.806533933794293e-05,
+      "loss": 0.3147,
+      "step": 5521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8062561333700166e-05,
+      "loss": 0.32,
+      "step": 5522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8059781426133465e-05,
+      "loss": 0.3227,
+      "step": 5523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.805699961550013e-05,
+      "loss": 0.3153,
+      "step": 5524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8054215902057664e-05,
+      "loss": 0.2952,
+      "step": 5525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.805143028606373e-05,
+      "loss": 0.3209,
+      "step": 5526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.804864276777615e-05,
+      "loss": 0.3181,
+      "step": 5527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.804585334745298e-05,
+      "loss": 0.3292,
+      "step": 5528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.80430620253524e-05,
+      "loss": 0.3375,
+      "step": 5529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.804026880173277e-05,
+      "loss": 0.3295,
+      "step": 5530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.803747367685266e-05,
+      "loss": 0.3346,
+      "step": 5531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8034676650970784e-05,
+      "loss": 0.3058,
+      "step": 5532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.803187772434604e-05,
+      "loss": 0.3367,
+      "step": 5533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8029076897237496e-05,
+      "loss": 0.3325,
+      "step": 5534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.802627416990442e-05,
+      "loss": 0.3202,
+      "step": 5535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.802346954260624e-05,
+      "loss": 0.318,
+      "step": 5536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.802066301560254e-05,
+      "loss": 0.3304,
+      "step": 5537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.801785458915312e-05,
+      "loss": 0.3375,
+      "step": 5538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.801504426351793e-05,
+      "loss": 0.3167,
+      "step": 5539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.801223203895709e-05,
+      "loss": 0.32,
+      "step": 5540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.800941791573092e-05,
+      "loss": 0.3343,
+      "step": 5541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8006601894099896e-05,
+      "loss": 0.3252,
+      "step": 5542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.800378397432468e-05,
+      "loss": 0.3222,
+      "step": 5543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8000964156666094e-05,
+      "loss": 0.335,
+      "step": 5544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.799814244138516e-05,
+      "loss": 0.3211,
+      "step": 5545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7995318828743074e-05,
+      "loss": 0.3277,
+      "step": 5546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.799249331900117e-05,
+      "loss": 0.3489,
+      "step": 5547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7989665912421e-05,
+      "loss": 0.3235,
+      "step": 5548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7986836609264285e-05,
+      "loss": 0.3169,
+      "step": 5549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.798400540979289e-05,
+      "loss": 0.3003,
+      "step": 5550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.798117231426889e-05,
+      "loss": 0.3153,
+      "step": 5551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7978337322954536e-05,
+      "loss": 0.3128,
+      "step": 5552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7975500436112214e-05,
+      "loss": 0.3159,
+      "step": 5553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7972661654004536e-05,
+      "loss": 0.3432,
+      "step": 5554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.796982097689426e-05,
+      "loss": 0.312,
+      "step": 5555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.796697840504433e-05,
+      "loss": 0.3307,
+      "step": 5556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.796413393871785e-05,
+      "loss": 0.3226,
+      "step": 5557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7961287578178126e-05,
+      "loss": 0.3287,
+      "step": 5558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.795843932368862e-05,
+      "loss": 0.3216,
+      "step": 5559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7955589175512974e-05,
+      "loss": 0.3223,
+      "step": 5560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7952737133914995e-05,
+      "loss": 0.3217,
+      "step": 5561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.794988319915868e-05,
+      "loss": 0.3211,
+      "step": 5562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.794702737150821e-05,
+      "loss": 0.3027,
+      "step": 5563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7944169651227915e-05,
+      "loss": 0.3177,
+      "step": 5564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.794131003858232e-05,
+      "loss": 0.3179,
+      "step": 5565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7938448533836104e-05,
+      "loss": 0.3057,
+      "step": 5566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7935585137254155e-05,
+      "loss": 0.3199,
+      "step": 5567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.79327198491015e-05,
+      "loss": 0.3232,
+      "step": 5568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.792985266964337e-05,
+      "loss": 0.317,
+      "step": 5569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.792698359914515e-05,
+      "loss": 0.316,
+      "step": 5570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.792411263787241e-05,
+      "loss": 0.3345,
+      "step": 5571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7921239786090896e-05,
+      "loss": 0.3156,
+      "step": 5572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7918365044066526e-05,
+      "loss": 0.3337,
+      "step": 5573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7915488412065394e-05,
+      "loss": 0.3332,
+      "step": 5574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.791260989035376e-05,
+      "loss": 0.3205,
+      "step": 5575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.790972947919809e-05,
+      "loss": 0.3114,
+      "step": 5576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7906847178864975e-05,
+      "loss": 0.3224,
+      "step": 5577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7903962989621224e-05,
+      "loss": 0.3262,
+      "step": 5578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.79010769117338e-05,
+      "loss": 0.3095,
+      "step": 5579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.789818894546986e-05,
+      "loss": 0.357,
+      "step": 5580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.78952990910967e-05,
+      "loss": 0.3147,
+      "step": 5581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.789240734888182e-05,
+      "loss": 0.3205,
+      "step": 5582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.788951371909289e-05,
+      "loss": 0.3123,
+      "step": 5583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7886618201997747e-05,
+      "loss": 0.3172,
+      "step": 5584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7883720797864404e-05,
+      "loss": 0.3186,
+      "step": 5585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.788082150696108e-05,
+      "loss": 0.3277,
+      "step": 5586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.787792032955611e-05,
+      "loss": 0.3351,
+      "step": 5587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7875017265918036e-05,
+      "loss": 0.3163,
+      "step": 5588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7872112316315596e-05,
+      "loss": 0.3313,
+      "step": 5589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.786920548101766e-05,
+      "loss": 0.3078,
+      "step": 5590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.786629676029329e-05,
+      "loss": 0.3214,
+      "step": 5591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.786338615441174e-05,
+      "loss": 0.3306,
+      "step": 5592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.786047366364242e-05,
+      "loss": 0.3448,
+      "step": 5593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.785755928825491e-05,
+      "loss": 0.3084,
+      "step": 5594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7854643028518974e-05,
+      "loss": 0.3444,
+      "step": 5595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.785172488470455e-05,
+      "loss": 0.3186,
+      "step": 5596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7848804857081746e-05,
+      "loss": 0.3004,
+      "step": 5597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.784588294592085e-05,
+      "loss": 0.3267,
+      "step": 5598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.784295915149233e-05,
+      "loss": 0.3082,
+      "step": 5599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.78400334740668e-05,
+      "loss": 0.3056,
+      "step": 5600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7837105913915086e-05,
+      "loss": 0.3214,
+      "step": 5601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.783417647130817e-05,
+      "loss": 0.3351,
+      "step": 5602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.78312451465172e-05,
+      "loss": 0.3261,
+      "step": 5603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7828311939813515e-05,
+      "loss": 0.3251,
+      "step": 5604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.782537685146861e-05,
+      "loss": 0.3214,
+      "step": 5605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.782243988175417e-05,
+      "loss": 0.3103,
+      "step": 5606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.781950103094205e-05,
+      "loss": 0.3055,
+      "step": 5607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.781656029930427e-05,
+      "loss": 0.3261,
+      "step": 5608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7813617687113044e-05,
+      "loss": 0.3112,
+      "step": 5609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.781067319464075e-05,
+      "loss": 0.3197,
+      "step": 5610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.780772682215992e-05,
+      "loss": 0.3329,
+      "step": 5611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.780477856994329e-05,
+      "loss": 0.3427,
+      "step": 5612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7801828438263755e-05,
+      "loss": 0.3375,
+      "step": 5613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.779887642739438e-05,
+      "loss": 0.3237,
+      "step": 5614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.779592253760843e-05,
+      "loss": 0.327,
+      "step": 5615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.779296676917931e-05,
+      "loss": 0.3313,
+      "step": 5616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7790009122380614e-05,
+      "loss": 0.3248,
+      "step": 5617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.778704959748611e-05,
+      "loss": 0.3509,
+      "step": 5618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.778408819476974e-05,
+      "loss": 0.3174,
+      "step": 5619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.778112491450562e-05,
+      "loss": 0.3172,
+      "step": 5620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7778159756968046e-05,
+      "loss": 0.3271,
+      "step": 5621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7775192722431474e-05,
+      "loss": 0.3303,
+      "step": 5622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.777222381117053e-05,
+      "loss": 0.3248,
+      "step": 5623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.776925302346004e-05,
+      "loss": 0.313,
+      "step": 5624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.776628035957498e-05,
+      "loss": 0.3147,
+      "step": 5625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7763305819790515e-05,
+      "loss": 0.3141,
+      "step": 5626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.776032940438197e-05,
+      "loss": 0.3358,
+      "step": 5627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7757351113624846e-05,
+      "loss": 0.3114,
+      "step": 5628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.775437094779483e-05,
+      "loss": 0.3262,
+      "step": 5629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7751388907167765e-05,
+      "loss": 0.3165,
+      "step": 5630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.774840499201969e-05,
+      "loss": 0.3406,
+      "step": 5631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.774541920262679e-05,
+      "loss": 0.3016,
+      "step": 5632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7742431539265434e-05,
+      "loss": 0.319,
+      "step": 5633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.773944200221218e-05,
+      "loss": 0.315,
+      "step": 5634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7736450591743745e-05,
+      "loss": 0.3313,
+      "step": 5635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.773345730813702e-05,
+      "loss": 0.3295,
+      "step": 5636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.773046215166908e-05,
+      "loss": 0.3219,
+      "step": 5637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7727465122617144e-05,
+      "loss": 0.326,
+      "step": 5638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.772446622125864e-05,
+      "loss": 0.3184,
+      "step": 5639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7721465447871156e-05,
+      "loss": 0.3183,
+      "step": 5640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7718462802732436e-05,
+      "loss": 0.3188,
+      "step": 5641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.771545828612044e-05,
+      "loss": 0.2976,
+      "step": 5642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.771245189831323e-05,
+      "loss": 0.3257,
+      "step": 5643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.770944363958912e-05,
+      "loss": 0.3178,
+      "step": 5644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7706433510226564e-05,
+      "loss": 0.3149,
+      "step": 5645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.770342151050417e-05,
+      "loss": 0.323,
+      "step": 5646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.770040764070074e-05,
+      "loss": 0.3209,
+      "step": 5647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.769739190109526e-05,
+      "loss": 0.3306,
+      "step": 5648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.769437429196687e-05,
+      "loss": 0.3026,
+      "step": 5649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7691354813594854e-05,
+      "loss": 0.3415,
+      "step": 5650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.768833346625875e-05,
+      "loss": 0.3454,
+      "step": 5651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7685310250238194e-05,
+      "loss": 0.3091,
+      "step": 5652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.768228516581304e-05,
+      "loss": 0.3214,
+      "step": 5653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.767925821326328e-05,
+      "loss": 0.3268,
+      "step": 5654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.76762293928691e-05,
+      "loss": 0.3115,
+      "step": 5655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.767319870491087e-05,
+      "loss": 0.3284,
+      "step": 5656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.767016614966911e-05,
+      "loss": 0.3226,
+      "step": 5657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.766713172742451e-05,
+      "loss": 0.3192,
+      "step": 5658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.766409543845796e-05,
+      "loss": 0.3054,
+      "step": 5659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7661057283050505e-05,
+      "loss": 0.3317,
+      "step": 5660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.765801726148336e-05,
+      "loss": 0.3326,
+      "step": 5661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7654975374037907e-05,
+      "loss": 0.3253,
+      "step": 5662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.765193162099573e-05,
+      "loss": 0.3235,
+      "step": 5663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7648886002638554e-05,
+      "loss": 0.319,
+      "step": 5664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7645838519248294e-05,
+      "loss": 0.3254,
+      "step": 5665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7642789171107026e-05,
+      "loss": 0.3083,
+      "step": 5666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.763973795849702e-05,
+      "loss": 0.314,
+      "step": 5667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.763668488170069e-05,
+      "loss": 0.3331,
+      "step": 5668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.763362994100066e-05,
+      "loss": 0.3355,
+      "step": 5669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7630573136679666e-05,
+      "loss": 0.3312,
+      "step": 5670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.762751446902068e-05,
+      "loss": 0.3156,
+      "step": 5671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7624453938306815e-05,
+      "loss": 0.3012,
+      "step": 5672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7621391544821356e-05,
+      "loss": 0.3302,
+      "step": 5673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.761832728884778e-05,
+      "loss": 0.307,
+      "step": 5674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.76152611706697e-05,
+      "loss": 0.3418,
+      "step": 5675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.761219319057095e-05,
+      "loss": 0.3184,
+      "step": 5676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.760912334883549e-05,
+      "loss": 0.3154,
+      "step": 5677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7606051645747484e-05,
+      "loss": 0.3268,
+      "step": 5678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7602978081591246e-05,
+      "loss": 0.3313,
+      "step": 5679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7599902656651285e-05,
+      "loss": 0.3166,
+      "step": 5680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.759682537121227e-05,
+      "loss": 0.3393,
+      "step": 5681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.759374622555903e-05,
+      "loss": 0.3043,
+      "step": 5682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.759066521997658e-05,
+      "loss": 0.3291,
+      "step": 5683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.758758235475011e-05,
+      "loss": 0.3199,
+      "step": 5684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7584497630165e-05,
+      "loss": 0.3312,
+      "step": 5685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7581411046506745e-05,
+      "loss": 0.3251,
+      "step": 5686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.757832260406106e-05,
+      "loss": 0.3291,
+      "step": 5687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.757523230311383e-05,
+      "loss": 0.305,
+      "step": 5688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.757214014395109e-05,
+      "loss": 0.3105,
+      "step": 5689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.756904612685906e-05,
+      "loss": 0.3084,
+      "step": 5690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.756595025212412e-05,
+      "loss": 0.3395,
+      "step": 5691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7562852520032856e-05,
+      "loss": 0.3197,
+      "step": 5692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7559752930871984e-05,
+      "loss": 0.3233,
+      "step": 5693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.755665148492842e-05,
+      "loss": 0.3096,
+      "step": 5694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.755354818248924e-05,
+      "loss": 0.3056,
+      "step": 5695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7550443023841676e-05,
+      "loss": 0.3504,
+      "step": 5696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7547336009273184e-05,
+      "loss": 0.326,
+      "step": 5697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.754422713907132e-05,
+      "loss": 0.3183,
+      "step": 5698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.754111641352388e-05,
+      "loss": 0.3218,
+      "step": 5699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.753800383291878e-05,
+      "loss": 0.3192,
+      "step": 5700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.753488939754415e-05,
+      "loss": 0.3173,
+      "step": 5701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.753177310768824e-05,
+      "loss": 0.3256,
+      "step": 5702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7528654963639527e-05,
+      "loss": 0.3168,
+      "step": 5703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7525534965686614e-05,
+      "loss": 0.3264,
+      "step": 5704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7522413114118316e-05,
+      "loss": 0.3153,
+      "step": 5705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.751928940922359e-05,
+      "loss": 0.3228,
+      "step": 5706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.751616385129158e-05,
+      "loss": 0.3276,
+      "step": 5707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.75130364406116e-05,
+      "loss": 0.3254,
+      "step": 5708
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7509907177473104e-05,
+      "loss": 0.3371,
+      "step": 5709
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.750677606216577e-05,
+      "loss": 0.3417,
+      "step": 5710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7503643094979425e-05,
+      "loss": 0.3198,
+      "step": 5711
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7500508276204046e-05,
+      "loss": 0.3085,
+      "step": 5712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.749737160612981e-05,
+      "loss": 0.3237,
+      "step": 5713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7494233085047055e-05,
+      "loss": 0.3323,
+      "step": 5714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7491092713246283e-05,
+      "loss": 0.3208,
+      "step": 5715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.748795049101819e-05,
+      "loss": 0.3044,
+      "step": 5716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.748480641865362e-05,
+      "loss": 0.3211,
+      "step": 5717
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.74816604964436e-05,
+      "loss": 0.3199,
+      "step": 5718
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.747851272467931e-05,
+      "loss": 0.3169,
+      "step": 5719
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7475363103652135e-05,
+      "loss": 0.3108,
+      "step": 5720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7472211633653604e-05,
+      "loss": 0.3204,
+      "step": 5721
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.746905831497543e-05,
+      "loss": 0.323,
+      "step": 5722
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.746590314790948e-05,
+      "loss": 0.3106,
+      "step": 5723
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.746274613274782e-05,
+      "loss": 0.3286,
+      "step": 5724
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.745958726978266e-05,
+      "loss": 0.3076,
+      "step": 5725
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7456426559306385e-05,
+      "loss": 0.3243,
+      "step": 5726
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.745326400161159e-05,
+      "loss": 0.3203,
+      "step": 5727
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7450099596990975e-05,
+      "loss": 0.3112,
+      "step": 5728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.744693334573747e-05,
+      "loss": 0.3345,
+      "step": 5729
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.744376524814413e-05,
+      "loss": 0.321,
+      "step": 5730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.744059530450422e-05,
+      "loss": 0.3236,
+      "step": 5731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.743742351511115e-05,
+      "loss": 0.324,
+      "step": 5732
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.743424988025852e-05,
+      "loss": 0.3439,
+      "step": 5733
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.743107440024007e-05,
+      "loss": 0.3125,
+      "step": 5734
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7427897075349746e-05,
+      "loss": 0.3337,
+      "step": 5735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7424717905881645e-05,
+      "loss": 0.3275,
+      "step": 5736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.742153689213004e-05,
+      "loss": 0.3316,
+      "step": 5737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7418354034389376e-05,
+      "loss": 0.3222,
+      "step": 5738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7415169332954254e-05,
+      "loss": 0.3153,
+      "step": 5739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.741198278811947e-05,
+      "loss": 0.2961,
+      "step": 5740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.740879440017999e-05,
+      "loss": 0.3043,
+      "step": 5741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.740560416943091e-05,
+      "loss": 0.2869,
+      "step": 5742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7402412096167545e-05,
+      "loss": 0.3005,
+      "step": 5743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.739921818068536e-05,
+      "loss": 0.3187,
+      "step": 5744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.739602242327999e-05,
+      "loss": 0.3194,
+      "step": 5745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.739282482424724e-05,
+      "loss": 0.3058,
+      "step": 5746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7389625383883104e-05,
+      "loss": 0.3116,
+      "step": 5747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.738642410248372e-05,
+      "loss": 0.3139,
+      "step": 5748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.738322098034539e-05,
+      "loss": 0.3172,
+      "step": 5749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.738001601776462e-05,
+      "loss": 0.3131,
+      "step": 5750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.737680921503807e-05,
+      "loss": 0.2983,
+      "step": 5751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7373600572462565e-05,
+      "loss": 0.3022,
+      "step": 5752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.737039009033512e-05,
+      "loss": 0.3097,
+      "step": 5753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7367177768952874e-05,
+      "loss": 0.3068,
+      "step": 5754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7363963608613205e-05,
+      "loss": 0.307,
+      "step": 5755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.73607476096136e-05,
+      "loss": 0.323,
+      "step": 5756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7357529772251743e-05,
+      "loss": 0.3146,
+      "step": 5757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.735431009682549e-05,
+      "loss": 0.3283,
+      "step": 5758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.735108858363286e-05,
+      "loss": 0.2971,
+      "step": 5759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.734786523297205e-05,
+      "loss": 0.3005,
+      "step": 5760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.734464004514141e-05,
+      "loss": 0.3212,
+      "step": 5761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.734141302043949e-05,
+      "loss": 0.2806,
+      "step": 5762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7338184159164975e-05,
+      "loss": 0.2995,
+      "step": 5763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.733495346161675e-05,
+      "loss": 0.312,
+      "step": 5764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7331720928093845e-05,
+      "loss": 0.2958,
+      "step": 5765
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7328486558895474e-05,
+      "loss": 0.2978,
+      "step": 5766
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.732525035432101e-05,
+      "loss": 0.3239,
+      "step": 5767
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.732201231467004e-05,
+      "loss": 0.3114,
+      "step": 5768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7318772440242245e-05,
+      "loss": 0.308,
+      "step": 5769
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.731553073133754e-05,
+      "loss": 0.3151,
+      "step": 5770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.731228718825597e-05,
+      "loss": 0.3148,
+      "step": 5771
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7309041811297775e-05,
+      "loss": 0.3,
+      "step": 5772
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7305794600763356e-05,
+      "loss": 0.3041,
+      "step": 5773
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.730254555695327e-05,
+      "loss": 0.2987,
+      "step": 5774
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.729929468016829e-05,
+      "loss": 0.3025,
+      "step": 5775
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.729604197070929e-05,
+      "loss": 0.3269,
+      "step": 5776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7292787428877366e-05,
+      "loss": 0.3162,
+      "step": 5777
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.728953105497376e-05,
+      "loss": 0.331,
+      "step": 5778
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7286272849299904e-05,
+      "loss": 0.301,
+      "step": 5779
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.728301281215736e-05,
+      "loss": 0.3192,
+      "step": 5780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7279750943847916e-05,
+      "loss": 0.3146,
+      "step": 5781
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.727648724467348e-05,
+      "loss": 0.3138,
+      "step": 5782
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.727322171493616e-05,
+      "loss": 0.2908,
+      "step": 5783
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.72699543549382e-05,
+      "loss": 0.3215,
+      "step": 5784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7266685164982053e-05,
+      "loss": 0.3024,
+      "step": 5785
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.726341414537033e-05,
+      "loss": 0.3096,
+      "step": 5786
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7260141296405794e-05,
+      "loss": 0.3083,
+      "step": 5787
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.725686661839139e-05,
+      "loss": 0.2997,
+      "step": 5788
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.725359011163023e-05,
+      "loss": 0.3277,
+      "step": 5789
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7250311776425596e-05,
+      "loss": 0.2997,
+      "step": 5790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.724703161308094e-05,
+      "loss": 0.3144,
+      "step": 5791
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.724374962189989e-05,
+      "loss": 0.2956,
+      "step": 5792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.724046580318622e-05,
+      "loss": 0.3062,
+      "step": 5793
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7237180157243895e-05,
+      "loss": 0.3096,
+      "step": 5794
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7233892684377055e-05,
+      "loss": 0.301,
+      "step": 5795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.723060338488997e-05,
+      "loss": 0.2995,
+      "step": 5796
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.722731225908714e-05,
+      "loss": 0.3137,
+      "step": 5797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.722401930727317e-05,
+      "loss": 0.3061,
+      "step": 5798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.722072452975289e-05,
+      "loss": 0.3063,
+      "step": 5799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.721742792683125e-05,
+      "loss": 0.3158,
+      "step": 5800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.721412949881341e-05,
+      "loss": 0.3045,
+      "step": 5801
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.721082924600467e-05,
+      "loss": 0.3022,
+      "step": 5802
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.720752716871052e-05,
+      "loss": 0.2978,
+      "step": 5803
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.720422326723659e-05,
+      "loss": 0.3231,
+      "step": 5804
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.720091754188872e-05,
+      "loss": 0.3124,
+      "step": 5805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.719760999297288e-05,
+      "loss": 0.3156,
+      "step": 5806
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7194300620795234e-05,
+      "loss": 0.2968,
+      "step": 5807
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.719098942566211e-05,
+      "loss": 0.3156,
+      "step": 5808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.718767640787999e-05,
+      "loss": 0.3231,
+      "step": 5809
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7184361567755534e-05,
+      "loss": 0.3013,
+      "step": 5810
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.718104490559559e-05,
+      "loss": 0.3153,
+      "step": 5811
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7177726421707135e-05,
+      "loss": 0.2974,
+      "step": 5812
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.717440611639736e-05,
+      "loss": 0.3089,
+      "step": 5813
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.717108398997357e-05,
+      "loss": 0.3197,
+      "step": 5814
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.71677600427433e-05,
+      "loss": 0.3175,
+      "step": 5815
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.716443427501421e-05,
+      "loss": 0.3041,
+      "step": 5816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7161106687094134e-05,
+      "loss": 0.3099,
+      "step": 5817
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.715777727929109e-05,
+      "loss": 0.3189,
+      "step": 5818
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.715444605191327e-05,
+      "loss": 0.3246,
+      "step": 5819
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.715111300526899e-05,
+      "loss": 0.3128,
+      "step": 5820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.71477781396668e-05,
+      "loss": 0.3105,
+      "step": 5821
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.714444145541536e-05,
+      "loss": 0.3007,
+      "step": 5822
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7141102952823534e-05,
+      "loss": 0.3025,
+      "step": 5823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.713776263220033e-05,
+      "loss": 0.3019,
+      "step": 5824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.713442049385495e-05,
+      "loss": 0.3205,
+      "step": 5825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.713107653809675e-05,
+      "loss": 0.3001,
+      "step": 5826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.712773076523524e-05,
+      "loss": 0.2909,
+      "step": 5827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.712438317558013e-05,
+      "loss": 0.2984,
+      "step": 5828
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.712103376944128e-05,
+      "loss": 0.3002,
+      "step": 5829
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.71176825471287e-05,
+      "loss": 0.3255,
+      "step": 5830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7114329508952624e-05,
+      "loss": 0.3136,
+      "step": 5831
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7110974655223386e-05,
+      "loss": 0.3098,
+      "step": 5832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.710761798625153e-05,
+      "loss": 0.3261,
+      "step": 5833
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.710425950234777e-05,
+      "loss": 0.3047,
+      "step": 5834
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.710089920382295e-05,
+      "loss": 0.2992,
+      "step": 5835
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7097537090988126e-05,
+      "loss": 0.307,
+      "step": 5836
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7094173164154504e-05,
+      "loss": 0.3133,
+      "step": 5837
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.709080742363345e-05,
+      "loss": 0.3272,
+      "step": 5838
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.708743986973652e-05,
+      "loss": 0.3272,
+      "step": 5839
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.708407050277541e-05,
+      "loss": 0.3058,
+      "step": 5840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7080699323062e-05,
+      "loss": 0.3107,
+      "step": 5841
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.707732633090834e-05,
+      "loss": 0.303,
+      "step": 5842
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.707395152662663e-05,
+      "loss": 0.3106,
+      "step": 5843
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7070574910529266e-05,
+      "loss": 0.3099,
+      "step": 5844
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.706719648292879e-05,
+      "loss": 0.315,
+      "step": 5845
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.706381624413792e-05,
+      "loss": 0.3157,
+      "step": 5846
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.706043419446953e-05,
+      "loss": 0.3089,
+      "step": 5847
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.70570503342367e-05,
+      "loss": 0.3237,
+      "step": 5848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.70536646637526e-05,
+      "loss": 0.3129,
+      "step": 5849
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.705027718333066e-05,
+      "loss": 0.3066,
+      "step": 5850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.704688789328442e-05,
+      "loss": 0.3059,
+      "step": 5851
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.704349679392759e-05,
+      "loss": 0.3242,
+      "step": 5852
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.704010388557408e-05,
+      "loss": 0.3136,
+      "step": 5853
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.703670916853792e-05,
+      "loss": 0.3088,
+      "step": 5854
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.703331264313336e-05,
+      "loss": 0.3203,
+      "step": 5855
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7029914309674785e-05,
+      "loss": 0.3122,
+      "step": 5856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7026514168476736e-05,
+      "loss": 0.341,
+      "step": 5857
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.702311221985397e-05,
+      "loss": 0.3235,
+      "step": 5858
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7019708464121344e-05,
+      "loss": 0.2984,
+      "step": 5859
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7016302901593935e-05,
+      "loss": 0.2984,
+      "step": 5860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.701289553258698e-05,
+      "loss": 0.3227,
+      "step": 5861
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.700948635741586e-05,
+      "loss": 0.3049,
+      "step": 5862
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7006075376396155e-05,
+      "loss": 0.3137,
+      "step": 5863
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.700266258984357e-05,
+      "loss": 0.2992,
+      "step": 5864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6999247998074023e-05,
+      "loss": 0.2977,
+      "step": 5865
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6995831601403563e-05,
+      "loss": 0.3047,
+      "step": 5866
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6992413400148423e-05,
+      "loss": 0.3311,
+      "step": 5867
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6988993394625e-05,
+      "loss": 0.3131,
+      "step": 5868
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6985571585149876e-05,
+      "loss": 0.3253,
+      "step": 5869
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.698214797203976e-05,
+      "loss": 0.3054,
+      "step": 5870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.697872255561155e-05,
+      "loss": 0.317,
+      "step": 5871
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.697529533618233e-05,
+      "loss": 0.3067,
+      "step": 5872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6971866314069324e-05,
+      "loss": 0.3248,
+      "step": 5873
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.696843548958992e-05,
+      "loss": 0.3087,
+      "step": 5874
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6965002863061705e-05,
+      "loss": 0.2985,
+      "step": 5875
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.696156843480239e-05,
+      "loss": 0.3015,
+      "step": 5876
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.695813220512989e-05,
+      "loss": 0.2968,
+      "step": 5877
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6954694174362264e-05,
+      "loss": 0.3166,
+      "step": 5878
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.695125434281775e-05,
+      "loss": 0.3193,
+      "step": 5879
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.694781271081474e-05,
+      "loss": 0.3115,
+      "step": 5880
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.694436927867182e-05,
+      "loss": 0.3095,
+      "step": 5881
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.694092404670769e-05,
+      "loss": 0.3033,
+      "step": 5882
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.693747701524128e-05,
+      "loss": 0.3157,
+      "step": 5883
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.693402818459164e-05,
+      "loss": 0.3174,
+      "step": 5884
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.693057755507801e-05,
+      "loss": 0.3313,
+      "step": 5885
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6927125127019775e-05,
+      "loss": 0.314,
+      "step": 5886
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.692367090073652e-05,
+      "loss": 0.3189,
+      "step": 5887
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6920214876547976e-05,
+      "loss": 0.3197,
+      "step": 5888
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.691675705477402e-05,
+      "loss": 0.2877,
+      "step": 5889
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6913297435734736e-05,
+      "loss": 0.2821,
+      "step": 5890
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.690983601975035e-05,
+      "loss": 0.3039,
+      "step": 5891
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6906372807141265e-05,
+      "loss": 0.3156,
+      "step": 5892
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.690290779822804e-05,
+      "loss": 0.3241,
+      "step": 5893
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6899440993331404e-05,
+      "loss": 0.3165,
+      "step": 5894
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.689597239277226e-05,
+      "loss": 0.3248,
+      "step": 5895
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6892501996871654e-05,
+      "loss": 0.3132,
+      "step": 5896
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.688902980595084e-05,
+      "loss": 0.3048,
+      "step": 5897
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.688555582033118e-05,
+      "loss": 0.3094,
+      "step": 5898
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.688208004033427e-05,
+      "loss": 0.3123,
+      "step": 5899
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.687860246628182e-05,
+      "loss": 0.3146,
+      "step": 5900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.687512309849572e-05,
+      "loss": 0.2945,
+      "step": 5901
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6871641937298036e-05,
+      "loss": 0.3204,
+      "step": 5902
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.686815898301099e-05,
+      "loss": 0.3158,
+      "step": 5903
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.686467423595698e-05,
+      "loss": 0.3083,
+      "step": 5904
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.686118769645856e-05,
+      "loss": 0.3043,
+      "step": 5905
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.685769936483845e-05,
+      "loss": 0.3118,
+      "step": 5906
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.685420924141953e-05,
+      "loss": 0.29,
+      "step": 5907
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.685071732652487e-05,
+      "loss": 0.3225,
+      "step": 5908
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6847223620477686e-05,
+      "loss": 0.3119,
+      "step": 5909
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.684372812360137e-05,
+      "loss": 0.2981,
+      "step": 5910
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.684023083621946e-05,
+      "loss": 0.3249,
+      "step": 5911
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.683673175865569e-05,
+      "loss": 0.3231,
+      "step": 5912
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.683323089123393e-05,
+      "loss": 0.2942,
+      "step": 5913
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6829728234278234e-05,
+      "loss": 0.3206,
+      "step": 5914
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.682622378811282e-05,
+      "loss": 0.3109,
+      "step": 5915
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.682271755306207e-05,
+      "loss": 0.3046,
+      "step": 5916
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6819209529450526e-05,
+      "loss": 0.2927,
+      "step": 5917
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.68156997176029e-05,
+      "loss": 0.3099,
+      "step": 5918
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6812188117844075e-05,
+      "loss": 0.3032,
+      "step": 5919
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.680867473049908e-05,
+      "loss": 0.3057,
+      "step": 5920
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.680515955589314e-05,
+      "loss": 0.3011,
+      "step": 5921
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.680164259435162e-05,
+      "loss": 0.314,
+      "step": 5922
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.679812384620006e-05,
+      "loss": 0.332,
+      "step": 5923
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.679460331176416e-05,
+      "loss": 0.3233,
+      "step": 5924
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.679108099136979e-05,
+      "loss": 0.3087,
+      "step": 5925
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6787556885343006e-05,
+      "loss": 0.3261,
+      "step": 5926
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.678403099400999e-05,
+      "loss": 0.2989,
+      "step": 5927
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.678050331769709e-05,
+      "loss": 0.3101,
+      "step": 5928
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6776973856730874e-05,
+      "loss": 0.3238,
+      "step": 5929
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.677344261143802e-05,
+      "loss": 0.2936,
+      "step": 5930
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6769909582145374e-05,
+      "loss": 0.3201,
+      "step": 5931
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.676637476917999e-05,
+      "loss": 0.315,
+      "step": 5932
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6762838172869046e-05,
+      "loss": 0.3173,
+      "step": 5933
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.67592997935399e-05,
+      "loss": 0.3211,
+      "step": 5934
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6755759631520076e-05,
+      "loss": 0.2899,
+      "step": 5935
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6752217687137246e-05,
+      "loss": 0.3105,
+      "step": 5936
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6748673960719285e-05,
+      "loss": 0.3142,
+      "step": 5937
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.67451284525942e-05,
+      "loss": 0.3294,
+      "step": 5938
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.674158116309016e-05,
+      "loss": 0.3108,
+      "step": 5939
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6738032092535523e-05,
+      "loss": 0.3086,
+      "step": 5940
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.67344812412588e-05,
+      "loss": 0.3215,
+      "step": 5941
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.673092860958867e-05,
+      "loss": 0.327,
+      "step": 5942
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6727374197853965e-05,
+      "loss": 0.3209,
+      "step": 5943
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.67238180063837e-05,
+      "loss": 0.3091,
+      "step": 5944
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.672026003550704e-05,
+      "loss": 0.3304,
+      "step": 5945
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.671670028555332e-05,
+      "loss": 0.3002,
+      "step": 5946
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6713138756852035e-05,
+      "loss": 0.3187,
+      "step": 5947
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6709575449732864e-05,
+      "loss": 0.3122,
+      "step": 5948
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.670601036452561e-05,
+      "loss": 0.3004,
+      "step": 5949
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.67024435015603e-05,
+      "loss": 0.3331,
+      "step": 5950
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.669887486116708e-05,
+      "loss": 0.3095,
+      "step": 5951
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.669530444367625e-05,
+      "loss": 0.3082,
+      "step": 5952
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.669173224941832e-05,
+      "loss": 0.2963,
+      "step": 5953
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.668815827872394e-05,
+      "loss": 0.3295,
+      "step": 5954
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.668458253192391e-05,
+      "loss": 0.3183,
+      "step": 5955
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.668100500934924e-05,
+      "loss": 0.2989,
+      "step": 5956
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.667742571133105e-05,
+      "loss": 0.3177,
+      "step": 5957
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.667384463820065e-05,
+      "loss": 0.3117,
+      "step": 5958
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.667026179028953e-05,
+      "loss": 0.3016,
+      "step": 5959
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.666667716792931e-05,
+      "loss": 0.316,
+      "step": 5960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.666309077145181e-05,
+      "loss": 0.3137,
+      "step": 5961
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6659502601188975e-05,
+      "loss": 0.3233,
+      "step": 5962
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6655912657472945e-05,
+      "loss": 0.3192,
+      "step": 5963
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.665232094063603e-05,
+      "loss": 0.2981,
+      "step": 5964
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.664872745101067e-05,
+      "loss": 0.3275,
+      "step": 5965
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.664513218892948e-05,
+      "loss": 0.327,
+      "step": 5966
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6641535154725265e-05,
+      "loss": 0.3266,
+      "step": 5967
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.663793634873097e-05,
+      "loss": 0.3155,
+      "step": 5968
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6634335771279705e-05,
+      "loss": 0.3054,
+      "step": 5969
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6630733422704756e-05,
+      "loss": 0.3109,
+      "step": 5970
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.662712930333956e-05,
+      "loss": 0.3176,
+      "step": 5971
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6623523413517725e-05,
+      "loss": 0.3062,
+      "step": 5972
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.661991575357304e-05,
+      "loss": 0.3033,
+      "step": 5973
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.66163063238394e-05,
+      "loss": 0.3211,
+      "step": 5974
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.661269512465094e-05,
+      "loss": 0.3291,
+      "step": 5975
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.66090821563419e-05,
+      "loss": 0.2994,
+      "step": 5976
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.660546741924672e-05,
+      "loss": 0.3275,
+      "step": 5977
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.660185091369997e-05,
+      "loss": 0.2968,
+      "step": 5978
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.659823264003642e-05,
+      "loss": 0.3041,
+      "step": 5979
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.659461259859099e-05,
+      "loss": 0.3018,
+      "step": 5980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.659099078969875e-05,
+      "loss": 0.3109,
+      "step": 5981
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.658736721369495e-05,
+      "loss": 0.3192,
+      "step": 5982
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.658374187091498e-05,
+      "loss": 0.333,
+      "step": 5983
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6580114761694435e-05,
+      "loss": 0.2999,
+      "step": 5984
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6576485886369045e-05,
+      "loss": 0.3115,
+      "step": 5985
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.65728552452747e-05,
+      "loss": 0.2999,
+      "step": 5986
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6569222838747455e-05,
+      "loss": 0.2886,
+      "step": 5987
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6565588667123557e-05,
+      "loss": 0.2981,
+      "step": 5988
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.656195273073938e-05,
+      "loss": 0.2984,
+      "step": 5989
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6558315029931467e-05,
+      "loss": 0.3176,
+      "step": 5990
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.655467556503655e-05,
+      "loss": 0.3034,
+      "step": 5991
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.655103433639151e-05,
+      "loss": 0.3127,
+      "step": 5992
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.654739134433338e-05,
+      "loss": 0.3127,
+      "step": 5993
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.654374658919935e-05,
+      "loss": 0.3015,
+      "step": 5994
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.65401000713268e-05,
+      "loss": 0.3169,
+      "step": 5995
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.653645179105328e-05,
+      "loss": 0.3026,
+      "step": 5996
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6532801748716456e-05,
+      "loss": 0.3118,
+      "step": 5997
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.65291499446542e-05,
+      "loss": 0.3229,
+      "step": 5998
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.652549637920453e-05,
+      "loss": 0.3172,
+      "step": 5999
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6521841052705634e-05,
+      "loss": 0.3185,
+      "step": 6000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.651818396549585e-05,
+      "loss": 0.2876,
+      "step": 6001
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6514525117913686e-05,
+      "loss": 0.3173,
+      "step": 6002
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.651086451029783e-05,
+      "loss": 0.2963,
+      "step": 6003
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.65072021429871e-05,
+      "loss": 0.32,
+      "step": 6004
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.65035380163205e-05,
+      "loss": 0.309,
+      "step": 6005
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.649987213063719e-05,
+      "loss": 0.3214,
+      "step": 6006
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.64962044862765e-05,
+      "loss": 0.3038,
+      "step": 6007
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.649253508357792e-05,
+      "loss": 0.3077,
+      "step": 6008
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.648886392288108e-05,
+      "loss": 0.3353,
+      "step": 6009
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.648519100452581e-05,
+      "loss": 0.3157,
+      "step": 6010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6481516328852074e-05,
+      "loss": 0.3184,
+      "step": 6011
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6477839896200016e-05,
+      "loss": 0.3038,
+      "step": 6012
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6474161706909936e-05,
+      "loss": 0.297,
+      "step": 6013
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6470481761322296e-05,
+      "loss": 0.3299,
+      "step": 6014
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.646680005977772e-05,
+      "loss": 0.3049,
+      "step": 6015
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6463116602617e-05,
+      "loss": 0.2998,
+      "step": 6016
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6459431390181074e-05,
+      "loss": 0.3106,
+      "step": 6017
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6455744422811075e-05,
+      "loss": 0.3227,
+      "step": 6018
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.645205570084825e-05,
+      "loss": 0.3035,
+      "step": 6019
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.644836522463407e-05,
+      "loss": 0.3129,
+      "step": 6020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6444672994510114e-05,
+      "loss": 0.3155,
+      "step": 6021
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.644097901081815e-05,
+      "loss": 0.3206,
+      "step": 6022
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.64372832739001e-05,
+      "loss": 0.3279,
+      "step": 6023
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.643358578409804e-05,
+      "loss": 0.3261,
+      "step": 6024
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.642988654175425e-05,
+      "loss": 0.3,
+      "step": 6025
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.642618554721111e-05,
+      "loss": 0.3051,
+      "step": 6026
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.642248280081122e-05,
+      "loss": 0.3256,
+      "step": 6027
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.64187783028973e-05,
+      "loss": 0.3254,
+      "step": 6028
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.641507205381224e-05,
+      "loss": 0.278,
+      "step": 6029
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.641136405389912e-05,
+      "loss": 0.3112,
+      "step": 6030
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6407654303501155e-05,
+      "loss": 0.3075,
+      "step": 6031
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.640394280296172e-05,
+      "loss": 0.3261,
+      "step": 6032
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.640022955262437e-05,
+      "loss": 0.2981,
+      "step": 6033
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.639651455283282e-05,
+      "loss": 0.3033,
+      "step": 6034
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6392797803930916e-05,
+      "loss": 0.3121,
+      "step": 6035
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.638907930626271e-05,
+      "loss": 0.2983,
+      "step": 6036
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.638535906017239e-05,
+      "loss": 0.3127,
+      "step": 6037
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6381637066004315e-05,
+      "loss": 0.3161,
+      "step": 6038
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6377913324103e-05,
+      "loss": 0.3082,
+      "step": 6039
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.637418783481312e-05,
+      "loss": 0.3046,
+      "step": 6040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6370460598479535e-05,
+      "loss": 0.3153,
+      "step": 6041
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.636673161544721e-05,
+      "loss": 0.3236,
+      "step": 6042
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.636300088606134e-05,
+      "loss": 0.3056,
+      "step": 6043
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.635926841066724e-05,
+      "loss": 0.3209,
+      "step": 6044
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.635553418961042e-05,
+      "loss": 0.3259,
+      "step": 6045
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.635179822323649e-05,
+      "loss": 0.305,
+      "step": 6046
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6348060511891285e-05,
+      "loss": 0.3231,
+      "step": 6047
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.634432105592077e-05,
+      "loss": 0.3175,
+      "step": 6048
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.634057985567108e-05,
+      "loss": 0.3234,
+      "step": 6049
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.633683691148852e-05,
+      "loss": 0.3088,
+      "step": 6050
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6333092223719545e-05,
+      "loss": 0.2907,
+      "step": 6051
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6329345792710746e-05,
+      "loss": 0.307,
+      "step": 6052
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.632559761880893e-05,
+      "loss": 0.321,
+      "step": 6053
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.632184770236104e-05,
+      "loss": 0.2989,
+      "step": 6054
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.631809604371415e-05,
+      "loss": 0.3062,
+      "step": 6055
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6314342643215556e-05,
+      "loss": 0.3211,
+      "step": 6056
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6310587501212666e-05,
+      "loss": 0.3052,
+      "step": 6057
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6306830618053064e-05,
+      "loss": 0.3208,
+      "step": 6058
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.63030719940845e-05,
+      "loss": 0.3221,
+      "step": 6059
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.629931162965489e-05,
+      "loss": 0.3222,
+      "step": 6060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.629554952511229e-05,
+      "loss": 0.2837,
+      "step": 6061
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.629178568080493e-05,
+      "loss": 0.2923,
+      "step": 6062
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.628802009708121e-05,
+      "loss": 0.3052,
+      "step": 6063
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.628425277428968e-05,
+      "loss": 0.3196,
+      "step": 6064
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.628048371277906e-05,
+      "loss": 0.3327,
+      "step": 6065
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.62767129128982e-05,
+      "loss": 0.2946,
+      "step": 6066
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.627294037499616e-05,
+      "loss": 0.3246,
+      "step": 6067
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6269166099422126e-05,
+      "loss": 0.3121,
+      "step": 6068
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.626539008652546e-05,
+      "loss": 0.3051,
+      "step": 6069
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.626161233665567e-05,
+      "loss": 0.3157,
+      "step": 6070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.625783285016243e-05,
+      "loss": 0.3028,
+      "step": 6071
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.625405162739561e-05,
+      "loss": 0.3243,
+      "step": 6072
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6250268668705176e-05,
+      "loss": 0.3104,
+      "step": 6073
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6246483974441304e-05,
+      "loss": 0.3189,
+      "step": 6074
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.624269754495431e-05,
+      "loss": 0.3039,
+      "step": 6075
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6238909380594685e-05,
+      "loss": 0.3069,
+      "step": 6076
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.623511948171305e-05,
+      "loss": 0.3134,
+      "step": 6077
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.623132784866023e-05,
+      "loss": 0.3025,
+      "step": 6078
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.622753448178718e-05,
+      "loss": 0.3283,
+      "step": 6079
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.622373938144503e-05,
+      "loss": 0.3064,
+      "step": 6080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6219942547985054e-05,
+      "loss": 0.2961,
+      "step": 6081
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.62161439817587e-05,
+      "loss": 0.3299,
+      "step": 6082
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.621234368311757e-05,
+      "loss": 0.2944,
+      "step": 6083
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.620854165241344e-05,
+      "loss": 0.2951,
+      "step": 6084
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.620473788999823e-05,
+      "loss": 0.296,
+      "step": 6085
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.620093239622403e-05,
+      "loss": 0.319,
+      "step": 6086
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.619712517144307e-05,
+      "loss": 0.303,
+      "step": 6087
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.619331621600777e-05,
+      "loss": 0.3098,
+      "step": 6088
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.61895055302707e-05,
+      "loss": 0.3083,
+      "step": 6089
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6185693114584586e-05,
+      "loss": 0.3212,
+      "step": 6090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.618187896930231e-05,
+      "loss": 0.3195,
+      "step": 6091
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.617806309477692e-05,
+      "loss": 0.3239,
+      "step": 6092
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6174245491361616e-05,
+      "loss": 0.3305,
+      "step": 6093
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6170426159409777e-05,
+      "loss": 0.3061,
+      "step": 6094
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.616660509927493e-05,
+      "loss": 0.3111,
+      "step": 6095
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.616278231131076e-05,
+      "loss": 0.3086,
+      "step": 6096
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.615895779587111e-05,
+      "loss": 0.3079,
+      "step": 6097
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6155131553309995e-05,
+      "loss": 0.3162,
+      "step": 6098
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.615130358398157e-05,
+      "loss": 0.2969,
+      "step": 6099
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.614747388824017e-05,
+      "loss": 0.3113,
+      "step": 6100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6143642466440286e-05,
+      "loss": 0.3254,
+      "step": 6101
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.613980931893655e-05,
+      "loss": 0.2881,
+      "step": 6102
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.613597444608377e-05,
+      "loss": 0.3241,
+      "step": 6103
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.613213784823694e-05,
+      "loss": 0.3025,
+      "step": 6104
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.612829952575114e-05,
+      "loss": 0.301,
+      "step": 6105
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6124459478981705e-05,
+      "loss": 0.3097,
+      "step": 6106
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.612061770828404e-05,
+      "loss": 0.3168,
+      "step": 6107
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.611677421401376e-05,
+      "loss": 0.3112,
+      "step": 6108
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6112928996526626e-05,
+      "loss": 0.3081,
+      "step": 6109
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.610908205617857e-05,
+      "loss": 0.3036,
+      "step": 6110
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.610523339332568e-05,
+      "loss": 0.3234,
+      "step": 6111
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6101383008324185e-05,
+      "loss": 0.3031,
+      "step": 6112
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.609753090153049e-05,
+      "loss": 0.3337,
+      "step": 6113
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.609367707330116e-05,
+      "loss": 0.3056,
+      "step": 6114
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6089821523992916e-05,
+      "loss": 0.3172,
+      "step": 6115
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.608596425396263e-05,
+      "loss": 0.3119,
+      "step": 6116
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.608210526356734e-05,
+      "loss": 0.3123,
+      "step": 6117
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6078244553164254e-05,
+      "loss": 0.3341,
+      "step": 6118
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.607438212311073e-05,
+      "loss": 0.3241,
+      "step": 6119
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.607051797376427e-05,
+      "loss": 0.3108,
+      "step": 6120
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6066652105482564e-05,
+      "loss": 0.3189,
+      "step": 6121
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.606278451862344e-05,
+      "loss": 0.3013,
+      "step": 6122
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6058915213544895e-05,
+      "loss": 0.312,
+      "step": 6123
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.605504419060508e-05,
+      "loss": 0.309,
+      "step": 6124
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.60511714501623e-05,
+      "loss": 0.3018,
+      "step": 6125
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.604729699257504e-05,
+      "loss": 0.3138,
+      "step": 6126
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.604342081820192e-05,
+      "loss": 0.3097,
+      "step": 6127
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6039542927401744e-05,
+      "loss": 0.3062,
+      "step": 6128
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.603566332053343e-05,
+      "loss": 0.2934,
+      "step": 6129
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.603178199795612e-05,
+      "loss": 0.308,
+      "step": 6130
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.602789896002905e-05,
+      "loss": 0.3149,
+      "step": 6131
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.602401420711165e-05,
+      "loss": 0.3105,
+      "step": 6132
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.602012773956352e-05,
+      "loss": 0.3213,
+      "step": 6133
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6016239557744384e-05,
+      "loss": 0.3184,
+      "step": 6134
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6012349662014145e-05,
+      "loss": 0.3182,
+      "step": 6135
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.600845805273287e-05,
+      "loss": 0.3148,
+      "step": 6136
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6004564730260766e-05,
+      "loss": 0.2959,
+      "step": 6137
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.600066969495822e-05,
+      "loss": 0.3152,
+      "step": 6138
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5996772947185754e-05,
+      "loss": 0.3281,
+      "step": 6139
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.599287448730407e-05,
+      "loss": 0.3077,
+      "step": 6140
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.598897431567401e-05,
+      "loss": 0.3136,
+      "step": 6141
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5985072432656614e-05,
+      "loss": 0.3176,
+      "step": 6142
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.598116883861301e-05,
+      "loss": 0.3075,
+      "step": 6143
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.597726353390455e-05,
+      "loss": 0.3019,
+      "step": 6144
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.59733565188927e-05,
+      "loss": 0.3161,
+      "step": 6145
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5969447793939115e-05,
+      "loss": 0.3145,
+      "step": 6146
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.59655373594056e-05,
+      "loss": 0.3016,
+      "step": 6147
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.596162521565412e-05,
+      "loss": 0.3184,
+      "step": 6148
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.595771136304678e-05,
+      "loss": 0.3136,
+      "step": 6149
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.595379580194586e-05,
+      "loss": 0.3024,
+      "step": 6150
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.594987853271379e-05,
+      "loss": 0.3047,
+      "step": 6151
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5945959555713175e-05,
+      "loss": 0.2996,
+      "step": 6152
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5942038871306756e-05,
+      "loss": 0.3435,
+      "step": 6153
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.593811647985745e-05,
+      "loss": 0.316,
+      "step": 6154
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5934192381728316e-05,
+      "loss": 0.2967,
+      "step": 6155
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5930266577282576e-05,
+      "loss": 0.3223,
+      "step": 6156
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.592633906688362e-05,
+      "loss": 0.307,
+      "step": 6157
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.592240985089498e-05,
+      "loss": 0.3135,
+      "step": 6158
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.591847892968037e-05,
+      "loss": 0.287,
+      "step": 6159
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5914546303603633e-05,
+      "loss": 0.3235,
+      "step": 6160
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.591061197302879e-05,
+      "loss": 0.3122,
+      "step": 6161
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5906675938319996e-05,
+      "loss": 0.3059,
+      "step": 6162
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.590273819984161e-05,
+      "loss": 0.3063,
+      "step": 6163
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.589879875795809e-05,
+      "loss": 0.2981,
+      "step": 6164
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.589485761303411e-05,
+      "loss": 0.305,
+      "step": 6165
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5890914765434445e-05,
+      "loss": 0.2922,
+      "step": 6166
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.588697021552407e-05,
+      "loss": 0.3155,
+      "step": 6167
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.58830239636681e-05,
+      "loss": 0.3084,
+      "step": 6168
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.587907601023181e-05,
+      "loss": 0.3148,
+      "step": 6169
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.587512635558064e-05,
+      "loss": 0.3302,
+      "step": 6170
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.587117500008017e-05,
+      "loss": 0.3095,
+      "step": 6171
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5867221944096144e-05,
+      "loss": 0.3102,
+      "step": 6172
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.586326718799448e-05,
+      "loss": 0.3197,
+      "step": 6173
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.585931073214124e-05,
+      "loss": 0.3147,
+      "step": 6174
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5855352576902624e-05,
+      "loss": 0.3485,
+      "step": 6175
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.585139272264503e-05,
+      "loss": 0.2994,
+      "step": 6176
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.584743116973499e-05,
+      "loss": 0.305,
+      "step": 6177
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.584346791853918e-05,
+      "loss": 0.2968,
+      "step": 6178
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.583950296942448e-05,
+      "loss": 0.3068,
+      "step": 6179
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.583553632275786e-05,
+      "loss": 0.3219,
+      "step": 6180
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5831567978906504e-05,
+      "loss": 0.3146,
+      "step": 6181
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.582759793823773e-05,
+      "loss": 0.3257,
+      "step": 6182
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5823626201119e-05,
+      "loss": 0.3132,
+      "step": 6183
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.581965276791798e-05,
+      "loss": 0.3004,
+      "step": 6184
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.581567763900243e-05,
+      "loss": 0.3142,
+      "step": 6185
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5811700814740305e-05,
+      "loss": 0.3148,
+      "step": 6186
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5807722295499734e-05,
+      "loss": 0.2974,
+      "step": 6187
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.580374208164894e-05,
+      "loss": 0.3083,
+      "step": 6188
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5799760173556386e-05,
+      "loss": 0.2916,
+      "step": 6189
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.579577657159061e-05,
+      "loss": 0.3294,
+      "step": 6190
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5791791276120353e-05,
+      "loss": 0.3108,
+      "step": 6191
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.578780428751452e-05,
+      "loss": 0.2991,
+      "step": 6192
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5783815606142145e-05,
+      "loss": 0.3155,
+      "step": 6193
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.577982523237243e-05,
+      "loss": 0.3184,
+      "step": 6194
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.577583316657474e-05,
+      "loss": 0.3252,
+      "step": 6195
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.577183940911859e-05,
+      "loss": 0.3315,
+      "step": 6196
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.576784396037364e-05,
+      "loss": 0.3063,
+      "step": 6197
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.576384682070974e-05,
+      "loss": 0.3113,
+      "step": 6198
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.575984799049686e-05,
+      "loss": 0.3246,
+      "step": 6199
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.575584747010515e-05,
+      "loss": 0.3134,
+      "step": 6200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5751845259904905e-05,
+      "loss": 0.2838,
+      "step": 6201
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.574784136026659e-05,
+      "loss": 0.3061,
+      "step": 6202
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.57438357715608e-05,
+      "loss": 0.2972,
+      "step": 6203
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5739828494158305e-05,
+      "loss": 0.3277,
+      "step": 6204
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.573581952843004e-05,
+      "loss": 0.3081,
+      "step": 6205
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.573180887474707e-05,
+      "loss": 0.2938,
+      "step": 6206
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.572779653348065e-05,
+      "loss": 0.3092,
+      "step": 6207
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5723782505002156e-05,
+      "loss": 0.3053,
+      "step": 6208
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5719766789683154e-05,
+      "loss": 0.3053,
+      "step": 6209
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.571574938789533e-05,
+      "loss": 0.3182,
+      "step": 6210
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.571173030001056e-05,
+      "loss": 0.299,
+      "step": 6211
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.570770952640085e-05,
+      "loss": 0.3035,
+      "step": 6212
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.570368706743838e-05,
+      "loss": 0.3215,
+      "step": 6213
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.569966292349548e-05,
+      "loss": 0.3103,
+      "step": 6214
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.569563709494464e-05,
+      "loss": 0.3223,
+      "step": 6215
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.569160958215848e-05,
+      "loss": 0.3182,
+      "step": 6216
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.568758038550981e-05,
+      "loss": 0.3074,
+      "step": 6217
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.56835495053716e-05,
+      "loss": 0.3251,
+      "step": 6218
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5679516942116935e-05,
+      "loss": 0.3177,
+      "step": 6219
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.567548269611909e-05,
+      "loss": 0.3334,
+      "step": 6220
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.567144676775148e-05,
+      "loss": 0.3254,
+      "step": 6221
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.566740915738769e-05,
+      "loss": 0.3078,
+      "step": 6222
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5663369865401435e-05,
+      "loss": 0.3177,
+      "step": 6223
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.565932889216661e-05,
+      "loss": 0.3393,
+      "step": 6224
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.565528623805727e-05,
+      "loss": 0.3098,
+      "step": 6225
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.56512419034476e-05,
+      "loss": 0.317,
+      "step": 6226
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5647195888711954e-05,
+      "loss": 0.3254,
+      "step": 6227
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.564314819422485e-05,
+      "loss": 0.3036,
+      "step": 6228
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.563909882036095e-05,
+      "loss": 0.3004,
+      "step": 6229
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.563504776749507e-05,
+      "loss": 0.3167,
+      "step": 6230
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.56309950360022e-05,
+      "loss": 0.301,
+      "step": 6231
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5626940626257446e-05,
+      "loss": 0.3096,
+      "step": 6232
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5622884538636114e-05,
+      "loss": 0.2871,
+      "step": 6233
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.561882677351365e-05,
+      "loss": 0.3073,
+      "step": 6234
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.561476733126563e-05,
+      "loss": 0.3111,
+      "step": 6235
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.561070621226782e-05,
+      "loss": 0.2782,
+      "step": 6236
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5606643416896134e-05,
+      "loss": 0.3039,
+      "step": 6237
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5602578945526625e-05,
+      "loss": 0.3153,
+      "step": 6238
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5598512798535513e-05,
+      "loss": 0.3173,
+      "step": 6239
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.559444497629917e-05,
+      "loss": 0.3176,
+      "step": 6240
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.559037547919414e-05,
+      "loss": 0.2993,
+      "step": 6241
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.558630430759707e-05,
+      "loss": 0.2908,
+      "step": 6242
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5582231461884834e-05,
+      "loss": 0.3313,
+      "step": 6243
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.557815694243441e-05,
+      "loss": 0.3397,
+      "step": 6244
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5574080749622945e-05,
+      "loss": 0.3071,
+      "step": 6245
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.557000288382774e-05,
+      "loss": 0.3102,
+      "step": 6246
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.556592334542626e-05,
+      "loss": 0.3127,
+      "step": 6247
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5561842134796115e-05,
+      "loss": 0.3291,
+      "step": 6248
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.555775925231507e-05,
+      "loss": 0.3034,
+      "step": 6249
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.555367469836105e-05,
+      "loss": 0.3215,
+      "step": 6250
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.554958847331214e-05,
+      "loss": 0.3041,
+      "step": 6251
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5545500577546555e-05,
+      "loss": 0.2958,
+      "step": 6252
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.55414110114427e-05,
+      "loss": 0.3252,
+      "step": 6253
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553731977537909e-05,
+      "loss": 0.2999,
+      "step": 6254
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5533226869734444e-05,
+      "loss": 0.3118,
+      "step": 6255
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.55291322948876e-05,
+      "loss": 0.3255,
+      "step": 6256
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.552503605121756e-05,
+      "loss": 0.3139,
+      "step": 6257
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.55209381391035e-05,
+      "loss": 0.3134,
+      "step": 6258
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551683855892472e-05,
+      "loss": 0.308,
+      "step": 6259
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551273731106069e-05,
+      "loss": 0.3053,
+      "step": 6260
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550863439589102e-05,
+      "loss": 0.3152,
+      "step": 6261
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550452981379551e-05,
+      "loss": 0.3219,
+      "step": 6262
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5500423565154076e-05,
+      "loss": 0.3258,
+      "step": 6263
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.549631565034681e-05,
+      "loss": 0.322,
+      "step": 6264
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5492206069753943e-05,
+      "loss": 0.2967,
+      "step": 6265
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.548809482375588e-05,
+      "loss": 0.3035,
+      "step": 6266
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5483981912733154e-05,
+      "loss": 0.3336,
+      "step": 6267
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5479867337066476e-05,
+      "loss": 0.2922,
+      "step": 6268
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.54757510971367e-05,
+      "loss": 0.3116,
+      "step": 6269
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.547163319332484e-05,
+      "loss": 0.3084,
+      "step": 6270
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5467513626012046e-05,
+      "loss": 0.3348,
+      "step": 6271
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5463392395579654e-05,
+      "loss": 0.299,
+      "step": 6272
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5459269502409125e-05,
+      "loss": 0.2951,
+      "step": 6273
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5455144946882084e-05,
+      "loss": 0.322,
+      "step": 6274
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.545101872938031e-05,
+      "loss": 0.3261,
+      "step": 6275
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5446890850285757e-05,
+      "loss": 0.3234,
+      "step": 6276
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5442761309980473e-05,
+      "loss": 0.3221,
+      "step": 6277
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.543863010884673e-05,
+      "loss": 0.3197,
+      "step": 6278
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5434497247266914e-05,
+      "loss": 0.3125,
+      "step": 6279
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5430362725623575e-05,
+      "loss": 0.3053,
+      "step": 6280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.542622654429941e-05,
+      "loss": 0.3223,
+      "step": 6281
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5422088703677284e-05,
+      "loss": 0.3137,
+      "step": 6282
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.54179492041402e-05,
+      "loss": 0.2983,
+      "step": 6283
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.541380804607131e-05,
+      "loss": 0.3176,
+      "step": 6284
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540966522985395e-05,
+      "loss": 0.3095,
+      "step": 6285
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540552075587158e-05,
+      "loss": 0.305,
+      "step": 6286
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540137462450782e-05,
+      "loss": 0.3254,
+      "step": 6287
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.539722683614645e-05,
+      "loss": 0.304,
+      "step": 6288
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5393077391171404e-05,
+      "loss": 0.3078,
+      "step": 6289
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.538892628996677e-05,
+      "loss": 0.2993,
+      "step": 6290
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5384773532916754e-05,
+      "loss": 0.3139,
+      "step": 6291
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.538061912040579e-05,
+      "loss": 0.3119,
+      "step": 6292
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537646305281839e-05,
+      "loss": 0.3269,
+      "step": 6293
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5372305330539266e-05,
+      "loss": 0.2957,
+      "step": 6294
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.536814595395326e-05,
+      "loss": 0.3217,
+      "step": 6295
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.536398492344537e-05,
+      "loss": 0.3091,
+      "step": 6296
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.535982223940077e-05,
+      "loss": 0.3092,
+      "step": 6297
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.535565790220475e-05,
+      "loss": 0.3114,
+      "step": 6298
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5351491912242785e-05,
+      "loss": 0.3107,
+      "step": 6299
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.534732426990047e-05,
+      "loss": 0.295,
+      "step": 6300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.53431549755636e-05,
+      "loss": 0.3058,
+      "step": 6301
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5338984029618076e-05,
+      "loss": 0.3094,
+      "step": 6302
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.533481143244998e-05,
+      "loss": 0.305,
+      "step": 6303
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.533063718444554e-05,
+      "loss": 0.3005,
+      "step": 6304
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.532646128599112e-05,
+      "loss": 0.299,
+      "step": 6305
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.532228373747327e-05,
+      "loss": 0.3194,
+      "step": 6306
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.531810453927868e-05,
+      "loss": 0.2974,
+      "step": 6307
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.531392369179416e-05,
+      "loss": 0.3094,
+      "step": 6308
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5309741195406716e-05,
+      "loss": 0.3074,
+      "step": 6309
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.53055570505035e-05,
+      "loss": 0.3269,
+      "step": 6310
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5301371257471785e-05,
+      "loss": 0.3844,
+      "step": 6311
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.529718381669905e-05,
+      "loss": 0.2931,
+      "step": 6312
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.529299472857286e-05,
+      "loss": 0.3022,
+      "step": 6313
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.528880399348099e-05,
+      "loss": 0.3012,
+      "step": 6314
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.528461161181134e-05,
+      "loss": 0.33,
+      "step": 6315
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.528041758395196e-05,
+      "loss": 0.325,
+      "step": 6316
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5276221910291085e-05,
+      "loss": 0.3064,
+      "step": 6317
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5272024591217046e-05,
+      "loss": 0.3184,
+      "step": 6318
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5267825627118375e-05,
+      "loss": 0.3187,
+      "step": 6319
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.526362501838374e-05,
+      "loss": 0.3242,
+      "step": 6320
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.525942276540195e-05,
+      "loss": 0.3197,
+      "step": 6321
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5255218868561986e-05,
+      "loss": 0.3021,
+      "step": 6322
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5251013328252975e-05,
+      "loss": 0.3016,
+      "step": 6323
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5246806144864174e-05,
+      "loss": 0.3091,
+      "step": 6324
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.524259731878503e-05,
+      "loss": 0.3213,
+      "step": 6325
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.523838685040511e-05,
+      "loss": 0.3082,
+      "step": 6326
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.523417474011416e-05,
+      "loss": 0.3007,
+      "step": 6327
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.522996098830205e-05,
+      "loss": 0.3095,
+      "step": 6328
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.522574559535882e-05,
+      "loss": 0.313,
+      "step": 6329
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.522152856167466e-05,
+      "loss": 0.3126,
+      "step": 6330
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.52173098876399e-05,
+      "loss": 0.3061,
+      "step": 6331
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5213089573645046e-05,
+      "loss": 0.3216,
+      "step": 6332
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.520886762008074e-05,
+      "loss": 0.3199,
+      "step": 6333
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.520464402733777e-05,
+      "loss": 0.3153,
+      "step": 6334
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5200418795807076e-05,
+      "loss": 0.3098,
+      "step": 6335
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.519619192587977e-05,
+      "loss": 0.3071,
+      "step": 6336
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5191963417947106e-05,
+      "loss": 0.3102,
+      "step": 6337
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.518773327240047e-05,
+      "loss": 0.2977,
+      "step": 6338
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5183501489631415e-05,
+      "loss": 0.3169,
+      "step": 6339
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5179268070031656e-05,
+      "loss": 0.3182,
+      "step": 6340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.517503301399304e-05,
+      "loss": 0.3001,
+      "step": 6341
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5170796321907594e-05,
+      "loss": 0.2919,
+      "step": 6342
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.516655799416746e-05,
+      "loss": 0.3078,
+      "step": 6343
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5162318031164946e-05,
+      "loss": 0.2831,
+      "step": 6344
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5158076433292526e-05,
+      "loss": 0.3128,
+      "step": 6345
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.51538332009428e-05,
+      "loss": 0.3306,
+      "step": 6346
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.514958833450855e-05,
+      "loss": 0.3074,
+      "step": 6347
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.514534183438268e-05,
+      "loss": 0.311,
+      "step": 6348
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.514109370095826e-05,
+      "loss": 0.33,
+      "step": 6349
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.51368439346285e-05,
+      "loss": 0.3078,
+      "step": 6350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.513259253578678e-05,
+      "loss": 0.3161,
+      "step": 6351
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5128339504826615e-05,
+      "loss": 0.3023,
+      "step": 6352
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5124084842141684e-05,
+      "loss": 0.3202,
+      "step": 6353
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.51198285481258e-05,
+      "loss": 0.3091,
+      "step": 6354
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.511557062317295e-05,
+      "loss": 0.3173,
+      "step": 6355
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5111311067677234e-05,
+      "loss": 0.2979,
+      "step": 6356
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.510704988203295e-05,
+      "loss": 0.3074,
+      "step": 6357
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5102787066634526e-05,
+      "loss": 0.3204,
+      "step": 6358
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5098522621876515e-05,
+      "loss": 0.3281,
+      "step": 6359
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.509425654815368e-05,
+      "loss": 0.3105,
+      "step": 6360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.508998884586087e-05,
+      "loss": 0.3141,
+      "step": 6361
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5085719515393114e-05,
+      "loss": 0.2979,
+      "step": 6362
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5081448557145614e-05,
+      "loss": 0.302,
+      "step": 6363
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5077175971513696e-05,
+      "loss": 0.3259,
+      "step": 6364
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.507290175889283e-05,
+      "loss": 0.3131,
+      "step": 6365
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.506862591967865e-05,
+      "loss": 0.3085,
+      "step": 6366
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5064348454266954e-05,
+      "loss": 0.3109,
+      "step": 6367
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.506006936305365e-05,
+      "loss": 0.3052,
+      "step": 6368
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.505578864643485e-05,
+      "loss": 0.3159,
+      "step": 6369
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.505150630480676e-05,
+      "loss": 0.3226,
+      "step": 6370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.504722233856579e-05,
+      "loss": 0.3051,
+      "step": 6371
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.504293674810846e-05,
+      "loss": 0.3057,
+      "step": 6372
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.503864953383146e-05,
+      "loss": 0.3255,
+      "step": 6373
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.503436069613162e-05,
+      "loss": 0.3028,
+      "step": 6374
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.503007023540594e-05,
+      "loss": 0.3126,
+      "step": 6375
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.502577815205154e-05,
+      "loss": 0.2858,
+      "step": 6376
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.502148444646571e-05,
+      "loss": 0.3191,
+      "step": 6377
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5017189119045896e-05,
+      "loss": 0.3112,
+      "step": 6378
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.501289217018968e-05,
+      "loss": 0.2981,
+      "step": 6379
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.500859360029479e-05,
+      "loss": 0.3102,
+      "step": 6380
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.500429340975913e-05,
+      "loss": 0.3158,
+      "step": 6381
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.499999159898071e-05,
+      "loss": 0.2922,
+      "step": 6382
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4995688168357747e-05,
+      "loss": 0.311,
+      "step": 6383
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.499138311828856e-05,
+      "loss": 0.3209,
+      "step": 6384
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.498707644917164e-05,
+      "loss": 0.2943,
+      "step": 6385
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.498276816140562e-05,
+      "loss": 0.2886,
+      "step": 6386
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.497845825538929e-05,
+      "loss": 0.3343,
+      "step": 6387
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.497414673152159e-05,
+      "loss": 0.2954,
+      "step": 6388
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.49698335902016e-05,
+      "loss": 0.3051,
+      "step": 6389
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.496551883182856e-05,
+      "loss": 0.3152,
+      "step": 6390
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.496120245680185e-05,
+      "loss": 0.3322,
+      "step": 6391
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4956884465521e-05,
+      "loss": 0.3092,
+      "step": 6392
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.495256485838571e-05,
+      "loss": 0.3012,
+      "step": 6393
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4948243635795794e-05,
+      "loss": 0.3096,
+      "step": 6394
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4943920798151255e-05,
+      "loss": 0.3042,
+      "step": 6395
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.493959634585222e-05,
+      "loss": 0.3192,
+      "step": 6396
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4935270279298965e-05,
+      "loss": 0.303,
+      "step": 6397
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.493094259889193e-05,
+      "loss": 0.3149,
+      "step": 6398
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4926613305031696e-05,
+      "loss": 0.3106,
+      "step": 6399
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.492228239811898e-05,
+      "loss": 0.3084,
+      "step": 6400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4917949878554675e-05,
+      "loss": 0.3192,
+      "step": 6401
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.491361574673982e-05,
+      "loss": 0.2978,
+      "step": 6402
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.490928000307557e-05,
+      "loss": 0.3114,
+      "step": 6403
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4904942647963256e-05,
+      "loss": 0.3138,
+      "step": 6404
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.490060368180438e-05,
+      "loss": 0.3258,
+      "step": 6405
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.489626310500054e-05,
+      "loss": 0.31,
+      "step": 6406
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.489192091795352e-05,
+      "loss": 0.3214,
+      "step": 6407
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4887577121065244e-05,
+      "loss": 0.3102,
+      "step": 6408
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.48832317147378e-05,
+      "loss": 0.3276,
+      "step": 6409
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.487888469937339e-05,
+      "loss": 0.3179,
+      "step": 6410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.487453607537438e-05,
+      "loss": 0.3143,
+      "step": 6411
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.487018584314331e-05,
+      "loss": 0.3123,
+      "step": 6412
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4865834003082854e-05,
+      "loss": 0.3088,
+      "step": 6413
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4861480555595794e-05,
+      "loss": 0.3139,
+      "step": 6414
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.485712550108513e-05,
+      "loss": 0.311,
+      "step": 6415
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.485276883995397e-05,
+      "loss": 0.2957,
+      "step": 6416
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4848410572605565e-05,
+      "loss": 0.318,
+      "step": 6417
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.484405069944334e-05,
+      "loss": 0.3131,
+      "step": 6418
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.483968922087085e-05,
+      "loss": 0.3152,
+      "step": 6419
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.483532613729181e-05,
+      "loss": 0.2962,
+      "step": 6420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.483096144911007e-05,
+      "loss": 0.3011,
+      "step": 6421
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.482659515672964e-05,
+      "loss": 0.3159,
+      "step": 6422
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.482222726055468e-05,
+      "loss": 0.3089,
+      "step": 6423
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.481785776098949e-05,
+      "loss": 0.3146,
+      "step": 6424
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.481348665843852e-05,
+      "loss": 0.3173,
+      "step": 6425
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.480911395330638e-05,
+      "loss": 0.3135,
+      "step": 6426
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.480473964599782e-05,
+      "loss": 0.3204,
+      "step": 6427
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.480036373691772e-05,
+      "loss": 0.33,
+      "step": 6428
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4795986226471136e-05,
+      "loss": 0.3014,
+      "step": 6429
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.479160711506326e-05,
+      "loss": 0.3096,
+      "step": 6430
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4787226403099444e-05,
+      "loss": 0.318,
+      "step": 6431
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.478284409098516e-05,
+      "loss": 0.2908,
+      "step": 6432
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.477846017912606e-05,
+      "loss": 0.3179,
+      "step": 6433
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4774074667927916e-05,
+      "loss": 0.3015,
+      "step": 6434
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.476968755779668e-05,
+      "loss": 0.3149,
+      "step": 6435
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.476529884913843e-05,
+      "loss": 0.3056,
+      "step": 6436
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4760908542359386e-05,
+      "loss": 0.2958,
+      "step": 6437
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4756516637865935e-05,
+      "loss": 0.3123,
+      "step": 6438
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.47521231360646e-05,
+      "loss": 0.3057,
+      "step": 6439
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.474772803736205e-05,
+      "loss": 0.3275,
+      "step": 6440
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.474333134216512e-05,
+      "loss": 0.294,
+      "step": 6441
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.473893305088077e-05,
+      "loss": 0.297,
+      "step": 6442
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4734533163916115e-05,
+      "loss": 0.3134,
+      "step": 6443
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.473013168167843e-05,
+      "loss": 0.3249,
+      "step": 6444
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4725728604575114e-05,
+      "loss": 0.3176,
+      "step": 6445
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4721323933013735e-05,
+      "loss": 0.3093,
+      "step": 6446
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4716917667402005e-05,
+      "loss": 0.3278,
+      "step": 6447
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.471250980814778e-05,
+      "loss": 0.3242,
+      "step": 6448
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.470810035565905e-05,
+      "loss": 0.3216,
+      "step": 6449
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.470368931034397e-05,
+      "loss": 0.2998,
+      "step": 6450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.469927667261085e-05,
+      "loss": 0.3126,
+      "step": 6451
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4694862442868114e-05,
+      "loss": 0.3157,
+      "step": 6452
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.469044662152437e-05,
+      "loss": 0.3285,
+      "step": 6453
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.468602920898836e-05,
+      "loss": 0.3089,
+      "step": 6454
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.468161020566896e-05,
+      "loss": 0.3023,
+      "step": 6455
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4677189611975216e-05,
+      "loss": 0.2861,
+      "step": 6456
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4672767428316304e-05,
+      "loss": 0.3171,
+      "step": 6457
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.466834365510155e-05,
+      "loss": 0.3532,
+      "step": 6458
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.466391829274043e-05,
+      "loss": 0.3037,
+      "step": 6459
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.465949134164257e-05,
+      "loss": 0.3112,
+      "step": 6460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4655062802217737e-05,
+      "loss": 0.2959,
+      "step": 6461
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4650632674875855e-05,
+      "loss": 0.3041,
+      "step": 6462
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.464620096002698e-05,
+      "loss": 0.3136,
+      "step": 6463
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.464176765808133e-05,
+      "loss": 0.3047,
+      "step": 6464
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.463733276944926e-05,
+      "loss": 0.3193,
+      "step": 6465
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.463289629454127e-05,
+      "loss": 0.3422,
+      "step": 6466
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.462845823376801e-05,
+      "loss": 0.3101,
+      "step": 6467
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.462401858754029e-05,
+      "loss": 0.3072,
+      "step": 6468
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.461957735626906e-05,
+      "loss": 0.33,
+      "step": 6469
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4615134540365386e-05,
+      "loss": 0.3042,
+      "step": 6470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.461069014024053e-05,
+      "loss": 0.3123,
+      "step": 6471
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.460624415630586e-05,
+      "loss": 0.3176,
+      "step": 6472
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.460179658897293e-05,
+      "loss": 0.3185,
+      "step": 6473
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4597347438653396e-05,
+      "loss": 0.317,
+      "step": 6474
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4592896705759094e-05,
+      "loss": 0.3107,
+      "step": 6475
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4588444390701985e-05,
+      "loss": 0.3047,
+      "step": 6476
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.458399049389421e-05,
+      "loss": 0.3182,
+      "step": 6477
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.457953501574801e-05,
+      "loss": 0.3064,
+      "step": 6478
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.457507795667581e-05,
+      "loss": 0.3084,
+      "step": 6479
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4570619317090153e-05,
+      "loss": 0.2825,
+      "step": 6480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4566159097403756e-05,
+      "loss": 0.3275,
+      "step": 6481
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.456169729802946e-05,
+      "loss": 0.3326,
+      "step": 6482
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.455723391938026e-05,
+      "loss": 0.321,
+      "step": 6483
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.45527689618693e-05,
+      "loss": 0.3085,
+      "step": 6484
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.454830242590988e-05,
+      "loss": 0.3114,
+      "step": 6485
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.454383431191541e-05,
+      "loss": 0.2933,
+      "step": 6486
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.453936462029949e-05,
+      "loss": 0.3114,
+      "step": 6487
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4534893351475834e-05,
+      "loss": 0.309,
+      "step": 6488
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.453042050585833e-05,
+      "loss": 0.3149,
+      "step": 6489
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4525946083860984e-05,
+      "loss": 0.3114,
+      "step": 6490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4521470085897965e-05,
+      "loss": 0.3111,
+      "step": 6491
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.451699251238358e-05,
+      "loss": 0.3098,
+      "step": 6492
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.451251336373227e-05,
+      "loss": 0.3083,
+      "step": 6493
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.450803264035867e-05,
+      "loss": 0.3185,
+      "step": 6494
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.45035503426775e-05,
+      "loss": 0.323,
+      "step": 6495
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.449906647110367e-05,
+      "loss": 0.3163,
+      "step": 6496
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.44945810260522e-05,
+      "loss": 0.3146,
+      "step": 6497
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.449009400793829e-05,
+      "loss": 0.3294,
+      "step": 6498
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.448560541717726e-05,
+      "loss": 0.3033,
+      "step": 6499
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4481115254184604e-05,
+      "loss": 0.3012,
+      "step": 6500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4476623519375914e-05,
+      "loss": 0.3122,
+      "step": 6501
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.447213021316699e-05,
+      "loss": 0.3145,
+      "step": 6502
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.446763533597371e-05,
+      "loss": 0.312,
+      "step": 6503
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.446313888821216e-05,
+      "loss": 0.3156,
+      "step": 6504
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.445864087029852e-05,
+      "loss": 0.2861,
+      "step": 6505
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.445414128264916e-05,
+      "loss": 0.3119,
+      "step": 6506
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.444964012568056e-05,
+      "loss": 0.2888,
+      "step": 6507
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4445137399809363e-05,
+      "loss": 0.3334,
+      "step": 6508
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.444063310545234e-05,
+      "loss": 0.3224,
+      "step": 6509
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4436127243026445e-05,
+      "loss": 0.2827,
+      "step": 6510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.443161981294873e-05,
+      "loss": 0.3107,
+      "step": 6511
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.442711081563643e-05,
+      "loss": 0.306,
+      "step": 6512
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.44226002515069e-05,
+      "loss": 0.303,
+      "step": 6513
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4418088120977645e-05,
+      "loss": 0.3303,
+      "step": 6514
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4413574424466336e-05,
+      "loss": 0.3125,
+      "step": 6515
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.440905916239076e-05,
+      "loss": 0.3063,
+      "step": 6516
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4404542335168863e-05,
+      "loss": 0.29,
+      "step": 6517
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.440002394321874e-05,
+      "loss": 0.313,
+      "step": 6518
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.439550398695862e-05,
+      "loss": 0.3113,
+      "step": 6519
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4390982466806884e-05,
+      "loss": 0.2988,
+      "step": 6520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.438645938318205e-05,
+      "loss": 0.3248,
+      "step": 6521
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.438193473650279e-05,
+      "loss": 0.3045,
+      "step": 6522
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.437740852718793e-05,
+      "loss": 0.3252,
+      "step": 6523
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.43728807556564e-05,
+      "loss": 0.3109,
+      "step": 6524
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.436835142232733e-05,
+      "loss": 0.3119,
+      "step": 6525
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4363820527619945e-05,
+      "loss": 0.3067,
+      "step": 6526
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.435928807195365e-05,
+      "loss": 0.2838,
+      "step": 6527
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4354754055747966e-05,
+      "loss": 0.3127,
+      "step": 6528
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4350218479422597e-05,
+      "loss": 0.3134,
+      "step": 6529
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.434568134339735e-05,
+      "loss": 0.3058,
+      "step": 6530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.43411426480922e-05,
+      "loss": 0.3151,
+      "step": 6531
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433660239392726e-05,
+      "loss": 0.321,
+      "step": 6532
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.433206058132278e-05,
+      "loss": 0.34,
+      "step": 6533
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.432751721069918e-05,
+      "loss": 0.3171,
+      "step": 6534
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.432297228247698e-05,
+      "loss": 0.3274,
+      "step": 6535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4318425797076896e-05,
+      "loss": 0.3094,
+      "step": 6536
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4313877754919754e-05,
+      "loss": 0.3161,
+      "step": 6537
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4309328156426524e-05,
+      "loss": 0.3151,
+      "step": 6538
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.430477700201833e-05,
+      "loss": 0.316,
+      "step": 6539
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.430022429211645e-05,
+      "loss": 0.2909,
+      "step": 6540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.429567002714229e-05,
+      "loss": 0.3076,
+      "step": 6541
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.42911142075174e-05,
+      "loss": 0.2941,
+      "step": 6542
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.428655683366348e-05,
+      "loss": 0.3103,
+      "step": 6543
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.428199790600238e-05,
+      "loss": 0.3081,
+      "step": 6544
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.427743742495607e-05,
+      "loss": 0.3212,
+      "step": 6545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.42728753909467e-05,
+      "loss": 0.307,
+      "step": 6546
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.426831180439652e-05,
+      "loss": 0.3313,
+      "step": 6547
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4263746665727964e-05,
+      "loss": 0.3099,
+      "step": 6548
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4259179975363596e-05,
+      "loss": 0.3136,
+      "step": 6549
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4254611733726114e-05,
+      "loss": 0.3005,
+      "step": 6550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.425004194123836e-05,
+      "loss": 0.3243,
+      "step": 6551
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.424547059832334e-05,
+      "loss": 0.3156,
+      "step": 6552
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.424089770540418e-05,
+      "loss": 0.294,
+      "step": 6553
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.423632326290416e-05,
+      "loss": 0.2963,
+      "step": 6554
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4231747271246696e-05,
+      "loss": 0.3206,
+      "step": 6555
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.422716973085537e-05,
+      "loss": 0.3136,
+      "step": 6556
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.422259064215388e-05,
+      "loss": 0.3042,
+      "step": 6557
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.421801000556608e-05,
+      "loss": 0.3018,
+      "step": 6558
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.421342782151596e-05,
+      "loss": 0.3199,
+      "step": 6559
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.420884409042766e-05,
+      "loss": 0.2997,
+      "step": 6560
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.420425881272548e-05,
+      "loss": 0.304,
+      "step": 6561
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.419967198883382e-05,
+      "loss": 0.3061,
+      "step": 6562
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.419508361917727e-05,
+      "loss": 0.2926,
+      "step": 6563
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4190493704180526e-05,
+      "loss": 0.3199,
+      "step": 6564
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4185902244268444e-05,
+      "loss": 0.2878,
+      "step": 6565
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.418130923986603e-05,
+      "loss": 0.3061,
+      "step": 6566
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.417671469139842e-05,
+      "loss": 0.3015,
+      "step": 6567
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4172118599290894e-05,
+      "loss": 0.3161,
+      "step": 6568
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.416752096396888e-05,
+      "loss": 0.3314,
+      "step": 6569
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4162921785857945e-05,
+      "loss": 0.3075,
+      "step": 6570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4158321065383806e-05,
+      "loss": 0.32,
+      "step": 6571
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.415371880297232e-05,
+      "loss": 0.3179,
+      "step": 6572
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.414911499904948e-05,
+      "loss": 0.3246,
+      "step": 6573
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.414450965404142e-05,
+      "loss": 0.3324,
+      "step": 6574
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.413990276837443e-05,
+      "loss": 0.305,
+      "step": 6575
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.413529434247494e-05,
+      "loss": 0.3343,
+      "step": 6576
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.41306843767695e-05,
+      "loss": 0.3223,
+      "step": 6577
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.412607287168484e-05,
+      "loss": 0.3022,
+      "step": 6578
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.41214598276478e-05,
+      "loss": 0.2996,
+      "step": 6579
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4116845245085386e-05,
+      "loss": 0.3107,
+      "step": 6580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.411222912442472e-05,
+      "loss": 0.3093,
+      "step": 6581
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.41076114660931e-05,
+      "loss": 0.3309,
+      "step": 6582
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.410299227051795e-05,
+      "loss": 0.3126,
+      "step": 6583
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4098371538126806e-05,
+      "loss": 0.3138,
+      "step": 6584
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.40937492693474e-05,
+      "loss": 0.3018,
+      "step": 6585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.408912546460757e-05,
+      "loss": 0.3244,
+      "step": 6586
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4084500124335324e-05,
+      "loss": 0.3236,
+      "step": 6587
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.407987324895878e-05,
+      "loss": 0.285,
+      "step": 6588
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4075244838906216e-05,
+      "loss": 0.3175,
+      "step": 6589
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4070614894606054e-05,
+      "loss": 0.315,
+      "step": 6590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.406598341648685e-05,
+      "loss": 0.3793,
+      "step": 6591
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4061350404977295e-05,
+      "loss": 0.3094,
+      "step": 6592
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4056715860506255e-05,
+      "loss": 0.3122,
+      "step": 6593
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.40520797835027e-05,
+      "loss": 0.2948,
+      "step": 6594
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.404744217439577e-05,
+      "loss": 0.3239,
+      "step": 6595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.404280303361472e-05,
+      "loss": 0.3118,
+      "step": 6596
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.403816236158896e-05,
+      "loss": 0.316,
+      "step": 6597
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4033520158748055e-05,
+      "loss": 0.2872,
+      "step": 6598
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4028876425521694e-05,
+      "loss": 0.306,
+      "step": 6599
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4024231162339715e-05,
+      "loss": 0.2945,
+      "step": 6600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.40195843696321e-05,
+      "loss": 0.3195,
+      "step": 6601
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4014936047828945e-05,
+      "loss": 0.3205,
+      "step": 6602
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.401028619736053e-05,
+      "loss": 0.3222,
+      "step": 6603
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.400563481865726e-05,
+      "loss": 0.303,
+      "step": 6604
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.400098191214967e-05,
+      "loss": 0.3042,
+      "step": 6605
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.399632747826845e-05,
+      "loss": 0.316,
+      "step": 6606
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.399167151744442e-05,
+      "loss": 0.3339,
+      "step": 6607
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.3987014030108554e-05,
+      "loss": 0.3252,
+      "step": 6608
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.398235501669196e-05,
+      "loss": 0.3061,
+      "step": 6609
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.3977694477625895e-05,
+      "loss": 0.3126,
+      "step": 6610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.397303241334173e-05,
+      "loss": 0.3354,
+      "step": 6611
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.3968368824271014e-05,
+      "loss": 0.3117,
+      "step": 6612
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.396370371084543e-05,
+      "loss": 0.3176,
+      "step": 6613
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.395903707349677e-05,
+      "loss": 0.31,
+      "step": 6614
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.3954368912657e-05,
+      "loss": 0.2922,
+      "step": 6615
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.394969922875822e-05,
+      "loss": 0.3128,
+      "step": 6616
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.394502802223267e-05,
+      "loss": 0.3279,
+      "step": 6617
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.394035529351272e-05,
+      "loss": 0.325,
+      "step": 6618
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.3935681043030895e-05,
+      "loss": 0.3125,
+      "step": 6619
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.393100527121986e-05,
+      "loss": 0.3107,
+      "step": 6620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.3926327978512404e-05,
+      "loss": 0.2978,
+      "step": 6621
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.3921649165341486e-05,
+      "loss": 0.319,
+      "step": 6622
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.3916968832140176e-05,
+      "loss": 0.3174,
+      "step": 6623
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.391228697934171e-05,
+      "loss": 0.2915,
+      "step": 6624
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.390760360737943e-05,
+      "loss": 0.325,
+      "step": 6625
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.390291871668687e-05,
+      "loss": 0.3022,
+      "step": 6626
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3898232307697654e-05,
+      "loss": 0.3275,
+      "step": 6627
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.389354438084558e-05,
+      "loss": 0.3134,
+      "step": 6628
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.388885493656456e-05,
+      "loss": 0.3117,
+      "step": 6629
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.388416397528868e-05,
+      "loss": 0.298,
+      "step": 6630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.387947149745214e-05,
+      "loss": 0.3083,
+      "step": 6631
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3874777503489285e-05,
+      "loss": 0.3216,
+      "step": 6632
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3870081993834606e-05,
+      "loss": 0.3097,
+      "step": 6633
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.386538496892272e-05,
+      "loss": 0.3135,
+      "step": 6634
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3860686429188416e-05,
+      "loss": 0.3152,
+      "step": 6635
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.385598637506659e-05,
+      "loss": 0.3262,
+      "step": 6636
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.38512848069923e-05,
+      "loss": 0.3031,
+      "step": 6637
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3846581725400726e-05,
+      "loss": 0.3327,
+      "step": 6638
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.38418771307272e-05,
+      "loss": 0.3041,
+      "step": 6639
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3837171023407195e-05,
+      "loss": 0.306,
+      "step": 6640
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.383246340387631e-05,
+      "loss": 0.3072,
+      "step": 6641
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.382775427257032e-05,
+      "loss": 0.3037,
+      "step": 6642
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.382304362992509e-05,
+      "loss": 0.3252,
+      "step": 6643
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.381833147637665e-05,
+      "loss": 0.3014,
+      "step": 6644
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.381361781236118e-05,
+      "loss": 0.3145,
+      "step": 6645
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3808902638314975e-05,
+      "loss": 0.2955,
+      "step": 6646
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3804185954674506e-05,
+      "loss": 0.3074,
+      "step": 6647
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3799467761876344e-05,
+      "loss": 0.3168,
+      "step": 6648
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3794748060357224e-05,
+      "loss": 0.3094,
+      "step": 6649
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3790026850554006e-05,
+      "loss": 0.316,
+      "step": 6650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3785304132903705e-05,
+      "loss": 0.3082,
+      "step": 6651
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.378057990784346e-05,
+      "loss": 0.3036,
+      "step": 6652
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3775854175810575e-05,
+      "loss": 0.3321,
+      "step": 6653
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3771126937242454e-05,
+      "loss": 0.3032,
+      "step": 6654
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3766398192576684e-05,
+      "loss": 0.319,
+      "step": 6655
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.376166794225096e-05,
+      "loss": 0.3091,
+      "step": 6656
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.375693618670311e-05,
+      "loss": 0.3097,
+      "step": 6657
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3752202926371145e-05,
+      "loss": 0.3148,
+      "step": 6658
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.374746816169317e-05,
+      "loss": 0.2856,
+      "step": 6659
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3742731893107455e-05,
+      "loss": 0.3139,
+      "step": 6660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3737994121052396e-05,
+      "loss": 0.3228,
+      "step": 6661
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.373325484596654e-05,
+      "loss": 0.2946,
+      "step": 6662
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.372851406828855e-05,
+      "loss": 0.3164,
+      "step": 6663
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.372377178845727e-05,
+      "loss": 0.3056,
+      "step": 6664
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.371902800691165e-05,
+      "loss": 0.3026,
+      "step": 6665
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.371428272409077e-05,
+      "loss": 0.3061,
+      "step": 6666
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.370953594043388e-05,
+      "loss": 0.309,
+      "step": 6667
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.370478765638036e-05,
+      "loss": 0.2946,
+      "step": 6668
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.370003787236971e-05,
+      "loss": 0.3241,
+      "step": 6669
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.369528658884158e-05,
+      "loss": 0.295,
+      "step": 6670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3690533806235776e-05,
+      "loss": 0.3211,
+      "step": 6671
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3685779524992224e-05,
+      "loss": 0.31,
+      "step": 6672
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.368102374555099e-05,
+      "loss": 0.3155,
+      "step": 6673
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.367626646835227e-05,
+      "loss": 0.3173,
+      "step": 6674
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.367150769383644e-05,
+      "loss": 0.2961,
+      "step": 6675
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.366674742244395e-05,
+      "loss": 0.3083,
+      "step": 6676
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.366198565461544e-05,
+      "loss": 0.3189,
+      "step": 6677
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.365722239079167e-05,
+      "loss": 0.3032,
+      "step": 6678
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.365245763141353e-05,
+      "loss": 0.2951,
+      "step": 6679
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.3647691376922087e-05,
+      "loss": 0.2966,
+      "step": 6680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.364292362775849e-05,
+      "loss": 0.3221,
+      "step": 6681
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.363815438436407e-05,
+      "loss": 0.2925,
+      "step": 6682
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.363338364718026e-05,
+      "loss": 0.2974,
+      "step": 6683
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.362861141664868e-05,
+      "loss": 0.316,
+      "step": 6684
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.362383769321104e-05,
+      "loss": 0.3432,
+      "step": 6685
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.361906247730921e-05,
+      "loss": 0.3021,
+      "step": 6686
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.361428576938521e-05,
+      "loss": 0.3173,
+      "step": 6687
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.360950756988117e-05,
+      "loss": 0.3035,
+      "step": 6688
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.360472787923937e-05,
+      "loss": 0.3219,
+      "step": 6689
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.359994669790224e-05,
+      "loss": 0.3171,
+      "step": 6690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.359516402631234e-05,
+      "loss": 0.3126,
+      "step": 6691
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.359037986491237e-05,
+      "loss": 0.323,
+      "step": 6692
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.358559421414515e-05,
+      "loss": 0.3095,
+      "step": 6693
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.358080707445365e-05,
+      "loss": 0.3273,
+      "step": 6694
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3576018446281006e-05,
+      "loss": 0.2941,
+      "step": 6695
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3571228330070434e-05,
+      "loss": 0.3007,
+      "step": 6696
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.356643672626535e-05,
+      "loss": 0.2997,
+      "step": 6697
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.356164363530925e-05,
+      "loss": 0.3039,
+      "step": 6698
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.355684905764581e-05,
+      "loss": 0.3151,
+      "step": 6699
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.355205299371883e-05,
+      "loss": 0.2988,
+      "step": 6700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.354725544397224e-05,
+      "loss": 0.309,
+      "step": 6701
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.35424564088501e-05,
+      "loss": 0.2799,
+      "step": 6702
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3537655888796664e-05,
+      "loss": 0.3318,
+      "step": 6703
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.353285388425623e-05,
+      "loss": 0.3142,
+      "step": 6704
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.352805039567332e-05,
+      "loss": 0.3153,
+      "step": 6705
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.352324542349254e-05,
+      "loss": 0.3188,
+      "step": 6706
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3518438968158656e-05,
+      "loss": 0.2968,
+      "step": 6707
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3513631030116556e-05,
+      "loss": 0.3162,
+      "step": 6708
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.35088216098113e-05,
+      "loss": 0.305,
+      "step": 6709
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.350401070768803e-05,
+      "loss": 0.3192,
+      "step": 6710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3499198324192075e-05,
+      "loss": 0.3231,
+      "step": 6711
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3494384459768885e-05,
+      "loss": 0.3192,
+      "step": 6712
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3489569114864023e-05,
+      "loss": 0.3093,
+      "step": 6713
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.348475228992323e-05,
+      "loss": 0.3074,
+      "step": 6714
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.347993398539235e-05,
+      "loss": 0.3097,
+      "step": 6715
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.347511420171739e-05,
+      "loss": 0.3026,
+      "step": 6716
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.347029293934448e-05,
+      "loss": 0.328,
+      "step": 6717
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3465470198719876e-05,
+      "loss": 0.32,
+      "step": 6718
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.346064598028999e-05,
+      "loss": 0.2945,
+      "step": 6719
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.345582028450137e-05,
+      "loss": 0.2882,
+      "step": 6720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.345099311180069e-05,
+      "loss": 0.3262,
+      "step": 6721
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3446164462634765e-05,
+      "loss": 0.2886,
+      "step": 6722
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.344133433745055e-05,
+      "loss": 0.3011,
+      "step": 6723
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.343650273669513e-05,
+      "loss": 0.2952,
+      "step": 6724
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.343166966081573e-05,
+      "loss": 0.2962,
+      "step": 6725
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.342683511025972e-05,
+      "loss": 0.3092,
+      "step": 6726
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.342199908547459e-05,
+      "loss": 0.3216,
+      "step": 6727
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.341716158690797e-05,
+      "loss": 0.2941,
+      "step": 6728
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3412322615007656e-05,
+      "loss": 0.3239,
+      "step": 6729
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3407482170221535e-05,
+      "loss": 0.3368,
+      "step": 6730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.340264025299765e-05,
+      "loss": 0.3083,
+      "step": 6731
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.339779686378419e-05,
+      "loss": 0.299,
+      "step": 6732
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.339295200302947e-05,
+      "loss": 0.3128,
+      "step": 6733
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.338810567118194e-05,
+      "loss": 0.3175,
+      "step": 6734
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.338325786869019e-05,
+      "loss": 0.3061,
+      "step": 6735
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.337840859600294e-05,
+      "loss": 0.3177,
+      "step": 6736
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.337355785356907e-05,
+      "loss": 0.3186,
+      "step": 6737
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.336870564183756e-05,
+      "loss": 0.3084,
+      "step": 6738
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.336385196125754e-05,
+      "loss": 0.3199,
+      "step": 6739
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.3358996812278296e-05,
+      "loss": 0.3096,
+      "step": 6740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.335414019534922e-05,
+      "loss": 0.3222,
+      "step": 6741
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.334928211091985e-05,
+      "loss": 0.3157,
+      "step": 6742
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3344422559439876e-05,
+      "loss": 0.3108,
+      "step": 6743
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3339561541359107e-05,
+      "loss": 0.3296,
+      "step": 6744
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.333469905712749e-05,
+      "loss": 0.2967,
+      "step": 6745
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.33298351071951e-05,
+      "loss": 0.323,
+      "step": 6746
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.332496969201217e-05,
+      "loss": 0.3385,
+      "step": 6747
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.332010281202905e-05,
+      "loss": 0.314,
+      "step": 6748
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.331523446769623e-05,
+      "loss": 0.2871,
+      "step": 6749
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.331036465946434e-05,
+      "loss": 0.2931,
+      "step": 6750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.330549338778414e-05,
+      "loss": 0.3337,
+      "step": 6751
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.330062065310652e-05,
+      "loss": 0.3189,
+      "step": 6752
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.329574645588253e-05,
+      "loss": 0.3053,
+      "step": 6753
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.329087079656333e-05,
+      "loss": 0.307,
+      "step": 6754
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3285993675600214e-05,
+      "loss": 0.3155,
+      "step": 6755
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3281115093444637e-05,
+      "loss": 0.3103,
+      "step": 6756
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.327623505054816e-05,
+      "loss": 0.3026,
+      "step": 6757
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3271353547362494e-05,
+      "loss": 0.332,
+      "step": 6758
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.32664705843395e-05,
+      "loss": 0.3035,
+      "step": 6759
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.326158616193113e-05,
+      "loss": 0.3188,
+      "step": 6760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3256700280589525e-05,
+      "loss": 0.2976,
+      "step": 6761
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3251812940766925e-05,
+      "loss": 0.3087,
+      "step": 6762
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.324692414291571e-05,
+      "loss": 0.3124,
+      "step": 6763
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.32420338874884e-05,
+      "loss": 0.3014,
+      "step": 6764
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3237142174937655e-05,
+      "loss": 0.3132,
+      "step": 6765
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.323224900571626e-05,
+      "loss": 0.3139,
+      "step": 6766
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.322735438027715e-05,
+      "loss": 0.3187,
+      "step": 6767
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3222458299073375e-05,
+      "loss": 0.3167,
+      "step": 6768
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.321756076255813e-05,
+      "loss": 0.3365,
+      "step": 6769
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.321266177118474e-05,
+      "loss": 0.2764,
+      "step": 6770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.320776132540667e-05,
+      "loss": 0.3144,
+      "step": 6771
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3202859425677526e-05,
+      "loss": 0.317,
+      "step": 6772
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.319795607245103e-05,
+      "loss": 0.3151,
+      "step": 6773
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.319305126618106e-05,
+      "loss": 0.3194,
+      "step": 6774
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3188145007321614e-05,
+      "loss": 0.2998,
+      "step": 6775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.318323729632682e-05,
+      "loss": 0.3222,
+      "step": 6776
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.317832813365095e-05,
+      "loss": 0.3234,
+      "step": 6777
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3173417519748424e-05,
+      "loss": 0.3036,
+      "step": 6778
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.316850545507376e-05,
+      "loss": 0.3125,
+      "step": 6779
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.316359194008165e-05,
+      "loss": 0.3228,
+      "step": 6780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.315867697522688e-05,
+      "loss": 0.3011,
+      "step": 6781
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.315376056096442e-05,
+      "loss": 0.2974,
+      "step": 6782
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.314884269774932e-05,
+      "loss": 0.3129,
+      "step": 6783
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.31439233860368e-05,
+      "loss": 0.3172,
+      "step": 6784
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.313900262628221e-05,
+      "loss": 0.3407,
+      "step": 6785
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.313408041894103e-05,
+      "loss": 0.3116,
+      "step": 6786
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.312915676446886e-05,
+      "loss": 0.307,
+      "step": 6787
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.312423166332144e-05,
+      "loss": 0.3066,
+      "step": 6788
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3119305115954676e-05,
+      "loss": 0.2997,
+      "step": 6789
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3114377122824574e-05,
+      "loss": 0.3225,
+      "step": 6790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.310944768438726e-05,
+      "loss": 0.3104,
+      "step": 6791
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3104516801099046e-05,
+      "loss": 0.3136,
+      "step": 6792
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.309958447341633e-05,
+      "loss": 0.2831,
+      "step": 6793
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.309465070179566e-05,
+      "loss": 0.3114,
+      "step": 6794
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.308971548669372e-05,
+      "loss": 0.3161,
+      "step": 6795
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3084778828567335e-05,
+      "loss": 0.3013,
+      "step": 6796
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3079840727873446e-05,
+      "loss": 0.3121,
+      "step": 6797
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.3074901185069136e-05,
+      "loss": 0.3033,
+      "step": 6798
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.306996020061163e-05,
+      "loss": 0.3089,
+      "step": 6799
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.3065017774958265e-05,
+      "loss": 0.3177,
+      "step": 6800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.306007390856655e-05,
+      "loss": 0.3181,
+      "step": 6801
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.3055128601894074e-05,
+      "loss": 0.2963,
+      "step": 6802
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.3050181855398595e-05,
+      "loss": 0.3116,
+      "step": 6803
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.3045233669538e-05,
+      "loss": 0.3123,
+      "step": 6804
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.304028404477031e-05,
+      "loss": 0.3025,
+      "step": 6805
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.3035332981553684e-05,
+      "loss": 0.3229,
+      "step": 6806
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.303038048034638e-05,
+      "loss": 0.3219,
+      "step": 6807
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.3025426541606826e-05,
+      "loss": 0.2905,
+      "step": 6808
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.302047116579357e-05,
+      "loss": 0.3205,
+      "step": 6809
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.3015514353365307e-05,
+      "loss": 0.3025,
+      "step": 6810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.301055610478083e-05,
+      "loss": 0.3142,
+      "step": 6811
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.300559642049911e-05,
+      "loss": 0.2923,
+      "step": 6812
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.300063530097921e-05,
+      "loss": 0.3127,
+      "step": 6813
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.299567274668036e-05,
+      "loss": 0.3265,
+      "step": 6814
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2990708758061884e-05,
+      "loss": 0.3055,
+      "step": 6815
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.298574333558329e-05,
+      "loss": 0.3143,
+      "step": 6816
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.298077647970417e-05,
+      "loss": 0.3076,
+      "step": 6817
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.297580819088428e-05,
+      "loss": 0.3147,
+      "step": 6818
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.297083846958349e-05,
+      "loss": 0.3239,
+      "step": 6819
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.296586731626181e-05,
+      "loss": 0.3111,
+      "step": 6820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2960894731379395e-05,
+      "loss": 0.3277,
+      "step": 6821
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.295592071539651e-05,
+      "loss": 0.2992,
+      "step": 6822
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2950945268773555e-05,
+      "loss": 0.3024,
+      "step": 6823
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.29459683919711e-05,
+      "loss": 0.3178,
+      "step": 6824
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2940990085449775e-05,
+      "loss": 0.2975,
+      "step": 6825
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.293601034967042e-05,
+      "loss": 0.3107,
+      "step": 6826
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.293102918509396e-05,
+      "loss": 0.2927,
+      "step": 6827
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2926046592181456e-05,
+      "loss": 0.3218,
+      "step": 6828
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.292106257139413e-05,
+      "loss": 0.319,
+      "step": 6829
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.291607712319329e-05,
+      "loss": 0.3154,
+      "step": 6830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.291109024804043e-05,
+      "loss": 0.3127,
+      "step": 6831
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.290610194639712e-05,
+      "loss": 0.3192,
+      "step": 6832
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.290111221872511e-05,
+      "loss": 0.3236,
+      "step": 6833
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.289612106548625e-05,
+      "loss": 0.3333,
+      "step": 6834
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.289112848714254e-05,
+      "loss": 0.3183,
+      "step": 6835
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.288613448415612e-05,
+      "loss": 0.3206,
+      "step": 6836
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.288113905698922e-05,
+      "loss": 0.3194,
+      "step": 6837
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.287614220610426e-05,
+      "loss": 0.2996,
+      "step": 6838
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.287114393196373e-05,
+      "loss": 0.3317,
+      "step": 6839
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2866144235030304e-05,
+      "loss": 0.3228,
+      "step": 6840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.286114311576676e-05,
+      "loss": 0.3206,
+      "step": 6841
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2856140574636015e-05,
+      "loss": 0.32,
+      "step": 6842
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2851136612101114e-05,
+      "loss": 0.3227,
+      "step": 6843
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2846131228625246e-05,
+      "loss": 0.3363,
+      "step": 6844
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2841124424671716e-05,
+      "loss": 0.3314,
+      "step": 6845
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.283611620070397e-05,
+      "loss": 0.3209,
+      "step": 6846
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.283110655718557e-05,
+      "loss": 0.3249,
+      "step": 6847
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.282609549458024e-05,
+      "loss": 0.3124,
+      "step": 6848
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.282108301335181e-05,
+      "loss": 0.3327,
+      "step": 6849
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2816069113964245e-05,
+      "loss": 0.2943,
+      "step": 6850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.281105379688164e-05,
+      "loss": 0.312,
+      "step": 6851
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.280603706256824e-05,
+      "loss": 0.3227,
+      "step": 6852
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.28010189114884e-05,
+      "loss": 0.2913,
+      "step": 6853
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2795999344106605e-05,
+      "loss": 0.321,
+      "step": 6854
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.2790978360887485e-05,
+      "loss": 0.307,
+      "step": 6855
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.27859559622958e-05,
+      "loss": 0.3341,
+      "step": 6856
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.278093214879644e-05,
+      "loss": 0.2956,
+      "step": 6857
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.27759069208544e-05,
+      "loss": 0.3145,
+      "step": 6858
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.277088027893485e-05,
+      "loss": 0.2966,
+      "step": 6859
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.276585222350306e-05,
+      "loss": 0.3277,
+      "step": 6860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2760822755024446e-05,
+      "loss": 0.3237,
+      "step": 6861
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.275579187396453e-05,
+      "loss": 0.3005,
+      "step": 6862
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.275075958078901e-05,
+      "loss": 0.3079,
+      "step": 6863
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2745725875963684e-05,
+      "loss": 0.3044,
+      "step": 6864
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.274069075995446e-05,
+      "loss": 0.3281,
+      "step": 6865
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2735654233227424e-05,
+      "loss": 0.3191,
+      "step": 6866
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.273061629624877e-05,
+      "loss": 0.3063,
+      "step": 6867
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2725576949484805e-05,
+      "loss": 0.32,
+      "step": 6868
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2720536193402e-05,
+      "loss": 0.306,
+      "step": 6869
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.271549402846693e-05,
+      "loss": 0.3055,
+      "step": 6870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.271045045514633e-05,
+      "loss": 0.2963,
+      "step": 6871
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2705405473907026e-05,
+      "loss": 0.3113,
+      "step": 6872
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2700359085216004e-05,
+      "loss": 0.3288,
+      "step": 6873
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2695311289540365e-05,
+      "loss": 0.3063,
+      "step": 6874
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.269026208734734e-05,
+      "loss": 0.3139,
+      "step": 6875
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.268521147910432e-05,
+      "loss": 0.3004,
+      "step": 6876
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.268015946527878e-05,
+      "loss": 0.3065,
+      "step": 6877
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.267510604633836e-05,
+      "loss": 0.3081,
+      "step": 6878
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.267005122275081e-05,
+      "loss": 0.315,
+      "step": 6879
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.266499499498402e-05,
+      "loss": 0.3099,
+      "step": 6880
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.265993736350601e-05,
+      "loss": 0.3288,
+      "step": 6881
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.265487832878492e-05,
+      "loss": 0.32,
+      "step": 6882
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.264981789128904e-05,
+      "loss": 0.3205,
+      "step": 6883
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2644756051486764e-05,
+      "loss": 0.3043,
+      "step": 6884
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.263969280984664e-05,
+      "loss": 0.3118,
+      "step": 6885
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2634628166837325e-05,
+      "loss": 0.3091,
+      "step": 6886
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.262956212292763e-05,
+      "loss": 0.3138,
+      "step": 6887
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2624494678586455e-05,
+      "loss": 0.3291,
+      "step": 6888
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.261942583428288e-05,
+      "loss": 0.3042,
+      "step": 6889
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2614355590486095e-05,
+      "loss": 0.3023,
+      "step": 6890
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.260928394766539e-05,
+      "loss": 0.3343,
+      "step": 6891
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2604210906290215e-05,
+      "loss": 0.314,
+      "step": 6892
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2599136466830156e-05,
+      "loss": 0.3284,
+      "step": 6893
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2594060629754925e-05,
+      "loss": 0.3298,
+      "step": 6894
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.258898339553432e-05,
+      "loss": 0.3183,
+      "step": 6895
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2583904764638334e-05,
+      "loss": 0.3034,
+      "step": 6896
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2578824737537044e-05,
+      "loss": 0.2911,
+      "step": 6897
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.257374331470068e-05,
+      "loss": 0.3099,
+      "step": 6898
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2568660496599576e-05,
+      "loss": 0.3132,
+      "step": 6899
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.256357628370423e-05,
+      "loss": 0.3072,
+      "step": 6900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.255849067648524e-05,
+      "loss": 0.2903,
+      "step": 6901
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.255340367541334e-05,
+      "loss": 0.2926,
+      "step": 6902
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2548315280959394e-05,
+      "loss": 0.3091,
+      "step": 6903
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2543225493594414e-05,
+      "loss": 0.3134,
+      "step": 6904
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2538134313789506e-05,
+      "loss": 0.3233,
+      "step": 6905
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2533041742015925e-05,
+      "loss": 0.3405,
+      "step": 6906
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.252794777874506e-05,
+      "loss": 0.2889,
+      "step": 6907
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.252285242444842e-05,
+      "loss": 0.3218,
+      "step": 6908
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2517755679597646e-05,
+      "loss": 0.3059,
+      "step": 6909
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2512657544664506e-05,
+      "loss": 0.2972,
+      "step": 6910
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2507558020120886e-05,
+      "loss": 0.3092,
+      "step": 6911
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.2502457106438815e-05,
+      "loss": 0.2945,
+      "step": 6912
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.249735480409045e-05,
+      "loss": 0.3062,
+      "step": 6913
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.249225111354808e-05,
+      "loss": 0.3074,
+      "step": 6914
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.24871460352841e-05,
+      "loss": 0.3039,
+      "step": 6915
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.248203956977107e-05,
+      "loss": 0.2939,
+      "step": 6916
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.247693171748164e-05,
+      "loss": 0.3007,
+      "step": 6917
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2471822478888605e-05,
+      "loss": 0.2923,
+      "step": 6918
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.246671185446491e-05,
+      "loss": 0.3076,
+      "step": 6919
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.246159984468359e-05,
+      "loss": 0.3226,
+      "step": 6920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.245648645001782e-05,
+      "loss": 0.31,
+      "step": 6921
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.245137167094093e-05,
+      "loss": 0.3032,
+      "step": 6922
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.244625550792634e-05,
+      "loss": 0.3173,
+      "step": 6923
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.244113796144763e-05,
+      "loss": 0.2968,
+      "step": 6924
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.243601903197847e-05,
+      "loss": 0.3116,
+      "step": 6925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.243089871999271e-05,
+      "loss": 0.3046,
+      "step": 6926
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2425777025964284e-05,
+      "loss": 0.3244,
+      "step": 6927
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2420653950367264e-05,
+      "loss": 0.2784,
+      "step": 6928
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.241552949367587e-05,
+      "loss": 0.3062,
+      "step": 6929
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.241040365636442e-05,
+      "loss": 0.2989,
+      "step": 6930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.240527643890738e-05,
+      "loss": 0.3235,
+      "step": 6931
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.240014784177935e-05,
+      "loss": 0.2886,
+      "step": 6932
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.239501786545503e-05,
+      "loss": 0.295,
+      "step": 6933
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.238988651040927e-05,
+      "loss": 0.3054,
+      "step": 6934
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2384753777117045e-05,
+      "loss": 0.3138,
+      "step": 6935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2379619666053443e-05,
+      "loss": 0.3102,
+      "step": 6936
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.237448417769371e-05,
+      "loss": 0.3214,
+      "step": 6937
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.236934731251318e-05,
+      "loss": 0.3201,
+      "step": 6938
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2364209070987346e-05,
+      "loss": 0.3195,
+      "step": 6939
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2359069453591816e-05,
+      "loss": 0.3098,
+      "step": 6940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2353928460802326e-05,
+      "loss": 0.3076,
+      "step": 6941
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2348786093094733e-05,
+      "loss": 0.3123,
+      "step": 6942
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2343642350945036e-05,
+      "loss": 0.3346,
+      "step": 6943
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.233849723482935e-05,
+      "loss": 0.3126,
+      "step": 6944
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2333350745223924e-05,
+      "loss": 0.2993,
+      "step": 6945
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2328202882605124e-05,
+      "loss": 0.3046,
+      "step": 6946
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.232305364744945e-05,
+      "loss": 0.306,
+      "step": 6947
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.231790304023353e-05,
+      "loss": 0.3118,
+      "step": 6948
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.231275106143412e-05,
+      "loss": 0.3198,
+      "step": 6949
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2307597711528106e-05,
+      "loss": 0.3084,
+      "step": 6950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.230244299099248e-05,
+      "loss": 0.3,
+      "step": 6951
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.229728690030438e-05,
+      "loss": 0.3021,
+      "step": 6952
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.229212943994108e-05,
+      "loss": 0.3016,
+      "step": 6953
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2286970610379966e-05,
+      "loss": 0.3153,
+      "step": 6954
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.228181041209853e-05,
+      "loss": 0.3162,
+      "step": 6955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2276648845574444e-05,
+      "loss": 0.2948,
+      "step": 6956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.227148591128546e-05,
+      "loss": 0.3112,
+      "step": 6957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.226632160970946e-05,
+      "loss": 0.3055,
+      "step": 6958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.226115594132449e-05,
+      "loss": 0.3234,
+      "step": 6959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.225598890660868e-05,
+      "loss": 0.3127,
+      "step": 6960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2250820506040326e-05,
+      "loss": 0.2956,
+      "step": 6961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.22456507400978e-05,
+      "loss": 0.3094,
+      "step": 6962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.224047960925965e-05,
+      "loss": 0.3172,
+      "step": 6963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.223530711400452e-05,
+      "loss": 0.3151,
+      "step": 6964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2230133254811184e-05,
+      "loss": 0.3389,
+      "step": 6965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.222495803215856e-05,
+      "loss": 0.3016,
+      "step": 6966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2219781446525676e-05,
+      "loss": 0.3285,
+      "step": 6967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.221460349839169e-05,
+      "loss": 0.3109,
+      "step": 6968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.220942418823588e-05,
+      "loss": 0.3396,
+      "step": 6969
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.2204243516537675e-05,
+      "loss": 0.3367,
+      "step": 6970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.219906148377659e-05,
+      "loss": 0.2974,
+      "step": 6971
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2193878090432296e-05,
+      "loss": 0.3078,
+      "step": 6972
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2188693336984576e-05,
+      "loss": 0.3084,
+      "step": 6973
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.218350722391335e-05,
+      "loss": 0.3382,
+      "step": 6974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2178319751698665e-05,
+      "loss": 0.3295,
+      "step": 6975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.217313092082068e-05,
+      "loss": 0.3432,
+      "step": 6976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2167940731759674e-05,
+      "loss": 0.3178,
+      "step": 6977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.216274918499608e-05,
+      "loss": 0.3176,
+      "step": 6978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.215755628101045e-05,
+      "loss": 0.3116,
+      "step": 6979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2152362020283426e-05,
+      "loss": 0.3217,
+      "step": 6980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.214716640329582e-05,
+      "loss": 0.3245,
+      "step": 6981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.214196943052855e-05,
+      "loss": 0.3069,
+      "step": 6982
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2136771102462655e-05,
+      "loss": 0.3058,
+      "step": 6983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.213157141957931e-05,
+      "loss": 0.3066,
+      "step": 6984
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2126370382359815e-05,
+      "loss": 0.2972,
+      "step": 6985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.212116799128559e-05,
+      "loss": 0.3176,
+      "step": 6986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.211596424683817e-05,
+      "loss": 0.3145,
+      "step": 6987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.211075914949924e-05,
+      "loss": 0.3154,
+      "step": 6988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2105552699750595e-05,
+      "loss": 0.3186,
+      "step": 6989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.210034489807416e-05,
+      "loss": 0.331,
+      "step": 6990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.209513574495197e-05,
+      "loss": 0.3209,
+      "step": 6991
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2089925240866206e-05,
+      "loss": 0.3184,
+      "step": 6992
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.208471338629918e-05,
+      "loss": 0.3167,
+      "step": 6993
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.207950018173328e-05,
+      "loss": 0.3163,
+      "step": 6994
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2074285627651086e-05,
+      "loss": 0.308,
+      "step": 6995
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.206906972453527e-05,
+      "loss": 0.3101,
+      "step": 6996
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2063852472868595e-05,
+      "loss": 0.2991,
+      "step": 6997
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2058633873134025e-05,
+      "loss": 0.2877,
+      "step": 6998
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.205341392581458e-05,
+      "loss": 0.3185,
+      "step": 6999
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.204819263139344e-05,
+      "loss": 0.3061,
+      "step": 7000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2042969990353916e-05,
+      "loss": 0.328,
+      "step": 7001
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.20377460031794e-05,
+      "loss": 0.3167,
+      "step": 7002
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2032520670353464e-05,
+      "loss": 0.3222,
+      "step": 7003
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.202729399235976e-05,
+      "loss": 0.3139,
+      "step": 7004
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.20220659696821e-05,
+      "loss": 0.3143,
+      "step": 7005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2016836602804386e-05,
+      "loss": 0.3167,
+      "step": 7006
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.201160589221068e-05,
+      "loss": 0.294,
+      "step": 7007
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.2006373838385125e-05,
+      "loss": 0.3028,
+      "step": 7008
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.200114044181204e-05,
+      "loss": 0.331,
+      "step": 7009
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.199590570297583e-05,
+      "loss": 0.2818,
+      "step": 7010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.199066962236103e-05,
+      "loss": 0.307,
+      "step": 7011
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.198543220045231e-05,
+      "loss": 0.2988,
+      "step": 7012
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.198019343773447e-05,
+      "loss": 0.3135,
+      "step": 7013
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.1974953334692406e-05,
+      "loss": 0.318,
+      "step": 7014
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.196971189181117e-05,
+      "loss": 0.3229,
+      "step": 7015
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.196446910957591e-05,
+      "loss": 0.3117,
+      "step": 7016
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.195922498847192e-05,
+      "loss": 0.3142,
+      "step": 7017
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.195397952898461e-05,
+      "loss": 0.2923,
+      "step": 7018
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.1948732731599515e-05,
+      "loss": 0.3265,
+      "step": 7019
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.194348459680228e-05,
+      "loss": 0.3168,
+      "step": 7020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.19382351250787e-05,
+      "loss": 0.2958,
+      "step": 7021
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.193298431691467e-05,
+      "loss": 0.3241,
+      "step": 7022
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.192773217279623e-05,
+      "loss": 0.3053,
+      "step": 7023
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.1922478693209525e-05,
+      "loss": 0.3318,
+      "step": 7024
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.1917223878640824e-05,
+      "loss": 0.2872,
+      "step": 7025
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.1911967729576526e-05,
+      "loss": 0.3095,
+      "step": 7026
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.190671024650317e-05,
+      "loss": 0.2959,
+      "step": 7027
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.190145142990739e-05,
+      "loss": 0.3033,
+      "step": 7028
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.189619128027596e-05,
+      "loss": 0.2939,
+      "step": 7029
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1890929798095776e-05,
+      "loss": 0.3307,
+      "step": 7030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1885666983853845e-05,
+      "loss": 0.3334,
+      "step": 7031
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.188040283803731e-05,
+      "loss": 0.322,
+      "step": 7032
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1875137361133435e-05,
+      "loss": 0.3119,
+      "step": 7033
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1869870553629606e-05,
+      "loss": 0.3091,
+      "step": 7034
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.186460241601333e-05,
+      "loss": 0.3106,
+      "step": 7035
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.185933294877225e-05,
+      "loss": 0.3119,
+      "step": 7036
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1854062152394115e-05,
+      "loss": 0.3162,
+      "step": 7037
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1848790027366805e-05,
+      "loss": 0.322,
+      "step": 7038
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.18435165741783e-05,
+      "loss": 0.3224,
+      "step": 7039
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.183824179331677e-05,
+      "loss": 0.3119,
+      "step": 7040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.183296568527042e-05,
+      "loss": 0.3056,
+      "step": 7041
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.182768825052763e-05,
+      "loss": 0.3231,
+      "step": 7042
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.182240948957691e-05,
+      "loss": 0.3034,
+      "step": 7043
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1817129402906864e-05,
+      "loss": 0.3184,
+      "step": 7044
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1811847991006234e-05,
+      "loss": 0.2972,
+      "step": 7045
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.180656525436387e-05,
+      "loss": 0.3617,
+      "step": 7046
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.180128119346878e-05,
+      "loss": 0.3309,
+      "step": 7047
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.179599580881004e-05,
+      "loss": 0.3263,
+      "step": 7048
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.17907091008769e-05,
+      "loss": 0.3068,
+      "step": 7049
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.17854210701587e-05,
+      "loss": 0.3173,
+      "step": 7050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.178013171714492e-05,
+      "loss": 0.312,
+      "step": 7051
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.177484104232516e-05,
+      "loss": 0.3014,
+      "step": 7052
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.176954904618913e-05,
+      "loss": 0.3166,
+      "step": 7053
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1764255729226676e-05,
+      "loss": 0.3138,
+      "step": 7054
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1758961091927765e-05,
+      "loss": 0.322,
+      "step": 7055
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.175366513478247e-05,
+      "loss": 0.3218,
+      "step": 7056
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.174836785828101e-05,
+      "loss": 0.3063,
+      "step": 7057
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1743069262913706e-05,
+      "loss": 0.323,
+      "step": 7058
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.173776934917102e-05,
+      "loss": 0.2998,
+      "step": 7059
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.173246811754351e-05,
+      "loss": 0.3022,
+      "step": 7060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.172716556852189e-05,
+      "loss": 0.2949,
+      "step": 7061
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.172186170259697e-05,
+      "loss": 0.3156,
+      "step": 7062
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1716556520259684e-05,
+      "loss": 0.3266,
+      "step": 7063
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.171125002200109e-05,
+      "loss": 0.3003,
+      "step": 7064
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.170594220831239e-05,
+      "loss": 0.3239,
+      "step": 7065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1700633079684884e-05,
+      "loss": 0.3067,
+      "step": 7066
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.169532263660998e-05,
+      "loss": 0.3133,
+      "step": 7067
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.169001087957924e-05,
+      "loss": 0.3125,
+      "step": 7068
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.168469780908434e-05,
+      "loss": 0.3083,
+      "step": 7069
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.167938342561706e-05,
+      "loss": 0.2929,
+      "step": 7070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.167406772966933e-05,
+      "loss": 0.3345,
+      "step": 7071
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1668750721733166e-05,
+      "loss": 0.3009,
+      "step": 7072
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.166343240230072e-05,
+      "loss": 0.3181,
+      "step": 7073
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.165811277186429e-05,
+      "loss": 0.3202,
+      "step": 7074
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1652791830916266e-05,
+      "loss": 0.2867,
+      "step": 7075
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.164746957994916e-05,
+      "loss": 0.3157,
+      "step": 7076
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.1642146019455633e-05,
+      "loss": 0.2992,
+      "step": 7077
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.163682114992842e-05,
+      "loss": 0.3132,
+      "step": 7078
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.163149497186043e-05,
+      "loss": 0.3224,
+      "step": 7079
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.162616748574465e-05,
+      "loss": 0.3154,
+      "step": 7080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.162083869207422e-05,
+      "loss": 0.3164,
+      "step": 7081
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.161550859134238e-05,
+      "loss": 0.3137,
+      "step": 7082
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.161017718404249e-05,
+      "loss": 0.3056,
+      "step": 7083
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.160484447066805e-05,
+      "loss": 0.314,
+      "step": 7084
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.159951045171267e-05,
+      "loss": 0.3139,
+      "step": 7085
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.159417512767008e-05,
+      "loss": 0.3233,
+      "step": 7086
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1588838499034124e-05,
+      "loss": 0.3094,
+      "step": 7087
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.158350056629878e-05,
+      "loss": 0.3005,
+      "step": 7088
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1578161329958135e-05,
+      "loss": 0.3075,
+      "step": 7089
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.157282079050642e-05,
+      "loss": 0.3046,
+      "step": 7090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.156747894843794e-05,
+      "loss": 0.335,
+      "step": 7091
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1562135804247174e-05,
+      "loss": 0.3186,
+      "step": 7092
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.15567913584287e-05,
+      "loss": 0.3051,
+      "step": 7093
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.155144561147719e-05,
+      "loss": 0.313,
+      "step": 7094
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.154609856388748e-05,
+      "loss": 0.3091,
+      "step": 7095
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.15407502161545e-05,
+      "loss": 0.3058,
+      "step": 7096
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1535400568773295e-05,
+      "loss": 0.3178,
+      "step": 7097
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.153004962223907e-05,
+      "loss": 0.3024,
+      "step": 7098
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.15246973770471e-05,
+      "loss": 0.3227,
+      "step": 7099
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.15193438336928e-05,
+      "loss": 0.3156,
+      "step": 7100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1513988992671734e-05,
+      "loss": 0.3157,
+      "step": 7101
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.150863285447954e-05,
+      "loss": 0.2943,
+      "step": 7102
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.150327541961199e-05,
+      "loss": 0.3015,
+      "step": 7103
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1497916688565e-05,
+      "loss": 0.3328,
+      "step": 7104
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1492556661834577e-05,
+      "loss": 0.308,
+      "step": 7105
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1487195339916854e-05,
+      "loss": 0.3151,
+      "step": 7106
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1481832723308106e-05,
+      "loss": 0.2939,
+      "step": 7107
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.147646881250471e-05,
+      "loss": 0.2908,
+      "step": 7108
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1471103608003134e-05,
+      "loss": 0.3241,
+      "step": 7109
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.146573711030002e-05,
+      "loss": 0.2886,
+      "step": 7110
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.14603693198921e-05,
+      "loss": 0.3076,
+      "step": 7111
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.145500023727624e-05,
+      "loss": 0.3182,
+      "step": 7112
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1449629862949404e-05,
+      "loss": 0.2976,
+      "step": 7113
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.144425819740868e-05,
+      "loss": 0.315,
+      "step": 7114
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.143888524115131e-05,
+      "loss": 0.3345,
+      "step": 7115
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.143351099467461e-05,
+      "loss": 0.311,
+      "step": 7116
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1428135458476024e-05,
+      "loss": 0.3071,
+      "step": 7117
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.142275863305315e-05,
+      "loss": 0.3051,
+      "step": 7118
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.141738051890367e-05,
+      "loss": 0.2987,
+      "step": 7119
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.141200111652539e-05,
+      "loss": 0.3069,
+      "step": 7120
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.140662042641625e-05,
+      "loss": 0.3208,
+      "step": 7121
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.14012384490743e-05,
+      "loss": 0.341,
+      "step": 7122
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.139585518499771e-05,
+      "loss": 0.3222,
+      "step": 7123
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1390470634684756e-05,
+      "loss": 0.3197,
+      "step": 7124
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1385084798633865e-05,
+      "loss": 0.3217,
+      "step": 7125
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1379697677343555e-05,
+      "loss": 0.3171,
+      "step": 7126
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.137430927131248e-05,
+      "loss": 0.3301,
+      "step": 7127
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.136891958103938e-05,
+      "loss": 0.3153,
+      "step": 7128
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.136352860702316e-05,
+      "loss": 0.2879,
+      "step": 7129
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.135813634976282e-05,
+      "loss": 0.3319,
+      "step": 7130
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.135274280975748e-05,
+      "loss": 0.3116,
+      "step": 7131
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1347347987506394e-05,
+      "loss": 0.323,
+      "step": 7132
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.134195188350888e-05,
+      "loss": 0.3136,
+      "step": 7133
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.133655449826446e-05,
+      "loss": 0.3094,
+      "step": 7134
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1331155832272705e-05,
+      "loss": 0.3055,
+      "step": 7135
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.132575588603334e-05,
+      "loss": 0.3204,
+      "step": 7136
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1320354660046195e-05,
+      "loss": 0.284,
+      "step": 7137
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.131495215481122e-05,
+      "loss": 0.311,
+      "step": 7138
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.130954837082848e-05,
+      "loss": 0.3235,
+      "step": 7139
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.130414330859817e-05,
+      "loss": 0.313,
+      "step": 7140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.12987369686206e-05,
+      "loss": 0.3337,
+      "step": 7141
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.1293329351396185e-05,
+      "loss": 0.3004,
+      "step": 7142
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1287920457425476e-05,
+      "loss": 0.3028,
+      "step": 7143
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.128251028720913e-05,
+      "loss": 0.3141,
+      "step": 7144
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.127709884124792e-05,
+      "loss": 0.3191,
+      "step": 7145
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.127168612004277e-05,
+      "loss": 0.3053,
+      "step": 7146
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.126627212409467e-05,
+      "loss": 0.2977,
+      "step": 7147
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.126085685390475e-05,
+      "loss": 0.301,
+      "step": 7148
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.125544030997428e-05,
+      "loss": 0.3086,
+      "step": 7149
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.125002249280462e-05,
+      "loss": 0.2931,
+      "step": 7150
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.124460340289726e-05,
+      "loss": 0.3088,
+      "step": 7151
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.12391830407538e-05,
+      "loss": 0.3559,
+      "step": 7152
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.123376140687597e-05,
+      "loss": 0.3169,
+      "step": 7153
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1228338501765605e-05,
+      "loss": 0.3208,
+      "step": 7154
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1222914325924664e-05,
+      "loss": 0.3319,
+      "step": 7155
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.121748887985522e-05,
+      "loss": 0.3023,
+      "step": 7156
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1212062164059474e-05,
+      "loss": 0.3063,
+      "step": 7157
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1206634179039725e-05,
+      "loss": 0.3138,
+      "step": 7158
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.120120492529842e-05,
+      "loss": 0.3156,
+      "step": 7159
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.119577440333809e-05,
+      "loss": 0.3177,
+      "step": 7160
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1190342613661404e-05,
+      "loss": 0.3059,
+      "step": 7161
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.118490955677113e-05,
+      "loss": 0.3285,
+      "step": 7162
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1179475233170185e-05,
+      "loss": 0.2972,
+      "step": 7163
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1174039643361574e-05,
+      "loss": 0.3077,
+      "step": 7164
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1168602787848434e-05,
+      "loss": 0.2987,
+      "step": 7165
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1163164667134006e-05,
+      "loss": 0.317,
+      "step": 7166
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.115772528172166e-05,
+      "loss": 0.307,
+      "step": 7167
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.115228463211489e-05,
+      "loss": 0.312,
+      "step": 7168
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.114684271881728e-05,
+      "loss": 0.3183,
+      "step": 7169
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.114139954233256e-05,
+      "loss": 0.302,
+      "step": 7170
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.113595510316456e-05,
+      "loss": 0.2971,
+      "step": 7171
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1130509401817224e-05,
+      "loss": 0.3251,
+      "step": 7172
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1125062438794623e-05,
+      "loss": 0.3307,
+      "step": 7173
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1119614214600964e-05,
+      "loss": 0.3058,
+      "step": 7174
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.111416472974052e-05,
+      "loss": 0.3023,
+      "step": 7175
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.110871398471773e-05,
+      "loss": 0.295,
+      "step": 7176
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.110326198003711e-05,
+      "loss": 0.3229,
+      "step": 7177
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.109780871620332e-05,
+      "loss": 0.3253,
+      "step": 7178
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.109235419372114e-05,
+      "loss": 0.3022,
+      "step": 7179
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.108689841309545e-05,
+      "loss": 0.3165,
+      "step": 7180
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.108144137483123e-05,
+      "loss": 0.2882,
+      "step": 7181
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1075983079433616e-05,
+      "loss": 0.308,
+      "step": 7182
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.107052352740784e-05,
+      "loss": 0.361,
+      "step": 7183
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.106506271925926e-05,
+      "loss": 0.3037,
+      "step": 7184
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.105960065549332e-05,
+      "loss": 0.3128,
+      "step": 7185
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.105413733661563e-05,
+      "loss": 0.3135,
+      "step": 7186
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.104867276313186e-05,
+      "loss": 0.2956,
+      "step": 7187
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.104320693554785e-05,
+      "loss": 0.3258,
+      "step": 7188
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.103773985436953e-05,
+      "loss": 0.3039,
+      "step": 7189
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1032271520102927e-05,
+      "loss": 0.3112,
+      "step": 7190
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.102680193325423e-05,
+      "loss": 0.3058,
+      "step": 7191
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1021331094329686e-05,
+      "loss": 0.3033,
+      "step": 7192
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.101585900383572e-05,
+      "loss": 0.2982,
+      "step": 7193
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.1010385662278834e-05,
+      "loss": 0.3064,
+      "step": 7194
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.100491107016564e-05,
+      "loss": 0.3124,
+      "step": 7195
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.09994352280029e-05,
+      "loss": 0.3315,
+      "step": 7196
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.0993958136297474e-05,
+      "loss": 0.3023,
+      "step": 7197
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.0988479795556314e-05,
+      "loss": 0.3221,
+      "step": 7198
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.098300020628652e-05,
+      "loss": 0.2988,
+      "step": 7199
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.0977519368995305e-05,
+      "loss": 0.3309,
+      "step": 7200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.097203728418998e-05,
+      "loss": 0.2992,
+      "step": 7201
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.096655395237799e-05,
+      "loss": 0.333,
+      "step": 7202
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.096106937406688e-05,
+      "loss": 0.3321,
+      "step": 7203
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.095558354976431e-05,
+      "loss": 0.323,
+      "step": 7204
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.095009647997807e-05,
+      "loss": 0.2959,
+      "step": 7205
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.094460816521606e-05,
+      "loss": 0.308,
+      "step": 7206
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.09391186059863e-05,
+      "loss": 0.3568,
+      "step": 7207
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.093362780279689e-05,
+      "loss": 0.2913,
+      "step": 7208
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0928135756156105e-05,
+      "loss": 0.298,
+      "step": 7209
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.092264246657228e-05,
+      "loss": 0.2955,
+      "step": 7210
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.09171479345539e-05,
+      "loss": 0.3035,
+      "step": 7211
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.091165216060955e-05,
+      "loss": 0.3108,
+      "step": 7212
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0906155145247924e-05,
+      "loss": 0.3244,
+      "step": 7213
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.090065688897786e-05,
+      "loss": 0.3152,
+      "step": 7214
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0895157392308266e-05,
+      "loss": 0.3078,
+      "step": 7215
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.088965665574822e-05,
+      "loss": 0.3157,
+      "step": 7216
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0884154679806856e-05,
+      "loss": 0.309,
+      "step": 7217
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0878651464993466e-05,
+      "loss": 0.3064,
+      "step": 7218
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.087314701181744e-05,
+      "loss": 0.2925,
+      "step": 7219
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0867641320788284e-05,
+      "loss": 0.295,
+      "step": 7220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0862134392415615e-05,
+      "loss": 0.3058,
+      "step": 7221
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.085662622720917e-05,
+      "loss": 0.3176,
+      "step": 7222
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.085111682567881e-05,
+      "loss": 0.3072,
+      "step": 7223
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.084560618833448e-05,
+      "loss": 0.3116,
+      "step": 7224
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.084009431568628e-05,
+      "loss": 0.2988,
+      "step": 7225
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0834581208244377e-05,
+      "loss": 0.3153,
+      "step": 7226
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.082906686651911e-05,
+      "loss": 0.3209,
+      "step": 7227
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.082355129102087e-05,
+      "loss": 0.307,
+      "step": 7228
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0818034482260215e-05,
+      "loss": 0.3051,
+      "step": 7229
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0812516440747787e-05,
+      "loss": 0.322,
+      "step": 7230
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.080699716699435e-05,
+      "loss": 0.3071,
+      "step": 7231
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.080147666151078e-05,
+      "loss": 0.3126,
+      "step": 7232
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0795954924808076e-05,
+      "loss": 0.3252,
+      "step": 7233
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.079043195739734e-05,
+      "loss": 0.3104,
+      "step": 7234
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.078490775978979e-05,
+      "loss": 0.2898,
+      "step": 7235
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0779382332496775e-05,
+      "loss": 0.328,
+      "step": 7236
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.077385567602972e-05,
+      "loss": 0.3043,
+      "step": 7237
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.07683277909002e-05,
+      "loss": 0.3027,
+      "step": 7238
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.076279867761988e-05,
+      "loss": 0.3206,
+      "step": 7239
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.075726833670057e-05,
+      "loss": 0.2998,
+      "step": 7240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0751736768654146e-05,
+      "loss": 0.3172,
+      "step": 7241
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.074620397399265e-05,
+      "loss": 0.3239,
+      "step": 7242
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0740669953228195e-05,
+      "loss": 0.3039,
+      "step": 7243
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0735134706873025e-05,
+      "loss": 0.3181,
+      "step": 7244
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.07295982354395e-05,
+      "loss": 0.3061,
+      "step": 7245
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.07240605394401e-05,
+      "loss": 0.3266,
+      "step": 7246
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.071852161938739e-05,
+      "loss": 0.3087,
+      "step": 7247
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.071298147579408e-05,
+      "loss": 0.2901,
+      "step": 7248
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.070744010917297e-05,
+      "loss": 0.3078,
+      "step": 7249
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0701897520037e-05,
+      "loss": 0.3011,
+      "step": 7250
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.069635370889919e-05,
+      "loss": 0.3036,
+      "step": 7251
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.069080867627269e-05,
+      "loss": 0.3025,
+      "step": 7252
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.068526242267078e-05,
+      "loss": 0.3168,
+      "step": 7253
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.0679714948606815e-05,
+      "loss": 0.3082,
+      "step": 7254
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.067416625459429e-05,
+      "loss": 0.3163,
+      "step": 7255
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.066861634114681e-05,
+      "loss": 0.3259,
+      "step": 7256
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.066306520877809e-05,
+      "loss": 0.3072,
+      "step": 7257
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.065751285800196e-05,
+      "loss": 0.3173,
+      "step": 7258
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.065195928933235e-05,
+      "loss": 0.3086,
+      "step": 7259
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0646404503283317e-05,
+      "loss": 0.3044,
+      "step": 7260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.064084850036902e-05,
+      "loss": 0.3042,
+      "step": 7261
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.063529128110376e-05,
+      "loss": 0.2922,
+      "step": 7262
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0629732846001894e-05,
+      "loss": 0.3204,
+      "step": 7263
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.062417319557796e-05,
+      "loss": 0.3245,
+      "step": 7264
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0618612330346546e-05,
+      "loss": 0.3151,
+      "step": 7265
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.061305025082239e-05,
+      "loss": 0.3072,
+      "step": 7266
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.060748695752034e-05,
+      "loss": 0.3115,
+      "step": 7267
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0601922450955335e-05,
+      "loss": 0.2923,
+      "step": 7268
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.059635673164245e-05,
+      "loss": 0.2996,
+      "step": 7269
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0590789800096854e-05,
+      "loss": 0.3199,
+      "step": 7270
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.058522165683385e-05,
+      "loss": 0.2877,
+      "step": 7271
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.057965230236883e-05,
+      "loss": 0.3206,
+      "step": 7272
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.057408173721731e-05,
+      "loss": 0.2984,
+      "step": 7273
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.056850996189491e-05,
+      "loss": 0.308,
+      "step": 7274
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.056293697691739e-05,
+      "loss": 0.3027,
+      "step": 7275
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0557362782800564e-05,
+      "loss": 0.3204,
+      "step": 7276
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.055178738006042e-05,
+      "loss": 0.3109,
+      "step": 7277
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.054621076921303e-05,
+      "loss": 0.2949,
+      "step": 7278
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.054063295077457e-05,
+      "loss": 0.3136,
+      "step": 7279
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.053505392526134e-05,
+      "loss": 0.2993,
+      "step": 7280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.052947369318975e-05,
+      "loss": 0.3203,
+      "step": 7281
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0523892255076327e-05,
+      "loss": 0.3043,
+      "step": 7282
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.051830961143769e-05,
+      "loss": 0.2979,
+      "step": 7283
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0512725762790605e-05,
+      "loss": 0.3309,
+      "step": 7284
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.050714070965191e-05,
+      "loss": 0.3181,
+      "step": 7285
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0501554452538566e-05,
+      "loss": 0.3324,
+      "step": 7286
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0495966991967664e-05,
+      "loss": 0.3083,
+      "step": 7287
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.049037832845639e-05,
+      "loss": 0.3112,
+      "step": 7288
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.048478846252205e-05,
+      "loss": 0.3163,
+      "step": 7289
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.047919739468206e-05,
+      "loss": 0.3166,
+      "step": 7290
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.047360512545392e-05,
+      "loss": 0.3042,
+      "step": 7291
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.046801165535529e-05,
+      "loss": 0.3195,
+      "step": 7292
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.046241698490391e-05,
+      "loss": 0.3005,
+      "step": 7293
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0456821114617625e-05,
+      "loss": 0.311,
+      "step": 7294
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.045122404501442e-05,
+      "loss": 0.3085,
+      "step": 7295
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.044562577661237e-05,
+      "loss": 0.3071,
+      "step": 7296
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.044002630992966e-05,
+      "loss": 0.286,
+      "step": 7297
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0434425645484583e-05,
+      "loss": 0.328,
+      "step": 7298
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.042882378379557e-05,
+      "loss": 0.3059,
+      "step": 7299
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.042322072538113e-05,
+      "loss": 0.3033,
+      "step": 7300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0417616470759914e-05,
+      "loss": 0.3062,
+      "step": 7301
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0412011020450644e-05,
+      "loss": 0.3059,
+      "step": 7302
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.040640437497218e-05,
+      "loss": 0.3214,
+      "step": 7303
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.04007965348435e-05,
+      "loss": 0.2997,
+      "step": 7304
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.039518750058368e-05,
+      "loss": 0.2886,
+      "step": 7305
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.038957727271189e-05,
+      "loss": 0.2887,
+      "step": 7306
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.038396585174743e-05,
+      "loss": 0.3121,
+      "step": 7307
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.037835323820972e-05,
+      "loss": 0.3086,
+      "step": 7308
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0372739432618284e-05,
+      "loss": 0.3277,
+      "step": 7309
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.036712443549273e-05,
+      "loss": 0.3037,
+      "step": 7310
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.03615082473528e-05,
+      "loss": 0.2982,
+      "step": 7311
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0355890868718356e-05,
+      "loss": 0.3097,
+      "step": 7312
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.0350272300109346e-05,
+      "loss": 0.326,
+      "step": 7313
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.034465254204585e-05,
+      "loss": 0.3074,
+      "step": 7314
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0339031595048024e-05,
+      "loss": 0.3386,
+      "step": 7315
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.033340945963619e-05,
+      "loss": 0.3183,
+      "step": 7316
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.032778613633073e-05,
+      "loss": 0.3082,
+      "step": 7317
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.032216162565215e-05,
+      "loss": 0.313,
+      "step": 7318
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0316535928121066e-05,
+      "loss": 0.2987,
+      "step": 7319
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.031090904425823e-05,
+      "loss": 0.3144,
+      "step": 7320
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.030528097458446e-05,
+      "loss": 0.3175,
+      "step": 7321
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.029965171962071e-05,
+      "loss": 0.3212,
+      "step": 7322
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.029402127988804e-05,
+      "loss": 0.3216,
+      "step": 7323
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.028838965590761e-05,
+      "loss": 0.3097,
+      "step": 7324
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.028275684820071e-05,
+      "loss": 0.3191,
+      "step": 7325
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.027712285728872e-05,
+      "loss": 0.3114,
+      "step": 7326
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.027148768369314e-05,
+      "loss": 0.2929,
+      "step": 7327
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0265851327935585e-05,
+      "loss": 0.3186,
+      "step": 7328
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.026021379053775e-05,
+      "loss": 0.2911,
+      "step": 7329
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.025457507202148e-05,
+      "loss": 0.3129,
+      "step": 7330
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.024893517290869e-05,
+      "loss": 0.3035,
+      "step": 7331
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0243294093721436e-05,
+      "loss": 0.3168,
+      "step": 7332
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0237651834981866e-05,
+      "loss": 0.2896,
+      "step": 7333
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.023200839721226e-05,
+      "loss": 0.3021,
+      "step": 7334
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.022636378093496e-05,
+      "loss": 0.2993,
+      "step": 7335
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.022071798667246e-05,
+      "loss": 0.3134,
+      "step": 7336
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.021507101494735e-05,
+      "loss": 0.3008,
+      "step": 7337
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.020942286628233e-05,
+      "loss": 0.3242,
+      "step": 7338
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0203773541200205e-05,
+      "loss": 0.3003,
+      "step": 7339
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0198123040223894e-05,
+      "loss": 0.3105,
+      "step": 7340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0192471363876415e-05,
+      "loss": 0.3215,
+      "step": 7341
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.018681851268091e-05,
+      "loss": 0.2962,
+      "step": 7342
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.018116448716062e-05,
+      "loss": 0.3141,
+      "step": 7343
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.01755092878389e-05,
+      "loss": 0.3168,
+      "step": 7344
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.016985291523919e-05,
+      "loss": 0.3215,
+      "step": 7345
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0164195369885075e-05,
+      "loss": 0.3015,
+      "step": 7346
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.015853665230024e-05,
+      "loss": 0.3018,
+      "step": 7347
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.015287676300845e-05,
+      "loss": 0.3137,
+      "step": 7348
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0147215702533614e-05,
+      "loss": 0.3141,
+      "step": 7349
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.014155347139972e-05,
+      "loss": 0.3202,
+      "step": 7350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.01358900701309e-05,
+      "loss": 0.3036,
+      "step": 7351
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.013022549925136e-05,
+      "loss": 0.3145,
+      "step": 7352
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.012455975928542e-05,
+      "loss": 0.3124,
+      "step": 7353
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.011889285075754e-05,
+      "loss": 0.3294,
+      "step": 7354
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.011322477419224e-05,
+      "loss": 0.3074,
+      "step": 7355
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.010755553011418e-05,
+      "loss": 0.3166,
+      "step": 7356
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.010188511904812e-05,
+      "loss": 0.2953,
+      "step": 7357
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0096213541518924e-05,
+      "loss": 0.3213,
+      "step": 7358
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0090540798051574e-05,
+      "loss": 0.3184,
+      "step": 7359
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0084866889171156e-05,
+      "loss": 0.3147,
+      "step": 7360
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.007919181540284e-05,
+      "loss": 0.3174,
+      "step": 7361
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0073515577271964e-05,
+      "loss": 0.2895,
+      "step": 7362
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.006783817530391e-05,
+      "loss": 0.3148,
+      "step": 7363
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.006215961002419e-05,
+      "loss": 0.3038,
+      "step": 7364
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.005647988195844e-05,
+      "loss": 0.3097,
+      "step": 7365
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.005079899163238e-05,
+      "loss": 0.3083,
+      "step": 7366
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.004511693957186e-05,
+      "loss": 0.3126,
+      "step": 7367
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.0039433726302805e-05,
+      "loss": 0.3141,
+      "step": 7368
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.003374935235128e-05,
+      "loss": 0.3076,
+      "step": 7369
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.002806381824346e-05,
+      "loss": 0.3094,
+      "step": 7370
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.002237712450559e-05,
+      "loss": 0.3262,
+      "step": 7371
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.001668927166406e-05,
+      "loss": 0.3108,
+      "step": 7372
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.001100026024534e-05,
+      "loss": 0.307,
+      "step": 7373
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.000531009077603e-05,
+      "loss": 0.3043,
+      "step": 7374
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9999618763782814e-05,
+      "loss": 0.3196,
+      "step": 7375
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9993926279792514e-05,
+      "loss": 0.3285,
+      "step": 7376
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.998823263933203e-05,
+      "loss": 0.3143,
+      "step": 7377
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.998253784292838e-05,
+      "loss": 0.3046,
+      "step": 7378
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9976841891108685e-05,
+      "loss": 0.3191,
+      "step": 7379
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.997114478440019e-05,
+      "loss": 0.3028,
+      "step": 7380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.996544652333023e-05,
+      "loss": 0.3159,
+      "step": 7381
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9959747108426246e-05,
+      "loss": 0.3139,
+      "step": 7382
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.995404654021579e-05,
+      "loss": 0.3042,
+      "step": 7383
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.994834481922652e-05,
+      "loss": 0.2943,
+      "step": 7384
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.99426419459862e-05,
+      "loss": 0.3152,
+      "step": 7385
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.993693792102273e-05,
+      "loss": 0.3162,
+      "step": 7386
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.993123274486405e-05,
+      "loss": 0.3111,
+      "step": 7387
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.992552641803827e-05,
+      "loss": 0.3119,
+      "step": 7388
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.991981894107358e-05,
+      "loss": 0.3012,
+      "step": 7389
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.991411031449827e-05,
+      "loss": 0.3168,
+      "step": 7390
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9908400538840754e-05,
+      "loss": 0.3013,
+      "step": 7391
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.990268961462953e-05,
+      "loss": 0.296,
+      "step": 7392
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9896977542393236e-05,
+      "loss": 0.2942,
+      "step": 7393
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.989126432266058e-05,
+      "loss": 0.3179,
+      "step": 7394
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.98855499559604e-05,
+      "loss": 0.3106,
+      "step": 7395
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.987983444282163e-05,
+      "loss": 0.2902,
+      "step": 7396
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.987411778377332e-05,
+      "loss": 0.3227,
+      "step": 7397
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.98683999793446e-05,
+      "loss": 0.3116,
+      "step": 7398
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9862681030064745e-05,
+      "loss": 0.3099,
+      "step": 7399
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.985696093646312e-05,
+      "loss": 0.3081,
+      "step": 7400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.985123969906916e-05,
+      "loss": 0.319,
+      "step": 7401
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9845517318412454e-05,
+      "loss": 0.3013,
+      "step": 7402
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9839793795022704e-05,
+      "loss": 0.3131,
+      "step": 7403
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.983406912942966e-05,
+      "loss": 0.3154,
+      "step": 7404
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.982834332216323e-05,
+      "loss": 0.3044,
+      "step": 7405
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.982261637375341e-05,
+      "loss": 0.3173,
+      "step": 7406
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.981688828473028e-05,
+      "loss": 0.3324,
+      "step": 7407
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.981115905562408e-05,
+      "loss": 0.31,
+      "step": 7408
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9805428686965106e-05,
+      "loss": 0.3218,
+      "step": 7409
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.979969717928376e-05,
+      "loss": 0.3063,
+      "step": 7410
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.979396453311059e-05,
+      "loss": 0.3015,
+      "step": 7411
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9788230748976214e-05,
+      "loss": 0.3048,
+      "step": 7412
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.978249582741136e-05,
+      "loss": 0.314,
+      "step": 7413
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9776759768946894e-05,
+      "loss": 0.2954,
+      "step": 7414
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9771022574113723e-05,
+      "loss": 0.3149,
+      "step": 7415
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.976528424344292e-05,
+      "loss": 0.3748,
+      "step": 7416
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.975954477746563e-05,
+      "loss": 0.3051,
+      "step": 7417
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.975380417671312e-05,
+      "loss": 0.3343,
+      "step": 7418
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.974806244171675e-05,
+      "loss": 0.3119,
+      "step": 7419
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.974231957300799e-05,
+      "loss": 0.3059,
+      "step": 7420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.973657557111842e-05,
+      "loss": 0.3248,
+      "step": 7421
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.973083043657971e-05,
+      "loss": 0.2979,
+      "step": 7422
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.972508416992365e-05,
+      "loss": 0.3269,
+      "step": 7423
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.971933677168214e-05,
+      "loss": 0.3052,
+      "step": 7424
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9713588242387164e-05,
+      "loss": 0.3201,
+      "step": 7425
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.970783858257081e-05,
+      "loss": 0.3073,
+      "step": 7426
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.97020877927653e-05,
+      "loss": 0.3247,
+      "step": 7427
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.969633587350293e-05,
+      "loss": 0.3117,
+      "step": 7428
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.969058282531611e-05,
+      "loss": 0.3157,
+      "step": 7429
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9684828648737384e-05,
+      "loss": 0.3218,
+      "step": 7430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.967907334429934e-05,
+      "loss": 0.3038,
+      "step": 7431
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.967331691253471e-05,
+      "loss": 0.2935,
+      "step": 7432
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.966755935397634e-05,
+      "loss": 0.3242,
+      "step": 7433
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9661800669157154e-05,
+      "loss": 0.3035,
+      "step": 7434
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.96560408586102e-05,
+      "loss": 0.3033,
+      "step": 7435
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.96502799228686e-05,
+      "loss": 0.3098,
+      "step": 7436
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9644517862465626e-05,
+      "loss": 0.3207,
+      "step": 7437
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.963875467793461e-05,
+      "loss": 0.3242,
+      "step": 7438
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.963299036980901e-05,
+      "loss": 0.3088,
+      "step": 7439
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.96272249386224e-05,
+      "loss": 0.3152,
+      "step": 7440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9621458384908446e-05,
+      "loss": 0.3052,
+      "step": 7441
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.961569070920088e-05,
+      "loss": 0.3019,
+      "step": 7442
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.960992191203361e-05,
+      "loss": 0.3122,
+      "step": 7443
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9604151993940595e-05,
+      "loss": 0.3191,
+      "step": 7444
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9598380955455904e-05,
+      "loss": 0.3271,
+      "step": 7445
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9592608797113744e-05,
+      "loss": 0.3146,
+      "step": 7446
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.958683551944838e-05,
+      "loss": 0.3174,
+      "step": 7447
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.958106112299422e-05,
+      "loss": 0.3195,
+      "step": 7448
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.957528560828573e-05,
+      "loss": 0.3083,
+      "step": 7449
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9569508975857534e-05,
+      "loss": 0.3222,
+      "step": 7450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9563731226244316e-05,
+      "loss": 0.3244,
+      "step": 7451
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.955795235998088e-05,
+      "loss": 0.321,
+      "step": 7452
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9552172377602145e-05,
+      "loss": 0.312,
+      "step": 7453
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9546391279643116e-05,
+      "loss": 0.3122,
+      "step": 7454
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.95406090666389e-05,
+      "loss": 0.3036,
+      "step": 7455
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.953482573912472e-05,
+      "loss": 0.3092,
+      "step": 7456
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9529041297635884e-05,
+      "loss": 0.2982,
+      "step": 7457
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9523255742707835e-05,
+      "loss": 0.3184,
+      "step": 7458
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.951746907487607e-05,
+      "loss": 0.3209,
+      "step": 7459
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9511681294676255e-05,
+      "loss": 0.3081,
+      "step": 7460
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.950589240264409e-05,
+      "loss": 0.3125,
+      "step": 7461
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.950010239931543e-05,
+      "loss": 0.305,
+      "step": 7462
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9494311285226204e-05,
+      "loss": 0.3086,
+      "step": 7463
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9488519060912455e-05,
+      "loss": 0.2981,
+      "step": 7464
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.948272572691032e-05,
+      "loss": 0.3129,
+      "step": 7465
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.947693128375605e-05,
+      "loss": 0.2945,
+      "step": 7466
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9471135731985996e-05,
+      "loss": 0.3282,
+      "step": 7467
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9465339072136615e-05,
+      "loss": 0.3037,
+      "step": 7468
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.945954130474444e-05,
+      "loss": 0.3151,
+      "step": 7469
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9453742430346154e-05,
+      "loss": 0.2962,
+      "step": 7470
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9447942449478495e-05,
+      "loss": 0.2991,
+      "step": 7471
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9442141362678324e-05,
+      "loss": 0.3143,
+      "step": 7472
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.943633917048262e-05,
+      "loss": 0.3085,
+      "step": 7473
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.943053587342844e-05,
+      "loss": 0.3137,
+      "step": 7474
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.942473147205295e-05,
+      "loss": 0.3114,
+      "step": 7475
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.941892596689343e-05,
+      "loss": 0.3068,
+      "step": 7476
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.941311935848724e-05,
+      "loss": 0.3322,
+      "step": 7477
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.940731164737187e-05,
+      "loss": 0.3094,
+      "step": 7478
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.940150283408488e-05,
+      "loss": 0.3068,
+      "step": 7479
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.939569291916396e-05,
+      "loss": 0.3074,
+      "step": 7480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.938988190314688e-05,
+      "loss": 0.3124,
+      "step": 7481
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9384069786571535e-05,
+      "loss": 0.3147,
+      "step": 7482
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.93782565699759e-05,
+      "loss": 0.3182,
+      "step": 7483
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.9372442253898076e-05,
+      "loss": 0.3063,
+      "step": 7484
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.936662683887624e-05,
+      "loss": 0.3107,
+      "step": 7485
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.936081032544867e-05,
+      "loss": 0.3136,
+      "step": 7486
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.935499271415378e-05,
+      "loss": 0.3141,
+      "step": 7487
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9349174005530056e-05,
+      "loss": 0.319,
+      "step": 7488
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.934335420011609e-05,
+      "loss": 0.3281,
+      "step": 7489
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9337533298450566e-05,
+      "loss": 0.3193,
+      "step": 7490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.933171130107231e-05,
+      "loss": 0.3176,
+      "step": 7491
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.93258882085202e-05,
+      "loss": 0.331,
+      "step": 7492
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.932006402133324e-05,
+      "loss": 0.3159,
+      "step": 7493
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.931423874005054e-05,
+      "loss": 0.3024,
+      "step": 7494
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.93084123652113e-05,
+      "loss": 0.3058,
+      "step": 7495
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9302584897354804e-05,
+      "loss": 0.3042,
+      "step": 7496
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9296756337020484e-05,
+      "loss": 0.3151,
+      "step": 7497
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.929092668474784e-05,
+      "loss": 0.3184,
+      "step": 7498
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9285095941076475e-05,
+      "loss": 0.297,
+      "step": 7499
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.92792641065461e-05,
+      "loss": 0.3002,
+      "step": 7500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.927343118169653e-05,
+      "loss": 0.3144,
+      "step": 7501
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.926759716706767e-05,
+      "loss": 0.3068,
+      "step": 7502
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.926176206319953e-05,
+      "loss": 0.3417,
+      "step": 7503
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.925592587063222e-05,
+      "loss": 0.3035,
+      "step": 7504
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.925008858990596e-05,
+      "loss": 0.3297,
+      "step": 7505
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.924425022156106e-05,
+      "loss": 0.3144,
+      "step": 7506
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9238410766137934e-05,
+      "loss": 0.3188,
+      "step": 7507
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.923257022417711e-05,
+      "loss": 0.3048,
+      "step": 7508
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.922672859621918e-05,
+      "loss": 0.299,
+      "step": 7509
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.922088588280488e-05,
+      "loss": 0.3173,
+      "step": 7510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.921504208447502e-05,
+      "loss": 0.3071,
+      "step": 7511
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.92091972017705e-05,
+      "loss": 0.3089,
+      "step": 7512
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.920335123523238e-05,
+      "loss": 0.3073,
+      "step": 7513
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9197504185401725e-05,
+      "loss": 0.3327,
+      "step": 7514
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.91916560528198e-05,
+      "loss": 0.3121,
+      "step": 7515
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.918580683802789e-05,
+      "loss": 0.3032,
+      "step": 7516
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.917995654156745e-05,
+      "loss": 0.2975,
+      "step": 7517
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9174105163979956e-05,
+      "loss": 0.3048,
+      "step": 7518
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.916825270580705e-05,
+      "loss": 0.3169,
+      "step": 7519
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.916239916759044e-05,
+      "loss": 0.3235,
+      "step": 7520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.915654454987196e-05,
+      "loss": 0.3166,
+      "step": 7521
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.915068885319352e-05,
+      "loss": 0.3075,
+      "step": 7522
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.914483207809713e-05,
+      "loss": 0.3194,
+      "step": 7523
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.913897422512492e-05,
+      "loss": 0.3215,
+      "step": 7524
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.913311529481911e-05,
+      "loss": 0.2861,
+      "step": 7525
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9127255287722006e-05,
+      "loss": 0.3282,
+      "step": 7526
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.912139420437604e-05,
+      "loss": 0.3102,
+      "step": 7527
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9115532045323706e-05,
+      "loss": 0.3191,
+      "step": 7528
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9109668811107646e-05,
+      "loss": 0.3083,
+      "step": 7529
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.910380450227056e-05,
+      "loss": 0.3062,
+      "step": 7530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9097939119355274e-05,
+      "loss": 0.3038,
+      "step": 7531
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.90920726629047e-05,
+      "loss": 0.3063,
+      "step": 7532
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.908620513346185e-05,
+      "loss": 0.3305,
+      "step": 7533
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.908033653156984e-05,
+      "loss": 0.3048,
+      "step": 7534
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.907446685777188e-05,
+      "loss": 0.3046,
+      "step": 7535
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.906859611261128e-05,
+      "loss": 0.3041,
+      "step": 7536
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.906272429663146e-05,
+      "loss": 0.3222,
+      "step": 7537
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9056851410375924e-05,
+      "loss": 0.3087,
+      "step": 7538
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9050977454388276e-05,
+      "loss": 0.3252,
+      "step": 7539
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.904510242921224e-05,
+      "loss": 0.295,
+      "step": 7540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.903922633539161e-05,
+      "loss": 0.3242,
+      "step": 7541
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9033349173470306e-05,
+      "loss": 0.3269,
+      "step": 7542
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.902747094399231e-05,
+      "loss": 0.3099,
+      "step": 7543
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9021591647501754e-05,
+      "loss": 0.298,
+      "step": 7544
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.901571128454282e-05,
+      "loss": 0.3039,
+      "step": 7545
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.900982985565983e-05,
+      "loss": 0.3005,
+      "step": 7546
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.900394736139716e-05,
+      "loss": 0.298,
+      "step": 7547
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.899806380229933e-05,
+      "loss": 0.3141,
+      "step": 7548
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.899217917891092e-05,
+      "loss": 0.3239,
+      "step": 7549
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.898629349177664e-05,
+      "loss": 0.3116,
+      "step": 7550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.898040674144128e-05,
+      "loss": 0.3198,
+      "step": 7551
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.897451892844973e-05,
+      "loss": 0.3011,
+      "step": 7552
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8968630053346976e-05,
+      "loss": 0.3311,
+      "step": 7553
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.896274011667812e-05,
+      "loss": 0.3068,
+      "step": 7554
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.895684911898834e-05,
+      "loss": 0.3096,
+      "step": 7555
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.895095706082293e-05,
+      "loss": 0.3136,
+      "step": 7556
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8945063942727273e-05,
+      "loss": 0.3105,
+      "step": 7557
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.893916976524685e-05,
+      "loss": 0.3054,
+      "step": 7558
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.893327452892724e-05,
+      "loss": 0.3336,
+      "step": 7559
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.892737823431411e-05,
+      "loss": 0.318,
+      "step": 7560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.892148088195326e-05,
+      "loss": 0.3143,
+      "step": 7561
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.891558247239055e-05,
+      "loss": 0.3196,
+      "step": 7562
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.890968300617195e-05,
+      "loss": 0.3064,
+      "step": 7563
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.890378248384354e-05,
+      "loss": 0.3051,
+      "step": 7564
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8897880905951466e-05,
+      "loss": 0.3189,
+      "step": 7565
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.889197827304203e-05,
+      "loss": 0.319,
+      "step": 7566
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.888607458566156e-05,
+      "loss": 0.3319,
+      "step": 7567
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8880169844356536e-05,
+      "loss": 0.3015,
+      "step": 7568
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.887426404967351e-05,
+      "loss": 0.3054,
+      "step": 7569
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8868357202159135e-05,
+      "loss": 0.3056,
+      "step": 7570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8862449302360164e-05,
+      "loss": 0.3225,
+      "step": 7571
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8856540350823455e-05,
+      "loss": 0.3002,
+      "step": 7572
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8850630348095954e-05,
+      "loss": 0.3143,
+      "step": 7573
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.884471929472471e-05,
+      "loss": 0.2994,
+      "step": 7574
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8838807191256845e-05,
+      "loss": 0.3018,
+      "step": 7575
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.883289403823962e-05,
+      "loss": 0.3285,
+      "step": 7576
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8826979836220364e-05,
+      "loss": 0.2974,
+      "step": 7577
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8821064585746514e-05,
+      "loss": 0.3091,
+      "step": 7578
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8815148287365595e-05,
+      "loss": 0.2951,
+      "step": 7579
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.880923094162524e-05,
+      "loss": 0.3092,
+      "step": 7580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.880331254907318e-05,
+      "loss": 0.3098,
+      "step": 7581
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.879739311025722e-05,
+      "loss": 0.3011,
+      "step": 7582
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.879147262572529e-05,
+      "loss": 0.3184,
+      "step": 7583
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.87855510960254e-05,
+      "loss": 0.3024,
+      "step": 7584
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8779628521705665e-05,
+      "loss": 0.316,
+      "step": 7585
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.877370490331429e-05,
+      "loss": 0.3162,
+      "step": 7586
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.876778024139959e-05,
+      "loss": 0.3122,
+      "step": 7587
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.876185453650996e-05,
+      "loss": 0.3166,
+      "step": 7588
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.87559277891939e-05,
+      "loss": 0.3016,
+      "step": 7589
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.875e-05,
+      "loss": 0.3073,
+      "step": 7590
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.874407116947697e-05,
+      "loss": 0.2925,
+      "step": 7591
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.873814129817357e-05,
+      "loss": 0.3276,
+      "step": 7592
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.873221038663871e-05,
+      "loss": 0.2965,
+      "step": 7593
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.872627843542135e-05,
+      "loss": 0.316,
+      "step": 7594
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.872034544507058e-05,
+      "loss": 0.2946,
+      "step": 7595
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.871441141613558e-05,
+      "loss": 0.2992,
+      "step": 7596
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8708476349165587e-05,
+      "loss": 0.2995,
+      "step": 7597
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.870254024471e-05,
+      "loss": 0.2976,
+      "step": 7598
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.869660310331826e-05,
+      "loss": 0.3267,
+      "step": 7599
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8690664925539946e-05,
+      "loss": 0.3056,
+      "step": 7600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.868472571192468e-05,
+      "loss": 0.3066,
+      "step": 7601
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.867878546302224e-05,
+      "loss": 0.294,
+      "step": 7602
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.867284417938245e-05,
+      "loss": 0.311,
+      "step": 7603
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.866690186155526e-05,
+      "loss": 0.3267,
+      "step": 7604
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.866095851009071e-05,
+      "loss": 0.2939,
+      "step": 7605
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.865501412553893e-05,
+      "loss": 0.3086,
+      "step": 7606
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.864906870845014e-05,
+      "loss": 0.3103,
+      "step": 7607
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.864312225937467e-05,
+      "loss": 0.3172,
+      "step": 7608
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8637174778862935e-05,
+      "loss": 0.3109,
+      "step": 7609
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8631226267465454e-05,
+      "loss": 0.2948,
+      "step": 7610
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.862527672573283e-05,
+      "loss": 0.3142,
+      "step": 7611
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.861932615421578e-05,
+      "loss": 0.2958,
+      "step": 7612
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8613374553465084e-05,
+      "loss": 0.3088,
+      "step": 7613
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.860742192403166e-05,
+      "loss": 0.3111,
+      "step": 7614
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.860146826646649e-05,
+      "loss": 0.3107,
+      "step": 7615
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8595513581320645e-05,
+      "loss": 0.3016,
+      "step": 7616
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8589557869145335e-05,
+      "loss": 0.3079,
+      "step": 7617
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.858360113049181e-05,
+      "loss": 0.3013,
+      "step": 7618
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8577643365911456e-05,
+      "loss": 0.2998,
+      "step": 7619
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.857168457595574e-05,
+      "loss": 0.3029,
+      "step": 7620
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.856572476117621e-05,
+      "loss": 0.314,
+      "step": 7621
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8559763922124534e-05,
+      "loss": 0.313,
+      "step": 7622
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.855380205935246e-05,
+      "loss": 0.3135,
+      "step": 7623
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.854783917341183e-05,
+      "loss": 0.3191,
+      "step": 7624
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.854187526485459e-05,
+      "loss": 0.3107,
+      "step": 7625
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.853591033423277e-05,
+      "loss": 0.3049,
+      "step": 7626
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.852994438209849e-05,
+      "loss": 0.314,
+      "step": 7627
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8523977409004004e-05,
+      "loss": 0.3239,
+      "step": 7628
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.85180094155016e-05,
+      "loss": 0.3393,
+      "step": 7629
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.85120404021437e-05,
+      "loss": 0.297,
+      "step": 7630
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.850607036948281e-05,
+      "loss": 0.3037,
+      "step": 7631
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.850009931807155e-05,
+      "loss": 0.3021,
+      "step": 7632
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.849412724846259e-05,
+      "loss": 0.3165,
+      "step": 7633
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8488154161208736e-05,
+      "loss": 0.2997,
+      "step": 7634
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.848218005686287e-05,
+      "loss": 0.3074,
+      "step": 7635
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.847620493597795e-05,
+      "loss": 0.322,
+      "step": 7636
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.847022879910709e-05,
+      "loss": 0.3047,
+      "step": 7637
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.846425164680342e-05,
+      "loss": 0.3022,
+      "step": 7638
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.845827347962021e-05,
+      "loss": 0.3156,
+      "step": 7639
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.845229429811084e-05,
+      "loss": 0.3119,
+      "step": 7640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.844631410282872e-05,
+      "loss": 0.3297,
+      "step": 7641
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.844033289432741e-05,
+      "loss": 0.3116,
+      "step": 7642
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.843435067316056e-05,
+      "loss": 0.3132,
+      "step": 7643
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.842836743988187e-05,
+      "loss": 0.3002,
+      "step": 7644
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.842238319504519e-05,
+      "loss": 0.3027,
+      "step": 7645
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.841639793920442e-05,
+      "loss": 0.2985,
+      "step": 7646
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8410411672913586e-05,
+      "loss": 0.3189,
+      "step": 7647
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.840442439672677e-05,
+      "loss": 0.3258,
+      "step": 7648
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.83984361111982e-05,
+      "loss": 0.3058,
+      "step": 7649
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.839244681688214e-05,
+      "loss": 0.3107,
+      "step": 7650
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.838645651433299e-05,
+      "loss": 0.3289,
+      "step": 7651
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.838046520410523e-05,
+      "loss": 0.3068,
+      "step": 7652
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8374472886753405e-05,
+      "loss": 0.3194,
+      "step": 7653
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.836847956283221e-05,
+      "loss": 0.3246,
+      "step": 7654
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8362485232896394e-05,
+      "loss": 0.3141,
+      "step": 7655
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8356489897500804e-05,
+      "loss": 0.3017,
+      "step": 7656
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.835049355720039e-05,
+      "loss": 0.2932,
+      "step": 7657
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8344496212550166e-05,
+      "loss": 0.2955,
+      "step": 7658
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.833849786410531e-05,
+      "loss": 0.3084,
+      "step": 7659
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8332498512420985e-05,
+      "loss": 0.3152,
+      "step": 7660
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.832649815805254e-05,
+      "loss": 0.305,
+      "step": 7661
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.832049680155539e-05,
+      "loss": 0.3114,
+      "step": 7662
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.831449444348502e-05,
+      "loss": 0.3081,
+      "step": 7663
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.830849108439703e-05,
+      "loss": 0.3195,
+      "step": 7664
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8302486724847106e-05,
+      "loss": 0.2931,
+      "step": 7665
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.829648136539103e-05,
+      "loss": 0.2993,
+      "step": 7666
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8290475006584665e-05,
+      "loss": 0.3369,
+      "step": 7667
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.828446764898398e-05,
+      "loss": 0.3064,
+      "step": 7668
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8278459293145036e-05,
+      "loss": 0.3134,
+      "step": 7669
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.827244993962397e-05,
+      "loss": 0.3078,
+      "step": 7670
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8266439588977045e-05,
+      "loss": 0.3199,
+      "step": 7671
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.826042824176057e-05,
+      "loss": 0.2921,
+      "step": 7672
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8254415898530984e-05,
+      "loss": 0.3203,
+      "step": 7673
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8248402559844804e-05,
+      "loss": 0.3064,
+      "step": 7674
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.824238822625863e-05,
+      "loss": 0.3161,
+      "step": 7675
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.823637289832919e-05,
+      "loss": 0.3093,
+      "step": 7676
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.823035657661325e-05,
+      "loss": 0.3166,
+      "step": 7677
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8224339261667714e-05,
+      "loss": 0.3235,
+      "step": 7678
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.821832095404955e-05,
+      "loss": 0.3154,
+      "step": 7679
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.821230165431583e-05,
+      "loss": 0.3036,
+      "step": 7680
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.820628136302373e-05,
+      "loss": 0.3063,
+      "step": 7681
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8200260080730485e-05,
+      "loss": 0.3109,
+      "step": 7682
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8194237807993445e-05,
+      "loss": 0.294,
+      "step": 7683
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.818821454537005e-05,
+      "loss": 0.3133,
+      "step": 7684
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8182190293417825e-05,
+      "loss": 0.292,
+      "step": 7685
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.817616505269441e-05,
+      "loss": 0.329,
+      "step": 7686
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.817013882375748e-05,
+      "loss": 0.3216,
+      "step": 7687
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.816411160716487e-05,
+      "loss": 0.3146,
+      "step": 7688
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8158083403474464e-05,
+      "loss": 0.3183,
+      "step": 7689
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.815205421324424e-05,
+      "loss": 0.2934,
+      "step": 7690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8146024037032303e-05,
+      "loss": 0.2961,
+      "step": 7691
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.813999287539678e-05,
+      "loss": 0.297,
+      "step": 7692
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.813396072889597e-05,
+      "loss": 0.315,
+      "step": 7693
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.81279275980882e-05,
+      "loss": 0.3104,
+      "step": 7694
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8121893483531915e-05,
+      "loss": 0.328,
+      "step": 7695
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8115858385785666e-05,
+      "loss": 0.2944,
+      "step": 7696
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.810982230540805e-05,
+      "loss": 0.2869,
+      "step": 7697
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8103785242957806e-05,
+      "loss": 0.2994,
+      "step": 7698
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.809774719899373e-05,
+      "loss": 0.3178,
+      "step": 7699
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.809170817407471e-05,
+      "loss": 0.3119,
+      "step": 7700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8085668168759754e-05,
+      "loss": 0.3214,
+      "step": 7701
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.807962718360792e-05,
+      "loss": 0.3326,
+      "step": 7702
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.80735852191784e-05,
+      "loss": 0.3068,
+      "step": 7703
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8067542276030435e-05,
+      "loss": 0.3035,
+      "step": 7704
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.806149835472338e-05,
+      "loss": 0.286,
+      "step": 7705
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.805545345581668e-05,
+      "loss": 0.3153,
+      "step": 7706
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.804940757986986e-05,
+      "loss": 0.3132,
+      "step": 7707
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.804336072744255e-05,
+      "loss": 0.3067,
+      "step": 7708
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8037312899094454e-05,
+      "loss": 0.3379,
+      "step": 7709
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8031264095385374e-05,
+      "loss": 0.3548,
+      "step": 7710
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.802521431687522e-05,
+      "loss": 0.3051,
+      "step": 7711
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.801916356412395e-05,
+      "loss": 0.3148,
+      "step": 7712
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8013111837691665e-05,
+      "loss": 0.3055,
+      "step": 7713
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.80070591381385e-05,
+      "loss": 0.3173,
+      "step": 7714
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.800100546602474e-05,
+      "loss": 0.3056,
+      "step": 7715
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7994950821910697e-05,
+      "loss": 0.3086,
+      "step": 7716
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7988895206356814e-05,
+      "loss": 0.2895,
+      "step": 7717
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.798283861992363e-05,
+      "loss": 0.3362,
+      "step": 7718
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.797678106317174e-05,
+      "loss": 0.2955,
+      "step": 7719
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.797072253666186e-05,
+      "loss": 0.3115,
+      "step": 7720
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.796466304095477e-05,
+      "loss": 0.3045,
+      "step": 7721
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.795860257661137e-05,
+      "loss": 0.3044,
+      "step": 7722
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.795254114419262e-05,
+      "loss": 0.3131,
+      "step": 7723
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.794647874425958e-05,
+      "loss": 0.3254,
+      "step": 7724
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.79404153773734e-05,
+      "loss": 0.3137,
+      "step": 7725
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7934351044095336e-05,
+      "loss": 0.2981,
+      "step": 7726
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.79282857449867e-05,
+      "loss": 0.3153,
+      "step": 7727
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7922219480608926e-05,
+      "loss": 0.3249,
+      "step": 7728
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.791615225152352e-05,
+      "loss": 0.3221,
+      "step": 7729
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.791008405829209e-05,
+      "loss": 0.2974,
+      "step": 7730
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.790401490147629e-05,
+      "loss": 0.313,
+      "step": 7731
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.789794478163793e-05,
+      "loss": 0.3101,
+      "step": 7732
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7891873699338856e-05,
+      "loss": 0.319,
+      "step": 7733
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.788580165514104e-05,
+      "loss": 0.2901,
+      "step": 7734
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7879728649606514e-05,
+      "loss": 0.3007,
+      "step": 7735
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.787365468329742e-05,
+      "loss": 0.3119,
+      "step": 7736
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7867579756775975e-05,
+      "loss": 0.2921,
+      "step": 7737
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.786150387060449e-05,
+      "loss": 0.3347,
+      "step": 7738
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.785542702534536e-05,
+      "loss": 0.3122,
+      "step": 7739
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.784934922156108e-05,
+      "loss": 0.3194,
+      "step": 7740
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.784327045981422e-05,
+      "loss": 0.301,
+      "step": 7741
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7837190740667466e-05,
+      "loss": 0.3137,
+      "step": 7742
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7831110064683554e-05,
+      "loss": 0.2992,
+      "step": 7743
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.782502843242532e-05,
+      "loss": 0.3111,
+      "step": 7744
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7818945844455725e-05,
+      "loss": 0.3002,
+      "step": 7745
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.781286230133775e-05,
+      "loss": 0.2998,
+      "step": 7746
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.780677780363454e-05,
+      "loss": 0.2994,
+      "step": 7747
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.780069235190928e-05,
+      "loss": 0.3156,
+      "step": 7748
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.779460594672524e-05,
+      "loss": 0.3168,
+      "step": 7749
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.778851858864581e-05,
+      "loss": 0.3,
+      "step": 7750
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.778243027823445e-05,
+      "loss": 0.3102,
+      "step": 7751
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.77763410160547e-05,
+      "loss": 0.3047,
+      "step": 7752
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.777025080267021e-05,
+      "loss": 0.3086,
+      "step": 7753
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.77641596386447e-05,
+      "loss": 0.3027,
+      "step": 7754
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7758067524541975e-05,
+      "loss": 0.3206,
+      "step": 7755
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.775197446092595e-05,
+      "loss": 0.3091,
+      "step": 7756
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.774588044836062e-05,
+      "loss": 0.3027,
+      "step": 7757
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.773978548741005e-05,
+      "loss": 0.3178,
+      "step": 7758
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.77336895786384e-05,
+      "loss": 0.3305,
+      "step": 7759
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7727592722609946e-05,
+      "loss": 0.2935,
+      "step": 7760
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7721494919888994e-05,
+      "loss": 0.3097,
+      "step": 7761
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.771539617104001e-05,
+      "loss": 0.3195,
+      "step": 7762
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.770929647662749e-05,
+      "loss": 0.298,
+      "step": 7763
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.770319583721604e-05,
+      "loss": 0.3113,
+      "step": 7764
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7697094253370356e-05,
+      "loss": 0.2959,
+      "step": 7765
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.7690991725655204e-05,
+      "loss": 0.3067,
+      "step": 7766
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.768488825463547e-05,
+      "loss": 0.3049,
+      "step": 7767
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.767878384087608e-05,
+      "loss": 0.3119,
+      "step": 7768
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.767267848494209e-05,
+      "loss": 0.3007,
+      "step": 7769
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.766657218739863e-05,
+      "loss": 0.3085,
+      "step": 7770
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.766046494881092e-05,
+      "loss": 0.3171,
+      "step": 7771
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.765435676974425e-05,
+      "loss": 0.3065,
+      "step": 7772
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.764824765076401e-05,
+      "loss": 0.305,
+      "step": 7773
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.764213759243568e-05,
+      "loss": 0.3148,
+      "step": 7774
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.763602659532481e-05,
+      "loss": 0.2972,
+      "step": 7775
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.762991465999707e-05,
+      "loss": 0.315,
+      "step": 7776
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.762380178701819e-05,
+      "loss": 0.329,
+      "step": 7777
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7617687976953986e-05,
+      "loss": 0.3144,
+      "step": 7778
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.761157323037037e-05,
+      "loss": 0.2902,
+      "step": 7779
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7605457547833345e-05,
+      "loss": 0.3198,
+      "step": 7780
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7599340929908996e-05,
+      "loss": 0.3237,
+      "step": 7781
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7593223377163484e-05,
+      "loss": 0.334,
+      "step": 7782
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.758710489016306e-05,
+      "loss": 0.2856,
+      "step": 7783
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7580985469474086e-05,
+      "loss": 0.311,
+      "step": 7784
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.757486511566298e-05,
+      "loss": 0.3158,
+      "step": 7785
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.756874382929626e-05,
+      "loss": 0.3201,
+      "step": 7786
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7562621610940534e-05,
+      "loss": 0.3105,
+      "step": 7787
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.755649846116247e-05,
+      "loss": 0.2803,
+      "step": 7788
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7550374380528875e-05,
+      "loss": 0.3122,
+      "step": 7789
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.754424936960658e-05,
+      "loss": 0.3214,
+      "step": 7790
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7538123428962545e-05,
+      "loss": 0.3199,
+      "step": 7791
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.75319965591638e-05,
+      "loss": 0.3183,
+      "step": 7792
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7525868760777466e-05,
+      "loss": 0.3081,
+      "step": 7793
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.751974003437075e-05,
+      "loss": 0.3293,
+      "step": 7794
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.751361038051094e-05,
+      "loss": 0.2933,
+      "step": 7795
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.750747979976541e-05,
+      "loss": 0.3221,
+      "step": 7796
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7501348292701624e-05,
+      "loss": 0.32,
+      "step": 7797
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.749521585988713e-05,
+      "loss": 0.3105,
+      "step": 7798
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.748908250188956e-05,
+      "loss": 0.3026,
+      "step": 7799
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.748294821927665e-05,
+      "loss": 0.2946,
+      "step": 7800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7476813012616185e-05,
+      "loss": 0.3154,
+      "step": 7801
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.747067688247606e-05,
+      "loss": 0.3102,
+      "step": 7802
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.746453982942425e-05,
+      "loss": 0.2941,
+      "step": 7803
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.745840185402883e-05,
+      "loss": 0.3211,
+      "step": 7804
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.745226295685793e-05,
+      "loss": 0.3138,
+      "step": 7805
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7446123138479794e-05,
+      "loss": 0.3087,
+      "step": 7806
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7439982399462735e-05,
+      "loss": 0.3122,
+      "step": 7807
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7433840740375145e-05,
+      "loss": 0.3156,
+      "step": 7808
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.742769816178553e-05,
+      "loss": 0.314,
+      "step": 7809
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7421554664262457e-05,
+      "loss": 0.2785,
+      "step": 7810
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7415410248374585e-05,
+      "loss": 0.3278,
+      "step": 7811
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.740926491469064e-05,
+      "loss": 0.3093,
+      "step": 7812
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7403118663779473e-05,
+      "loss": 0.2915,
+      "step": 7813
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7396971496209995e-05,
+      "loss": 0.3209,
+      "step": 7814
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.739082341255118e-05,
+      "loss": 0.2941,
+      "step": 7815
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7384674413372146e-05,
+      "loss": 0.2995,
+      "step": 7816
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7378524499242026e-05,
+      "loss": 0.3111,
+      "step": 7817
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7372373670730094e-05,
+      "loss": 0.3106,
+      "step": 7818
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.736622192840569e-05,
+      "loss": 0.3285,
+      "step": 7819
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.736006927283821e-05,
+      "loss": 0.3136,
+      "step": 7820
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7353915704597184e-05,
+      "loss": 0.3002,
+      "step": 7821
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.734776122425219e-05,
+      "loss": 0.295,
+      "step": 7822
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.734160583237291e-05,
+      "loss": 0.2933,
+      "step": 7823
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.73354495295291e-05,
+      "loss": 0.3267,
+      "step": 7824
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.73292923162906e-05,
+      "loss": 0.3223,
+      "step": 7825
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.732313419322734e-05,
+      "loss": 0.3139,
+      "step": 7826
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.731697516090933e-05,
+      "loss": 0.3,
+      "step": 7827
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.731081521990667e-05,
+      "loss": 0.3178,
+      "step": 7828
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7304654370789536e-05,
+      "loss": 0.3076,
+      "step": 7829
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7298492614128194e-05,
+      "loss": 0.2826,
+      "step": 7830
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7292329950493e-05,
+      "loss": 0.2953,
+      "step": 7831
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.728616638045437e-05,
+      "loss": 0.3183,
+      "step": 7832
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.728000190458284e-05,
+      "loss": 0.3189,
+      "step": 7833
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.727383652344899e-05,
+      "loss": 0.3064,
+      "step": 7834
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7267670237623506e-05,
+      "loss": 0.3141,
+      "step": 7835
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7261503047677164e-05,
+      "loss": 0.2984,
+      "step": 7836
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7255334954180815e-05,
+      "loss": 0.3135,
+      "step": 7837
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.72491659577054e-05,
+      "loss": 0.3225,
+      "step": 7838
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.724299605882191e-05,
+      "loss": 0.3215,
+      "step": 7839
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.723682525810148e-05,
+      "loss": 0.3037,
+      "step": 7840
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7230653556115274e-05,
+      "loss": 0.3061,
+      "step": 7841
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.722448095343457e-05,
+      "loss": 0.2903,
+      "step": 7842
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7218307450630725e-05,
+      "loss": 0.3136,
+      "step": 7843
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.721213304827515e-05,
+      "loss": 0.3226,
+      "step": 7844
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.720595774693939e-05,
+      "loss": 0.2934,
+      "step": 7845
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.719978154719504e-05,
+      "loss": 0.3026,
+      "step": 7846
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7193604449613776e-05,
+      "loss": 0.3288,
+      "step": 7847
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7187426454767375e-05,
+      "loss": 0.309,
+      "step": 7848
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.718124756322769e-05,
+      "loss": 0.3106,
+      "step": 7849
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.717506777556665e-05,
+      "loss": 0.2984,
+      "step": 7850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7168887092356265e-05,
+      "loss": 0.3066,
+      "step": 7851
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7162705514168646e-05,
+      "loss": 0.3199,
+      "step": 7852
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.715652304157598e-05,
+      "loss": 0.3016,
+      "step": 7853
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.715033967515052e-05,
+      "loss": 0.2904,
+      "step": 7854
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.714415541546462e-05,
+      "loss": 0.3187,
+      "step": 7855
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.713797026309071e-05,
+      "loss": 0.3042,
+      "step": 7856
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.713178421860131e-05,
+      "loss": 0.3381,
+      "step": 7857
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.712559728256901e-05,
+      "loss": 0.3084,
+      "step": 7858
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7119409455566486e-05,
+      "loss": 0.3149,
+      "step": 7859
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.711322073816652e-05,
+      "loss": 0.3196,
+      "step": 7860
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7107031130941916e-05,
+      "loss": 0.2953,
+      "step": 7861
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7100840634465636e-05,
+      "loss": 0.3126,
+      "step": 7862
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.709464924931067e-05,
+      "loss": 0.3018,
+      "step": 7863
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7088456976050114e-05,
+      "loss": 0.328,
+      "step": 7864
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.708226381525715e-05,
+      "loss": 0.3103,
+      "step": 7865
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7076069767505015e-05,
+      "loss": 0.3275,
+      "step": 7866
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.706987483336706e-05,
+      "loss": 0.314,
+      "step": 7867
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7063679013416695e-05,
+      "loss": 0.3191,
+      "step": 7868
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.705748230822743e-05,
+      "loss": 0.2988,
+      "step": 7869
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.705128471837285e-05,
+      "loss": 0.3124,
+      "step": 7870
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7045086244426594e-05,
+      "loss": 0.3034,
+      "step": 7871
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.703888688696245e-05,
+      "loss": 0.307,
+      "step": 7872
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.703268664655422e-05,
+      "loss": 0.3109,
+      "step": 7873
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.702648552377581e-05,
+      "loss": 0.2976,
+      "step": 7874
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.702028351920123e-05,
+      "loss": 0.3005,
+      "step": 7875
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.701408063340454e-05,
+      "loss": 0.3152,
+      "step": 7876
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7007876866959895e-05,
+      "loss": 0.3126,
+      "step": 7877
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.700167222044155e-05,
+      "loss": 0.31,
+      "step": 7878
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.699546669442379e-05,
+      "loss": 0.3206,
+      "step": 7879
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.698926028948105e-05,
+      "loss": 0.3134,
+      "step": 7880
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.698305300618779e-05,
+      "loss": 0.313,
+      "step": 7881
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.697684484511859e-05,
+      "loss": 0.321,
+      "step": 7882
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.697063580684806e-05,
+      "loss": 0.3036,
+      "step": 7883
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.696442589195096e-05,
+      "loss": 0.2839,
+      "step": 7884
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.695821510100208e-05,
+      "loss": 0.3016,
+      "step": 7885
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.695200343457631e-05,
+      "loss": 0.3184,
+      "step": 7886
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.694579089324861e-05,
+      "loss": 0.3213,
+      "step": 7887
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.693957747759404e-05,
+      "loss": 0.324,
+      "step": 7888
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.693336318818772e-05,
+      "loss": 0.322,
+      "step": 7889
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6927148025604876e-05,
+      "loss": 0.3006,
+      "step": 7890
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.692093199042078e-05,
+      "loss": 0.3202,
+      "step": 7891
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.691471508321082e-05,
+      "loss": 0.325,
+      "step": 7892
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.690849730455043e-05,
+      "loss": 0.3161,
+      "step": 7893
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6902278655015166e-05,
+      "loss": 0.3257,
+      "step": 7894
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.689605913518064e-05,
+      "loss": 0.2991,
+      "step": 7895
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6889838745622535e-05,
+      "loss": 0.3075,
+      "step": 7896
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6883617486916626e-05,
+      "loss": 0.3024,
+      "step": 7897
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.687739535963878e-05,
+      "loss": 0.3132,
+      "step": 7898
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.687117236436493e-05,
+      "loss": 0.3155,
+      "step": 7899
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.686494850167107e-05,
+      "loss": 0.3006,
+      "step": 7900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.685872377213334e-05,
+      "loss": 0.3028,
+      "step": 7901
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.685249817632788e-05,
+      "loss": 0.3059,
+      "step": 7902
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6846271714830966e-05,
+      "loss": 0.3082,
+      "step": 7903
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.684004438821893e-05,
+      "loss": 0.327,
+      "step": 7904
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.683381619706818e-05,
+      "loss": 0.3133,
+      "step": 7905
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.682758714195524e-05,
+      "loss": 0.292,
+      "step": 7906
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.682135722345666e-05,
+      "loss": 0.3066,
+      "step": 7907
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.681512644214911e-05,
+      "loss": 0.3206,
+      "step": 7908
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.680889479860933e-05,
+      "loss": 0.2973,
+      "step": 7909
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.680266229341412e-05,
+      "loss": 0.3133,
+      "step": 7910
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6796428927140406e-05,
+      "loss": 0.3076,
+      "step": 7911
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.679019470036514e-05,
+      "loss": 0.3027,
+      "step": 7912
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.678395961366538e-05,
+      "loss": 0.3208,
+      "step": 7913
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.677772366761827e-05,
+      "loss": 0.32,
+      "step": 7914
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.677148686280103e-05,
+      "loss": 0.3202,
+      "step": 7915
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6765249199790955e-05,
+      "loss": 0.3141,
+      "step": 7916
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6759010679165395e-05,
+      "loss": 0.3261,
+      "step": 7917
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.675277130150183e-05,
+      "loss": 0.3031,
+      "step": 7918
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.674653106737778e-05,
+      "loss": 0.3082,
+      "step": 7919
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6740289977370855e-05,
+      "loss": 0.3108,
+      "step": 7920
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.673404803205876e-05,
+      "loss": 0.2949,
+      "step": 7921
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.672780523201926e-05,
+      "loss": 0.3074,
+      "step": 7922
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.672156157783021e-05,
+      "loss": 0.2896,
+      "step": 7923
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6715317070069524e-05,
+      "loss": 0.3135,
+      "step": 7924
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.670907170931521e-05,
+      "loss": 0.3208,
+      "step": 7925
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6702825496145367e-05,
+      "loss": 0.3078,
+      "step": 7926
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6696578431138154e-05,
+      "loss": 0.2956,
+      "step": 7927
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.669033051487182e-05,
+      "loss": 0.2884,
+      "step": 7928
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.668408174792469e-05,
+      "loss": 0.3083,
+      "step": 7929
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.667783213087515e-05,
+      "loss": 0.3115,
+      "step": 7930
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.667158166430171e-05,
+      "loss": 0.318,
+      "step": 7931
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.66653303487829e-05,
+      "loss": 0.3068,
+      "step": 7932
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6659078184897375e-05,
+      "loss": 0.3284,
+      "step": 7933
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6652825173223846e-05,
+      "loss": 0.2934,
+      "step": 7934
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.66465713143411e-05,
+      "loss": 0.3029,
+      "step": 7935
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.664031660882803e-05,
+      "loss": 0.329,
+      "step": 7936
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.663406105726356e-05,
+      "loss": 0.3096,
+      "step": 7937
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.662780466022675e-05,
+      "loss": 0.3288,
+      "step": 7938
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.66215474182967e-05,
+      "loss": 0.3061,
+      "step": 7939
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.661528933205258e-05,
+      "loss": 0.3092,
+      "step": 7940
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6609030402073675e-05,
+      "loss": 0.3314,
+      "step": 7941
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.66027706289393e-05,
+      "loss": 0.3091,
+      "step": 7942
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.659651001322892e-05,
+      "loss": 0.3318,
+      "step": 7943
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6590248555522e-05,
+      "loss": 0.2986,
+      "step": 7944
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.658398625639811e-05,
+      "loss": 0.304,
+      "step": 7945
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6577723116436934e-05,
+      "loss": 0.2988,
+      "step": 7946
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.657145913621819e-05,
+      "loss": 0.3132,
+      "step": 7947
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.656519431632168e-05,
+      "loss": 0.32,
+      "step": 7948
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.655892865732731e-05,
+      "loss": 0.3204,
+      "step": 7949
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6552662159815035e-05,
+      "loss": 0.3102,
+      "step": 7950
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.65463948243649e-05,
+      "loss": 0.3211,
+      "step": 7951
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.654012665155702e-05,
+      "loss": 0.308,
+      "step": 7952
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.65338576419716e-05,
+      "loss": 0.3029,
+      "step": 7953
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.652758779618891e-05,
+      "loss": 0.3122,
+      "step": 7954
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6521317114789314e-05,
+      "loss": 0.315,
+      "step": 7955
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.651504559835324e-05,
+      "loss": 0.3104,
+      "step": 7956
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6508773247461175e-05,
+      "loss": 0.3067,
+      "step": 7957
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.650250006269374e-05,
+      "loss": 0.3131,
+      "step": 7958
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6496226044631564e-05,
+      "loss": 0.3058,
+      "step": 7959
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6489951193855413e-05,
+      "loss": 0.3168,
+      "step": 7960
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.648367551094608e-05,
+      "loss": 0.2954,
+      "step": 7961
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6477398996484476e-05,
+      "loss": 0.3081,
+      "step": 7962
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.647112165105156e-05,
+      "loss": 0.3123,
+      "step": 7963
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6464843475228385e-05,
+      "loss": 0.2917,
+      "step": 7964
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6458564469596074e-05,
+      "loss": 0.3021,
+      "step": 7965
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.645228463473583e-05,
+      "loss": 0.2994,
+      "step": 7966
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.644600397122893e-05,
+      "loss": 0.3159,
+      "step": 7967
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.643972247965672e-05,
+      "loss": 0.3275,
+      "step": 7968
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6433440160600644e-05,
+      "loss": 0.321,
+      "step": 7969
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.642715701464221e-05,
+      "loss": 0.2885,
+      "step": 7970
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.642087304236299e-05,
+      "loss": 0.3315,
+      "step": 7971
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.641458824434465e-05,
+      "loss": 0.303,
+      "step": 7972
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6408302621168936e-05,
+      "loss": 0.2982,
+      "step": 7973
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.640201617341765e-05,
+      "loss": 0.3038,
+      "step": 7974
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.639572890167268e-05,
+      "loss": 0.3213,
+      "step": 7975
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.638944080651601e-05,
+      "loss": 0.3052,
+      "step": 7976
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.638315188852966e-05,
+      "loss": 0.3237,
+      "step": 7977
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.637686214829577e-05,
+      "loss": 0.3277,
+      "step": 7978
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.637057158639651e-05,
+      "loss": 0.2995,
+      "step": 7979
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6364280203414175e-05,
+      "loss": 0.3297,
+      "step": 7980
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.635798799993109e-05,
+      "loss": 0.3102,
+      "step": 7981
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6351694976529706e-05,
+      "loss": 0.3068,
+      "step": 7982
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6345401133792495e-05,
+      "loss": 0.305,
+      "step": 7983
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6339106472302036e-05,
+      "loss": 0.3069,
+      "step": 7984
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6332810992640994e-05,
+      "loss": 0.3245,
+      "step": 7985
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.632651469539208e-05,
+      "loss": 0.309,
+      "step": 7986
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6320217581138095e-05,
+      "loss": 0.3015,
+      "step": 7987
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.631391965046193e-05,
+      "loss": 0.3014,
+      "step": 7988
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.630762090394652e-05,
+      "loss": 0.2977,
+      "step": 7989
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.630132134217491e-05,
+      "loss": 0.3189,
+      "step": 7990
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.62950209657302e-05,
+      "loss": 0.3025,
+      "step": 7991
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6288719775195564e-05,
+      "loss": 0.3014,
+      "step": 7992
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6282417771154255e-05,
+      "loss": 0.3171,
+      "step": 7993
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.62761149541896e-05,
+      "loss": 0.3169,
+      "step": 7994
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.626981132488502e-05,
+      "loss": 0.3321,
+      "step": 7995
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.626350688382398e-05,
+      "loss": 0.3292,
+      "step": 7996
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.625720163159004e-05,
+      "loss": 0.3156,
+      "step": 7997
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.625089556876684e-05,
+      "loss": 0.322,
+      "step": 7998
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.624458869593807e-05,
+      "loss": 0.3311,
+      "step": 7999
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.623828101368752e-05,
+      "loss": 0.3205,
+      "step": 8000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.623197252259904e-05,
+      "loss": 0.3122,
+      "step": 8001
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.622566322325656e-05,
+      "loss": 0.2939,
+      "step": 8002
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.621935311624409e-05,
+      "loss": 0.3115,
+      "step": 8003
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.621304220214571e-05,
+      "loss": 0.2944,
+      "step": 8004
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.620673048154558e-05,
+      "loss": 0.3073,
+      "step": 8005
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6200417955027906e-05,
+      "loss": 0.2996,
+      "step": 8006
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.619410462317702e-05,
+      "loss": 0.3155,
+      "step": 8007
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6187790486577276e-05,
+      "loss": 0.3273,
+      "step": 8008
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.618147554581315e-05,
+      "loss": 0.3133,
+      "step": 8009
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.617515980146914e-05,
+      "loss": 0.3099,
+      "step": 8010
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6168843254129884e-05,
+      "loss": 0.3062,
+      "step": 8011
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.616252590438004e-05,
+      "loss": 0.3236,
+      "step": 8012
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.615620775280435e-05,
+      "loss": 0.3128,
+      "step": 8013
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6149888799987644e-05,
+      "loss": 0.3048,
+      "step": 8014
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.614356904651483e-05,
+      "loss": 0.3204,
+      "step": 8015
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6137248492970866e-05,
+      "loss": 0.3183,
+      "step": 8016
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.613092713994081e-05,
+      "loss": 0.3106,
+      "step": 8017
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6124604988009784e-05,
+      "loss": 0.2944,
+      "step": 8018
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6118282037762965e-05,
+      "loss": 0.3118,
+      "step": 8019
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.611195828978565e-05,
+      "loss": 0.2899,
+      "step": 8020
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.610563374466317e-05,
+      "loss": 0.3135,
+      "step": 8021
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.609930840298092e-05,
+      "loss": 0.313,
+      "step": 8022
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6092982265324416e-05,
+      "loss": 0.293,
+      "step": 8023
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608665533227921e-05,
+      "loss": 0.2967,
+      "step": 8024
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608032760443094e-05,
+      "loss": 0.3018,
+      "step": 8025
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6073999082365326e-05,
+      "loss": 0.2957,
+      "step": 8026
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.606766976666815e-05,
+      "loss": 0.3015,
+      "step": 8027
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.606133965792526e-05,
+      "loss": 0.3055,
+      "step": 8028
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6055008756722594e-05,
+      "loss": 0.3104,
+      "step": 8029
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.604867706364615e-05,
+      "loss": 0.3215,
+      "step": 8030
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.604234457928202e-05,
+      "loss": 0.3066,
+      "step": 8031
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.603601130421633e-05,
+      "loss": 0.3204,
+      "step": 8032
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.602967723903533e-05,
+      "loss": 0.3223,
+      "step": 8033
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.602334238432532e-05,
+      "loss": 0.3216,
+      "step": 8034
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.601700674067264e-05,
+      "loss": 0.3136,
+      "step": 8035
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.601067030866377e-05,
+      "loss": 0.3061,
+      "step": 8036
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.60043330888852e-05,
+      "loss": 0.3116,
+      "step": 8037
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5997995081923525e-05,
+      "loss": 0.3113,
+      "step": 8038
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5991656288365424e-05,
+      "loss": 0.3096,
+      "step": 8039
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5985316708797604e-05,
+      "loss": 0.3199,
+      "step": 8040
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5978976343806895e-05,
+      "loss": 0.2949,
+      "step": 8041
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.597263519398017e-05,
+      "loss": 0.3065,
+      "step": 8042
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5966293259904385e-05,
+      "loss": 0.3112,
+      "step": 8043
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.595995054216656e-05,
+      "loss": 0.313,
+      "step": 8044
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5953607041353795e-05,
+      "loss": 0.3107,
+      "step": 8045
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.594726275805327e-05,
+      "loss": 0.2914,
+      "step": 8046
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.594091769285222e-05,
+      "loss": 0.315,
+      "step": 8047
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.593457184633796e-05,
+      "loss": 0.3071,
+      "step": 8048
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.592822521909788e-05,
+      "loss": 0.299,
+      "step": 8049
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.592187781171944e-05,
+      "loss": 0.296,
+      "step": 8050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5915529624790175e-05,
+      "loss": 0.3057,
+      "step": 8051
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5909180658897684e-05,
+      "loss": 0.3096,
+      "step": 8052
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5902830914629646e-05,
+      "loss": 0.3003,
+      "step": 8053
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5896480392573814e-05,
+      "loss": 0.3035,
+      "step": 8054
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5890129093318e-05,
+      "loss": 0.3065,
+      "step": 8055
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.588377701745011e-05,
+      "loss": 0.3051,
+      "step": 8056
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.587742416555811e-05,
+      "loss": 0.3111,
+      "step": 8057
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.587107053823e-05,
+      "loss": 0.3078,
+      "step": 8058
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.586471613605394e-05,
+      "loss": 0.3147,
+      "step": 8059
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.5858360959618065e-05,
+      "loss": 0.3121,
+      "step": 8060
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5852005009510664e-05,
+      "loss": 0.3116,
+      "step": 8061
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.584564828632004e-05,
+      "loss": 0.3165,
+      "step": 8062
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.583929079063458e-05,
+      "loss": 0.3079,
+      "step": 8063
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.583293252304277e-05,
+      "loss": 0.3156,
+      "step": 8064
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.582657348413313e-05,
+      "loss": 0.3164,
+      "step": 8065
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.58202136744943e-05,
+      "loss": 0.3108,
+      "step": 8066
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5813853094714923e-05,
+      "loss": 0.3189,
+      "step": 8067
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.580749174538377e-05,
+      "loss": 0.3126,
+      "step": 8068
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.580112962708965e-05,
+      "loss": 0.3079,
+      "step": 8069
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.579476674042148e-05,
+      "loss": 0.2934,
+      "step": 8070
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5788403085968206e-05,
+      "loss": 0.3106,
+      "step": 8071
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5782038664318875e-05,
+      "loss": 0.3185,
+      "step": 8072
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5775673476062586e-05,
+      "loss": 0.2963,
+      "step": 8073
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5769307521788534e-05,
+      "loss": 0.3125,
+      "step": 8074
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5762940802085936e-05,
+      "loss": 0.3242,
+      "step": 8075
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.575657331754414e-05,
+      "loss": 0.3204,
+      "step": 8076
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5750205068752535e-05,
+      "loss": 0.3163,
+      "step": 8077
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.574383605630058e-05,
+      "loss": 0.3235,
+      "step": 8078
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.573746628077779e-05,
+      "loss": 0.2891,
+      "step": 8079
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.573109574277378e-05,
+      "loss": 0.3039,
+      "step": 8080
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5724724442878243e-05,
+      "loss": 0.2858,
+      "step": 8081
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5718352381680895e-05,
+      "loss": 0.3102,
+      "step": 8082
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.571197955977156e-05,
+      "loss": 0.303,
+      "step": 8083
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5705605977740125e-05,
+      "loss": 0.3281,
+      "step": 8084
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.569923163617655e-05,
+      "loss": 0.29,
+      "step": 8085
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.569285653567085e-05,
+      "loss": 0.3074,
+      "step": 8086
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5686480676813115e-05,
+      "loss": 0.3223,
+      "step": 8087
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.568010406019353e-05,
+      "loss": 0.3156,
+      "step": 8088
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.567372668640232e-05,
+      "loss": 0.3175,
+      "step": 8089
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.566734855602979e-05,
+      "loss": 0.3212,
+      "step": 8090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.566096966966633e-05,
+      "loss": 0.3127,
+      "step": 8091
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.565459002790237e-05,
+      "loss": 0.292,
+      "step": 8092
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.564820963132844e-05,
+      "loss": 0.3057,
+      "step": 8093
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5641828480535104e-05,
+      "loss": 0.3148,
+      "step": 8094
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5635446576113044e-05,
+      "loss": 0.3235,
+      "step": 8095
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.562906391865297e-05,
+      "loss": 0.3205,
+      "step": 8096
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.562268050874568e-05,
+      "loss": 0.31,
+      "step": 8097
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.561629634698205e-05,
+      "loss": 0.2992,
+      "step": 8098
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5609911433952986e-05,
+      "loss": 0.3236,
+      "step": 8099
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5603525770249526e-05,
+      "loss": 0.3166,
+      "step": 8100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.559713935646272e-05,
+      "loss": 0.3025,
+      "step": 8101
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5590752193183726e-05,
+      "loss": 0.3137,
+      "step": 8102
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.558436428100375e-05,
+      "loss": 0.3163,
+      "step": 8103
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5577975620514066e-05,
+      "loss": 0.356,
+      "step": 8104
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5571586212306035e-05,
+      "loss": 0.2839,
+      "step": 8105
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.556519605697107e-05,
+      "loss": 0.3154,
+      "step": 8106
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.555880515510068e-05,
+      "loss": 0.3036,
+      "step": 8107
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.555241350728639e-05,
+      "loss": 0.3052,
+      "step": 8108
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5546021114119853e-05,
+      "loss": 0.3073,
+      "step": 8109
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5539627976192754e-05,
+      "loss": 0.3048,
+      "step": 8110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.553323409409687e-05,
+      "loss": 0.3186,
+      "step": 8111
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5526839468424026e-05,
+      "loss": 0.3431,
+      "step": 8112
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5520444099766136e-05,
+      "loss": 0.3166,
+      "step": 8113
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5514047988715147e-05,
+      "loss": 0.3183,
+      "step": 8114
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.550765113586313e-05,
+      "loss": 0.3255,
+      "step": 8115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.550125354180217e-05,
+      "loss": 0.2896,
+      "step": 8116
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.549485520712445e-05,
+      "loss": 0.3179,
+      "step": 8117
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5488456132422235e-05,
+      "loss": 0.3,
+      "step": 8118
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.548205631828781e-05,
+      "loss": 0.2894,
+      "step": 8119
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5475655765313594e-05,
+      "loss": 0.3353,
+      "step": 8120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5469254474092e-05,
+      "loss": 0.2962,
+      "step": 8121
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.546285244521558e-05,
+      "loss": 0.3045,
+      "step": 8122
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.54564496792769e-05,
+      "loss": 0.2956,
+      "step": 8123
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.545004617686863e-05,
+      "loss": 0.315,
+      "step": 8124
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.54436419385835e-05,
+      "loss": 0.3035,
+      "step": 8125
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.543723696501428e-05,
+      "loss": 0.3143,
+      "step": 8126
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.543083125675386e-05,
+      "loss": 0.3344,
+      "step": 8127
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5424424814395146e-05,
+      "loss": 0.3137,
+      "step": 8128
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5418017638531144e-05,
+      "loss": 0.3024,
+      "step": 8129
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.541160972975491e-05,
+      "loss": 0.3009,
+      "step": 8130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.540520108865959e-05,
+      "loss": 0.29,
+      "step": 8131
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.539879171583838e-05,
+      "loss": 0.3181,
+      "step": 8132
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5392381611884544e-05,
+      "loss": 0.3128,
+      "step": 8133
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5385970777391424e-05,
+      "loss": 0.2968,
+      "step": 8134
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5379559212952415e-05,
+      "loss": 0.3148,
+      "step": 8135
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5373146919161e-05,
+      "loss": 0.3158,
+      "step": 8136
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5366733896610706e-05,
+      "loss": 0.2841,
+      "step": 8137
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5360320145895146e-05,
+      "loss": 0.3005,
+      "step": 8138
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5353905667607985e-05,
+      "loss": 0.3062,
+      "step": 8139
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.534749046234297e-05,
+      "loss": 0.3069,
+      "step": 8140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.534107453069391e-05,
+      "loss": 0.2957,
+      "step": 8141
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.533465787325469e-05,
+      "loss": 0.3111,
+      "step": 8142
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.532824049061924e-05,
+      "loss": 0.3051,
+      "step": 8143
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.532182238338155e-05,
+      "loss": 0.3117,
+      "step": 8144
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5315403552135735e-05,
+      "loss": 0.3088,
+      "step": 8145
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5308983997475916e-05,
+      "loss": 0.3252,
+      "step": 8146
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.530256371999631e-05,
+      "loss": 0.3181,
+      "step": 8147
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5296142720291195e-05,
+      "loss": 0.3254,
+      "step": 8148
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.528972099895491e-05,
+      "loss": 0.3076,
+      "step": 8149
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5283298556581876e-05,
+      "loss": 0.3211,
+      "step": 8150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.527687539376656e-05,
+      "loss": 0.2989,
+      "step": 8151
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.527045151110351e-05,
+      "loss": 0.3512,
+      "step": 8152
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5264026909187335e-05,
+      "loss": 0.2917,
+      "step": 8153
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.525760158861273e-05,
+      "loss": 0.3145,
+      "step": 8154
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.525117554997441e-05,
+      "loss": 0.3239,
+      "step": 8155
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.524474879386721e-05,
+      "loss": 0.312,
+      "step": 8156
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5238321320886006e-05,
+      "loss": 0.2899,
+      "step": 8157
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.523189313162572e-05,
+      "loss": 0.2969,
+      "step": 8158
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.522546422668139e-05,
+      "loss": 0.2947,
+      "step": 8159
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.521903460664806e-05,
+      "loss": 0.3233,
+      "step": 8160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.521260427212091e-05,
+      "loss": 0.3255,
+      "step": 8161
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5206173223695114e-05,
+      "loss": 0.316,
+      "step": 8162
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.519974146196596e-05,
+      "loss": 0.3145,
+      "step": 8163
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.519330898752881e-05,
+      "loss": 0.2807,
+      "step": 8164
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5186875800979026e-05,
+      "loss": 0.3061,
+      "step": 8165
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.518044190291212e-05,
+      "loss": 0.3137,
+      "step": 8166
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.517400729392361e-05,
+      "loss": 0.3062,
+      "step": 8167
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5167571974609105e-05,
+      "loss": 0.3137,
+      "step": 8168
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.516113594556428e-05,
+      "loss": 0.3077,
+      "step": 8169
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.515469920738486e-05,
+      "loss": 0.3071,
+      "step": 8170
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5148261760666646e-05,
+      "loss": 0.2924,
+      "step": 8171
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.514182360600553e-05,
+      "loss": 0.3014,
+      "step": 8172
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.51353847439974e-05,
+      "loss": 0.32,
+      "step": 8173
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5128945175238294e-05,
+      "loss": 0.3187,
+      "step": 8174
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5122504900324254e-05,
+      "loss": 0.3048,
+      "step": 8175
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5116063919851414e-05,
+      "loss": 0.327,
+      "step": 8176
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5109622234415975e-05,
+      "loss": 0.2955,
+      "step": 8177
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.510317984461418e-05,
+      "loss": 0.3209,
+      "step": 8178
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.509673675104237e-05,
+      "loss": 0.2981,
+      "step": 8179
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5090292954296914e-05,
+      "loss": 0.3199,
+      "step": 8180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.50838484549743e-05,
+      "loss": 0.3334,
+      "step": 8181
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.507740325367102e-05,
+      "loss": 0.3153,
+      "step": 8182
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.507095735098365e-05,
+      "loss": 0.2941,
+      "step": 8183
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.506451074750887e-05,
+      "loss": 0.3077,
+      "step": 8184
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.505806344384337e-05,
+      "loss": 0.316,
+      "step": 8185
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5051615440583946e-05,
+      "loss": 0.304,
+      "step": 8186
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.504516673832743e-05,
+      "loss": 0.2994,
+      "step": 8187
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5038717337670735e-05,
+      "loss": 0.2996,
+      "step": 8188
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5032267239210835e-05,
+      "loss": 0.3015,
+      "step": 8189
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.502581644354476e-05,
+      "loss": 0.3182,
+      "step": 8190
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5019364951269635e-05,
+      "loss": 0.312,
+      "step": 8191
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.50129127629826e-05,
+      "loss": 0.3,
+      "step": 8192
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.500645987928089e-05,
+      "loss": 0.3171,
+      "step": 8193
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.500000630076183e-05,
+      "loss": 0.3287,
+      "step": 8194
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4993552028022735e-05,
+      "loss": 0.3112,
+      "step": 8195
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.498709706166106e-05,
+      "loss": 0.2906,
+      "step": 8196
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.498064140227429e-05,
+      "loss": 0.2986,
+      "step": 8197
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.497418505045997e-05,
+      "loss": 0.3138,
+      "step": 8198
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.496772800681572e-05,
+      "loss": 0.3126,
+      "step": 8199
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4961270271939214e-05,
+      "loss": 0.3174,
+      "step": 8200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4954811846428216e-05,
+      "loss": 0.3129,
+      "step": 8201
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.494835273088051e-05,
+      "loss": 0.2886,
+      "step": 8202
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.494189292589399e-05,
+      "loss": 0.3061,
+      "step": 8203
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.493543243206657e-05,
+      "loss": 0.3252,
+      "step": 8204
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.492897124999627e-05,
+      "loss": 0.3142,
+      "step": 8205
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.492250938028115e-05,
+      "loss": 0.3021,
+      "step": 8206
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.491604682351933e-05,
+      "loss": 0.3129,
+      "step": 8207
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4909583580309e-05,
+      "loss": 0.3169,
+      "step": 8208
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.490311965124843e-05,
+      "loss": 0.3194,
+      "step": 8209
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.489665503693591e-05,
+      "loss": 0.3032,
+      "step": 8210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4890189737969856e-05,
+      "loss": 0.3182,
+      "step": 8211
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4883723754948684e-05,
+      "loss": 0.3047,
+      "step": 8212
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.487725708847092e-05,
+      "loss": 0.3221,
+      "step": 8213
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.487078973913512e-05,
+      "loss": 0.297,
+      "step": 8214
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4864321707539945e-05,
+      "loss": 0.3158,
+      "step": 8215
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.485785299428406e-05,
+      "loss": 0.2946,
+      "step": 8216
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.485138359996624e-05,
+      "loss": 0.3159,
+      "step": 8217
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.484491352518532e-05,
+      "loss": 0.3067,
+      "step": 8218
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.483844277054017e-05,
+      "loss": 0.3203,
+      "step": 8219
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.483197133662975e-05,
+      "loss": 0.3043,
+      "step": 8220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4825499224053074e-05,
+      "loss": 0.31,
+      "step": 8221
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.48190264334092e-05,
+      "loss": 0.3093,
+      "step": 8222
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.481255296529729e-05,
+      "loss": 0.3026,
+      "step": 8223
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.480607882031653e-05,
+      "loss": 0.3144,
+      "step": 8224
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4799603999066186e-05,
+      "loss": 0.2824,
+      "step": 8225
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.479312850214559e-05,
+      "loss": 0.3229,
+      "step": 8226
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4786652330154116e-05,
+      "loss": 0.3023,
+      "step": 8227
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.478017548369124e-05,
+      "loss": 0.321,
+      "step": 8228
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.477369796335645e-05,
+      "loss": 0.3222,
+      "step": 8229
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.476721976974935e-05,
+      "loss": 0.3104,
+      "step": 8230
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.476074090346954e-05,
+      "loss": 0.3106,
+      "step": 8231
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.4754261365116764e-05,
+      "loss": 0.3084,
+      "step": 8232
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.474778115529076e-05,
+      "loss": 0.3296,
+      "step": 8233
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.474130027459134e-05,
+      "loss": 0.3099,
+      "step": 8234
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.473481872361843e-05,
+      "loss": 0.3205,
+      "step": 8235
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4728336502971934e-05,
+      "loss": 0.3229,
+      "step": 8236
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.47218536132519e-05,
+      "loss": 0.2901,
+      "step": 8237
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4715370055058396e-05,
+      "loss": 0.2987,
+      "step": 8238
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4708885828991526e-05,
+      "loss": 0.3073,
+      "step": 8239
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.470240093565153e-05,
+      "loss": 0.3116,
+      "step": 8240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.469591537563863e-05,
+      "loss": 0.3128,
+      "step": 8241
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4689429149553174e-05,
+      "loss": 0.2989,
+      "step": 8242
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4682942257995524e-05,
+      "loss": 0.3123,
+      "step": 8243
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.467645470156613e-05,
+      "loss": 0.3154,
+      "step": 8244
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.46699664808655e-05,
+      "loss": 0.3185,
+      "step": 8245
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4663477596494196e-05,
+      "loss": 0.3258,
+      "step": 8246
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4656988049052855e-05,
+      "loss": 0.3261,
+      "step": 8247
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.465049783914215e-05,
+      "loss": 0.3142,
+      "step": 8248
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4644006967362846e-05,
+      "loss": 0.337,
+      "step": 8249
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.463751543431576e-05,
+      "loss": 0.3072,
+      "step": 8250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.463102324060174e-05,
+      "loss": 0.3208,
+      "step": 8251
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.462453038682175e-05,
+      "loss": 0.3171,
+      "step": 8252
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.461803687357676e-05,
+      "loss": 0.3081,
+      "step": 8253
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.461154270146784e-05,
+      "loss": 0.3134,
+      "step": 8254
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.460504787109611e-05,
+      "loss": 0.2925,
+      "step": 8255
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4598552383062735e-05,
+      "loss": 0.3061,
+      "step": 8256
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.459205623796897e-05,
+      "loss": 0.2955,
+      "step": 8257
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4585559436416105e-05,
+      "loss": 0.308,
+      "step": 8258
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.45790619790055e-05,
+      "loss": 0.3007,
+      "step": 8259
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4572563866338584e-05,
+      "loss": 0.2923,
+      "step": 8260
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.456606509901683e-05,
+      "loss": 0.3286,
+      "step": 8261
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.455956567764179e-05,
+      "loss": 0.3168,
+      "step": 8262
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.455306560281506e-05,
+      "loss": 0.313,
+      "step": 8263
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.454656487513831e-05,
+      "loss": 0.2935,
+      "step": 8264
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.454006349521326e-05,
+      "loss": 0.3169,
+      "step": 8265
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4533561463641696e-05,
+      "loss": 0.3169,
+      "step": 8266
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.452705878102546e-05,
+      "loss": 0.3091,
+      "step": 8267
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.452055544796646e-05,
+      "loss": 0.3066,
+      "step": 8268
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.451405146506665e-05,
+      "loss": 0.2952,
+      "step": 8269
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.450754683292809e-05,
+      "loss": 0.2986,
+      "step": 8270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4501041552152826e-05,
+      "loss": 0.3111,
+      "step": 8271
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.449453562334303e-05,
+      "loss": 0.324,
+      "step": 8272
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4488029047100885e-05,
+      "loss": 0.2973,
+      "step": 8273
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4481521824028673e-05,
+      "loss": 0.3035,
+      "step": 8274
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4475013954728714e-05,
+      "loss": 0.333,
+      "step": 8275
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4468505439803403e-05,
+      "loss": 0.3092,
+      "step": 8276
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4461996279855165e-05,
+      "loss": 0.3066,
+      "step": 8277
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.445548647548652e-05,
+      "loss": 0.3071,
+      "step": 8278
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.444897602730004e-05,
+      "loss": 0.3016,
+      "step": 8279
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.4442464935898314e-05,
+      "loss": 0.3209,
+      "step": 8280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.443595320188406e-05,
+      "loss": 0.2999,
+      "step": 8281
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.442944082586001e-05,
+      "loss": 0.3123,
+      "step": 8282
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.442292780842897e-05,
+      "loss": 0.3132,
+      "step": 8283
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.44164141501938e-05,
+      "loss": 0.3164,
+      "step": 8284
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.440989985175741e-05,
+      "loss": 0.3129,
+      "step": 8285
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.44033849137228e-05,
+      "loss": 0.316,
+      "step": 8286
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.439686933669299e-05,
+      "loss": 0.2964,
+      "step": 8287
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.439035312127109e-05,
+      "loss": 0.307,
+      "step": 8288
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.438383626806025e-05,
+      "loss": 0.2892,
+      "step": 8289
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.43773187776637e-05,
+      "loss": 0.2878,
+      "step": 8290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4370800650684706e-05,
+      "loss": 0.3028,
+      "step": 8291
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.436428188772661e-05,
+      "loss": 0.3087,
+      "step": 8292
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4357762489392795e-05,
+      "loss": 0.3351,
+      "step": 8293
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.435124245628672e-05,
+      "loss": 0.3211,
+      "step": 8294
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.434472178901189e-05,
+      "loss": 0.3075,
+      "step": 8295
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.433820048817189e-05,
+      "loss": 0.3187,
+      "step": 8296
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.433167855437033e-05,
+      "loss": 0.3027,
+      "step": 8297
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.432515598821091e-05,
+      "loss": 0.3295,
+      "step": 8298
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.431863279029737e-05,
+      "loss": 0.3154,
+      "step": 8299
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.43121089612335e-05,
+      "loss": 0.3196,
+      "step": 8300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.43055845016232e-05,
+      "loss": 0.3065,
+      "step": 8301
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.429905941207036e-05,
+      "loss": 0.3047,
+      "step": 8302
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.429253369317896e-05,
+      "loss": 0.3279,
+      "step": 8303
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.428600734555306e-05,
+      "loss": 0.3097,
+      "step": 8304
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.427948036979673e-05,
+      "loss": 0.3009,
+      "step": 8305
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4272952766514135e-05,
+      "loss": 0.3196,
+      "step": 8306
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4266424536309495e-05,
+      "loss": 0.3064,
+      "step": 8307
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.425989567978706e-05,
+      "loss": 0.3135,
+      "step": 8308
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.425336619755118e-05,
+      "loss": 0.303,
+      "step": 8309
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.424683609020622e-05,
+      "loss": 0.3245,
+      "step": 8310
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.424030535835664e-05,
+      "loss": 0.299,
+      "step": 8311
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.423377400260693e-05,
+      "loss": 0.3081,
+      "step": 8312
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4227242023561656e-05,
+      "loss": 0.3066,
+      "step": 8313
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4220709421825444e-05,
+      "loss": 0.3115,
+      "step": 8314
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.421417619800294e-05,
+      "loss": 0.3205,
+      "step": 8315
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.420764235269891e-05,
+      "loss": 0.3216,
+      "step": 8316
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.420110788651811e-05,
+      "loss": 0.3231,
+      "step": 8317
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.419457280006542e-05,
+      "loss": 0.3139,
+      "step": 8318
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4188037093945725e-05,
+      "loss": 0.2976,
+      "step": 8319
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4181500768763984e-05,
+      "loss": 0.3042,
+      "step": 8320
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.417496382512523e-05,
+      "loss": 0.3349,
+      "step": 8321
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4168426263634527e-05,
+      "loss": 0.3055,
+      "step": 8322
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4161888084897024e-05,
+      "loss": 0.3089,
+      "step": 8323
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4155349289517896e-05,
+      "loss": 0.3115,
+      "step": 8324
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.414880987810239e-05,
+      "loss": 0.3118,
+      "step": 8325
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4142269851255835e-05,
+      "loss": 0.3161,
+      "step": 8326
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.413572920958356e-05,
+      "loss": 0.3159,
+      "step": 8327
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4129187953691014e-05,
+      "loss": 0.3029,
+      "step": 8328
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.412264608418365e-05,
+      "loss": 0.3077,
+      "step": 8329
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4116103601666996e-05,
+      "loss": 0.3255,
+      "step": 8330
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.410956050674667e-05,
+      "loss": 0.319,
+      "step": 8331
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.410301680002829e-05,
+      "loss": 0.3112,
+      "step": 8332
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.409647248211758e-05,
+      "loss": 0.2917,
+      "step": 8333
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.408992755362029e-05,
+      "loss": 0.3027,
+      "step": 8334
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.408338201514222e-05,
+      "loss": 0.2963,
+      "step": 8335
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4076835867289274e-05,
+      "loss": 0.3126,
+      "step": 8336
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.407028911066735e-05,
+      "loss": 0.3035,
+      "step": 8337
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.406374174588245e-05,
+      "loss": 0.2813,
+      "step": 8338
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4057193773540605e-05,
+      "loss": 0.3037,
+      "step": 8339
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.405064519424793e-05,
+      "loss": 0.3238,
+      "step": 8340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.404409600861056e-05,
+      "loss": 0.2919,
+      "step": 8341
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4037546217234704e-05,
+      "loss": 0.3022,
+      "step": 8342
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4030995820726645e-05,
+      "loss": 0.3085,
+      "step": 8343
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.402444481969269e-05,
+      "loss": 0.3269,
+      "step": 8344
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.401789321473921e-05,
+      "loss": 0.294,
+      "step": 8345
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.401134100647266e-05,
+      "loss": 0.3054,
+      "step": 8346
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.400478819549952e-05,
+      "loss": 0.3213,
+      "step": 8347
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.399823478242634e-05,
+      "loss": 0.2927,
+      "step": 8348
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3991680767859694e-05,
+      "loss": 0.2938,
+      "step": 8349
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.398512615240628e-05,
+      "loss": 0.3259,
+      "step": 8350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3978570936672775e-05,
+      "loss": 0.2919,
+      "step": 8351
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.397201512126597e-05,
+      "loss": 0.2919,
+      "step": 8352
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3965458706792674e-05,
+      "loss": 0.3216,
+      "step": 8353
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.395890169385977e-05,
+      "loss": 0.3122,
+      "step": 8354
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.39523440830742e-05,
+      "loss": 0.2952,
+      "step": 8355
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3945785875042946e-05,
+      "loss": 0.3127,
+      "step": 8356
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3939227070373045e-05,
+      "loss": 0.3063,
+      "step": 8357
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.393266766967161e-05,
+      "loss": 0.3118,
+      "step": 8358
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.392610767354579e-05,
+      "loss": 0.3181,
+      "step": 8359
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3919547082602803e-05,
+      "loss": 0.3069,
+      "step": 8360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.39129858974499e-05,
+      "loss": 0.302,
+      "step": 8361
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.390642411869442e-05,
+      "loss": 0.298,
+      "step": 8362
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3899861746943716e-05,
+      "loss": 0.2967,
+      "step": 8363
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3893298782805235e-05,
+      "loss": 0.2935,
+      "step": 8364
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.388673522688646e-05,
+      "loss": 0.3366,
+      "step": 8365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.388017107979492e-05,
+      "loss": 0.3096,
+      "step": 8366
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.387360634213823e-05,
+      "loss": 0.3044,
+      "step": 8367
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.386704101452402e-05,
+      "loss": 0.3175,
+      "step": 8368
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3860475097559985e-05,
+      "loss": 0.3138,
+      "step": 8369
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.385390859185391e-05,
+      "loss": 0.3092,
+      "step": 8370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.384734149801359e-05,
+      "loss": 0.3062,
+      "step": 8371
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.384077381664691e-05,
+      "loss": 0.2995,
+      "step": 8372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.383420554836177e-05,
+      "loss": 0.292,
+      "step": 8373
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.382763669376615e-05,
+      "loss": 0.3085,
+      "step": 8374
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3821067253468094e-05,
+      "loss": 0.3257,
+      "step": 8375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.381449722807568e-05,
+      "loss": 0.32,
+      "step": 8376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3807926618197045e-05,
+      "loss": 0.3106,
+      "step": 8377
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3801355424440385e-05,
+      "loss": 0.2959,
+      "step": 8378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.379478364741393e-05,
+      "loss": 0.3328,
+      "step": 8379
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3788211287726e-05,
+      "loss": 0.3041,
+      "step": 8380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3781638345984945e-05,
+      "loss": 0.3126,
+      "step": 8381
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3775064822799175e-05,
+      "loss": 0.289,
+      "step": 8382
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.376849071877715e-05,
+      "loss": 0.3173,
+      "step": 8383
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.376191603452738e-05,
+      "loss": 0.3268,
+      "step": 8384
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.375534077065844e-05,
+      "loss": 0.2996,
+      "step": 8385
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.374876492777895e-05,
+      "loss": 0.3042,
+      "step": 8386
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.37421885064976e-05,
+      "loss": 0.3227,
+      "step": 8387
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.37356115074231e-05,
+      "loss": 0.3001,
+      "step": 8388
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.372903393116425e-05,
+      "loss": 0.3162,
+      "step": 8389
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3722455778329886e-05,
+      "loss": 0.3271,
+      "step": 8390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3715877049528894e-05,
+      "loss": 0.323,
+      "step": 8391
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.370929774537022e-05,
+      "loss": 0.315,
+      "step": 8392
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.370271786646287e-05,
+      "loss": 0.3098,
+      "step": 8393
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3696137413415875e-05,
+      "loss": 0.3142,
+      "step": 8394
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.368955638683835e-05,
+      "loss": 0.3057,
+      "step": 8395
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.368297478733945e-05,
+      "loss": 0.3032,
+      "step": 8396
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3676392615528394e-05,
+      "loss": 0.3136,
+      "step": 8397
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3669809872014434e-05,
+      "loss": 0.3068,
+      "step": 8398
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3663226557406895e-05,
+      "loss": 0.3189,
+      "step": 8399
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.365664267231513e-05,
+      "loss": 0.316,
+      "step": 8400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3650058217348576e-05,
+      "loss": 0.3099,
+      "step": 8401
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.36434731931167e-05,
+      "loss": 0.3021,
+      "step": 8402
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.363688760022903e-05,
+      "loss": 0.2922,
+      "step": 8403
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3630301439295144e-05,
+      "loss": 0.3223,
+      "step": 8404
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.362371471092467e-05,
+      "loss": 0.3296,
+      "step": 8405
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3617127415727315e-05,
+      "loss": 0.3211,
+      "step": 8406
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3610539554312794e-05,
+      "loss": 0.3099,
+      "step": 8407
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3603951127290896e-05,
+      "loss": 0.2968,
+      "step": 8408
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3597362135271474e-05,
+      "loss": 0.3313,
+      "step": 8409
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.359077257886442e-05,
+      "loss": 0.3041,
+      "step": 8410
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.358418245867967e-05,
+      "loss": 0.2964,
+      "step": 8411
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.357759177532724e-05,
+      "loss": 0.3164,
+      "step": 8412
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3571000529417155e-05,
+      "loss": 0.3044,
+      "step": 8413
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3564408721559546e-05,
+      "loss": 0.3119,
+      "step": 8414
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.355781635236456e-05,
+      "loss": 0.3178,
+      "step": 8415
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.355122342244239e-05,
+      "loss": 0.2982,
+      "step": 8416
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.354462993240331e-05,
+      "loss": 0.2932,
+      "step": 8417
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.353803588285762e-05,
+      "loss": 0.2984,
+      "step": 8418
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.353144127441569e-05,
+      "loss": 0.3115,
+      "step": 8419
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.352484610768793e-05,
+      "loss": 0.3106,
+      "step": 8420
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.351825038328481e-05,
+      "loss": 0.3149,
+      "step": 8421
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3511654101816844e-05,
+      "loss": 0.3016,
+      "step": 8422
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.350505726389459e-05,
+      "loss": 0.2984,
+      "step": 8423
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3498459870128696e-05,
+      "loss": 0.3032,
+      "step": 8424
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.34918619211298e-05,
+      "loss": 0.2955,
+      "step": 8425
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3485263417508653e-05,
+      "loss": 0.312,
+      "step": 8426
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.347866435987602e-05,
+      "loss": 0.2934,
+      "step": 8427
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.347206474884272e-05,
+      "loss": 0.3164,
+      "step": 8428
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3465464585019644e-05,
+      "loss": 0.3083,
+      "step": 8429
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3458863869017706e-05,
+      "loss": 0.2952,
+      "step": 8430
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3452262601447895e-05,
+      "loss": 0.3154,
+      "step": 8431
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.344566078292124e-05,
+      "loss": 0.3183,
+      "step": 8432
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3439058414048815e-05,
+      "loss": 0.2898,
+      "step": 8433
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.343245549544176e-05,
+      "loss": 0.3157,
+      "step": 8434
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.342585202771125e-05,
+      "loss": 0.3185,
+      "step": 8435
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.341924801146853e-05,
+      "loss": 0.3094,
+      "step": 8436
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3412643447324886e-05,
+      "loss": 0.2883,
+      "step": 8437
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.340603833589164e-05,
+      "loss": 0.2764,
+      "step": 8438
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.339943267778019e-05,
+      "loss": 0.3377,
+      "step": 8439
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3392826473601966e-05,
+      "loss": 0.338,
+      "step": 8440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3386219723968466e-05,
+      "loss": 0.3083,
+      "step": 8441
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.337961242949122e-05,
+      "loss": 0.3353,
+      "step": 8442
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.33730045907818e-05,
+      "loss": 0.3287,
+      "step": 8443
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.336639620845187e-05,
+      "loss": 0.3318,
+      "step": 8444
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3359787283113115e-05,
+      "loss": 0.3064,
+      "step": 8445
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.335317781537726e-05,
+      "loss": 0.3127,
+      "step": 8446
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.334656780585611e-05,
+      "loss": 0.3231,
+      "step": 8447
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.33399572551615e-05,
+      "loss": 0.3162,
+      "step": 8448
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3333346163905314e-05,
+      "loss": 0.3177,
+      "step": 8449
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3326734532699495e-05,
+      "loss": 0.3001,
+      "step": 8450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.332012236215604e-05,
+      "loss": 0.3215,
+      "step": 8451
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.331350965288697e-05,
+      "loss": 0.3098,
+      "step": 8452
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.330689640550439e-05,
+      "loss": 0.3052,
+      "step": 8453
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3300282620620445e-05,
+      "loss": 0.3214,
+      "step": 8454
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.32936682988473e-05,
+      "loss": 0.3054,
+      "step": 8455
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.328705344079722e-05,
+      "loss": 0.3054,
+      "step": 8456
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.328043804708248e-05,
+      "loss": 0.3082,
+      "step": 8457
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3273822118315426e-05,
+      "loss": 0.2958,
+      "step": 8458
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3267205655108427e-05,
+      "loss": 0.2948,
+      "step": 8459
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3260588658073936e-05,
+      "loss": 0.3163,
+      "step": 8460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3253971127824446e-05,
+      "loss": 0.3213,
+      "step": 8461
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.324735306497247e-05,
+      "loss": 0.3208,
+      "step": 8462
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.324073447013061e-05,
+      "loss": 0.3092,
+      "step": 8463
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.32341153439115e-05,
+      "loss": 0.3014,
+      "step": 8464
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.322749568692781e-05,
+      "loss": 0.2971,
+      "step": 8465
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.322087549979229e-05,
+      "loss": 0.3112,
+      "step": 8466
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.321425478311771e-05,
+      "loss": 0.3177,
+      "step": 8467
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.320763353751691e-05,
+      "loss": 0.299,
+      "step": 8468
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.320101176360275e-05,
+      "loss": 0.2983,
+      "step": 8469
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.319438946198819e-05,
+      "loss": 0.3018,
+      "step": 8470
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.318776663328618e-05,
+      "loss": 0.3122,
+      "step": 8471
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3181143278109754e-05,
+      "loss": 0.2964,
+      "step": 8472
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3174519397071994e-05,
+      "loss": 0.3084,
+      "step": 8473
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.316789499078602e-05,
+      "loss": 0.2824,
+      "step": 8474
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3161270059865e-05,
+      "loss": 0.3053,
+      "step": 8475
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.315464460492216e-05,
+      "loss": 0.3147,
+      "step": 8476
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.314801862657076e-05,
+      "loss": 0.3158,
+      "step": 8477
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.314139212542413e-05,
+      "loss": 0.3039,
+      "step": 8478
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.313476510209563e-05,
+      "loss": 0.2989,
+      "step": 8479
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3128137557198676e-05,
+      "loss": 0.2863,
+      "step": 8480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.312150949134673e-05,
+      "loss": 0.3104,
+      "step": 8481
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.31148809051533e-05,
+      "loss": 0.2999,
+      "step": 8482
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3108251799231946e-05,
+      "loss": 0.3248,
+      "step": 8483
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3101622174196285e-05,
+      "loss": 0.3085,
+      "step": 8484
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3094992030659956e-05,
+      "loss": 0.3055,
+      "step": 8485
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.308836136923668e-05,
+      "loss": 0.3023,
+      "step": 8486
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.308173019054019e-05,
+      "loss": 0.3246,
+      "step": 8487
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.30750984951843e-05,
+      "loss": 0.3157,
+      "step": 8488
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.306846628378285e-05,
+      "loss": 0.2975,
+      "step": 8489
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.306183355694973e-05,
+      "loss": 0.296,
+      "step": 8490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.305520031529889e-05,
+      "loss": 0.2992,
+      "step": 8491
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3048566559444324e-05,
+      "loss": 0.319,
+      "step": 8492
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.304193229000006e-05,
+      "loss": 0.3042,
+      "step": 8493
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.303529750758018e-05,
+      "loss": 0.2908,
+      "step": 8494
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3028662212798834e-05,
+      "loss": 0.3152,
+      "step": 8495
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.302202640627018e-05,
+      "loss": 0.2995,
+      "step": 8496
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.301539008860847e-05,
+      "loss": 0.3198,
+      "step": 8497
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3008753260427963e-05,
+      "loss": 0.2935,
+      "step": 8498
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.300211592234297e-05,
+      "loss": 0.3192,
+      "step": 8499
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.299547807496789e-05,
+      "loss": 0.3294,
+      "step": 8500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.298883971891711e-05,
+      "loss": 0.3005,
+      "step": 8501
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2982200854805116e-05,
+      "loss": 0.3057,
+      "step": 8502
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.297556148324641e-05,
+      "loss": 0.302,
+      "step": 8503
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.296892160485554e-05,
+      "loss": 0.2826,
+      "step": 8504
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.296228122024713e-05,
+      "loss": 0.3204,
+      "step": 8505
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.295564033003581e-05,
+      "loss": 0.3128,
+      "step": 8506
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2948998934836295e-05,
+      "loss": 0.2889,
+      "step": 8507
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2942357035263324e-05,
+      "loss": 0.327,
+      "step": 8508
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.293571463193168e-05,
+      "loss": 0.3088,
+      "step": 8509
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.292907172545621e-05,
+      "loss": 0.3214,
+      "step": 8510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.29224283164518e-05,
+      "loss": 0.3007,
+      "step": 8511
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.291578440553338e-05,
+      "loss": 0.3123,
+      "step": 8512
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2909139993315916e-05,
+      "loss": 0.2996,
+      "step": 8513
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.290249508041445e-05,
+      "loss": 0.3138,
+      "step": 8514
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.289584966744404e-05,
+      "loss": 0.3122,
+      "step": 8515
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2889203755019804e-05,
+      "loss": 0.3291,
+      "step": 8516
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2882557343756915e-05,
+      "loss": 0.3198,
+      "step": 8517
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2875910434270575e-05,
+      "loss": 0.3055,
+      "step": 8518
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2869263027176035e-05,
+      "loss": 0.3084,
+      "step": 8519
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2862615123088606e-05,
+      "loss": 0.2993,
+      "step": 8520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.285596672262361e-05,
+      "loss": 0.3201,
+      "step": 8521
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.284931782639648e-05,
+      "loss": 0.3014,
+      "step": 8522
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2842668435022626e-05,
+      "loss": 0.2947,
+      "step": 8523
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2836018549117546e-05,
+      "loss": 0.3098,
+      "step": 8524
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.282936816929677e-05,
+      "loss": 0.305,
+      "step": 8525
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.282271729617585e-05,
+      "loss": 0.3288,
+      "step": 8526
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.281606593037045e-05,
+      "loss": 0.331,
+      "step": 8527
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2809414072496204e-05,
+      "loss": 0.3068,
+      "step": 8528
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.280276172316884e-05,
+      "loss": 0.311,
+      "step": 8529
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.279610888300412e-05,
+      "loss": 0.3127,
+      "step": 8530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.278945555261783e-05,
+      "loss": 0.3026,
+      "step": 8531
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2782801732625844e-05,
+      "loss": 0.3036,
+      "step": 8532
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.277614742364404e-05,
+      "loss": 0.328,
+      "step": 8533
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.276949262628836e-05,
+      "loss": 0.3126,
+      "step": 8534
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2762837341174793e-05,
+      "loss": 0.3332,
+      "step": 8535
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.275618156891937e-05,
+      "loss": 0.3217,
+      "step": 8536
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2749525310138156e-05,
+      "loss": 0.3202,
+      "step": 8537
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2742868565447286e-05,
+      "loss": 0.2999,
+      "step": 8538
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.273621133546292e-05,
+      "loss": 0.3134,
+      "step": 8539
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.272955362080127e-05,
+      "loss": 0.32,
+      "step": 8540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.272289542207859e-05,
+      "loss": 0.2962,
+      "step": 8541
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.271623673991118e-05,
+      "loss": 0.312,
+      "step": 8542
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2709577574915386e-05,
+      "loss": 0.2811,
+      "step": 8543
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.270291792770759e-05,
+      "loss": 0.313,
+      "step": 8544
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.269625779890423e-05,
+      "loss": 0.2968,
+      "step": 8545
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2689597189121794e-05,
+      "loss": 0.2887,
+      "step": 8546
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.26829360989768e-05,
+      "loss": 0.3152,
+      "step": 8547
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.267627452908581e-05,
+      "loss": 0.3023,
+      "step": 8548
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.266961248006544e-05,
+      "loss": 0.2999,
+      "step": 8549
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.266294995253235e-05,
+      "loss": 0.3169,
+      "step": 8550
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.265628694710324e-05,
+      "loss": 0.3009,
+      "step": 8551
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.264962346439485e-05,
+      "loss": 0.3166,
+      "step": 8552
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.264295950502396e-05,
+      "loss": 0.3195,
+      "step": 8553
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.263629506960743e-05,
+      "loss": 0.3082,
+      "step": 8554
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2629630158762115e-05,
+      "loss": 0.3231,
+      "step": 8555
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2622964773104946e-05,
+      "loss": 0.3159,
+      "step": 8556
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2616298913252895e-05,
+      "loss": 0.2906,
+      "step": 8557
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2609632579822954e-05,
+      "loss": 0.325,
+      "step": 8558
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2602965773432185e-05,
+      "loss": 0.2957,
+      "step": 8559
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.259629849469768e-05,
+      "loss": 0.3153,
+      "step": 8560
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2589630744236594e-05,
+      "loss": 0.3103,
+      "step": 8561
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.25829625226661e-05,
+      "loss": 0.3098,
+      "step": 8562
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2576293830603425e-05,
+      "loss": 0.3251,
+      "step": 8563
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.256962466866584e-05,
+      "loss": 0.3201,
+      "step": 8564
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.256295503747066e-05,
+      "loss": 0.3028,
+      "step": 8565
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.255628493763526e-05,
+      "loss": 0.2982,
+      "step": 8566
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.254961436977701e-05,
+      "loss": 0.3098,
+      "step": 8567
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.254294333451338e-05,
+      "loss": 0.3136,
+      "step": 8568
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.253627183246186e-05,
+      "loss": 0.2983,
+      "step": 8569
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.252959986423996e-05,
+      "loss": 0.301,
+      "step": 8570
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2522927430465275e-05,
+      "loss": 0.312,
+      "step": 8571
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.251625453175541e-05,
+      "loss": 0.3146,
+      "step": 8572
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2509581168728034e-05,
+      "loss": 0.3131,
+      "step": 8573
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.250290734200085e-05,
+      "loss": 0.3233,
+      "step": 8574
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.24962330521916e-05,
+      "loss": 0.2998,
+      "step": 8575
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.248955829991807e-05,
+      "loss": 0.3166,
+      "step": 8576
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.24828830857981e-05,
+      "loss": 0.3178,
+      "step": 8577
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.247620741044957e-05,
+      "loss": 0.3142,
+      "step": 8578
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.246953127449039e-05,
+      "loss": 0.2984,
+      "step": 8579
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2462854678538515e-05,
+      "loss": 0.3158,
+      "step": 8580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.245617762321196e-05,
+      "loss": 0.3028,
+      "step": 8581
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.244950010912876e-05,
+      "loss": 0.3145,
+      "step": 8582
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.244282213690701e-05,
+      "loss": 0.2914,
+      "step": 8583
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2436143707164847e-05,
+      "loss": 0.3091,
+      "step": 8584
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.242946482052043e-05,
+      "loss": 0.3181,
+      "step": 8585
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2422785477591975e-05,
+      "loss": 0.3207,
+      "step": 8586
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.241610567899774e-05,
+      "loss": 0.3078,
+      "step": 8587
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.240942542535604e-05,
+      "loss": 0.3244,
+      "step": 8588
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.24027447172852e-05,
+      "loss": 0.295,
+      "step": 8589
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.239606355540361e-05,
+      "loss": 0.327,
+      "step": 8590
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.238938194032969e-05,
+      "loss": 0.3019,
+      "step": 8591
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.238269987268191e-05,
+      "loss": 0.3108,
+      "step": 8592
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.237601735307879e-05,
+      "loss": 0.3104,
+      "step": 8593
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2369334382138855e-05,
+      "loss": 0.3052,
+      "step": 8594
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2362650960480736e-05,
+      "loss": 0.308,
+      "step": 8595
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.235596708872304e-05,
+      "loss": 0.3184,
+      "step": 8596
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.234928276748444e-05,
+      "loss": 0.3068,
+      "step": 8597
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.234259799738368e-05,
+      "loss": 0.3152,
+      "step": 8598
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2335912779039494e-05,
+      "loss": 0.2954,
+      "step": 8599
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2329227113070695e-05,
+      "loss": 0.3189,
+      "step": 8600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.232254100009613e-05,
+      "loss": 0.3041,
+      "step": 8601
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.231585444073467e-05,
+      "loss": 0.2869,
+      "step": 8602
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2309167435605255e-05,
+      "loss": 0.3176,
+      "step": 8603
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.230247998532683e-05,
+      "loss": 0.3228,
+      "step": 8604
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2295792090518435e-05,
+      "loss": 0.2975,
+      "step": 8605
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.228910375179909e-05,
+      "loss": 0.3185,
+      "step": 8606
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2282414969787895e-05,
+      "loss": 0.3,
+      "step": 8607
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2275725745103986e-05,
+      "loss": 0.3061,
+      "step": 8608
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.226903607836653e-05,
+      "loss": 0.3154,
+      "step": 8609
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.226234597019474e-05,
+      "loss": 0.3217,
+      "step": 8610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.225565542120786e-05,
+      "loss": 0.3121,
+      "step": 8611
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.224896443202521e-05,
+      "loss": 0.3086,
+      "step": 8612
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.22422730032661e-05,
+      "loss": 0.3049,
+      "step": 8613
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.223558113554992e-05,
+      "loss": 0.3184,
+      "step": 8614
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.222888882949608e-05,
+      "loss": 0.3128,
+      "step": 8615
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.222219608572404e-05,
+      "loss": 0.3196,
+      "step": 8616
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2215502904853296e-05,
+      "loss": 0.316,
+      "step": 8617
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2208809287503386e-05,
+      "loss": 0.315,
+      "step": 8618
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2202115234293896e-05,
+      "loss": 0.316,
+      "step": 8619
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.219542074584444e-05,
+      "loss": 0.3024,
+      "step": 8620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2188725822774677e-05,
+      "loss": 0.3115,
+      "step": 8621
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.218203046570429e-05,
+      "loss": 0.3195,
+      "step": 8622
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2175334675253056e-05,
+      "loss": 0.2939,
+      "step": 8623
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2168638452040714e-05,
+      "loss": 0.3086,
+      "step": 8624
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.216194179668711e-05,
+      "loss": 0.2856,
+      "step": 8625
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.215524470981209e-05,
+      "loss": 0.3232,
+      "step": 8626
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.214854719203558e-05,
+      "loss": 0.3006,
+      "step": 8627
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.2141849243977485e-05,
+      "loss": 0.3186,
+      "step": 8628
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.21351508662578e-05,
+      "loss": 0.3192,
+      "step": 8629
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.212845205949655e-05,
+      "loss": 0.3137,
+      "step": 8630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.212175282431378e-05,
+      "loss": 0.2887,
+      "step": 8631
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.211505316132959e-05,
+      "loss": 0.3214,
+      "step": 8632
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.210835307116414e-05,
+      "loss": 0.2946,
+      "step": 8633
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.210165255443758e-05,
+      "loss": 0.3111,
+      "step": 8634
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.209495161177014e-05,
+      "loss": 0.2953,
+      "step": 8635
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.208825024378208e-05,
+      "loss": 0.3113,
+      "step": 8636
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.2081548451093683e-05,
+      "loss": 0.3059,
+      "step": 8637
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.207484623432531e-05,
+      "loss": 0.2979,
+      "step": 8638
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.20681435940973e-05,
+      "loss": 0.3025,
+      "step": 8639
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.206144053103009e-05,
+      "loss": 0.3313,
+      "step": 8640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.205473704574413e-05,
+      "loss": 0.3198,
+      "step": 8641
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.204803313885991e-05,
+      "loss": 0.3058,
+      "step": 8642
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.204132881099796e-05,
+      "loss": 0.3113,
+      "step": 8643
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.203462406277885e-05,
+      "loss": 0.3114,
+      "step": 8644
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.202791889482319e-05,
+      "loss": 0.3049,
+      "step": 8645
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.202121330775163e-05,
+      "loss": 0.3011,
+      "step": 8646
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.2014507302184854e-05,
+      "loss": 0.3118,
+      "step": 8647
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.200780087874358e-05,
+      "loss": 0.3047,
+      "step": 8648
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.200109403804859e-05,
+      "loss": 0.3021,
+      "step": 8649
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.199438678072068e-05,
+      "loss": 0.3115,
+      "step": 8650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.198767910738067e-05,
+      "loss": 0.2874,
+      "step": 8651
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.198097101864946e-05,
+      "loss": 0.3063,
+      "step": 8652
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.197426251514797e-05,
+      "loss": 0.309,
+      "step": 8653
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.196755359749715e-05,
+      "loss": 0.3028,
+      "step": 8654
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.196084426631799e-05,
+      "loss": 0.3045,
+      "step": 8655
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1954134522231536e-05,
+      "loss": 0.3016,
+      "step": 8656
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.194742436585885e-05,
+      "loss": 0.2886,
+      "step": 8657
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1940713797821044e-05,
+      "loss": 0.3095,
+      "step": 8658
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.193400281873927e-05,
+      "loss": 0.3075,
+      "step": 8659
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.192729142923469e-05,
+      "loss": 0.3141,
+      "step": 8660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.192057962992857e-05,
+      "loss": 0.3125,
+      "step": 8661
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1913867421442126e-05,
+      "loss": 0.3113,
+      "step": 8662
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1907154804396696e-05,
+      "loss": 0.3059,
+      "step": 8663
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.190044177941359e-05,
+      "loss": 0.2903,
+      "step": 8664
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1893728347114196e-05,
+      "loss": 0.3058,
+      "step": 8665
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.188701450811993e-05,
+      "loss": 0.3191,
+      "step": 8666
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1880300263052235e-05,
+      "loss": 0.2984,
+      "step": 8667
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.187358561253259e-05,
+      "loss": 0.3166,
+      "step": 8668
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1866870557182545e-05,
+      "loss": 0.3091,
+      "step": 8669
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.186015509762363e-05,
+      "loss": 0.2972,
+      "step": 8670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1853439234477486e-05,
+      "loss": 0.3105,
+      "step": 8671
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.184672296836572e-05,
+      "loss": 0.3013,
+      "step": 8672
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.184000629991001e-05,
+      "loss": 0.314,
+      "step": 8673
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1833289229732075e-05,
+      "loss": 0.3106,
+      "step": 8674
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.182657175845367e-05,
+      "loss": 0.3251,
+      "step": 8675
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1819853886696574e-05,
+      "loss": 0.304,
+      "step": 8676
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1813135615082615e-05,
+      "loss": 0.2866,
+      "step": 8677
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1806416944233635e-05,
+      "loss": 0.3022,
+      "step": 8678
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.179969787477156e-05,
+      "loss": 0.3155,
+      "step": 8679
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.179297840731831e-05,
+      "loss": 0.3337,
+      "step": 8680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1786258542495854e-05,
+      "loss": 0.3013,
+      "step": 8681
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.177953828092621e-05,
+      "loss": 0.3098,
+      "step": 8682
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.177281762323141e-05,
+      "loss": 0.3153,
+      "step": 8683
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1766096570033546e-05,
+      "loss": 0.3027,
+      "step": 8684
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1759375121954724e-05,
+      "loss": 0.3096,
+      "step": 8685
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.175265327961711e-05,
+      "loss": 0.3122,
+      "step": 8686
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1745931043642893e-05,
+      "loss": 0.3067,
+      "step": 8687
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.17392084146543e-05,
+      "loss": 0.2914,
+      "step": 8688
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.173248539327359e-05,
+      "loss": 0.3094,
+      "step": 8689
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1725761980123074e-05,
+      "loss": 0.3166,
+      "step": 8690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.171903817582508e-05,
+      "loss": 0.3057,
+      "step": 8691
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.171231398100198e-05,
+      "loss": 0.3093,
+      "step": 8692
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.170558939627619e-05,
+      "loss": 0.322,
+      "step": 8693
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.169886442227015e-05,
+      "loss": 0.3101,
+      "step": 8694
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1692139059606347e-05,
+      "loss": 0.307,
+      "step": 8695
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.168541330890729e-05,
+      "loss": 0.3199,
+      "step": 8696
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.167868717079553e-05,
+      "loss": 0.2971,
+      "step": 8697
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.167196064589366e-05,
+      "loss": 0.296,
+      "step": 8698
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1665233734824304e-05,
+      "loss": 0.308,
+      "step": 8699
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.165850643821013e-05,
+      "loss": 0.2971,
+      "step": 8700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1651778756673824e-05,
+      "loss": 0.3174,
+      "step": 8701
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1645050690838116e-05,
+      "loss": 0.3073,
+      "step": 8702
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1638322241325784e-05,
+      "loss": 0.305,
+      "step": 8703
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.163159340875963e-05,
+      "loss": 0.3197,
+      "step": 8704
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.162486419376248e-05,
+      "loss": 0.3051,
+      "step": 8705
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.161813459695722e-05,
+      "loss": 0.313,
+      "step": 8706
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.161140461896674e-05,
+      "loss": 0.3004,
+      "step": 8707
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.160467426041401e-05,
+      "loss": 0.2974,
+      "step": 8708
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1597943521922e-05,
+      "loss": 0.3012,
+      "step": 8709
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.159121240411371e-05,
+      "loss": 0.3126,
+      "step": 8710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1584480907612216e-05,
+      "loss": 0.3135,
+      "step": 8711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1577749033040575e-05,
+      "loss": 0.3089,
+      "step": 8712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1571016781021925e-05,
+      "loss": 0.2981,
+      "step": 8713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.156428415217942e-05,
+      "loss": 0.2956,
+      "step": 8714
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.155755114713624e-05,
+      "loss": 0.3199,
+      "step": 8715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.155081776651562e-05,
+      "loss": 0.3101,
+      "step": 8716
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1544084010940804e-05,
+      "loss": 0.3243,
+      "step": 8717
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.15373498810351e-05,
+      "loss": 0.3014,
+      "step": 8718
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1530615377421835e-05,
+      "loss": 0.3129,
+      "step": 8719
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1523880500724364e-05,
+      "loss": 0.3093,
+      "step": 8720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1517145251566095e-05,
+      "loss": 0.3276,
+      "step": 8721
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.151040963057045e-05,
+      "loss": 0.3142,
+      "step": 8722
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1503673638360906e-05,
+      "loss": 0.2897,
+      "step": 8723
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1496937275560954e-05,
+      "loss": 0.2945,
+      "step": 8724
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1490200542794134e-05,
+      "loss": 0.3009,
+      "step": 8725
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1483463440684025e-05,
+      "loss": 0.2793,
+      "step": 8726
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.147672596985421e-05,
+      "loss": 0.3254,
+      "step": 8727
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.146998813092834e-05,
+      "loss": 0.3021,
+      "step": 8728
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.146324992453008e-05,
+      "loss": 0.2914,
+      "step": 8729
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.145651135128315e-05,
+      "loss": 0.326,
+      "step": 8730
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1449772411811266e-05,
+      "loss": 0.3324,
+      "step": 8731
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.144303310673822e-05,
+      "loss": 0.3096,
+      "step": 8732
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1436293436687826e-05,
+      "loss": 0.322,
+      "step": 8733
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1429553402283905e-05,
+      "loss": 0.3083,
+      "step": 8734
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.142281300415034e-05,
+      "loss": 0.3284,
+      "step": 8735
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.141607224291105e-05,
+      "loss": 0.3095,
+      "step": 8736
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.140933111918995e-05,
+      "loss": 0.3214,
+      "step": 8737
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.140258963361106e-05,
+      "loss": 0.3227,
+      "step": 8738
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.139584778679835e-05,
+      "loss": 0.32,
+      "step": 8739
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.138910557937588e-05,
+      "loss": 0.3187,
+      "step": 8740
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.138236301196771e-05,
+      "loss": 0.3064,
+      "step": 8741
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.137562008519798e-05,
+      "loss": 0.3016,
+      "step": 8742
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.136887679969081e-05,
+      "loss": 0.301,
+      "step": 8743
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1362133156070374e-05,
+      "loss": 0.3137,
+      "step": 8744
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1355389154960894e-05,
+      "loss": 0.2897,
+      "step": 8745
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.13486447969866e-05,
+      "loss": 0.3068,
+      "step": 8746
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.134190008277179e-05,
+      "loss": 0.3047,
+      "step": 8747
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1335155012940734e-05,
+      "loss": 0.3022,
+      "step": 8748
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.132840958811781e-05,
+      "loss": 0.3175,
+      "step": 8749
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1321663808927375e-05,
+      "loss": 0.3113,
+      "step": 8750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1314917675993834e-05,
+      "loss": 0.3073,
+      "step": 8751
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.130817118994163e-05,
+      "loss": 0.3268,
+      "step": 8752
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.130142435139524e-05,
+      "loss": 0.2996,
+      "step": 8753
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1294677160979155e-05,
+      "loss": 0.3019,
+      "step": 8754
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1287929619317926e-05,
+      "loss": 0.3293,
+      "step": 8755
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1281181727036117e-05,
+      "loss": 0.3216,
+      "step": 8756
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.127443348475834e-05,
+      "loss": 0.2923,
+      "step": 8757
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.126768489310923e-05,
+      "loss": 0.3224,
+      "step": 8758
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1260935952713426e-05,
+      "loss": 0.3033,
+      "step": 8759
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.125418666419566e-05,
+      "loss": 0.3067,
+      "step": 8760
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.124743702818064e-05,
+      "loss": 0.313,
+      "step": 8761
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.124068704529314e-05,
+      "loss": 0.3182,
+      "step": 8762
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1233936716157966e-05,
+      "loss": 0.2981,
+      "step": 8763
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1227186041399935e-05,
+      "loss": 0.3048,
+      "step": 8764
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1220435021643914e-05,
+      "loss": 0.3108,
+      "step": 8765
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.121368365751478e-05,
+      "loss": 0.3014,
+      "step": 8766
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.120693194963747e-05,
+      "loss": 0.3076,
+      "step": 8767
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.120017989863694e-05,
+      "loss": 0.308,
+      "step": 8768
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.119342750513816e-05,
+      "loss": 0.306,
+      "step": 8769
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1186674769766175e-05,
+      "loss": 0.3119,
+      "step": 8770
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.117992169314602e-05,
+      "loss": 0.3145,
+      "step": 8771
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1173168275902776e-05,
+      "loss": 0.3186,
+      "step": 8772
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.116641451866157e-05,
+      "loss": 0.3055,
+      "step": 8773
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1159660422047525e-05,
+      "loss": 0.2867,
+      "step": 8774
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1152905986685836e-05,
+      "loss": 0.3257,
+      "step": 8775
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.114615121320172e-05,
+      "loss": 0.3014,
+      "step": 8776
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.113939610222039e-05,
+      "loss": 0.3187,
+      "step": 8777
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.113264065436712e-05,
+      "loss": 0.3277,
+      "step": 8778
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.112588487026723e-05,
+      "loss": 0.3306,
+      "step": 8779
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.111912875054605e-05,
+      "loss": 0.3331,
+      "step": 8780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.111237229582892e-05,
+      "loss": 0.2878,
+      "step": 8781
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.110561550674127e-05,
+      "loss": 0.3128,
+      "step": 8782
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1098858383908495e-05,
+      "loss": 0.3052,
+      "step": 8783
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1092100927956076e-05,
+      "loss": 0.3185,
+      "step": 8784
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.108534313950948e-05,
+      "loss": 0.2928,
+      "step": 8785
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1078585019194244e-05,
+      "loss": 0.3128,
+      "step": 8786
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1071826567635905e-05,
+      "loss": 0.2931,
+      "step": 8787
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.106506778546004e-05,
+      "loss": 0.3257,
+      "step": 8788
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1058308673292276e-05,
+      "loss": 0.2932,
+      "step": 8789
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.105154923175823e-05,
+      "loss": 0.2996,
+      "step": 8790
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1044789461483595e-05,
+      "loss": 0.3195,
+      "step": 8791
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1038029363094065e-05,
+      "loss": 0.3042,
+      "step": 8792
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.103126893721537e-05,
+      "loss": 0.3203,
+      "step": 8793
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1024508184473275e-05,
+      "loss": 0.3028,
+      "step": 8794
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.101774710549357e-05,
+      "loss": 0.3173,
+      "step": 8795
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.101098570090208e-05,
+      "loss": 0.3023,
+      "step": 8796
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.1004223971324665e-05,
+      "loss": 0.3105,
+      "step": 8797
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.09974619173872e-05,
+      "loss": 0.3033,
+      "step": 8798
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.0990699539715596e-05,
+      "loss": 0.3038,
+      "step": 8799
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.09839368389358e-05,
+      "loss": 0.2806,
+      "step": 8800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.097717381567378e-05,
+      "loss": 0.3172,
+      "step": 8801
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.097041047055555e-05,
+      "loss": 0.3008,
+      "step": 8802
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.096364680420713e-05,
+      "loss": 0.3136,
+      "step": 8803
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.095688281725461e-05,
+      "loss": 0.3237,
+      "step": 8804
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.095011851032403e-05,
+      "loss": 0.3118,
+      "step": 8805
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.094335388404156e-05,
+      "loss": 0.3147,
+      "step": 8806
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.093658893903334e-05,
+      "loss": 0.3006,
+      "step": 8807
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.092982367592554e-05,
+      "loss": 0.3049,
+      "step": 8808
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0923058095344374e-05,
+      "loss": 0.3034,
+      "step": 8809
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.091629219791608e-05,
+      "loss": 0.3096,
+      "step": 8810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.090952598426694e-05,
+      "loss": 0.3057,
+      "step": 8811
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.090275945502323e-05,
+      "loss": 0.3183,
+      "step": 8812
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0895992610811296e-05,
+      "loss": 0.3113,
+      "step": 8813
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.088922545225749e-05,
+      "loss": 0.3119,
+      "step": 8814
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0882457979988196e-05,
+      "loss": 0.3123,
+      "step": 8815
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0875690194629834e-05,
+      "loss": 0.2958,
+      "step": 8816
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.086892209680883e-05,
+      "loss": 0.3182,
+      "step": 8817
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.086215368715167e-05,
+      "loss": 0.2934,
+      "step": 8818
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0855384966284865e-05,
+      "loss": 0.3125,
+      "step": 8819
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0848615934834935e-05,
+      "loss": 0.3177,
+      "step": 8820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.084184659342843e-05,
+      "loss": 0.3072,
+      "step": 8821
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.083507694269196e-05,
+      "loss": 0.3011,
+      "step": 8822
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0828306983252116e-05,
+      "loss": 0.2938,
+      "step": 8823
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0821536715735564e-05,
+      "loss": 0.3041,
+      "step": 8824
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0814766140768957e-05,
+      "loss": 0.3081,
+      "step": 8825
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.080799525897902e-05,
+      "loss": 0.3289,
+      "step": 8826
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.080122407099247e-05,
+      "loss": 0.3137,
+      "step": 8827
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0794452577436075e-05,
+      "loss": 0.3046,
+      "step": 8828
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.07876807789366e-05,
+      "loss": 0.3116,
+      "step": 8829
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.078090867612089e-05,
+      "loss": 0.3097,
+      "step": 8830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.077413626961576e-05,
+      "loss": 0.3111,
+      "step": 8831
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.076736356004811e-05,
+      "loss": 0.3095,
+      "step": 8832
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0760590548044816e-05,
+      "loss": 0.3175,
+      "step": 8833
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0753817234232816e-05,
+      "loss": 0.3178,
+      "step": 8834
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.074704361923906e-05,
+      "loss": 0.3098,
+      "step": 8835
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.074026970369055e-05,
+      "loss": 0.3021,
+      "step": 8836
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.073349548821426e-05,
+      "loss": 0.2927,
+      "step": 8837
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.072672097343726e-05,
+      "loss": 0.3006,
+      "step": 8838
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.071994615998662e-05,
+      "loss": 0.3055,
+      "step": 8839
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.071317104848941e-05,
+      "loss": 0.3263,
+      "step": 8840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.070639563957277e-05,
+      "loss": 0.31,
+      "step": 8841
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0699619933863825e-05,
+      "loss": 0.311,
+      "step": 8842
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.069284393198979e-05,
+      "loss": 0.3134,
+      "step": 8843
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.068606763457784e-05,
+      "loss": 0.3249,
+      "step": 8844
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.067929104225522e-05,
+      "loss": 0.2917,
+      "step": 8845
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.067251415564919e-05,
+      "loss": 0.2939,
+      "step": 8846
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.066573697538702e-05,
+      "loss": 0.3237,
+      "step": 8847
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.065895950209604e-05,
+      "loss": 0.3098,
+      "step": 8848
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.065218173640359e-05,
+      "loss": 0.3243,
+      "step": 8849
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.064540367893703e-05,
+      "loss": 0.2964,
+      "step": 8850
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.063862533032375e-05,
+      "loss": 0.3152,
+      "step": 8851
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.063184669119119e-05,
+      "loss": 0.2881,
+      "step": 8852
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0625067762166774e-05,
+      "loss": 0.2938,
+      "step": 8853
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0618288543878e-05,
+      "loss": 0.3062,
+      "step": 8854
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0611509036952356e-05,
+      "loss": 0.316,
+      "step": 8855
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0604729242017386e-05,
+      "loss": 0.3071,
+      "step": 8856
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0597949159700616e-05,
+      "loss": 0.3201,
+      "step": 8857
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.059116879062966e-05,
+      "loss": 0.3112,
+      "step": 8858
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.058438813543211e-05,
+      "loss": 0.3012,
+      "step": 8859
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.05776071947356e-05,
+      "loss": 0.3077,
+      "step": 8860
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.05708259691678e-05,
+      "loss": 0.3067,
+      "step": 8861
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.056404445935639e-05,
+      "loss": 0.2996,
+      "step": 8862
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.055726266592909e-05,
+      "loss": 0.303,
+      "step": 8863
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.055048058951364e-05,
+      "loss": 0.3152,
+      "step": 8864
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.05436982307378e-05,
+      "loss": 0.2903,
+      "step": 8865
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.053691559022937e-05,
+      "loss": 0.303,
+      "step": 8866
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0530132668616165e-05,
+      "loss": 0.3058,
+      "step": 8867
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.052334946652603e-05,
+      "loss": 0.3045,
+      "step": 8868
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0516565984586825e-05,
+      "loss": 0.3158,
+      "step": 8869
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.050978222342647e-05,
+      "loss": 0.3333,
+      "step": 8870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0502998183672878e-05,
+      "loss": 0.3313,
+      "step": 8871
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0496213865953993e-05,
+      "loss": 0.2788,
+      "step": 8872
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0489429270897785e-05,
+      "loss": 0.2967,
+      "step": 8873
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.048264439913226e-05,
+      "loss": 0.3051,
+      "step": 8874
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0475859251285448e-05,
+      "loss": 0.3107,
+      "step": 8875
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0469073827985394e-05,
+      "loss": 0.3197,
+      "step": 8876
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.046228812986017e-05,
+      "loss": 0.3113,
+      "step": 8877
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.045550215753789e-05,
+      "loss": 0.3124,
+      "step": 8878
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.044871591164667e-05,
+      "loss": 0.3041,
+      "step": 8879
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0441929392814667e-05,
+      "loss": 0.2875,
+      "step": 8880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0435142601670058e-05,
+      "loss": 0.2931,
+      "step": 8881
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0428355538841048e-05,
+      "loss": 0.2889,
+      "step": 8882
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0421568204955857e-05,
+      "loss": 0.3088,
+      "step": 8883
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.041478060064275e-05,
+      "loss": 0.3151,
+      "step": 8884
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.040799272653e-05,
+      "loss": 0.3118,
+      "step": 8885
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0401204583245905e-05,
+      "loss": 0.3457,
+      "step": 8886
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0394416171418806e-05,
+      "loss": 0.3386,
+      "step": 8887
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.038762749167705e-05,
+      "loss": 0.3065,
+      "step": 8888
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0380838544649005e-05,
+      "loss": 0.3229,
+      "step": 8889
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.037404933096308e-05,
+      "loss": 0.3014,
+      "step": 8890
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.036725985124771e-05,
+      "loss": 0.3135,
+      "step": 8891
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0360470106131334e-05,
+      "loss": 0.3158,
+      "step": 8892
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0353680096242437e-05,
+      "loss": 0.2938,
+      "step": 8893
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0346889822209524e-05,
+      "loss": 0.3154,
+      "step": 8894
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.034009928466111e-05,
+      "loss": 0.3193,
+      "step": 8895
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0333308484225742e-05,
+      "loss": 0.3096,
+      "step": 8896
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.032651742153201e-05,
+      "loss": 0.3091,
+      "step": 8897
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.03197260972085e-05,
+      "loss": 0.3079,
+      "step": 8898
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0312934511883835e-05,
+      "loss": 0.2976,
+      "step": 8899
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0306142666186672e-05,
+      "loss": 0.2844,
+      "step": 8900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0299350560745664e-05,
+      "loss": 0.3033,
+      "step": 8901
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.029255819618953e-05,
+      "loss": 0.3007,
+      "step": 8902
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.028576557314697e-05,
+      "loss": 0.2943,
+      "step": 8903
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0278972692246725e-05,
+      "loss": 0.3201,
+      "step": 8904
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.027217955411758e-05,
+      "loss": 0.3113,
+      "step": 8905
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.02653861593883e-05,
+      "loss": 0.3212,
+      "step": 8906
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0258592508687727e-05,
+      "loss": 0.2962,
+      "step": 8907
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0251798602644672e-05,
+      "loss": 0.3026,
+      "step": 8908
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.024500444188802e-05,
+      "loss": 0.3142,
+      "step": 8909
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.023821002704664e-05,
+      "loss": 0.2971,
+      "step": 8910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0231415358749456e-05,
+      "loss": 0.3127,
+      "step": 8911
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0224620437625383e-05,
+      "loss": 0.2947,
+      "step": 8912
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.021782526430338e-05,
+      "loss": 0.2942,
+      "step": 8913
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0211029839412434e-05,
+      "loss": 0.3024,
+      "step": 8914
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.020423416358154e-05,
+      "loss": 0.3199,
+      "step": 8915
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0197438237439727e-05,
+      "loss": 0.3232,
+      "step": 8916
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0190642061616053e-05,
+      "loss": 0.3158,
+      "step": 8917
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0183845636739573e-05,
+      "loss": 0.2988,
+      "step": 8918
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0177048963439385e-05,
+      "loss": 0.3119,
+      "step": 8919
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.017025204234461e-05,
+      "loss": 0.3003,
+      "step": 8920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0163454874084396e-05,
+      "loss": 0.3115,
+      "step": 8921
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0156657459287896e-05,
+      "loss": 0.3182,
+      "step": 8922
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0149859798584294e-05,
+      "loss": 0.3026,
+      "step": 8923
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0143061892602808e-05,
+      "loss": 0.3165,
+      "step": 8924
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0136263741972666e-05,
+      "loss": 0.3279,
+      "step": 8925
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.012946534732312e-05,
+      "loss": 0.3208,
+      "step": 8926
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0122666709283458e-05,
+      "loss": 0.3065,
+      "step": 8927
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.011586782848297e-05,
+      "loss": 0.2926,
+      "step": 8928
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.010906870555097e-05,
+      "loss": 0.3187,
+      "step": 8929
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0102269341116817e-05,
+      "loss": 0.3085,
+      "step": 8930
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0095469735809873e-05,
+      "loss": 0.2965,
+      "step": 8931
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0088669890259536e-05,
+      "loss": 0.3151,
+      "step": 8932
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0081869805095193e-05,
+      "loss": 0.3039,
+      "step": 8933
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0075069480946295e-05,
+      "loss": 0.2904,
+      "step": 8934
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.00682689184423e-05,
+      "loss": 0.2941,
+      "step": 8935
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0061468118212676e-05,
+      "loss": 0.3186,
+      "step": 8936
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0054667080886932e-05,
+      "loss": 0.2993,
+      "step": 8937
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0047865807094583e-05,
+      "loss": 0.2813,
+      "step": 8938
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.004106429746517e-05,
+      "loss": 0.3358,
+      "step": 8939
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0034262552628267e-05,
+      "loss": 0.3018,
+      "step": 8940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.002746057321346e-05,
+      "loss": 0.301,
+      "step": 8941
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.002065835985036e-05,
+      "loss": 0.2909,
+      "step": 8942
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0013855913168576e-05,
+      "loss": 0.2946,
+      "step": 8943
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0007053233797794e-05,
+      "loss": 0.3085,
+      "step": 8944
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0000250322367656e-05,
+      "loss": 0.291,
+      "step": 8945
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9993447179507883e-05,
+      "loss": 0.3102,
+      "step": 8946
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9986643805848175e-05,
+      "loss": 0.3212,
+      "step": 8947
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9979840202018266e-05,
+      "loss": 0.3157,
+      "step": 8948
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9973036368647933e-05,
+      "loss": 0.3224,
+      "step": 8949
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9966232306366944e-05,
+      "loss": 0.322,
+      "step": 8950
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.995942801580511e-05,
+      "loss": 0.3035,
+      "step": 8951
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.995262349759224e-05,
+      "loss": 0.3306,
+      "step": 8952
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9945818752358184e-05,
+      "loss": 0.3145,
+      "step": 8953
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9939013780732817e-05,
+      "loss": 0.2958,
+      "step": 8954
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9932208583346007e-05,
+      "loss": 0.3193,
+      "step": 8955
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9925403160827682e-05,
+      "loss": 0.3129,
+      "step": 8956
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9918597513807743e-05,
+      "loss": 0.3266,
+      "step": 8957
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.991179164291616e-05,
+      "loss": 0.3228,
+      "step": 8958
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.99049855487829e-05,
+      "loss": 0.292,
+      "step": 8959
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9898179232037937e-05,
+      "loss": 0.2875,
+      "step": 8960
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9891372693311297e-05,
+      "loss": 0.3076,
+      "step": 8961
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9884565933233e-05,
+      "loss": 0.298,
+      "step": 8962
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.987775895243311e-05,
+      "loss": 0.3134,
+      "step": 8963
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9870951751541694e-05,
+      "loss": 0.3118,
+      "step": 8964
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9864144331188837e-05,
+      "loss": 0.3048,
+      "step": 8965
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9857336692004655e-05,
+      "loss": 0.3036,
+      "step": 8966
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9850528834619285e-05,
+      "loss": 0.3236,
+      "step": 8967
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.984372075966288e-05,
+      "loss": 0.3177,
+      "step": 8968
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.98369124677656e-05,
+      "loss": 0.311,
+      "step": 8969
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9830103959557655e-05,
+      "loss": 0.2885,
+      "step": 8970
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9823295235669258e-05,
+      "loss": 0.2951,
+      "step": 8971
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9816486296730624e-05,
+      "loss": 0.3134,
+      "step": 8972
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9809677143372023e-05,
+      "loss": 0.2986,
+      "step": 8973
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9802867776223716e-05,
+      "loss": 0.3159,
+      "step": 8974
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9796058195916006e-05,
+      "loss": 0.3278,
+      "step": 8975
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9789248403079202e-05,
+      "loss": 0.3096,
+      "step": 8976
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.9782438398343632e-05,
+      "loss": 0.3083,
+      "step": 8977
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.977562818233966e-05,
+      "loss": 0.3129,
+      "step": 8978
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.976881775569764e-05,
+      "loss": 0.319,
+      "step": 8979
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9762007119047973e-05,
+      "loss": 0.3039,
+      "step": 8980
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.975519627302107e-05,
+      "loss": 0.3085,
+      "step": 8981
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9748385218247353e-05,
+      "loss": 0.2858,
+      "step": 8982
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9741573955357283e-05,
+      "loss": 0.3038,
+      "step": 8983
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9734762484981316e-05,
+      "loss": 0.286,
+      "step": 8984
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9727950807749945e-05,
+      "loss": 0.3147,
+      "step": 8985
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9721138924293675e-05,
+      "loss": 0.3049,
+      "step": 8986
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9714326835243027e-05,
+      "loss": 0.3097,
+      "step": 8987
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.970751454122856e-05,
+      "loss": 0.3095,
+      "step": 8988
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9700702042880817e-05,
+      "loss": 0.3061,
+      "step": 8989
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9693889340830405e-05,
+      "loss": 0.3032,
+      "step": 8990
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.968707643570791e-05,
+      "loss": 0.3063,
+      "step": 8991
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.968026332814395e-05,
+      "loss": 0.3051,
+      "step": 8992
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9673450018769166e-05,
+      "loss": 0.3064,
+      "step": 8993
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.966663650821422e-05,
+      "loss": 0.2978,
+      "step": 8994
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9659822797109795e-05,
+      "loss": 0.2924,
+      "step": 8995
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9653008886086564e-05,
+      "loss": 0.2947,
+      "step": 8996
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.964619477577526e-05,
+      "loss": 0.3126,
+      "step": 8997
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9639380466806604e-05,
+      "loss": 0.2986,
+      "step": 8998
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9632565959811355e-05,
+      "loss": 0.308,
+      "step": 8999
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.962575125542027e-05,
+      "loss": 0.2834,
+      "step": 9000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9618936354264145e-05,
+      "loss": 0.3155,
+      "step": 9001
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9612121256973785e-05,
+      "loss": 0.3175,
+      "step": 9002
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.960530596418001e-05,
+      "loss": 0.2864,
+      "step": 9003
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.959849047651366e-05,
+      "loss": 0.2919,
+      "step": 9004
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.959167479460559e-05,
+      "loss": 0.2856,
+      "step": 9005
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9584858919086683e-05,
+      "loss": 0.2969,
+      "step": 9006
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9578042850587838e-05,
+      "loss": 0.3098,
+      "step": 9007
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9571226589739964e-05,
+      "loss": 0.2857,
+      "step": 9008
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9564410137173986e-05,
+      "loss": 0.3195,
+      "step": 9009
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9557593493520867e-05,
+      "loss": 0.3132,
+      "step": 9010
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.955077665941155e-05,
+      "loss": 0.313,
+      "step": 9011
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.954395963547704e-05,
+      "loss": 0.3149,
+      "step": 9012
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.953714242234834e-05,
+      "loss": 0.3205,
+      "step": 9013
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9530325020656453e-05,
+      "loss": 0.2965,
+      "step": 9014
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9523507431032423e-05,
+      "loss": 0.3001,
+      "step": 9015
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.95166896541073e-05,
+      "loss": 0.3074,
+      "step": 9016
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.950987169051216e-05,
+      "loss": 0.3112,
+      "step": 9017
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.950305354087809e-05,
+      "loss": 0.3215,
+      "step": 9018
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9496235205836192e-05,
+      "loss": 0.3099,
+      "step": 9019
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9489416686017594e-05,
+      "loss": 0.2942,
+      "step": 9020
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9482597982053434e-05,
+      "loss": 0.3107,
+      "step": 9021
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9475779094574873e-05,
+      "loss": 0.3084,
+      "step": 9022
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9468960024213083e-05,
+      "loss": 0.3183,
+      "step": 9023
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9462140771599245e-05,
+      "loss": 0.312,
+      "step": 9024
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9455321337364576e-05,
+      "loss": 0.3,
+      "step": 9025
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9448501722140298e-05,
+      "loss": 0.3182,
+      "step": 9026
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9441681926557656e-05,
+      "loss": 0.3096,
+      "step": 9027
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9434861951247905e-05,
+      "loss": 0.3103,
+      "step": 9028
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.942804179684232e-05,
+      "loss": 0.3104,
+      "step": 9029
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9421221463972186e-05,
+      "loss": 0.3098,
+      "step": 9030
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9414400953268823e-05,
+      "loss": 0.3151,
+      "step": 9031
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9407580265363545e-05,
+      "loss": 0.2992,
+      "step": 9032
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9400759400887696e-05,
+      "loss": 0.3086,
+      "step": 9033
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9393938360472635e-05,
+      "loss": 0.3201,
+      "step": 9034
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.9387117144749726e-05,
+      "loss": 0.2991,
+      "step": 9035
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.938029575435037e-05,
+      "loss": 0.3407,
+      "step": 9036
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9373474189905974e-05,
+      "loss": 0.3145,
+      "step": 9037
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9366652452047938e-05,
+      "loss": 0.2878,
+      "step": 9038
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9359830541407734e-05,
+      "loss": 0.319,
+      "step": 9039
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.935300845861678e-05,
+      "loss": 0.3071,
+      "step": 9040
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9346186204306582e-05,
+      "loss": 0.294,
+      "step": 9041
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.933936377910859e-05,
+      "loss": 0.3008,
+      "step": 9042
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9332541183654327e-05,
+      "loss": 0.3137,
+      "step": 9043
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.932571841857531e-05,
+      "loss": 0.3117,
+      "step": 9044
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9318895484503057e-05,
+      "loss": 0.3132,
+      "step": 9045
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9312072382069132e-05,
+      "loss": 0.2959,
+      "step": 9046
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9305249111905096e-05,
+      "loss": 0.3254,
+      "step": 9047
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9298425674642517e-05,
+      "loss": 0.2987,
+      "step": 9048
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9291602070913003e-05,
+      "loss": 0.3171,
+      "step": 9049
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.928477830134816e-05,
+      "loss": 0.3074,
+      "step": 9050
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9277954366579618e-05,
+      "loss": 0.3022,
+      "step": 9051
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.927113026723901e-05,
+      "loss": 0.3011,
+      "step": 9052
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9264306003958e-05,
+      "loss": 0.3042,
+      "step": 9053
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9257481577368252e-05,
+      "loss": 0.2747,
+      "step": 9054
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.925065698810145e-05,
+      "loss": 0.3259,
+      "step": 9055
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9243832236789314e-05,
+      "loss": 0.2953,
+      "step": 9056
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.923700732406354e-05,
+      "loss": 0.3002,
+      "step": 9057
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.923018225055587e-05,
+      "loss": 0.2863,
+      "step": 9058
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9223357016898058e-05,
+      "loss": 0.3014,
+      "step": 9059
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9216531623721844e-05,
+      "loss": 0.3117,
+      "step": 9060
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.920970607165902e-05,
+      "loss": 0.312,
+      "step": 9061
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.920288036134138e-05,
+      "loss": 0.3146,
+      "step": 9062
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.919605449340071e-05,
+      "loss": 0.3286,
+      "step": 9063
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.918922846846885e-05,
+      "loss": 0.3033,
+      "step": 9064
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9182402287177623e-05,
+      "loss": 0.3034,
+      "step": 9065
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.917557595015889e-05,
+      "loss": 0.3122,
+      "step": 9066
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9168749458044503e-05,
+      "loss": 0.316,
+      "step": 9067
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9161922811466335e-05,
+      "loss": 0.3028,
+      "step": 9068
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9155096011056294e-05,
+      "loss": 0.2995,
+      "step": 9069
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9148269057446275e-05,
+      "loss": 0.3111,
+      "step": 9070
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9141441951268206e-05,
+      "loss": 0.3174,
+      "step": 9071
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9134614693154016e-05,
+      "loss": 0.2935,
+      "step": 9072
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9127787283735658e-05,
+      "loss": 0.3159,
+      "step": 9073
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.912095972364509e-05,
+      "loss": 0.2983,
+      "step": 9074
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.911413201351429e-05,
+      "loss": 0.3006,
+      "step": 9075
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.910730415397525e-05,
+      "loss": 0.3111,
+      "step": 9076
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9100476145659978e-05,
+      "loss": 0.3018,
+      "step": 9077
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.909364798920048e-05,
+      "loss": 0.2969,
+      "step": 9078
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.90868196852288e-05,
+      "loss": 0.2936,
+      "step": 9079
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9079991234376976e-05,
+      "loss": 0.3138,
+      "step": 9080
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9073162637277073e-05,
+      "loss": 0.3247,
+      "step": 9081
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.906633389456116e-05,
+      "loss": 0.2985,
+      "step": 9082
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9059505006861325e-05,
+      "loss": 0.3143,
+      "step": 9083
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.905267597480967e-05,
+      "loss": 0.3102,
+      "step": 9084
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.90458467990383e-05,
+      "loss": 0.3036,
+      "step": 9085
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9039017480179347e-05,
+      "loss": 0.313,
+      "step": 9086
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.903218801886495e-05,
+      "loss": 0.3037,
+      "step": 9087
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.902535841572726e-05,
+      "loss": 0.3189,
+      "step": 9088
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9018528671398443e-05,
+      "loss": 0.3028,
+      "step": 9089
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9011698786510678e-05,
+      "loss": 0.2932,
+      "step": 9090
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.9004868761696163e-05,
+      "loss": 0.3246,
+      "step": 9091
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.899803859758709e-05,
+      "loss": 0.3164,
+      "step": 9092
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.8991208294815685e-05,
+      "loss": 0.3131,
+      "step": 9093
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8984377854014176e-05,
+      "loss": 0.3133,
+      "step": 9094
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8977547275814816e-05,
+      "loss": 0.2925,
+      "step": 9095
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8970716560849846e-05,
+      "loss": 0.3253,
+      "step": 9096
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8963885709751544e-05,
+      "loss": 0.2995,
+      "step": 9097
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.895705472315218e-05,
+      "loss": 0.3089,
+      "step": 9098
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8950223601684063e-05,
+      "loss": 0.3162,
+      "step": 9099
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.894339234597949e-05,
+      "loss": 0.2778,
+      "step": 9100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.893656095667079e-05,
+      "loss": 0.3145,
+      "step": 9101
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8929729434390274e-05,
+      "loss": 0.3309,
+      "step": 9102
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8922897779770307e-05,
+      "loss": 0.3156,
+      "step": 9103
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8916065993443233e-05,
+      "loss": 0.3052,
+      "step": 9104
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8909234076041418e-05,
+      "loss": 0.2914,
+      "step": 9105
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8902402028197258e-05,
+      "loss": 0.3095,
+      "step": 9106
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.889556985054312e-05,
+      "loss": 0.3017,
+      "step": 9107
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8888737543711426e-05,
+      "loss": 0.3038,
+      "step": 9108
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8881905108334588e-05,
+      "loss": 0.3018,
+      "step": 9109
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8875072545045035e-05,
+      "loss": 0.3215,
+      "step": 9110
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8868239854475205e-05,
+      "loss": 0.3087,
+      "step": 9111
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8861407037257544e-05,
+      "loss": 0.3088,
+      "step": 9112
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.885457409402453e-05,
+      "loss": 0.3125,
+      "step": 9113
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.884774102540862e-05,
+      "loss": 0.3114,
+      "step": 9114
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8840907832042313e-05,
+      "loss": 0.3057,
+      "step": 9115
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8834074514558107e-05,
+      "loss": 0.3111,
+      "step": 9116
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.88272410735885e-05,
+      "loss": 0.3314,
+      "step": 9117
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.882040750976603e-05,
+      "loss": 0.3104,
+      "step": 9118
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.881357382372321e-05,
+      "loss": 0.3086,
+      "step": 9119
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8806740016092605e-05,
+      "loss": 0.32,
+      "step": 9120
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8799906087506752e-05,
+      "loss": 0.3219,
+      "step": 9121
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8793072038598228e-05,
+      "loss": 0.3108,
+      "step": 9122
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.87862378699996e-05,
+      "loss": 0.3064,
+      "step": 9123
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8779403582343468e-05,
+      "loss": 0.289,
+      "step": 9124
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8772569176262425e-05,
+      "loss": 0.3069,
+      "step": 9125
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.876573465238908e-05,
+      "loss": 0.3115,
+      "step": 9126
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.875890001135606e-05,
+      "loss": 0.3051,
+      "step": 9127
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8752065253795997e-05,
+      "loss": 0.2992,
+      "step": 9128
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8745230380341526e-05,
+      "loss": 0.3186,
+      "step": 9129
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8738395391625306e-05,
+      "loss": 0.3139,
+      "step": 9130
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8731560288279995e-05,
+      "loss": 0.2951,
+      "step": 9131
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.872472507093828e-05,
+      "loss": 0.3121,
+      "step": 9132
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8717889740232838e-05,
+      "loss": 0.3223,
+      "step": 9133
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8711054296796365e-05,
+      "loss": 0.3196,
+      "step": 9134
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8704218741261573e-05,
+      "loss": 0.3103,
+      "step": 9135
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8697383074261175e-05,
+      "loss": 0.3013,
+      "step": 9136
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8690547296427895e-05,
+      "loss": 0.3173,
+      "step": 9137
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8683711408394477e-05,
+      "loss": 0.3115,
+      "step": 9138
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8676875410793657e-05,
+      "loss": 0.3326,
+      "step": 9139
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8670039304258207e-05,
+      "loss": 0.3254,
+      "step": 9140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.866320308942088e-05,
+      "loss": 0.3126,
+      "step": 9141
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8656366766914472e-05,
+      "loss": 0.3228,
+      "step": 9142
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8649530337371762e-05,
+      "loss": 0.3088,
+      "step": 9143
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.864269380142553e-05,
+      "loss": 0.3219,
+      "step": 9144
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.863585715970861e-05,
+      "loss": 0.3105,
+      "step": 9145
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8629020412853807e-05,
+      "loss": 0.2964,
+      "step": 9146
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8622183561493955e-05,
+      "loss": 0.3077,
+      "step": 9147
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8615346606261876e-05,
+      "loss": 0.2975,
+      "step": 9148
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8608509547790434e-05,
+      "loss": 0.2947,
+      "step": 9149
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.8601672386712476e-05,
+      "loss": 0.306,
+      "step": 9150
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8594835123660863e-05,
+      "loss": 0.3043,
+      "step": 9151
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8587997759268482e-05,
+      "loss": 0.2993,
+      "step": 9152
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.85811602941682e-05,
+      "loss": 0.3016,
+      "step": 9153
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8574322728992935e-05,
+      "loss": 0.3131,
+      "step": 9154
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8567485064375564e-05,
+      "loss": 0.316,
+      "step": 9155
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8560647300949005e-05,
+      "loss": 0.2913,
+      "step": 9156
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8553809439346197e-05,
+      "loss": 0.3072,
+      "step": 9157
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.854697148020005e-05,
+      "loss": 0.3066,
+      "step": 9158
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8540133424143517e-05,
+      "loss": 0.3198,
+      "step": 9159
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8533295271809535e-05,
+      "loss": 0.3063,
+      "step": 9160
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.852645702383106e-05,
+      "loss": 0.3056,
+      "step": 9161
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.851961868084107e-05,
+      "loss": 0.3163,
+      "step": 9162
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.851278024347253e-05,
+      "loss": 0.2942,
+      "step": 9163
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8505941712358435e-05,
+      "loss": 0.3026,
+      "step": 9164
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8499103088131763e-05,
+      "loss": 0.3221,
+      "step": 9165
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.849226437142552e-05,
+      "loss": 0.3103,
+      "step": 9166
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.848542556287272e-05,
+      "loss": 0.3092,
+      "step": 9167
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.847858666310637e-05,
+      "loss": 0.3082,
+      "step": 9168
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.847174767275952e-05,
+      "loss": 0.3135,
+      "step": 9169
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.846490859246517e-05,
+      "loss": 0.3169,
+      "step": 9170
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8458069422856394e-05,
+      "loss": 0.3128,
+      "step": 9171
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.845123016456622e-05,
+      "loss": 0.2853,
+      "step": 9172
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8444390818227723e-05,
+      "loss": 0.2965,
+      "step": 9173
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8437551384473963e-05,
+      "loss": 0.2934,
+      "step": 9174
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.843071186393802e-05,
+      "loss": 0.323,
+      "step": 9175
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.842387225725298e-05,
+      "loss": 0.333,
+      "step": 9176
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8417032565051932e-05,
+      "loss": 0.3214,
+      "step": 9177
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.841019278796796e-05,
+      "loss": 0.3193,
+      "step": 9178
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.84033529266342e-05,
+      "loss": 0.2846,
+      "step": 9179
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8396512981683736e-05,
+      "loss": 0.3194,
+      "step": 9180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8389672953749724e-05,
+      "loss": 0.3107,
+      "step": 9181
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8382832843465278e-05,
+      "loss": 0.3111,
+      "step": 9182
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8375992651463528e-05,
+      "loss": 0.2949,
+      "step": 9183
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8369152378377635e-05,
+      "loss": 0.3219,
+      "step": 9184
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8362312024840743e-05,
+      "loss": 0.3172,
+      "step": 9185
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.835547159148602e-05,
+      "loss": 0.3133,
+      "step": 9186
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.834863107894663e-05,
+      "loss": 0.2866,
+      "step": 9187
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8341790487855747e-05,
+      "loss": 0.3298,
+      "step": 9188
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8334949818846557e-05,
+      "loss": 0.3073,
+      "step": 9189
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8328109072552243e-05,
+      "loss": 0.2899,
+      "step": 9190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.832126824960602e-05,
+      "loss": 0.3076,
+      "step": 9191
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.831442735064107e-05,
+      "loss": 0.3122,
+      "step": 9192
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8307586376290617e-05,
+      "loss": 0.3003,
+      "step": 9193
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8300745327187877e-05,
+      "loss": 0.322,
+      "step": 9194
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.829390420396607e-05,
+      "loss": 0.3311,
+      "step": 9195
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8287063007258447e-05,
+      "loss": 0.3152,
+      "step": 9196
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8280221737698227e-05,
+      "loss": 0.2974,
+      "step": 9197
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8273380395918654e-05,
+      "loss": 0.3143,
+      "step": 9198
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8266538982553e-05,
+      "loss": 0.2938,
+      "step": 9199
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8259697498234505e-05,
+      "loss": 0.2912,
+      "step": 9200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8252855943596447e-05,
+      "loss": 0.319,
+      "step": 9201
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8246014319272083e-05,
+      "loss": 0.3085,
+      "step": 9202
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8239172625894704e-05,
+      "loss": 0.3088,
+      "step": 9203
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8232330864097594e-05,
+      "loss": 0.3138,
+      "step": 9204
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8225489034514047e-05,
+      "loss": 0.3049,
+      "step": 9205
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.821864713777735e-05,
+      "loss": 0.3141,
+      "step": 9206
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.8211805174520812e-05,
+      "loss": 0.2918,
+      "step": 9207
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.820496314537774e-05,
+      "loss": 0.319,
+      "step": 9208
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8198121050981464e-05,
+      "loss": 0.31,
+      "step": 9209
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8191278891965283e-05,
+      "loss": 0.3076,
+      "step": 9210
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8184436668962545e-05,
+      "loss": 0.3054,
+      "step": 9211
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.817759438260657e-05,
+      "loss": 0.3272,
+      "step": 9212
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8170752033530707e-05,
+      "loss": 0.3039,
+      "step": 9213
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.81639096223683e-05,
+      "loss": 0.3132,
+      "step": 9214
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8157067149752692e-05,
+      "loss": 0.3139,
+      "step": 9215
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8150224616317255e-05,
+      "loss": 0.3162,
+      "step": 9216
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8143382022695343e-05,
+      "loss": 0.3221,
+      "step": 9217
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.813653936952032e-05,
+      "loss": 0.3061,
+      "step": 9218
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8129696657425562e-05,
+      "loss": 0.3291,
+      "step": 9219
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.812285388704446e-05,
+      "loss": 0.3035,
+      "step": 9220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8116011059010376e-05,
+      "loss": 0.2742,
+      "step": 9221
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.810916817395672e-05,
+      "loss": 0.3033,
+      "step": 9222
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.810232523251688e-05,
+      "loss": 0.3096,
+      "step": 9223
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8095482235324245e-05,
+      "loss": 0.2891,
+      "step": 9224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8088639183012243e-05,
+      "loss": 0.3274,
+      "step": 9225
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.808179607621427e-05,
+      "loss": 0.2931,
+      "step": 9226
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8074952915563747e-05,
+      "loss": 0.3123,
+      "step": 9227
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.806810970169409e-05,
+      "loss": 0.3114,
+      "step": 9228
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8061266435238724e-05,
+      "loss": 0.2898,
+      "step": 9229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.805442311683109e-05,
+      "loss": 0.2931,
+      "step": 9230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.804757974710462e-05,
+      "loss": 0.2912,
+      "step": 9231
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.804073632669274e-05,
+      "loss": 0.3152,
+      "step": 9232
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.803389285622891e-05,
+      "loss": 0.304,
+      "step": 9233
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8027049336346573e-05,
+      "loss": 0.3266,
+      "step": 9234
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.802020576767918e-05,
+      "loss": 0.297,
+      "step": 9235
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.80133621508602e-05,
+      "loss": 0.3503,
+      "step": 9236
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.8006518486523087e-05,
+      "loss": 0.309,
+      "step": 9237
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.799967477530131e-05,
+      "loss": 0.3044,
+      "step": 9238
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.799283101782834e-05,
+      "loss": 0.3107,
+      "step": 9239
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7985987214737652e-05,
+      "loss": 0.2934,
+      "step": 9240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.797914336666274e-05,
+      "loss": 0.3189,
+      "step": 9241
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7972299474237068e-05,
+      "loss": 0.3164,
+      "step": 9242
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.796545553809413e-05,
+      "loss": 0.2952,
+      "step": 9243
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7958611558867426e-05,
+      "loss": 0.3049,
+      "step": 9244
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.795176753719045e-05,
+      "loss": 0.3024,
+      "step": 9245
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7944923473696698e-05,
+      "loss": 0.3015,
+      "step": 9246
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7938079369019677e-05,
+      "loss": 0.3054,
+      "step": 9247
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7931235223792895e-05,
+      "loss": 0.3013,
+      "step": 9248
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.792439103864986e-05,
+      "loss": 0.3464,
+      "step": 9249
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.79175468142241e-05,
+      "loss": 0.3109,
+      "step": 9250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7910702551149115e-05,
+      "loss": 0.306,
+      "step": 9251
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7903858250058445e-05,
+      "loss": 0.2916,
+      "step": 9252
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7897013911585606e-05,
+      "loss": 0.3304,
+      "step": 9253
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.789016953636413e-05,
+      "loss": 0.3203,
+      "step": 9254
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7883325125027555e-05,
+      "loss": 0.3136,
+      "step": 9255
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.787648067820941e-05,
+      "loss": 0.3072,
+      "step": 9256
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7869636196543235e-05,
+      "loss": 0.3112,
+      "step": 9257
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7862791680662587e-05,
+      "loss": 0.316,
+      "step": 9258
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7855947131200993e-05,
+      "loss": 0.3016,
+      "step": 9259
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7849102548792012e-05,
+      "loss": 0.3099,
+      "step": 9260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7842257934069193e-05,
+      "loss": 0.3112,
+      "step": 9261
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.78354132876661e-05,
+      "loss": 0.299,
+      "step": 9262
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.7828568610216276e-05,
+      "loss": 0.2958,
+      "step": 9263
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.782172390235329e-05,
+      "loss": 0.2957,
+      "step": 9264
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.781487916471071e-05,
+      "loss": 0.3021,
+      "step": 9265
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7808034397922094e-05,
+      "loss": 0.3276,
+      "step": 9266
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7801189602621025e-05,
+      "loss": 0.3155,
+      "step": 9267
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7794344779441066e-05,
+      "loss": 0.3059,
+      "step": 9268
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.778749992901578e-05,
+      "loss": 0.2971,
+      "step": 9269
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7780655051978766e-05,
+      "loss": 0.3095,
+      "step": 9270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7773810148963587e-05,
+      "loss": 0.305,
+      "step": 9271
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.776696522060384e-05,
+      "loss": 0.3091,
+      "step": 9272
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7760120267533097e-05,
+      "loss": 0.3115,
+      "step": 9273
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7753275290384943e-05,
+      "loss": 0.3251,
+      "step": 9274
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7746430289792975e-05,
+      "loss": 0.3144,
+      "step": 9275
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.773958526639078e-05,
+      "loss": 0.3056,
+      "step": 9276
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7732740220811954e-05,
+      "loss": 0.2973,
+      "step": 9277
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7725895153690094e-05,
+      "loss": 0.3093,
+      "step": 9278
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7719050065658786e-05,
+      "loss": 0.2998,
+      "step": 9279
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7712204957351646e-05,
+      "loss": 0.2948,
+      "step": 9280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7705359829402257e-05,
+      "loss": 0.3159,
+      "step": 9281
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.769851468244423e-05,
+      "loss": 0.308,
+      "step": 9282
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7691669517111168e-05,
+      "loss": 0.3276,
+      "step": 9283
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.768482433403669e-05,
+      "loss": 0.3191,
+      "step": 9284
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7677979133854388e-05,
+      "loss": 0.2947,
+      "step": 9285
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7671133917197867e-05,
+      "loss": 0.3198,
+      "step": 9286
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7664288684700756e-05,
+      "loss": 0.3076,
+      "step": 9287
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7657443436996662e-05,
+      "loss": 0.3012,
+      "step": 9288
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7650598174719187e-05,
+      "loss": 0.3194,
+      "step": 9289
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7643752898501952e-05,
+      "loss": 0.3252,
+      "step": 9290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7636907608978585e-05,
+      "loss": 0.3017,
+      "step": 9291
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.76300623067827e-05,
+      "loss": 0.2989,
+      "step": 9292
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7623216992547902e-05,
+      "loss": 0.3081,
+      "step": 9293
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7616371666907826e-05,
+      "loss": 0.314,
+      "step": 9294
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7609526330496073e-05,
+      "loss": 0.3147,
+      "step": 9295
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7602680983946294e-05,
+      "loss": 0.3142,
+      "step": 9296
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.759583562789209e-05,
+      "loss": 0.3179,
+      "step": 9297
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7588990262967097e-05,
+      "loss": 0.2927,
+      "step": 9298
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7582144889804928e-05,
+      "loss": 0.311,
+      "step": 9299
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.757529950903922e-05,
+      "loss": 0.297,
+      "step": 9300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7568454121303595e-05,
+      "loss": 0.3064,
+      "step": 9301
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.756160872723168e-05,
+      "loss": 0.3289,
+      "step": 9302
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7554763327457094e-05,
+      "loss": 0.2898,
+      "step": 9303
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.754791792261348e-05,
+      "loss": 0.3207,
+      "step": 9304
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7541072513334455e-05,
+      "loss": 0.2938,
+      "step": 9305
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7534227100253653e-05,
+      "loss": 0.3155,
+      "step": 9306
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7527381684004704e-05,
+      "loss": 0.2985,
+      "step": 9307
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7520536265221232e-05,
+      "loss": 0.3085,
+      "step": 9308
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7513690844536877e-05,
+      "loss": 0.3126,
+      "step": 9309
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7506845422585248e-05,
+      "loss": 0.3251,
+      "step": 9310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.2903,
+      "step": 9311
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7493154577414754e-05,
+      "loss": 0.3133,
+      "step": 9312
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.748630915546313e-05,
+      "loss": 0.3152,
+      "step": 9313
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7479463734778777e-05,
+      "loss": 0.2972,
+      "step": 9314
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.74726183159953e-05,
+      "loss": 0.3061,
+      "step": 9315
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.746577289974635e-05,
+      "loss": 0.305,
+      "step": 9316
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7458927486665554e-05,
+      "loss": 0.3014,
+      "step": 9317
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7452082077386528e-05,
+      "loss": 0.2905,
+      "step": 9318
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.744523667254291e-05,
+      "loss": 0.3097,
+      "step": 9319
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7438391272768326e-05,
+      "loss": 0.3044,
+      "step": 9320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7431545878696414e-05,
+      "loss": 0.3149,
+      "step": 9321
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7424700490960786e-05,
+      "loss": 0.3043,
+      "step": 9322
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.7417855110195074e-05,
+      "loss": 0.3464,
+      "step": 9323
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7411009737032912e-05,
+      "loss": 0.3311,
+      "step": 9324
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7404164372107915e-05,
+      "loss": 0.3152,
+      "step": 9325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7397319016053712e-05,
+      "loss": 0.3091,
+      "step": 9326
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7390473669503936e-05,
+      "loss": 0.3049,
+      "step": 9327
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7383628333092186e-05,
+      "loss": 0.3036,
+      "step": 9328
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7376783007452106e-05,
+      "loss": 0.3152,
+      "step": 9329
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7369937693217305e-05,
+      "loss": 0.3065,
+      "step": 9330
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.736309239102142e-05,
+      "loss": 0.3215,
+      "step": 9331
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.735624710149805e-05,
+      "loss": 0.2957,
+      "step": 9332
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7349401825280825e-05,
+      "loss": 0.3262,
+      "step": 9333
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7342556563003353e-05,
+      "loss": 0.3168,
+      "step": 9334
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.733571131529925e-05,
+      "loss": 0.3064,
+      "step": 9335
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7328866082802135e-05,
+      "loss": 0.3293,
+      "step": 9336
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.732202086614562e-05,
+      "loss": 0.3139,
+      "step": 9337
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.731517566596332e-05,
+      "loss": 0.3072,
+      "step": 9338
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7308330482888834e-05,
+      "loss": 0.3092,
+      "step": 9339
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7301485317555774e-05,
+      "loss": 0.3008,
+      "step": 9340
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7294640170597752e-05,
+      "loss": 0.3162,
+      "step": 9341
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7287795042648363e-05,
+      "loss": 0.3178,
+      "step": 9342
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7280949934341216e-05,
+      "loss": 0.312,
+      "step": 9343
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7274104846309918e-05,
+      "loss": 0.2887,
+      "step": 9344
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7267259779188048e-05,
+      "loss": 0.3203,
+      "step": 9345
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7260414733609217e-05,
+      "loss": 0.3314,
+      "step": 9346
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7253569710207027e-05,
+      "loss": 0.292,
+      "step": 9347
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7246724709615062e-05,
+      "loss": 0.2944,
+      "step": 9348
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7239879732466912e-05,
+      "loss": 0.3143,
+      "step": 9349
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7233034779396165e-05,
+      "loss": 0.3158,
+      "step": 9350
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.722618985103641e-05,
+      "loss": 0.316,
+      "step": 9351
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7219344948021236e-05,
+      "loss": 0.3208,
+      "step": 9352
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.721250007098422e-05,
+      "loss": 0.2993,
+      "step": 9353
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.720565522055894e-05,
+      "loss": 0.3102,
+      "step": 9354
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7198810397378977e-05,
+      "loss": 0.3022,
+      "step": 9355
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7191965602077908e-05,
+      "loss": 0.3128,
+      "step": 9356
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.718512083528929e-05,
+      "loss": 0.3289,
+      "step": 9357
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7178276097646708e-05,
+      "loss": 0.3027,
+      "step": 9358
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.717143138978373e-05,
+      "loss": 0.2932,
+      "step": 9359
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7164586712333905e-05,
+      "loss": 0.2989,
+      "step": 9360
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.715774206593081e-05,
+      "loss": 0.313,
+      "step": 9361
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.715089745120799e-05,
+      "loss": 0.3078,
+      "step": 9362
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.714405286879901e-05,
+      "loss": 0.2996,
+      "step": 9363
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7137208319337422e-05,
+      "loss": 0.307,
+      "step": 9364
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.713036380345676e-05,
+      "loss": 0.2978,
+      "step": 9365
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7123519321790592e-05,
+      "loss": 0.3142,
+      "step": 9366
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7116674874972447e-05,
+      "loss": 0.3225,
+      "step": 9367
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7109830463635872e-05,
+      "loss": 0.2989,
+      "step": 9368
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.71029860884144e-05,
+      "loss": 0.3143,
+      "step": 9369
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7096141749941557e-05,
+      "loss": 0.2984,
+      "step": 9370
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7089297448850887e-05,
+      "loss": 0.2986,
+      "step": 9371
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7082453185775903e-05,
+      "loss": 0.3107,
+      "step": 9372
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.707560896135014e-05,
+      "loss": 0.3057,
+      "step": 9373
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7068764776207117e-05,
+      "loss": 0.3302,
+      "step": 9374
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7061920630980335e-05,
+      "loss": 0.3121,
+      "step": 9375
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7055076526303314e-05,
+      "loss": 0.3323,
+      "step": 9376
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7048232462809563e-05,
+      "loss": 0.3038,
+      "step": 9377
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7041388441132583e-05,
+      "loss": 0.3227,
+      "step": 9378
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.703454446190588e-05,
+      "loss": 0.3172,
+      "step": 9379
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.7027700525762944e-05,
+      "loss": 0.2973,
+      "step": 9380
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.7020856633337272e-05,
+      "loss": 0.3023,
+      "step": 9381
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.7014012785262353e-05,
+      "loss": 0.2879,
+      "step": 9382
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.7007168982171667e-05,
+      "loss": 0.3146,
+      "step": 9383
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.7000325224698703e-05,
+      "loss": 0.3212,
+      "step": 9384
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6993481513476932e-05,
+      "loss": 0.3075,
+      "step": 9385
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.698663784913981e-05,
+      "loss": 0.2999,
+      "step": 9386
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.697979423232083e-05,
+      "loss": 0.2825,
+      "step": 9387
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.697295066365344e-05,
+      "loss": 0.2999,
+      "step": 9388
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6966107143771102e-05,
+      "loss": 0.3034,
+      "step": 9389
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6959263673307272e-05,
+      "loss": 0.2903,
+      "step": 9390
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6952420252895392e-05,
+      "loss": 0.3334,
+      "step": 9391
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6945576883168917e-05,
+      "loss": 0.2975,
+      "step": 9392
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.693873356476128e-05,
+      "loss": 0.2974,
+      "step": 9393
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6931890298305917e-05,
+      "loss": 0.3136,
+      "step": 9394
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.692504708443627e-05,
+      "loss": 0.2917,
+      "step": 9395
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.691820392378574e-05,
+      "loss": 0.3053,
+      "step": 9396
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.691136081698777e-05,
+      "loss": 0.3202,
+      "step": 9397
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6904517764675764e-05,
+      "loss": 0.2987,
+      "step": 9398
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6897674767483137e-05,
+      "loss": 0.2945,
+      "step": 9399
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6890831826043296e-05,
+      "loss": 0.3021,
+      "step": 9400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6883988940989636e-05,
+      "loss": 0.3019,
+      "step": 9401
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.687714611295556e-05,
+      "loss": 0.3134,
+      "step": 9402
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6870303342574443e-05,
+      "loss": 0.2985,
+      "step": 9403
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6863460630479693e-05,
+      "loss": 0.3024,
+      "step": 9404
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6856617977304673e-05,
+      "loss": 0.3037,
+      "step": 9405
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.684977538368275e-05,
+      "loss": 0.2942,
+      "step": 9406
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.684293285024731e-05,
+      "loss": 0.2966,
+      "step": 9407
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6836090377631708e-05,
+      "loss": 0.3067,
+      "step": 9408
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6829247966469295e-05,
+      "loss": 0.3196,
+      "step": 9409
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6822405617393438e-05,
+      "loss": 0.3041,
+      "step": 9410
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6815563331037464e-05,
+      "loss": 0.301,
+      "step": 9411
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6808721108034723e-05,
+      "loss": 0.2883,
+      "step": 9412
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6801878949018545e-05,
+      "loss": 0.3043,
+      "step": 9413
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6795036854622263e-05,
+      "loss": 0.3014,
+      "step": 9414
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6788194825479197e-05,
+      "loss": 0.3311,
+      "step": 9415
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.678135286222266e-05,
+      "loss": 0.3046,
+      "step": 9416
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6774510965485965e-05,
+      "loss": 0.3071,
+      "step": 9417
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.676766913590241e-05,
+      "loss": 0.3146,
+      "step": 9418
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.67608273741053e-05,
+      "loss": 0.3246,
+      "step": 9419
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6753985680727922e-05,
+      "loss": 0.3179,
+      "step": 9420
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6747144056403565e-05,
+      "loss": 0.3297,
+      "step": 9421
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6740302501765507e-05,
+      "loss": 0.3072,
+      "step": 9422
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.673346101744701e-05,
+      "loss": 0.2923,
+      "step": 9423
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6726619604081348e-05,
+      "loss": 0.3047,
+      "step": 9424
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.671977826230178e-05,
+      "loss": 0.2911,
+      "step": 9425
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6712936992741562e-05,
+      "loss": 0.3102,
+      "step": 9426
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6706095796033932e-05,
+      "loss": 0.2977,
+      "step": 9427
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.669925467281213e-05,
+      "loss": 0.3074,
+      "step": 9428
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6692413623709388e-05,
+      "loss": 0.3141,
+      "step": 9429
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6685572649358938e-05,
+      "loss": 0.2975,
+      "step": 9430
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6678731750393993e-05,
+      "loss": 0.3185,
+      "step": 9431
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6671890927447762e-05,
+      "loss": 0.3146,
+      "step": 9432
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.666505018115345e-05,
+      "loss": 0.3057,
+      "step": 9433
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.665820951214426e-05,
+      "loss": 0.286,
+      "step": 9434
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6651368921053377e-05,
+      "loss": 0.3151,
+      "step": 9435
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6644528408513987e-05,
+      "loss": 0.3065,
+      "step": 9436
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6637687975159266e-05,
+      "loss": 0.3266,
+      "step": 9437
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6630847621622373e-05,
+      "loss": 0.3043,
+      "step": 9438
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6624007348536477e-05,
+      "loss": 0.3106,
+      "step": 9439
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6617167156534727e-05,
+      "loss": 0.34,
+      "step": 9440
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.661032704625028e-05,
+      "loss": 0.3101,
+      "step": 9441
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6603487018316266e-05,
+      "loss": 0.3185,
+      "step": 9442
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.659664707336581e-05,
+      "loss": 0.3056,
+      "step": 9443
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.658980721203205e-05,
+      "loss": 0.301,
+      "step": 9444
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.658296743494808e-05,
+      "loss": 0.3062,
+      "step": 9445
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6576127742747032e-05,
+      "loss": 0.3037,
+      "step": 9446
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6569288136061986e-05,
+      "loss": 0.3053,
+      "step": 9447
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6562448615526042e-05,
+      "loss": 0.341,
+      "step": 9448
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6555609181772283e-05,
+      "loss": 0.2899,
+      "step": 9449
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6548769835433784e-05,
+      "loss": 0.3007,
+      "step": 9450
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6541930577143615e-05,
+      "loss": 0.324,
+      "step": 9451
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6535091407534834e-05,
+      "loss": 0.3243,
+      "step": 9452
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6528252327240493e-05,
+      "loss": 0.3087,
+      "step": 9453
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.652141333689363e-05,
+      "loss": 0.2881,
+      "step": 9454
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.651457443712728e-05,
+      "loss": 0.3188,
+      "step": 9455
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6507735628574483e-05,
+      "loss": 0.2989,
+      "step": 9456
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6500896911868245e-05,
+      "loss": 0.2934,
+      "step": 9457
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.649405828764157e-05,
+      "loss": 0.2957,
+      "step": 9458
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.648721975652747e-05,
+      "loss": 0.3109,
+      "step": 9459
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6480381319158938e-05,
+      "loss": 0.3107,
+      "step": 9460
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6473542976168946e-05,
+      "loss": 0.3024,
+      "step": 9461
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6466704728190474e-05,
+      "loss": 0.3066,
+      "step": 9462
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6459866575856495e-05,
+      "loss": 0.2963,
+      "step": 9463
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.645302851979996e-05,
+      "loss": 0.3058,
+      "step": 9464
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6446190560653812e-05,
+      "loss": 0.3173,
+      "step": 9465
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6439352699050994e-05,
+      "loss": 0.3147,
+      "step": 9466
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.643251493562444e-05,
+      "loss": 0.3066,
+      "step": 9467
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6425677271007077e-05,
+      "loss": 0.3111,
+      "step": 9468
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6418839705831805e-05,
+      "loss": 0.3059,
+      "step": 9469
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6412002240731527e-05,
+      "loss": 0.3045,
+      "step": 9470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.640516487633914e-05,
+      "loss": 0.305,
+      "step": 9471
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6398327613287526e-05,
+      "loss": 0.3044,
+      "step": 9472
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6391490452209568e-05,
+      "loss": 0.3135,
+      "step": 9473
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6384653393738122e-05,
+      "loss": 0.3129,
+      "step": 9474
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6377816438506047e-05,
+      "loss": 0.299,
+      "step": 9475
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6370979587146195e-05,
+      "loss": 0.2983,
+      "step": 9476
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.636414284029139e-05,
+      "loss": 0.2893,
+      "step": 9477
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6357306198574473e-05,
+      "loss": 0.3197,
+      "step": 9478
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6350469662628253e-05,
+      "loss": 0.3321,
+      "step": 9479
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.634363323308553e-05,
+      "loss": 0.317,
+      "step": 9480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.633679691057912e-05,
+      "loss": 0.2858,
+      "step": 9481
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6329960695741795e-05,
+      "loss": 0.3209,
+      "step": 9482
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6323124589206345e-05,
+      "loss": 0.3076,
+      "step": 9483
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6316288591605532e-05,
+      "loss": 0.2846,
+      "step": 9484
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6309452703572107e-05,
+      "loss": 0.3018,
+      "step": 9485
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6302616925738827e-05,
+      "loss": 0.317,
+      "step": 9486
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.629578125873843e-05,
+      "loss": 0.3027,
+      "step": 9487
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6288945703203634e-05,
+      "loss": 0.323,
+      "step": 9488
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6282110259767167e-05,
+      "loss": 0.3049,
+      "step": 9489
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6275274929061723e-05,
+      "loss": 0.3238,
+      "step": 9490
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6268439711720004e-05,
+      "loss": 0.3093,
+      "step": 9491
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6261604608374696e-05,
+      "loss": 0.2941,
+      "step": 9492
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.6254769619658476e-05,
+      "loss": 0.3118,
+      "step": 9493
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.624793474620401e-05,
+      "loss": 0.3103,
+      "step": 9494
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.624109998864394e-05,
+      "loss": 0.3266,
+      "step": 9495
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.623426534761092e-05,
+      "loss": 0.3241,
+      "step": 9496
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6227430823737574e-05,
+      "loss": 0.3102,
+      "step": 9497
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6220596417656534e-05,
+      "loss": 0.3064,
+      "step": 9498
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6213762130000403e-05,
+      "loss": 0.301,
+      "step": 9499
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.620692796140179e-05,
+      "loss": 0.3146,
+      "step": 9500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6200093912493257e-05,
+      "loss": 0.3276,
+      "step": 9501
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.619325998390741e-05,
+      "loss": 0.3157,
+      "step": 9502
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6186426176276797e-05,
+      "loss": 0.3086,
+      "step": 9503
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6179592490233983e-05,
+      "loss": 0.3176,
+      "step": 9504
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.617275892641151e-05,
+      "loss": 0.2883,
+      "step": 9505
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6165925485441905e-05,
+      "loss": 0.3231,
+      "step": 9506
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.61590921679577e-05,
+      "loss": 0.2953,
+      "step": 9507
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6152258974591388e-05,
+      "loss": 0.3132,
+      "step": 9508
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6145425905975484e-05,
+      "loss": 0.3097,
+      "step": 9509
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6138592962742465e-05,
+      "loss": 0.3032,
+      "step": 9510
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6131760145524807e-05,
+      "loss": 0.3092,
+      "step": 9511
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6124927454954977e-05,
+      "loss": 0.3291,
+      "step": 9512
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6118094891665417e-05,
+      "loss": 0.2879,
+      "step": 9513
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6111262456288583e-05,
+      "loss": 0.2889,
+      "step": 9514
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6104430149456892e-05,
+      "loss": 0.3191,
+      "step": 9515
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6097597971802758e-05,
+      "loss": 0.3108,
+      "step": 9516
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6090765923958594e-05,
+      "loss": 0.3003,
+      "step": 9517
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6083934006556776e-05,
+      "loss": 0.2947,
+      "step": 9518
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6077102220229705e-05,
+      "loss": 0.303,
+      "step": 9519
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6070270565609735e-05,
+      "loss": 0.3035,
+      "step": 9520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6063439043329224e-05,
+      "loss": 0.3213,
+      "step": 9521
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.605660765402052e-05,
+      "loss": 0.3076,
+      "step": 9522
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6049776398315946e-05,
+      "loss": 0.2877,
+      "step": 9523
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.604294527684783e-05,
+      "loss": 0.3007,
+      "step": 9524
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6036114290248475e-05,
+      "loss": 0.3141,
+      "step": 9525
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6029283439150166e-05,
+      "loss": 0.308,
+      "step": 9526
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.6022452724185196e-05,
+      "loss": 0.3233,
+      "step": 9527
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.601562214598583e-05,
+      "loss": 0.3169,
+      "step": 9528
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.600879170518432e-05,
+      "loss": 0.3146,
+      "step": 9529
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.600196140241292e-05,
+      "loss": 0.3288,
+      "step": 9530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5995131238303842e-05,
+      "loss": 0.3178,
+      "step": 9531
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.598830121348933e-05,
+      "loss": 0.3118,
+      "step": 9532
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5981471328601566e-05,
+      "loss": 0.3126,
+      "step": 9533
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5974641584272745e-05,
+      "loss": 0.2956,
+      "step": 9534
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.596781198113506e-05,
+      "loss": 0.3225,
+      "step": 9535
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.596098251982066e-05,
+      "loss": 0.3112,
+      "step": 9536
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5954153200961705e-05,
+      "loss": 0.316,
+      "step": 9537
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5947324025190333e-05,
+      "loss": 0.3029,
+      "step": 9538
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.594049499313868e-05,
+      "loss": 0.3029,
+      "step": 9539
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5933666105438847e-05,
+      "loss": 0.3075,
+      "step": 9540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5926837362722932e-05,
+      "loss": 0.295,
+      "step": 9541
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5920008765623033e-05,
+      "loss": 0.3246,
+      "step": 9542
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5913180314771206e-05,
+      "loss": 0.321,
+      "step": 9543
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.590635201079953e-05,
+      "loss": 0.3171,
+      "step": 9544
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.589952385434003e-05,
+      "loss": 0.3262,
+      "step": 9545
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5892695846024754e-05,
+      "loss": 0.3071,
+      "step": 9546
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.588586798648572e-05,
+      "loss": 0.3075,
+      "step": 9547
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5879040276354914e-05,
+      "loss": 0.3278,
+      "step": 9548
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.587221271626435e-05,
+      "loss": 0.321,
+      "step": 9549
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5865385306845986e-05,
+      "loss": 0.3167,
+      "step": 9550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5858558048731803e-05,
+      "loss": 0.3136,
+      "step": 9551
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.585173094255373e-05,
+      "loss": 0.312,
+      "step": 9552
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5844903988943708e-05,
+      "loss": 0.2835,
+      "step": 9553
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.583807718853367e-05,
+      "loss": 0.3017,
+      "step": 9554
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5831250541955503e-05,
+      "loss": 0.3074,
+      "step": 9555
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5824424049841118e-05,
+      "loss": 0.301,
+      "step": 9556
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5817597712822383e-05,
+      "loss": 0.3121,
+      "step": 9557
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.581077153153116e-05,
+      "loss": 0.3223,
+      "step": 9558
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5803945506599298e-05,
+      "loss": 0.3281,
+      "step": 9559
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.579711963865863e-05,
+      "loss": 0.3049,
+      "step": 9560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5790293928340985e-05,
+      "loss": 0.2963,
+      "step": 9561
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5783468376278165e-05,
+      "loss": 0.312,
+      "step": 9562
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.577664298310195e-05,
+      "loss": 0.3175,
+      "step": 9563
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5769817749444135e-05,
+      "loss": 0.2962,
+      "step": 9564
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5762992675936465e-05,
+      "loss": 0.3349,
+      "step": 9565
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5756167763210698e-05,
+      "loss": 0.3026,
+      "step": 9566
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.574934301189856e-05,
+      "loss": 0.3073,
+      "step": 9567
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.574251842263176e-05,
+      "loss": 0.2924,
+      "step": 9568
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5735693996042017e-05,
+      "loss": 0.2952,
+      "step": 9569
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5728869732760995e-05,
+      "loss": 0.3229,
+      "step": 9570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5722045633420395e-05,
+      "loss": 0.2964,
+      "step": 9571
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5715221698651848e-05,
+      "loss": 0.2968,
+      "step": 9572
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5708397929087002e-05,
+      "loss": 0.3075,
+      "step": 9573
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.570157432535749e-05,
+      "loss": 0.2956,
+      "step": 9574
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5694750888094916e-05,
+      "loss": 0.3138,
+      "step": 9575
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5687927617930873e-05,
+      "loss": 0.3085,
+      "step": 9576
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5681104515496952e-05,
+      "loss": 0.3174,
+      "step": 9577
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5674281581424696e-05,
+      "loss": 0.2972,
+      "step": 9578
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.566745881634568e-05,
+      "loss": 0.326,
+      "step": 9579
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5660636220891415e-05,
+      "loss": 0.315,
+      "step": 9580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5653813795693434e-05,
+      "loss": 0.2981,
+      "step": 9581
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5646991541383225e-05,
+      "loss": 0.3105,
+      "step": 9582
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5640169458592278e-05,
+      "loss": 0.3155,
+      "step": 9583
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5633347547952064e-05,
+      "loss": 0.3031,
+      "step": 9584
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5626525810094038e-05,
+      "loss": 0.3252,
+      "step": 9585
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5619704245649633e-05,
+      "loss": 0.3047,
+      "step": 9586
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5612882855250282e-05,
+      "loss": 0.2983,
+      "step": 9587
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.560606163952738e-05,
+      "loss": 0.3241,
+      "step": 9588
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5599240599112312e-05,
+      "loss": 0.3056,
+      "step": 9589
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5592419734636464e-05,
+      "loss": 0.3107,
+      "step": 9590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5585599046731186e-05,
+      "loss": 0.3211,
+      "step": 9591
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5578778536027816e-05,
+      "loss": 0.3142,
+      "step": 9592
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5571958203157688e-05,
+      "loss": 0.3262,
+      "step": 9593
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5565138048752104e-05,
+      "loss": 0.3147,
+      "step": 9594
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5558318073442346e-05,
+      "loss": 0.3007,
+      "step": 9595
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5551498277859704e-05,
+      "loss": 0.3196,
+      "step": 9596
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5544678662635422e-05,
+      "loss": 0.3275,
+      "step": 9597
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5537859228400756e-05,
+      "loss": 0.3074,
+      "step": 9598
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5531039975786926e-05,
+      "loss": 0.291,
+      "step": 9599
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.552422090542513e-05,
+      "loss": 0.3021,
+      "step": 9600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5517402017946568e-05,
+      "loss": 0.3144,
+      "step": 9601
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5510583313982404e-05,
+      "loss": 0.293,
+      "step": 9602
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.550376479416381e-05,
+      "loss": 0.3046,
+      "step": 9603
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.549694645912192e-05,
+      "loss": 0.3131,
+      "step": 9604
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5490128309487843e-05,
+      "loss": 0.3171,
+      "step": 9605
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5483310345892702e-05,
+      "loss": 0.3172,
+      "step": 9606
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.547649256896758e-05,
+      "loss": 0.3015,
+      "step": 9607
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.546967497934355e-05,
+      "loss": 0.3043,
+      "step": 9608
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5462857577651668e-05,
+      "loss": 0.2989,
+      "step": 9609
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5456040364522954e-05,
+      "loss": 0.3101,
+      "step": 9610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5449223340588445e-05,
+      "loss": 0.2925,
+      "step": 9611
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5442406506479138e-05,
+      "loss": 0.3096,
+      "step": 9612
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5435589862826016e-05,
+      "loss": 0.3167,
+      "step": 9613
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5428773410260044e-05,
+      "loss": 0.2905,
+      "step": 9614
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5421957149412164e-05,
+      "loss": 0.3132,
+      "step": 9615
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.541514108091332e-05,
+      "loss": 0.3073,
+      "step": 9616
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.540832520539441e-05,
+      "loss": 0.3084,
+      "step": 9617
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.540150952348635e-05,
+      "loss": 0.3052,
+      "step": 9618
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.539469403582e-05,
+      "loss": 0.3244,
+      "step": 9619
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5387878743026217e-05,
+      "loss": 0.309,
+      "step": 9620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5381063645735854e-05,
+      "loss": 0.3035,
+      "step": 9621
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.537424874457973e-05,
+      "loss": 0.3039,
+      "step": 9622
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5367434040188647e-05,
+      "loss": 0.3004,
+      "step": 9623
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5360619533193404e-05,
+      "loss": 0.2964,
+      "step": 9624
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.535380522422474e-05,
+      "loss": 0.3043,
+      "step": 9625
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5346991113913445e-05,
+      "loss": 0.2981,
+      "step": 9626
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.534017720289022e-05,
+      "loss": 0.3184,
+      "step": 9627
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5333363491785783e-05,
+      "loss": 0.2958,
+      "step": 9628
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.532654998123084e-05,
+      "loss": 0.2933,
+      "step": 9629
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5319736671856064e-05,
+      "loss": 0.3207,
+      "step": 9630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5312923564292102e-05,
+      "loss": 0.322,
+      "step": 9631
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5306110659169607e-05,
+      "loss": 0.3065,
+      "step": 9632
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5299297957119182e-05,
+      "loss": 0.3042,
+      "step": 9633
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5292485458771453e-05,
+      "loss": 0.2983,
+      "step": 9634
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5285673164756985e-05,
+      "loss": 0.3192,
+      "step": 9635
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.527886107570634e-05,
+      "loss": 0.3191,
+      "step": 9636
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.527204919225007e-05,
+      "loss": 0.2956,
+      "step": 9637
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5265237515018696e-05,
+      "loss": 0.3093,
+      "step": 9638
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.525842604464273e-05,
+      "loss": 0.3091,
+      "step": 9639
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.525161478175266e-05,
+      "loss": 0.2986,
+      "step": 9640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5244803726978943e-05,
+      "loss": 0.3209,
+      "step": 9641
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.523799288095204e-05,
+      "loss": 0.3127,
+      "step": 9642
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.523118224430237e-05,
+      "loss": 0.3068,
+      "step": 9643
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.522437181766035e-05,
+      "loss": 0.2905,
+      "step": 9644
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5217561601656377e-05,
+      "loss": 0.3012,
+      "step": 9645
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5210751596920806e-05,
+      "loss": 0.3115,
+      "step": 9646
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.520394180408401e-05,
+      "loss": 0.3111,
+      "step": 9647
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5197132223776293e-05,
+      "loss": 0.3074,
+      "step": 9648
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5190322856627996e-05,
+      "loss": 0.3014,
+      "step": 9649
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5183513703269395e-05,
+      "loss": 0.306,
+      "step": 9650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.517670476433076e-05,
+      "loss": 0.3044,
+      "step": 9651
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5169896040442354e-05,
+      "loss": 0.3266,
+      "step": 9652
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5163087532234407e-05,
+      "loss": 0.3181,
+      "step": 9653
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5156279240337133e-05,
+      "loss": 0.3055,
+      "step": 9654
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5149471165380728e-05,
+      "loss": 0.3059,
+      "step": 9655
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.514266330799535e-05,
+      "loss": 0.3331,
+      "step": 9656
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5135855668811175e-05,
+      "loss": 0.304,
+      "step": 9657
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5129048248458315e-05,
+      "loss": 0.3231,
+      "step": 9658
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5122241047566897e-05,
+      "loss": 0.2891,
+      "step": 9659
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5115434066767006e-05,
+      "loss": 0.3112,
+      "step": 9660
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5108627306688708e-05,
+      "loss": 0.3203,
+      "step": 9661
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5101820767962075e-05,
+      "loss": 0.3044,
+      "step": 9662
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5095014451217115e-05,
+      "loss": 0.3034,
+      "step": 9663
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.508820835708385e-05,
+      "loss": 0.3213,
+      "step": 9664
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5081402486192262e-05,
+      "loss": 0.316,
+      "step": 9665
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.507459683917233e-05,
+      "loss": 0.3137,
+      "step": 9666
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5067791416653995e-05,
+      "loss": 0.3296,
+      "step": 9667
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5060986219267192e-05,
+      "loss": 0.3086,
+      "step": 9668
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.505418124764182e-05,
+      "loss": 0.3051,
+      "step": 9669
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5047376502407765e-05,
+      "loss": 0.2887,
+      "step": 9670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.50405719841949e-05,
+      "loss": 0.3499,
+      "step": 9671
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5033767693633065e-05,
+      "loss": 0.3207,
+      "step": 9672
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5026963631352075e-05,
+      "loss": 0.3173,
+      "step": 9673
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.502015979798174e-05,
+      "loss": 0.3129,
+      "step": 9674
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5013356194151833e-05,
+      "loss": 0.3183,
+      "step": 9675
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.500655282049213e-05,
+      "loss": 0.3124,
+      "step": 9676
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.499974967763235e-05,
+      "loss": 0.306,
+      "step": 9677
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4992946766202214e-05,
+      "loss": 0.2989,
+      "step": 9678
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4986144086831426e-05,
+      "loss": 0.2947,
+      "step": 9679
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.497934164014965e-05,
+      "loss": 0.3072,
+      "step": 9680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4972539426786544e-05,
+      "loss": 0.3204,
+      "step": 9681
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4965737447371735e-05,
+      "loss": 0.2856,
+      "step": 9682
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.495893570253483e-05,
+      "loss": 0.3078,
+      "step": 9683
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4952134192905426e-05,
+      "loss": 0.3271,
+      "step": 9684
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4945332919113073e-05,
+      "loss": 0.3247,
+      "step": 9685
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.493853188178733e-05,
+      "loss": 0.3161,
+      "step": 9686
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4931731081557708e-05,
+      "loss": 0.3034,
+      "step": 9687
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4924930519053707e-05,
+      "loss": 0.3019,
+      "step": 9688
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4918130194904815e-05,
+      "loss": 0.3024,
+      "step": 9689
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4911330109740476e-05,
+      "loss": 0.3111,
+      "step": 9690
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.490453026419013e-05,
+      "loss": 0.3131,
+      "step": 9691
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4897730658883188e-05,
+      "loss": 0.3029,
+      "step": 9692
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4890931294449033e-05,
+      "loss": 0.3106,
+      "step": 9693
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4884132171517045e-05,
+      "loss": 0.3104,
+      "step": 9694
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.487733329071654e-05,
+      "loss": 0.3279,
+      "step": 9695
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.487053465267688e-05,
+      "loss": 0.3207,
+      "step": 9696
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4863736258027343e-05,
+      "loss": 0.2868,
+      "step": 9697
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4856938107397194e-05,
+      "loss": 0.3017,
+      "step": 9698
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.485014020141571e-05,
+      "loss": 0.3127,
+      "step": 9699
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4843342540712113e-05,
+      "loss": 0.3216,
+      "step": 9700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.483654512591561e-05,
+      "loss": 0.305,
+      "step": 9701
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.482974795765539e-05,
+      "loss": 0.305,
+      "step": 9702
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.482295103656062e-05,
+      "loss": 0.3028,
+      "step": 9703
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4816154363260436e-05,
+      "loss": 0.3036,
+      "step": 9704
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4809357938383952e-05,
+      "loss": 0.3041,
+      "step": 9705
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.480256176256027e-05,
+      "loss": 0.3048,
+      "step": 9706
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4795765836418467e-05,
+      "loss": 0.3161,
+      "step": 9707
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.478897016058757e-05,
+      "loss": 0.3169,
+      "step": 9708
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.478217473569663e-05,
+      "loss": 0.333,
+      "step": 9709
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.477537956237463e-05,
+      "loss": 0.2989,
+      "step": 9710
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.476858464125056e-05,
+      "loss": 0.3049,
+      "step": 9711
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4761789972953364e-05,
+      "loss": 0.3235,
+      "step": 9712
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4754995558111983e-05,
+      "loss": 0.3033,
+      "step": 9713
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.474820139735533e-05,
+      "loss": 0.3042,
+      "step": 9714
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.474140749131228e-05,
+      "loss": 0.2836,
+      "step": 9715
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4734613840611703e-05,
+      "loss": 0.3046,
+      "step": 9716
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4727820445882428e-05,
+      "loss": 0.3137,
+      "step": 9717
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4721027307753273e-05,
+      "loss": 0.2991,
+      "step": 9718
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.471423442685304e-05,
+      "loss": 0.2856,
+      "step": 9719
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.470744180381047e-05,
+      "loss": 0.2884,
+      "step": 9720
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.470064943925433e-05,
+      "loss": 0.2946,
+      "step": 9721
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.469385733381333e-05,
+      "loss": 0.3194,
+      "step": 9722
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4687065488116163e-05,
+      "loss": 0.3095,
+      "step": 9723
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4680273902791508e-05,
+      "loss": 0.3221,
+      "step": 9724
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.467348257846799e-05,
+      "loss": 0.317,
+      "step": 9725
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.466669151577426e-05,
+      "loss": 0.3094,
+      "step": 9726
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4659900715338895e-05,
+      "loss": 0.3229,
+      "step": 9727
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4653110177790484e-05,
+      "loss": 0.3076,
+      "step": 9728
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4646319903757565e-05,
+      "loss": 0.3103,
+      "step": 9729
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4639529893868668e-05,
+      "loss": 0.3061,
+      "step": 9730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.46327401487523e-05,
+      "loss": 0.3007,
+      "step": 9731
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.462595066903692e-05,
+      "loss": 0.3088,
+      "step": 9732
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4619161455351004e-05,
+      "loss": 0.2934,
+      "step": 9733
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4612372508322962e-05,
+      "loss": 0.3029,
+      "step": 9734
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4605583828581192e-05,
+      "loss": 0.2856,
+      "step": 9735
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4598795416754093e-05,
+      "loss": 0.315,
+      "step": 9736
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4592007273469998e-05,
+      "loss": 0.289,
+      "step": 9737
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4585219399357253e-05,
+      "loss": 0.3235,
+      "step": 9738
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4578431795044145e-05,
+      "loss": 0.306,
+      "step": 9739
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4571644461158954e-05,
+      "loss": 0.3305,
+      "step": 9740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.456485739832995e-05,
+      "loss": 0.3001,
+      "step": 9741
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4558070607185335e-05,
+      "loss": 0.3071,
+      "step": 9742
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4551284088353332e-05,
+      "loss": 0.3091,
+      "step": 9743
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.454449784246211e-05,
+      "loss": 0.3087,
+      "step": 9744
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.453771187013983e-05,
+      "loss": 0.3008,
+      "step": 9745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4530926172014608e-05,
+      "loss": 0.312,
+      "step": 9746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4524140748714554e-05,
+      "loss": 0.3079,
+      "step": 9747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4517355600867738e-05,
+      "loss": 0.3134,
+      "step": 9748
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4510570729102217e-05,
+      "loss": 0.2948,
+      "step": 9749
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.450378613404601e-05,
+      "loss": 0.3118,
+      "step": 9750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4497001816327124e-05,
+      "loss": 0.315,
+      "step": 9751
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4490217776573536e-05,
+      "loss": 0.3057,
+      "step": 9752
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4483434015413177e-05,
+      "loss": 0.3238,
+      "step": 9753
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4476650533473983e-05,
+      "loss": 0.3049,
+      "step": 9754
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.446986733138385e-05,
+      "loss": 0.3051,
+      "step": 9755
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4463084409770643e-05,
+      "loss": 0.3241,
+      "step": 9756
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.445630176926221e-05,
+      "loss": 0.3144,
+      "step": 9757
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.444951941048637e-05,
+      "loss": 0.3235,
+      "step": 9758
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.444273733407092e-05,
+      "loss": 0.3048,
+      "step": 9759
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.443595554064362e-05,
+      "loss": 0.3058,
+      "step": 9760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4429174030832208e-05,
+      "loss": 0.2948,
+      "step": 9761
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.442239280526441e-05,
+      "loss": 0.3197,
+      "step": 9762
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.44156118645679e-05,
+      "loss": 0.3177,
+      "step": 9763
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4408831209370347e-05,
+      "loss": 0.3193,
+      "step": 9764
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.440205084029939e-05,
+      "loss": 0.3134,
+      "step": 9765
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.439527075798263e-05,
+      "loss": 0.3116,
+      "step": 9766
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.438849096304765e-05,
+      "loss": 0.323,
+      "step": 9767
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4381711456122e-05,
+      "loss": 0.3087,
+      "step": 9768
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.437493223783323e-05,
+      "loss": 0.2983,
+      "step": 9769
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.436815330880883e-05,
+      "loss": 0.3054,
+      "step": 9770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4361374669676258e-05,
+      "loss": 0.3053,
+      "step": 9771
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4354596321062985e-05,
+      "loss": 0.3154,
+      "step": 9772
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4347818263596423e-05,
+      "loss": 0.3186,
+      "step": 9773
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4341040497903966e-05,
+      "loss": 0.3002,
+      "step": 9774
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4334263024612992e-05,
+      "loss": 0.3094,
+      "step": 9775
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4327485844350824e-05,
+      "loss": 0.2972,
+      "step": 9776
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4320708957744786e-05,
+      "loss": 0.2998,
+      "step": 9777
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4313932365422165e-05,
+      "loss": 0.3163,
+      "step": 9778
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4307156068010216e-05,
+      "loss": 0.3133,
+      "step": 9779
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.430038006613618e-05,
+      "loss": 0.312,
+      "step": 9780
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4293604360427245e-05,
+      "loss": 0.2911,
+      "step": 9781
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4286828951510605e-05,
+      "loss": 0.3217,
+      "step": 9782
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4280053840013395e-05,
+      "loss": 0.3309,
+      "step": 9783
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4273279026562746e-05,
+      "loss": 0.3042,
+      "step": 9784
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.426650451178575e-05,
+      "loss": 0.3179,
+      "step": 9785
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.425973029630946e-05,
+      "loss": 0.3139,
+      "step": 9786
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4252956380760943e-05,
+      "loss": 0.3162,
+      "step": 9787
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.424618276576719e-05,
+      "loss": 0.2832,
+      "step": 9788
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4239409451955193e-05,
+      "loss": 0.3024,
+      "step": 9789
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.42326364399519e-05,
+      "loss": 0.3238,
+      "step": 9790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4225863730384235e-05,
+      "loss": 0.2853,
+      "step": 9791
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4219091323879116e-05,
+      "loss": 0.3228,
+      "step": 9792
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4212319221063396e-05,
+      "loss": 0.3105,
+      "step": 9793
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4205547422563933e-05,
+      "loss": 0.2997,
+      "step": 9794
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4198775929007533e-05,
+      "loss": 0.3242,
+      "step": 9795
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.419200474102098e-05,
+      "loss": 0.3144,
+      "step": 9796
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4185233859231042e-05,
+      "loss": 0.3096,
+      "step": 9797
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.417846328426444e-05,
+      "loss": 0.2954,
+      "step": 9798
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4171693016747886e-05,
+      "loss": 0.3157,
+      "step": 9799
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4164923057308046e-05,
+      "loss": 0.3203,
+      "step": 9800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.415815340657157e-05,
+      "loss": 0.3109,
+      "step": 9801
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4151384065165077e-05,
+      "loss": 0.3215,
+      "step": 9802
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.414461503371514e-05,
+      "loss": 0.2974,
+      "step": 9803
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4137846312848326e-05,
+      "loss": 0.2816,
+      "step": 9804
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4131077903191175e-05,
+      "loss": 0.3051,
+      "step": 9805
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4124309805370178e-05,
+      "loss": 0.3026,
+      "step": 9806
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4117542020011813e-05,
+      "loss": 0.32,
+      "step": 9807
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4110774547742514e-05,
+      "loss": 0.3117,
+      "step": 9808
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.410400738918871e-05,
+      "loss": 0.2936,
+      "step": 9809
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.409724054497677e-05,
+      "loss": 0.2984,
+      "step": 9810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4090474015733073e-05,
+      "loss": 0.3194,
+      "step": 9811
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4083707802083928e-05,
+      "loss": 0.2991,
+      "step": 9812
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.407694190465563e-05,
+      "loss": 0.3007,
+      "step": 9813
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4070176324074467e-05,
+      "loss": 0.3026,
+      "step": 9814
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.406341106096666e-05,
+      "loss": 0.3065,
+      "step": 9815
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4056646115958436e-05,
+      "loss": 0.3058,
+      "step": 9816
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4049881489675967e-05,
+      "loss": 0.3048,
+      "step": 9817
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4043117182745407e-05,
+      "loss": 0.3041,
+      "step": 9818
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4036353195792866e-05,
+      "loss": 0.3078,
+      "step": 9819
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4029589529444453e-05,
+      "loss": 0.3015,
+      "step": 9820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4022826184326225e-05,
+      "loss": 0.335,
+      "step": 9821
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.401606316106421e-05,
+      "loss": 0.3051,
+      "step": 9822
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4009300460284413e-05,
+      "loss": 0.3067,
+      "step": 9823
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4002538082612807e-05,
+      "loss": 0.2916,
+      "step": 9824
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3995776028675337e-05,
+      "loss": 0.3099,
+      "step": 9825
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3989014299097923e-05,
+      "loss": 0.3276,
+      "step": 9826
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3982252894506435e-05,
+      "loss": 0.3021,
+      "step": 9827
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.397549181552673e-05,
+      "loss": 0.2833,
+      "step": 9828
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3968731062784635e-05,
+      "loss": 0.305,
+      "step": 9829
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3961970636905933e-05,
+      "loss": 0.3132,
+      "step": 9830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.395521053851641e-05,
+      "loss": 0.2886,
+      "step": 9831
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3948450768241776e-05,
+      "loss": 0.3176,
+      "step": 9832
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3941691326707733e-05,
+      "loss": 0.3037,
+      "step": 9833
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.393493221453997e-05,
+      "loss": 0.3194,
+      "step": 9834
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3928173432364104e-05,
+      "loss": 0.3199,
+      "step": 9835
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.392141498080576e-05,
+      "loss": 0.3075,
+      "step": 9836
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3914656860490525e-05,
+      "loss": 0.3007,
+      "step": 9837
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3907899072043932e-05,
+      "loss": 0.3066,
+      "step": 9838
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.390114161609151e-05,
+      "loss": 0.3011,
+      "step": 9839
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3894384493258736e-05,
+      "loss": 0.2995,
+      "step": 9840
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3887627704171078e-05,
+      "loss": 0.3045,
+      "step": 9841
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3880871249453955e-05,
+      "loss": 0.3034,
+      "step": 9842
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3874115129732772e-05,
+      "loss": 0.3011,
+      "step": 9843
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3867359345632884e-05,
+      "loss": 0.3165,
+      "step": 9844
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.386060389777962e-05,
+      "loss": 0.315,
+      "step": 9845
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3853848786798293e-05,
+      "loss": 0.3163,
+      "step": 9846
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.384709401331416e-05,
+      "loss": 0.3099,
+      "step": 9847
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3840339577952477e-05,
+      "loss": 0.3034,
+      "step": 9848
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.383358548133844e-05,
+      "loss": 0.31,
+      "step": 9849
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.382683172409722e-05,
+      "loss": 0.3054,
+      "step": 9850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3820078306853988e-05,
+      "loss": 0.2989,
+      "step": 9851
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3813325230233824e-05,
+      "loss": 0.3321,
+      "step": 9852
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.380657249486184e-05,
+      "loss": 0.2957,
+      "step": 9853
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3799820101363067e-05,
+      "loss": 0.3158,
+      "step": 9854
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3793068050362533e-05,
+      "loss": 0.3073,
+      "step": 9855
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.378631634248522e-05,
+      "loss": 0.3094,
+      "step": 9856
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.377956497835609e-05,
+      "loss": 0.3014,
+      "step": 9857
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3772813958600064e-05,
+      "loss": 0.3302,
+      "step": 9858
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3766063283842033e-05,
+      "loss": 0.324,
+      "step": 9859
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.375931295470685e-05,
+      "loss": 0.3056,
+      "step": 9860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3752562971819365e-05,
+      "loss": 0.3158,
+      "step": 9861
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3745813335804346e-05,
+      "loss": 0.3113,
+      "step": 9862
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.373906404728658e-05,
+      "loss": 0.2981,
+      "step": 9863
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3732315106890785e-05,
+      "loss": 0.2983,
+      "step": 9864
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3725566515241654e-05,
+      "loss": 0.3258,
+      "step": 9865
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.371881827296388e-05,
+      "loss": 0.3069,
+      "step": 9866
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3712070380682066e-05,
+      "loss": 0.2949,
+      "step": 9867
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3705322839020844e-05,
+      "loss": 0.3212,
+      "step": 9868
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.369857564860477e-05,
+      "loss": 0.2988,
+      "step": 9869
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3691828810058376e-05,
+      "loss": 0.2949,
+      "step": 9870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3685082324006168e-05,
+      "loss": 0.3142,
+      "step": 9871
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.367833619107263e-05,
+      "loss": 0.3097,
+      "step": 9872
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3671590411882193e-05,
+      "loss": 0.2971,
+      "step": 9873
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3664844987059265e-05,
+      "loss": 0.3061,
+      "step": 9874
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3658099917228215e-05,
+      "loss": 0.3027,
+      "step": 9875
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3651355203013396e-05,
+      "loss": 0.3345,
+      "step": 9876
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3644610845039104e-05,
+      "loss": 0.3166,
+      "step": 9877
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3637866843929635e-05,
+      "loss": 0.301,
+      "step": 9878
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3631123200309202e-05,
+      "loss": 0.3116,
+      "step": 9879
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.362437991480203e-05,
+      "loss": 0.2923,
+      "step": 9880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3617636988032292e-05,
+      "loss": 0.3079,
+      "step": 9881
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3610894420624135e-05,
+      "loss": 0.3009,
+      "step": 9882
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3604152213201658e-05,
+      "loss": 0.3075,
+      "step": 9883
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.359741036638895e-05,
+      "loss": 0.3088,
+      "step": 9884
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3590668880810053e-05,
+      "loss": 0.3083,
+      "step": 9885
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3583927757088958e-05,
+      "loss": 0.3061,
+      "step": 9886
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.357718699584967e-05,
+      "loss": 0.3146,
+      "step": 9887
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3570446597716107e-05,
+      "loss": 0.3019,
+      "step": 9888
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3563706563312183e-05,
+      "loss": 0.2934,
+      "step": 9889
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3556966893261785e-05,
+      "loss": 0.291,
+      "step": 9890
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.355022758818874e-05,
+      "loss": 0.304,
+      "step": 9891
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3543488648716865e-05,
+      "loss": 0.3082,
+      "step": 9892
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3536750075469925e-05,
+      "loss": 0.2989,
+      "step": 9893
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.353001186907167e-05,
+      "loss": 0.3134,
+      "step": 9894
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3523274030145802e-05,
+      "loss": 0.32,
+      "step": 9895
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3516536559315987e-05,
+      "loss": 0.2984,
+      "step": 9896
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.350979945720587e-05,
+      "loss": 0.3085,
+      "step": 9897
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.350306272443905e-05,
+      "loss": 0.3076,
+      "step": 9898
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3496326361639103e-05,
+      "loss": 0.3021,
+      "step": 9899
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.348959036942956e-05,
+      "loss": 0.2955,
+      "step": 9900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3482854748433914e-05,
+      "loss": 0.302,
+      "step": 9901
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3476119499275645e-05,
+      "loss": 0.3064,
+      "step": 9902
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3469384622578173e-05,
+      "loss": 0.3289,
+      "step": 9903
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3462650118964908e-05,
+      "loss": 0.2956,
+      "step": 9904
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3455915989059208e-05,
+      "loss": 0.3018,
+      "step": 9905
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3449182233484395e-05,
+      "loss": 0.3118,
+      "step": 9906
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3442448852863775e-05,
+      "loss": 0.3142,
+      "step": 9907
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3435715847820587e-05,
+      "loss": 0.3185,
+      "step": 9908
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3428983218978084e-05,
+      "loss": 0.3093,
+      "step": 9909
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3422250966959437e-05,
+      "loss": 0.3003,
+      "step": 9910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3415519092387793e-05,
+      "loss": 0.2957,
+      "step": 9911
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3408787595886296e-05,
+      "loss": 0.297,
+      "step": 9912
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.340205647807801e-05,
+      "loss": 0.3176,
+      "step": 9913
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3395325739585992e-05,
+      "loss": 0.3028,
+      "step": 9914
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3388595381033263e-05,
+      "loss": 0.3085,
+      "step": 9915
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.338186540304279e-05,
+      "loss": 0.2928,
+      "step": 9916
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3375135806237525e-05,
+      "loss": 0.321,
+      "step": 9917
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3368406591240376e-05,
+      "loss": 0.3178,
+      "step": 9918
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3361677758674215e-05,
+      "loss": 0.2861,
+      "step": 9919
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.335494930916189e-05,
+      "loss": 0.3138,
+      "step": 9920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.334822124332618e-05,
+      "loss": 0.3174,
+      "step": 9921
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.334149356178988e-05,
+      "loss": 0.3055,
+      "step": 9922
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.33347662651757e-05,
+      "loss": 0.3096,
+      "step": 9923
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3328039354106346e-05,
+      "loss": 0.2947,
+      "step": 9924
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.332131282920448e-05,
+      "loss": 0.3056,
+      "step": 9925
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3314586691092722e-05,
+      "loss": 0.3114,
+      "step": 9926
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3307860940393662e-05,
+      "loss": 0.3064,
+      "step": 9927
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.330113557772985e-05,
+      "loss": 0.3103,
+      "step": 9928
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.329441060372381e-05,
+      "loss": 0.2963,
+      "step": 9929
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3287686018998016e-05,
+      "loss": 0.3198,
+      "step": 9930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3280961824174922e-05,
+      "loss": 0.2851,
+      "step": 9931
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.327423801987693e-05,
+      "loss": 0.2919,
+      "step": 9932
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.326751460672641e-05,
+      "loss": 0.3071,
+      "step": 9933
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3260791585345708e-05,
+      "loss": 0.3108,
+      "step": 9934
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3254068956357112e-05,
+      "loss": 0.3064,
+      "step": 9935
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3247346720382896e-05,
+      "loss": 0.3141,
+      "step": 9936
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3240624878045285e-05,
+      "loss": 0.3168,
+      "step": 9937
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.323390342996647e-05,
+      "loss": 0.3092,
+      "step": 9938
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3227182376768604e-05,
+      "loss": 0.3241,
+      "step": 9939
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.32204617190738e-05,
+      "loss": 0.3158,
+      "step": 9940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.321374145750415e-05,
+      "loss": 0.3006,
+      "step": 9941
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.32070215926817e-05,
+      "loss": 0.3106,
+      "step": 9942
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3200302125228445e-05,
+      "loss": 0.3273,
+      "step": 9943
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3193583055766367e-05,
+      "loss": 0.305,
+      "step": 9944
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3186864384917394e-05,
+      "loss": 0.3085,
+      "step": 9945
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.318014611330343e-05,
+      "loss": 0.3297,
+      "step": 9946
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3173428241546336e-05,
+      "loss": 0.2873,
+      "step": 9947
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3166710770267924e-05,
+      "loss": 0.3064,
+      "step": 9948
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3159993700089993e-05,
+      "loss": 0.3104,
+      "step": 9949
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3153277031634286e-05,
+      "loss": 0.2995,
+      "step": 9950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3146560765522523e-05,
+      "loss": 0.3138,
+      "step": 9951
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3139844902376368e-05,
+      "loss": 0.3027,
+      "step": 9952
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3133129442817464e-05,
+      "loss": 0.282,
+      "step": 9953
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3126414387467416e-05,
+      "loss": 0.2918,
+      "step": 9954
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3119699736947774e-05,
+      "loss": 0.296,
+      "step": 9955
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.311298549188008e-05,
+      "loss": 0.3215,
+      "step": 9956
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.310627165288581e-05,
+      "loss": 0.2941,
+      "step": 9957
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3099558220586414e-05,
+      "loss": 0.2995,
+      "step": 9958
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3092845195603313e-05,
+      "loss": 0.3103,
+      "step": 9959
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3086132578557872e-05,
+      "loss": 0.3223,
+      "step": 9960
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3079420370071442e-05,
+      "loss": 0.3016,
+      "step": 9961
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.307270857076531e-05,
+      "loss": 0.3087,
+      "step": 9962
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3065997181260745e-05,
+      "loss": 0.2898,
+      "step": 9963
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3059286202178965e-05,
+      "loss": 0.3228,
+      "step": 9964
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3052575634141155e-05,
+      "loss": 0.2872,
+      "step": 9965
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.304586547776847e-05,
+      "loss": 0.2832,
+      "step": 9966
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3039155733682017e-05,
+      "loss": 0.2972,
+      "step": 9967
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3032446402502856e-05,
+      "loss": 0.3181,
+      "step": 9968
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3025737484852033e-05,
+      "loss": 0.2933,
+      "step": 9969
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3019028981350544e-05,
+      "loss": 0.3023,
+      "step": 9970
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3012320892619337e-05,
+      "loss": 0.2981,
+      "step": 9971
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.300561321927933e-05,
+      "loss": 0.3148,
+      "step": 9972
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2998905961951413e-05,
+      "loss": 0.311,
+      "step": 9973
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.299219912125642e-05,
+      "loss": 0.3109,
+      "step": 9974
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2985492697815148e-05,
+      "loss": 0.3037,
+      "step": 9975
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2978786692248372e-05,
+      "loss": 0.3243,
+      "step": 9976
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2972081105176807e-05,
+      "loss": 0.3117,
+      "step": 9977
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.296537593722115e-05,
+      "loss": 0.3034,
+      "step": 9978
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2958671189002044e-05,
+      "loss": 0.3194,
+      "step": 9979
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2951966861140088e-05,
+      "loss": 0.3404,
+      "step": 9980
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2945262954255868e-05,
+      "loss": 0.3201,
+      "step": 9981
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2938559468969907e-05,
+      "loss": 0.3197,
+      "step": 9982
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.29318564059027e-05,
+      "loss": 0.2945,
+      "step": 9983
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.29251537656747e-05,
+      "loss": 0.3039,
+      "step": 9984
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2918451548906312e-05,
+      "loss": 0.3224,
+      "step": 9985
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2911749756217927e-05,
+      "loss": 0.2985,
+      "step": 9986
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.290504838822986e-05,
+      "loss": 0.296,
+      "step": 9987
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2898347445562423e-05,
+      "loss": 0.3321,
+      "step": 9988
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2891646928835868e-05,
+      "loss": 0.3079,
+      "step": 9989
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2884946838670405e-05,
+      "loss": 0.3095,
+      "step": 9990
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.287824717568623e-05,
+      "loss": 0.2872,
+      "step": 9991
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2871547940503458e-05,
+      "loss": 0.3138,
+      "step": 9992
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.28648491337422e-05,
+      "loss": 0.3249,
+      "step": 9993
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2858150756022517e-05,
+      "loss": 0.3101,
+      "step": 9994
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2851452807964425e-05,
+      "loss": 0.2957,
+      "step": 9995
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2844755290187903e-05,
+      "loss": 0.3091,
+      "step": 9996
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.283805820331289e-05,
+      "loss": 0.3202,
+      "step": 9997
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2831361547959284e-05,
+      "loss": 0.2968,
+      "step": 9998
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2824665324746953e-05,
+      "loss": 0.3105,
+      "step": 9999
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2817969534295707e-05,
+      "loss": 0.3131,
+      "step": 10000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2811274177225332e-05,
+      "loss": 0.3247,
+      "step": 10001
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.280457925415556e-05,
+      "loss": 0.3166,
+      "step": 10002
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2797884765706102e-05,
+      "loss": 0.313,
+      "step": 10003
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.279119071249662e-05,
+      "loss": 0.3302,
+      "step": 10004
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2784497095146713e-05,
+      "loss": 0.3089,
+      "step": 10005
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2777803914275968e-05,
+      "loss": 0.297,
+      "step": 10006
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2771111170503934e-05,
+      "loss": 0.3154,
+      "step": 10007
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2764418864450087e-05,
+      "loss": 0.2924,
+      "step": 10008
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2757726996733907e-05,
+      "loss": 0.3054,
+      "step": 10009
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2751035567974804e-05,
+      "loss": 0.3097,
+      "step": 10010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2744344578792143e-05,
+      "loss": 0.3048,
+      "step": 10011
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.273765402980527e-05,
+      "loss": 0.3117,
+      "step": 10012
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.273096392163348e-05,
+      "loss": 0.304,
+      "step": 10013
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2724274254896023e-05,
+      "loss": 0.3166,
+      "step": 10014
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2717585030212113e-05,
+      "loss": 0.3228,
+      "step": 10015
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.271089624820092e-05,
+      "loss": 0.3129,
+      "step": 10016
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2704207909481577e-05,
+      "loss": 0.3056,
+      "step": 10017
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2697520014673167e-05,
+      "loss": 0.2919,
+      "step": 10018
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2690832564394753e-05,
+      "loss": 0.296,
+      "step": 10019
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.268414555926534e-05,
+      "loss": 0.2804,
+      "step": 10020
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2677458999903877e-05,
+      "loss": 0.3044,
+      "step": 10021
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2670772886929314e-05,
+      "loss": 0.3122,
+      "step": 10022
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.266408722096051e-05,
+      "loss": 0.3015,
+      "step": 10023
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2657402002616335e-05,
+      "loss": 0.3273,
+      "step": 10024
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2650717232515568e-05,
+      "loss": 0.3128,
+      "step": 10025
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2644032911276974e-05,
+      "loss": 0.3123,
+      "step": 10026
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.263734903951928e-05,
+      "loss": 0.3308,
+      "step": 10027
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2630665617861147e-05,
+      "loss": 0.3137,
+      "step": 10028
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.262398264692122e-05,
+      "loss": 0.3147,
+      "step": 10029
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2617300127318096e-05,
+      "loss": 0.3038,
+      "step": 10030
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.261061805967032e-05,
+      "loss": 0.3097,
+      "step": 10031
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.26039364445964e-05,
+      "loss": 0.3148,
+      "step": 10032
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.259725528271481e-05,
+      "loss": 0.2904,
+      "step": 10033
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.259057457464397e-05,
+      "loss": 0.3181,
+      "step": 10034
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2583894321002264e-05,
+      "loss": 0.2878,
+      "step": 10035
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.257721452240803e-05,
+      "loss": 0.2894,
+      "step": 10036
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.257053517947958e-05,
+      "loss": 0.3048,
+      "step": 10037
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2563856292835155e-05,
+      "loss": 0.3026,
+      "step": 10038
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2557177863092992e-05,
+      "loss": 0.3078,
+      "step": 10039
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2550499890871247e-05,
+      "loss": 0.3067,
+      "step": 10040
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2543822376788044e-05,
+      "loss": 0.3103,
+      "step": 10041
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.253714532146149e-05,
+      "loss": 0.3077,
+      "step": 10042
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2530468725509618e-05,
+      "loss": 0.3001,
+      "step": 10043
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2523792589550436e-05,
+      "loss": 0.3173,
+      "step": 10044
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2517116914201903e-05,
+      "loss": 0.3209,
+      "step": 10045
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2510441700081936e-05,
+      "loss": 0.2944,
+      "step": 10046
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2503766947808414e-05,
+      "loss": 0.2914,
+      "step": 10047
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2497092657999158e-05,
+      "loss": 0.3093,
+      "step": 10048
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.249041883127197e-05,
+      "loss": 0.3039,
+      "step": 10049
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.24837454682446e-05,
+      "loss": 0.3124,
+      "step": 10050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.247707256953473e-05,
+      "loss": 0.3049,
+      "step": 10051
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2470400135760046e-05,
+      "loss": 0.3207,
+      "step": 10052
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.246372816753815e-05,
+      "loss": 0.303,
+      "step": 10053
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.245705666548662e-05,
+      "loss": 0.3039,
+      "step": 10054
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2450385630222992e-05,
+      "loss": 0.3011,
+      "step": 10055
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.244371506236475e-05,
+      "loss": 0.3119,
+      "step": 10056
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.243704496252934e-05,
+      "loss": 0.3062,
+      "step": 10057
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2430375331334162e-05,
+      "loss": 0.3059,
+      "step": 10058
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2423706169396584e-05,
+      "loss": 0.3241,
+      "step": 10059
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2417037477333908e-05,
+      "loss": 0.3463,
+      "step": 10060
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2410369255763414e-05,
+      "loss": 0.2915,
+      "step": 10061
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2403701505302322e-05,
+      "loss": 0.2975,
+      "step": 10062
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.239703422656782e-05,
+      "loss": 0.3062,
+      "step": 10063
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2390367420177054e-05,
+      "loss": 0.3059,
+      "step": 10064
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2383701086747117e-05,
+      "loss": 0.3024,
+      "step": 10065
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2377035226895053e-05,
+      "loss": 0.3138,
+      "step": 10066
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2370369841237887e-05,
+      "loss": 0.2895,
+      "step": 10067
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.2363704930392575e-05,
+      "loss": 0.325,
+      "step": 10068
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2357040494976038e-05,
+      "loss": 0.3035,
+      "step": 10069
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.235037653560516e-05,
+      "loss": 0.3142,
+      "step": 10070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.234371305289677e-05,
+      "loss": 0.3007,
+      "step": 10071
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2337050047467655e-05,
+      "loss": 0.3249,
+      "step": 10072
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.233038751993456e-05,
+      "loss": 0.3079,
+      "step": 10073
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2323725470914197e-05,
+      "loss": 0.2988,
+      "step": 10074
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2317063901023206e-05,
+      "loss": 0.3057,
+      "step": 10075
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2310402810878204e-05,
+      "loss": 0.3083,
+      "step": 10076
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2303742201095772e-05,
+      "loss": 0.3273,
+      "step": 10077
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2297082072292413e-05,
+      "loss": 0.287,
+      "step": 10078
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2290422425084622e-05,
+      "loss": 0.3367,
+      "step": 10079
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2283763260088826e-05,
+      "loss": 0.3241,
+      "step": 10080
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2277104577921417e-05,
+      "loss": 0.291,
+      "step": 10081
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2270446379198733e-05,
+      "loss": 0.2994,
+      "step": 10082
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.226378866453708e-05,
+      "loss": 0.3516,
+      "step": 10083
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.225713143455272e-05,
+      "loss": 0.301,
+      "step": 10084
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2250474689861846e-05,
+      "loss": 0.3058,
+      "step": 10085
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2243818431080642e-05,
+      "loss": 0.3001,
+      "step": 10086
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2237162658825215e-05,
+      "loss": 0.3111,
+      "step": 10087
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2230507373711647e-05,
+      "loss": 0.3168,
+      "step": 10088
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.222385257635597e-05,
+      "loss": 0.3126,
+      "step": 10089
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2217198267374162e-05,
+      "loss": 0.3212,
+      "step": 10090
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.221054444738217e-05,
+      "loss": 0.3208,
+      "step": 10091
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.220389111699589e-05,
+      "loss": 0.3196,
+      "step": 10092
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2197238276831162e-05,
+      "loss": 0.3088,
+      "step": 10093
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2190585927503804e-05,
+      "loss": 0.3153,
+      "step": 10094
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.218393406962956e-05,
+      "loss": 0.2989,
+      "step": 10095
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2177282703824148e-05,
+      "loss": 0.2983,
+      "step": 10096
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2170631830703242e-05,
+      "loss": 0.3149,
+      "step": 10097
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.216398145088246e-05,
+      "loss": 0.3109,
+      "step": 10098
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.215733156497738e-05,
+      "loss": 0.3007,
+      "step": 10099
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.215068217360352e-05,
+      "loss": 0.2999,
+      "step": 10100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2144033277376385e-05,
+      "loss": 0.3235,
+      "step": 10101
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.21373848769114e-05,
+      "loss": 0.3192,
+      "step": 10102
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2130736972823967e-05,
+      "loss": 0.3145,
+      "step": 10103
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2124089565729426e-05,
+      "loss": 0.3212,
+      "step": 10104
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.211744265624308e-05,
+      "loss": 0.3018,
+      "step": 10105
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.211079624498019e-05,
+      "loss": 0.2933,
+      "step": 10106
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.210415033255596e-05,
+      "loss": 0.307,
+      "step": 10107
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2097504919585555e-05,
+      "loss": 0.2934,
+      "step": 10108
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.209086000668409e-05,
+      "loss": 0.2931,
+      "step": 10109
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2084215594466627e-05,
+      "loss": 0.3023,
+      "step": 10110
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2077571683548203e-05,
+      "loss": 0.3232,
+      "step": 10111
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.207092827454379e-05,
+      "loss": 0.3262,
+      "step": 10112
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2064285368068322e-05,
+      "loss": 0.298,
+      "step": 10113
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2057642964736685e-05,
+      "loss": 0.3161,
+      "step": 10114
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2051001065163707e-05,
+      "loss": 0.2928,
+      "step": 10115
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2044359669964192e-05,
+      "loss": 0.3299,
+      "step": 10116
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.203771877975287e-05,
+      "loss": 0.3169,
+      "step": 10117
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2031078395144454e-05,
+      "loss": 0.2988,
+      "step": 10118
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2024438516753594e-05,
+      "loss": 0.3073,
+      "step": 10119
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2017799145194883e-05,
+      "loss": 0.3113,
+      "step": 10120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2011160281082887e-05,
+      "loss": 0.311,
+      "step": 10121
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.200452192503211e-05,
+      "loss": 0.2959,
+      "step": 10122
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.199788407765703e-05,
+      "loss": 0.3189,
+      "step": 10123
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1991246739572045e-05,
+      "loss": 0.3032,
+      "step": 10124
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1984609911391535e-05,
+      "loss": 0.3194,
+      "step": 10125
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1977973593729816e-05,
+      "loss": 0.322,
+      "step": 10126
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1971337787201168e-05,
+      "loss": 0.2909,
+      "step": 10127
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.196470249241982e-05,
+      "loss": 0.2984,
+      "step": 10128
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1958067709999946e-05,
+      "loss": 0.3256,
+      "step": 10129
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1951433440555685e-05,
+      "loss": 0.3145,
+      "step": 10130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1944799684701117e-05,
+      "loss": 0.3075,
+      "step": 10131
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.193816644305028e-05,
+      "loss": 0.3073,
+      "step": 10132
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1931533716217167e-05,
+      "loss": 0.3285,
+      "step": 10133
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.192490150481571e-05,
+      "loss": 0.3039,
+      "step": 10134
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1918269809459818e-05,
+      "loss": 0.306,
+      "step": 10135
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.191163863076333e-05,
+      "loss": 0.3018,
+      "step": 10136
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1905007969340053e-05,
+      "loss": 0.2968,
+      "step": 10137
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1898377825803724e-05,
+      "loss": 0.3114,
+      "step": 10138
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.189174820076806e-05,
+      "loss": 0.2939,
+      "step": 10139
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.188511909484671e-05,
+      "loss": 0.2911,
+      "step": 10140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1878490508653283e-05,
+      "loss": 0.3076,
+      "step": 10141
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1871862442801333e-05,
+      "loss": 0.31,
+      "step": 10142
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.186523489790438e-05,
+      "loss": 0.3031,
+      "step": 10143
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.185860787457588e-05,
+      "loss": 0.3121,
+      "step": 10144
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.185198137342925e-05,
+      "loss": 0.2983,
+      "step": 10145
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1845355395077854e-05,
+      "loss": 0.3045,
+      "step": 10146
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.183872994013501e-05,
+      "loss": 0.3212,
+      "step": 10147
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1832105009213988e-05,
+      "loss": 0.2921,
+      "step": 10148
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.182548060292801e-05,
+      "loss": 0.3109,
+      "step": 10149
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.181885672189026e-05,
+      "loss": 0.3075,
+      "step": 10150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.181223336671383e-05,
+      "loss": 0.3206,
+      "step": 10151
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1805610538011822e-05,
+      "loss": 0.3079,
+      "step": 10152
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.179898823639725e-05,
+      "loss": 0.2875,
+      "step": 10153
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1792366462483098e-05,
+      "loss": 0.306,
+      "step": 10154
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.17857452168823e-05,
+      "loss": 0.3303,
+      "step": 10155
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.177912450020772e-05,
+      "loss": 0.3166,
+      "step": 10156
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.17725043130722e-05,
+      "loss": 0.3193,
+      "step": 10157
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.176588465608851e-05,
+      "loss": 0.3188,
+      "step": 10158
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.17592655298694e-05,
+      "loss": 0.2979,
+      "step": 10159
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.175264693502754e-05,
+      "loss": 0.3085,
+      "step": 10160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1746028872175567e-05,
+      "loss": 0.3147,
+      "step": 10161
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.173941134192607e-05,
+      "loss": 0.3209,
+      "step": 10162
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.173279434489158e-05,
+      "loss": 0.3069,
+      "step": 10163
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.172617788168459e-05,
+      "loss": 0.3087,
+      "step": 10164
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.171956195291753e-05,
+      "loss": 0.3283,
+      "step": 10165
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1712946559202786e-05,
+      "loss": 0.3116,
+      "step": 10166
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1706331701152705e-05,
+      "loss": 0.2997,
+      "step": 10167
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1699717379379564e-05,
+      "loss": 0.2905,
+      "step": 10168
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1693103594495613e-05,
+      "loss": 0.3075,
+      "step": 10169
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1686490347113037e-05,
+      "loss": 0.3249,
+      "step": 10170
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.167987763784397e-05,
+      "loss": 0.3188,
+      "step": 10171
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1673265467300514e-05,
+      "loss": 0.3174,
+      "step": 10172
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.166665383609469e-05,
+      "loss": 0.3236,
+      "step": 10173
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.166004274483851e-05,
+      "loss": 0.3226,
+      "step": 10174
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1653432194143898e-05,
+      "loss": 0.2829,
+      "step": 10175
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1646822184622738e-05,
+      "loss": 0.314,
+      "step": 10176
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1640212716886893e-05,
+      "loss": 0.2958,
+      "step": 10177
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.163360379154813e-05,
+      "loss": 0.3023,
+      "step": 10178
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1626995409218203e-05,
+      "loss": 0.2985,
+      "step": 10179
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1620387570508794e-05,
+      "loss": 0.3012,
+      "step": 10180
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1613780276031542e-05,
+      "loss": 0.3045,
+      "step": 10181
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1607173526398032e-05,
+      "loss": 0.3203,
+      "step": 10182
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.160056732221981e-05,
+      "loss": 0.3,
+      "step": 10183
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1593961664108363e-05,
+      "loss": 0.3107,
+      "step": 10184
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1587356552675116e-05,
+      "loss": 0.3252,
+      "step": 10185
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1580751988531466e-05,
+      "loss": 0.3097,
+      "step": 10186
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1574147972288757e-05,
+      "loss": 0.3011,
+      "step": 10187
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1567544504558245e-05,
+      "loss": 0.3023,
+      "step": 10188
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1560941585951194e-05,
+      "loss": 0.3246,
+      "step": 10189
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1554339217078767e-05,
+      "loss": 0.2941,
+      "step": 10190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.154773739855211e-05,
+      "loss": 0.3249,
+      "step": 10191
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1541136130982296e-05,
+      "loss": 0.2974,
+      "step": 10192
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.153453541498036e-05,
+      "loss": 0.3162,
+      "step": 10193
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1527935251157283e-05,
+      "loss": 0.3117,
+      "step": 10194
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1521335640123987e-05,
+      "loss": 0.3016,
+      "step": 10195
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1514736582491352e-05,
+      "loss": 0.2994,
+      "step": 10196
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1508138078870205e-05,
+      "loss": 0.2826,
+      "step": 10197
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.150154012987131e-05,
+      "loss": 0.3102,
+      "step": 10198
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1494942736105416e-05,
+      "loss": 0.3276,
+      "step": 10199
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1488345898183165e-05,
+      "loss": 0.3085,
+      "step": 10200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1481749616715197e-05,
+      "loss": 0.2936,
+      "step": 10201
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1475153892312078e-05,
+      "loss": 0.2962,
+      "step": 10202
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1468558725584314e-05,
+      "loss": 0.3155,
+      "step": 10203
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1461964117142387e-05,
+      "loss": 0.3279,
+      "step": 10204
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1455370067596698e-05,
+      "loss": 0.2997,
+      "step": 10205
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1448776577557616e-05,
+      "loss": 0.2981,
+      "step": 10206
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1442183647635448e-05,
+      "loss": 0.297,
+      "step": 10207
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.143559127844046e-05,
+      "loss": 0.3241,
+      "step": 10208
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.142899947058285e-05,
+      "loss": 0.3166,
+      "step": 10209
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1422408224672767e-05,
+      "loss": 0.3108,
+      "step": 10210
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1415817541320338e-05,
+      "loss": 0.3112,
+      "step": 10211
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.140922742113559e-05,
+      "loss": 0.3079,
+      "step": 10212
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.140263786472853e-05,
+      "loss": 0.291,
+      "step": 10213
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1396048872709112e-05,
+      "loss": 0.2966,
+      "step": 10214
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1389460445687215e-05,
+      "loss": 0.3127,
+      "step": 10215
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.138287258427269e-05,
+      "loss": 0.3035,
+      "step": 10216
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1376285289075333e-05,
+      "loss": 0.2981,
+      "step": 10217
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1369698560704858e-05,
+      "loss": 0.2883,
+      "step": 10218
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.136311239977098e-05,
+      "loss": 0.3127,
+      "step": 10219
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1356526806883303e-05,
+      "loss": 0.3025,
+      "step": 10220
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1349941782651432e-05,
+      "loss": 0.3032,
+      "step": 10221
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1343357327684882e-05,
+      "loss": 0.2864,
+      "step": 10222
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1336773442593117e-05,
+      "loss": 0.3049,
+      "step": 10223
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.133019012798557e-05,
+      "loss": 0.3058,
+      "step": 10224
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1323607384471605e-05,
+      "loss": 0.3049,
+      "step": 10225
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.131702521266055e-05,
+      "loss": 0.2966,
+      "step": 10226
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.131044361316165e-05,
+      "loss": 0.305,
+      "step": 10227
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1303862586584133e-05,
+      "loss": 0.3163,
+      "step": 10228
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1297282133537137e-05,
+      "loss": 0.2796,
+      "step": 10229
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.129070225462978e-05,
+      "loss": 0.2968,
+      "step": 10230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.128412295047111e-05,
+      "loss": 0.3097,
+      "step": 10231
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1277544221670116e-05,
+      "loss": 0.2937,
+      "step": 10232
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1270966068835746e-05,
+      "loss": 0.3058,
+      "step": 10233
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1264388492576903e-05,
+      "loss": 0.3165,
+      "step": 10234
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1257811493502407e-05,
+      "loss": 0.3271,
+      "step": 10235
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1251235072221055e-05,
+      "loss": 0.3059,
+      "step": 10236
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.124465922934156e-05,
+      "loss": 0.3406,
+      "step": 10237
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.123808396547263e-05,
+      "loss": 0.3056,
+      "step": 10238
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1231509281222856e-05,
+      "loss": 0.3039,
+      "step": 10239
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1224935177200827e-05,
+      "loss": 0.3061,
+      "step": 10240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1218361654015057e-05,
+      "loss": 0.3015,
+      "step": 10241
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1211788712273993e-05,
+      "loss": 0.2979,
+      "step": 10242
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1205216352586072e-05,
+      "loss": 0.3191,
+      "step": 10243
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1198644575559624e-05,
+      "loss": 0.3233,
+      "step": 10244
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1192073381802953e-05,
+      "loss": 0.3242,
+      "step": 10245
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1185502771924314e-05,
+      "loss": 0.2988,
+      "step": 10246
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1178932746531898e-05,
+      "loss": 0.318,
+      "step": 10247
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1172363306233845e-05,
+      "loss": 0.3127,
+      "step": 10248
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1165794451638232e-05,
+      "loss": 0.2934,
+      "step": 10249
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1159226183353092e-05,
+      "loss": 0.3056,
+      "step": 10250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1152658501986412e-05,
+      "loss": 0.2755,
+      "step": 10251
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.114609140814609e-05,
+      "loss": 0.3116,
+      "step": 10252
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1139524902440017e-05,
+      "loss": 0.3416,
+      "step": 10253
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.113295898547599e-05,
+      "loss": 0.2993,
+      "step": 10254
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.112639365786178e-05,
+      "loss": 0.3062,
+      "step": 10255
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.111982892020509e-05,
+      "loss": 0.3013,
+      "step": 10256
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.111326477311355e-05,
+      "loss": 0.3134,
+      "step": 10257
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1106701217194774e-05,
+      "loss": 0.304,
+      "step": 10258
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.110013825305629e-05,
+      "loss": 0.3238,
+      "step": 10259
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1093575881305593e-05,
+      "loss": 0.3013,
+      "step": 10260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.10870141025501e-05,
+      "loss": 0.2991,
+      "step": 10261
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.108045291739721e-05,
+      "loss": 0.3111,
+      "step": 10262
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1073892326454216e-05,
+      "loss": 0.3012,
+      "step": 10263
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1067332330328394e-05,
+      "loss": 0.3179,
+      "step": 10264
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1060772929626964e-05,
+      "loss": 0.2947,
+      "step": 10265
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1054214124957066e-05,
+      "loss": 0.3039,
+      "step": 10266
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1047655916925812e-05,
+      "loss": 0.2942,
+      "step": 10267
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.104109830614024e-05,
+      "loss": 0.2942,
+      "step": 10268
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1034541293207338e-05,
+      "loss": 0.2885,
+      "step": 10269
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1027984878734042e-05,
+      "loss": 0.2947,
+      "step": 10270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.102142906332723e-05,
+      "loss": 0.3056,
+      "step": 10271
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1014873847593734e-05,
+      "loss": 0.3007,
+      "step": 10272
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.100831923214031e-05,
+      "loss": 0.3013,
+      "step": 10273
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1001765217573677e-05,
+      "loss": 0.3063,
+      "step": 10274
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.099521180450049e-05,
+      "loss": 0.2874,
+      "step": 10275
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0988658993527342e-05,
+      "loss": 0.3094,
+      "step": 10276
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0982106785260796e-05,
+      "loss": 0.2999,
+      "step": 10277
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.097555518030732e-05,
+      "loss": 0.2761,
+      "step": 10278
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.096900417927337e-05,
+      "loss": 0.2944,
+      "step": 10279
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0962453782765308e-05,
+      "loss": 0.3063,
+      "step": 10280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0955903991389447e-05,
+      "loss": 0.2967,
+      "step": 10281
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0949354805752082e-05,
+      "loss": 0.2994,
+      "step": 10282
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0942806226459393e-05,
+      "loss": 0.3071,
+      "step": 10283
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0936258254117558e-05,
+      "loss": 0.2944,
+      "step": 10284
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.092971088933266e-05,
+      "loss": 0.3019,
+      "step": 10285
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.092316413271074e-05,
+      "loss": 0.2945,
+      "step": 10286
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0916617984857783e-05,
+      "loss": 0.3161,
+      "step": 10287
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.091007244637972e-05,
+      "loss": 0.2983,
+      "step": 10288
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0903527517882427e-05,
+      "loss": 0.3215,
+      "step": 10289
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.089698319997171e-05,
+      "loss": 0.3131,
+      "step": 10290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0890439493253332e-05,
+      "loss": 0.323,
+      "step": 10291
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0883896398333002e-05,
+      "loss": 0.3,
+      "step": 10292
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0877353915816362e-05,
+      "loss": 0.334,
+      "step": 10293
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0870812046308995e-05,
+      "loss": 0.3009,
+      "step": 10294
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0864270790416446e-05,
+      "loss": 0.3137,
+      "step": 10295
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0857730148744177e-05,
+      "loss": 0.3068,
+      "step": 10296
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0851190121897606e-05,
+      "loss": 0.3005,
+      "step": 10297
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0844650710482112e-05,
+      "loss": 0.3142,
+      "step": 10298
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.083811191510298e-05,
+      "loss": 0.3108,
+      "step": 10299
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.083157373636548e-05,
+      "loss": 0.3084,
+      "step": 10300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0825036174874773e-05,
+      "loss": 0.3196,
+      "step": 10301
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.081849923123602e-05,
+      "loss": 0.3289,
+      "step": 10302
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0811962906054284e-05,
+      "loss": 0.3165,
+      "step": 10303
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.080542719993459e-05,
+      "loss": 0.3172,
+      "step": 10304
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.079889211348189e-05,
+      "loss": 0.3122,
+      "step": 10305
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.07923576473011e-05,
+      "loss": 0.3054,
+      "step": 10306
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0785823801997062e-05,
+      "loss": 0.311,
+      "step": 10307
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.077929057817457e-05,
+      "loss": 0.2907,
+      "step": 10308
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0772757976438346e-05,
+      "loss": 0.2999,
+      "step": 10309
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.076622599739307e-05,
+      "loss": 0.3168,
+      "step": 10310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.075969464164337e-05,
+      "loss": 0.3184,
+      "step": 10311
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0753163909793788e-05,
+      "loss": 0.2925,
+      "step": 10312
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.074663380244883e-05,
+      "loss": 0.2975,
+      "step": 10313
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0740104320212944e-05,
+      "loss": 0.3042,
+      "step": 10314
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0733575463690514e-05,
+      "loss": 0.3166,
+      "step": 10315
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.072704723348587e-05,
+      "loss": 0.2928,
+      "step": 10316
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0720519630203278e-05,
+      "loss": 0.3118,
+      "step": 10317
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0713992654446952e-05,
+      "loss": 0.3257,
+      "step": 10318
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0707466306821037e-05,
+      "loss": 0.2878,
+      "step": 10319
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0700940587929644e-05,
+      "loss": 0.3101,
+      "step": 10320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0694415498376806e-05,
+      "loss": 0.2893,
+      "step": 10321
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.06878910387665e-05,
+      "loss": 0.3106,
+      "step": 10322
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.068136720970264e-05,
+      "loss": 0.2942,
+      "step": 10323
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.06748440117891e-05,
+      "loss": 0.3091,
+      "step": 10324
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.066832144562968e-05,
+      "loss": 0.2967,
+      "step": 10325
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.066179951182812e-05,
+      "loss": 0.304,
+      "step": 10326
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0655278210988116e-05,
+      "loss": 0.3195,
+      "step": 10327
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.064875754371329e-05,
+      "loss": 0.294,
+      "step": 10328
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0642237510607214e-05,
+      "loss": 0.3138,
+      "step": 10329
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0635718112273396e-05,
+      "loss": 0.3233,
+      "step": 10330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.06291993493153e-05,
+      "loss": 0.3059,
+      "step": 10331
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0622681222336306e-05,
+      "loss": 0.3144,
+      "step": 10332
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0616163731939747e-05,
+      "loss": 0.3221,
+      "step": 10333
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0609646878728916e-05,
+      "loss": 0.311,
+      "step": 10334
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.060313066330702e-05,
+      "loss": 0.3011,
+      "step": 10335
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.059661508627721e-05,
+      "loss": 0.2955,
+      "step": 10336
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.05901001482426e-05,
+      "loss": 0.2931,
+      "step": 10337
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0583585849806207e-05,
+      "loss": 0.3043,
+      "step": 10338
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0577072191571033e-05,
+      "loss": 0.3074,
+      "step": 10339
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0570559174139993e-05,
+      "loss": 0.288,
+      "step": 10340
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0564046798115945e-05,
+      "loss": 0.3091,
+      "step": 10341
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.055753506410169e-05,
+      "loss": 0.2996,
+      "step": 10342
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0551023972699978e-05,
+      "loss": 0.3233,
+      "step": 10343
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0544513524513486e-05,
+      "loss": 0.3174,
+      "step": 10344
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0538003720144837e-05,
+      "loss": 0.3187,
+      "step": 10345
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0531494560196605e-05,
+      "loss": 0.3064,
+      "step": 10346
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0524986045271288e-05,
+      "loss": 0.3192,
+      "step": 10347
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.051847817597133e-05,
+      "loss": 0.2998,
+      "step": 10348
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0511970952899124e-05,
+      "loss": 0.3308,
+      "step": 10349
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0505464376656978e-05,
+      "loss": 0.2822,
+      "step": 10350
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0498958447847176e-05,
+      "loss": 0.3119,
+      "step": 10351
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.049245316707191e-05,
+      "loss": 0.3181,
+      "step": 10352
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0485948534933342e-05,
+      "loss": 0.3171,
+      "step": 10353
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0479444552033543e-05,
+      "loss": 0.3117,
+      "step": 10354
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0472941218974545e-05,
+      "loss": 0.3158,
+      "step": 10355
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0466438536358306e-05,
+      "loss": 0.2859,
+      "step": 10356
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0459936504786743e-05,
+      "loss": 0.2875,
+      "step": 10357
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.045343512486169e-05,
+      "loss": 0.3027,
+      "step": 10358
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0446934397184942e-05,
+      "loss": 0.3142,
+      "step": 10359
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.044043432235821e-05,
+      "loss": 0.3139,
+      "step": 10360
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.043393490098317e-05,
+      "loss": 0.3103,
+      "step": 10361
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0427436133661422e-05,
+      "loss": 0.3176,
+      "step": 10362
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.04209380209945e-05,
+      "loss": 0.301,
+      "step": 10363
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0414440563583903e-05,
+      "loss": 0.2903,
+      "step": 10364
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0407943762031035e-05,
+      "loss": 0.3048,
+      "step": 10365
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0401447616937263e-05,
+      "loss": 0.2888,
+      "step": 10366
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0394952128903892e-05,
+      "loss": 0.3259,
+      "step": 10367
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.038845729853216e-05,
+      "loss": 0.3207,
+      "step": 10368
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0381963126423243e-05,
+      "loss": 0.3069,
+      "step": 10369
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0375469613178254e-05,
+      "loss": 0.292,
+      "step": 10370
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0368976759398263e-05,
+      "loss": 0.3136,
+      "step": 10371
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0362484565684242e-05,
+      "loss": 0.3193,
+      "step": 10372
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.035599303263715e-05,
+      "loss": 0.2991,
+      "step": 10373
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0349502160857846e-05,
+      "loss": 0.2922,
+      "step": 10374
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.034301195094715e-05,
+      "loss": 0.3106,
+      "step": 10375
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0336522403505802e-05,
+      "loss": 0.2997,
+      "step": 10376
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.03300335191345e-05,
+      "loss": 0.3134,
+      "step": 10377
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.032354529843387e-05,
+      "loss": 0.3082,
+      "step": 10378
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0317057742004482e-05,
+      "loss": 0.3085,
+      "step": 10379
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0310570850446832e-05,
+      "loss": 0.2947,
+      "step": 10380
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.030408462436137e-05,
+      "loss": 0.2917,
+      "step": 10381
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0297599064348484e-05,
+      "loss": 0.3183,
+      "step": 10382
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.029111417100848e-05,
+      "loss": 0.3184,
+      "step": 10383
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.028462994494162e-05,
+      "loss": 0.3041,
+      "step": 10384
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0278146386748102e-05,
+      "loss": 0.3106,
+      "step": 10385
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0271663497028064e-05,
+      "loss": 0.3015,
+      "step": 10386
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0265181276381584e-05,
+      "loss": 0.2767,
+      "step": 10387
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.025869972540866e-05,
+      "loss": 0.3166,
+      "step": 10388
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0252218844709253e-05,
+      "loss": 0.3158,
+      "step": 10389
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0245738634883248e-05,
+      "loss": 0.2924,
+      "step": 10390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0239259096530462e-05,
+      "loss": 0.2998,
+      "step": 10391
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0232780230250663e-05,
+      "loss": 0.2959,
+      "step": 10392
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0226302036643552e-05,
+      "loss": 0.2828,
+      "step": 10393
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0219824516308772e-05,
+      "loss": 0.3233,
+      "step": 10394
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0213347669845893e-05,
+      "loss": 0.2947,
+      "step": 10395
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0206871497854425e-05,
+      "loss": 0.3202,
+      "step": 10396
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0200396000933823e-05,
+      "loss": 0.2971,
+      "step": 10397
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0193921179683485e-05,
+      "loss": 0.3242,
+      "step": 10398
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.018744703470272e-05,
+      "loss": 0.2913,
+      "step": 10399
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0180973566590816e-05,
+      "loss": 0.3179,
+      "step": 10400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0174500775946942e-05,
+      "loss": 0.2979,
+      "step": 10401
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0168028663370265e-05,
+      "loss": 0.3111,
+      "step": 10402
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.016155722945984e-05,
+      "loss": 0.317,
+      "step": 10403
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0155086474814688e-05,
+      "loss": 0.3069,
+      "step": 10404
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0148616400033772e-05,
+      "loss": 0.2928,
+      "step": 10405
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.014214700571595e-05,
+      "loss": 0.3078,
+      "step": 10406
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.013567829246007e-05,
+      "loss": 0.3136,
+      "step": 10407
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0129210260864878e-05,
+      "loss": 0.3111,
+      "step": 10408
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0122742911529086e-05,
+      "loss": 0.2855,
+      "step": 10409
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0116276245051324e-05,
+      "loss": 0.2945,
+      "step": 10410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0109810262030153e-05,
+      "loss": 0.3184,
+      "step": 10411
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0103344963064092e-05,
+      "loss": 0.3051,
+      "step": 10412
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.009688034875158e-05,
+      "loss": 0.3045,
+      "step": 10413
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0090416419691006e-05,
+      "loss": 0.3096,
+      "step": 10414
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.008395317648068e-05,
+      "loss": 0.3088,
+      "step": 10415
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0077490619718854e-05,
+      "loss": 0.3198,
+      "step": 10416
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0071028750003736e-05,
+      "loss": 0.3071,
+      "step": 10417
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.006456756793343e-05,
+      "loss": 0.3133,
+      "step": 10418
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.005810707410602e-05,
+      "loss": 0.3026,
+      "step": 10419
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0051647269119496e-05,
+      "loss": 0.3131,
+      "step": 10420
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0045188153571793e-05,
+      "loss": 0.3005,
+      "step": 10421
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0038729728060784e-05,
+      "loss": 0.3222,
+      "step": 10422
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0032271993184283e-05,
+      "loss": 0.3214,
+      "step": 10423
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0025814949540033e-05,
+      "loss": 0.298,
+      "step": 10424
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0019358597725718e-05,
+      "loss": 0.3133,
+      "step": 10425
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.001290293833894e-05,
+      "loss": 0.3193,
+      "step": 10426
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.000644797197727e-05,
+      "loss": 0.3158,
+      "step": 10427
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.999999369923818e-05,
+      "loss": 0.3203,
+      "step": 10428
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9993540120719112e-05,
+      "loss": 0.3048,
+      "step": 10429
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.998708723701741e-05,
+      "loss": 0.3034,
+      "step": 10430
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9980635048730373e-05,
+      "loss": 0.3127,
+      "step": 10431
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9974183556455245e-05,
+      "loss": 0.3047,
+      "step": 10432
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.996773276078917e-05,
+      "loss": 0.2878,
+      "step": 10433
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9961282662329274e-05,
+      "loss": 0.3179,
+      "step": 10434
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.995483326167257e-05,
+      "loss": 0.3062,
+      "step": 10435
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9948384559416062e-05,
+      "loss": 0.2942,
+      "step": 10436
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9941936556156637e-05,
+      "loss": 0.3035,
+      "step": 10437
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9935489252491135e-05,
+      "loss": 0.3031,
+      "step": 10438
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9929042649016356e-05,
+      "loss": 0.3202,
+      "step": 10439
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.992259674632899e-05,
+      "loss": 0.3135,
+      "step": 10440
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.991615154502571e-05,
+      "loss": 0.2968,
+      "step": 10441
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9909707045703085e-05,
+      "loss": 0.3184,
+      "step": 10442
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.990326324895764e-05,
+      "loss": 0.2962,
+      "step": 10443
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9896820155385826e-05,
+      "loss": 0.3131,
+      "step": 10444
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9890377765584033e-05,
+      "loss": 0.2789,
+      "step": 10445
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.988393608014859e-05,
+      "loss": 0.2988,
+      "step": 10446
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.987749509967575e-05,
+      "loss": 0.3134,
+      "step": 10447
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9871054824761715e-05,
+      "loss": 0.2998,
+      "step": 10448
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9864615256002607e-05,
+      "loss": 0.2976,
+      "step": 10449
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9858176393994482e-05,
+      "loss": 0.3079,
+      "step": 10450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9851738239333356e-05,
+      "loss": 0.3127,
+      "step": 10451
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.984530079261515e-05,
+      "loss": 0.3182,
+      "step": 10452
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.983886405443573e-05,
+      "loss": 0.3176,
+      "step": 10453
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9832428025390904e-05,
+      "loss": 0.3143,
+      "step": 10454
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9825992706076396e-05,
+      "loss": 0.3064,
+      "step": 10455
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9819558097087886e-05,
+      "loss": 0.3119,
+      "step": 10456
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9813124199020983e-05,
+      "loss": 0.2945,
+      "step": 10457
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9806691012471208e-05,
+      "loss": 0.2937,
+      "step": 10458
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9800258538034043e-05,
+      "loss": 0.2973,
+      "step": 10459
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.979382677630489e-05,
+      "loss": 0.3225,
+      "step": 10460
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9787395727879104e-05,
+      "loss": 0.2988,
+      "step": 10461
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9780965393351947e-05,
+      "loss": 0.3239,
+      "step": 10462
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.977453577331862e-05,
+      "loss": 0.2955,
+      "step": 10463
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9768106868374286e-05,
+      "loss": 0.3009,
+      "step": 10464
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9761678679114002e-05,
+      "loss": 0.313,
+      "step": 10465
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9755251206132793e-05,
+      "loss": 0.2984,
+      "step": 10466
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9748824450025595e-05,
+      "loss": 0.3179,
+      "step": 10467
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.974239841138728e-05,
+      "loss": 0.2932,
+      "step": 10468
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9735973090812664e-05,
+      "loss": 0.308,
+      "step": 10469
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9729548488896493e-05,
+      "loss": 0.3075,
+      "step": 10470
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.972312460623345e-05,
+      "loss": 0.2908,
+      "step": 10471
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9716701443418133e-05,
+      "loss": 0.317,
+      "step": 10472
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9710279001045096e-05,
+      "loss": 0.3224,
+      "step": 10473
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9703857279708814e-05,
+      "loss": 0.3113,
+      "step": 10474
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.969743628000369e-05,
+      "loss": 0.3107,
+      "step": 10475
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.969101600252409e-05,
+      "loss": 0.3157,
+      "step": 10476
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9684596447864267e-05,
+      "loss": 0.3095,
+      "step": 10477
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.967817761661845e-05,
+      "loss": 0.3218,
+      "step": 10478
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.967175950938077e-05,
+      "loss": 0.316,
+      "step": 10479
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.966534212674532e-05,
+      "loss": 0.2989,
+      "step": 10480
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.965892546930609e-05,
+      "loss": 0.307,
+      "step": 10481
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9652509537657026e-05,
+      "loss": 0.3086,
+      "step": 10482
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9646094332392017e-05,
+      "loss": 0.3084,
+      "step": 10483
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.963967985410486e-05,
+      "loss": 0.3273,
+      "step": 10484
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9633266103389296e-05,
+      "loss": 0.3026,
+      "step": 10485
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.962685308083901e-05,
+      "loss": 0.2759,
+      "step": 10486
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9620440787047584e-05,
+      "loss": 0.3013,
+      "step": 10487
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.961402922260858e-05,
+      "loss": 0.2994,
+      "step": 10488
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.960761838811546e-05,
+      "loss": 0.3046,
+      "step": 10489
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.960120828416162e-05,
+      "loss": 0.318,
+      "step": 10490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9594798911340412e-05,
+      "loss": 0.3017,
+      "step": 10491
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.958839027024509e-05,
+      "loss": 0.3112,
+      "step": 10492
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9581982361468865e-05,
+      "loss": 0.3106,
+      "step": 10493
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9575575185604862e-05,
+      "loss": 0.3166,
+      "step": 10494
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.956916874324614e-05,
+      "loss": 0.2945,
+      "step": 10495
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9562763034985714e-05,
+      "loss": 0.2911,
+      "step": 10496
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9556358061416502e-05,
+      "loss": 0.2944,
+      "step": 10497
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9549953823131367e-05,
+      "loss": 0.3169,
+      "step": 10498
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9543550320723097e-05,
+      "loss": 0.3281,
+      "step": 10499
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9537147554784424e-05,
+      "loss": 0.3162,
+      "step": 10500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9530745525908e-05,
+      "loss": 0.3285,
+      "step": 10501
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9524344234686415e-05,
+      "loss": 0.3133,
+      "step": 10502
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9517943681712186e-05,
+      "loss": 0.324,
+      "step": 10503
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9511543867577774e-05,
+      "loss": 0.304,
+      "step": 10504
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9505144792875545e-05,
+      "loss": 0.2863,
+      "step": 10505
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9498746458197837e-05,
+      "loss": 0.2966,
+      "step": 10506
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9492348864136878e-05,
+      "loss": 0.3226,
+      "step": 10507
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9485952011284862e-05,
+      "loss": 0.3025,
+      "step": 10508
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9479555900233876e-05,
+      "loss": 0.3222,
+      "step": 10509
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.947316053157598e-05,
+      "loss": 0.3211,
+      "step": 10510
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9466765905903135e-05,
+      "loss": 0.3043,
+      "step": 10511
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.946037202380725e-05,
+      "loss": 0.2835,
+      "step": 10512
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.945397888588016e-05,
+      "loss": 0.291,
+      "step": 10513
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9447586492713616e-05,
+      "loss": 0.3194,
+      "step": 10514
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9441194844899342e-05,
+      "loss": 0.31,
+      "step": 10515
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9434803943028933e-05,
+      "loss": 0.2972,
+      "step": 10516
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9428413787693977e-05,
+      "loss": 0.3137,
+      "step": 10517
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9422024379485943e-05,
+      "loss": 0.2937,
+      "step": 10518
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9415635718996267e-05,
+      "loss": 0.3039,
+      "step": 10519
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9409247806816283e-05,
+      "loss": 0.307,
+      "step": 10520
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9402860643537284e-05,
+      "loss": 0.2951,
+      "step": 10521
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9396474229750482e-05,
+      "loss": 0.3017,
+      "step": 10522
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9390088566047016e-05,
+      "loss": 0.3057,
+      "step": 10523
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9383703653017964e-05,
+      "loss": 0.2876,
+      "step": 10524
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9377319491254325e-05,
+      "loss": 0.2953,
+      "step": 10525
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9370936081347035e-05,
+      "loss": 0.3091,
+      "step": 10526
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.936455342388696e-05,
+      "loss": 0.3012,
+      "step": 10527
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9358171519464898e-05,
+      "loss": 0.3033,
+      "step": 10528
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9351790368671567e-05,
+      "loss": 0.3179,
+      "step": 10529
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.934540997209764e-05,
+      "loss": 0.2877,
+      "step": 10530
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9339030330333675e-05,
+      "loss": 0.3227,
+      "step": 10531
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9332651443970213e-05,
+      "loss": 0.2954,
+      "step": 10532
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.932627331359768e-05,
+      "loss": 0.3013,
+      "step": 10533
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9319895939806475e-05,
+      "loss": 0.3066,
+      "step": 10534
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.931351932318689e-05,
+      "loss": 0.3048,
+      "step": 10535
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9307143464329165e-05,
+      "loss": 0.3086,
+      "step": 10536
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.930076836382347e-05,
+      "loss": 0.2964,
+      "step": 10537
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9294394022259883e-05,
+      "loss": 0.3319,
+      "step": 10538
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9288020440228444e-05,
+      "loss": 0.3176,
+      "step": 10539
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9281647618319114e-05,
+      "loss": 0.3065,
+      "step": 10540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9275275557121762e-05,
+      "loss": 0.2815,
+      "step": 10541
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9268904257226215e-05,
+      "loss": 0.3037,
+      "step": 10542
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9262533719222216e-05,
+      "loss": 0.3037,
+      "step": 10543
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.925616394369943e-05,
+      "loss": 0.3034,
+      "step": 10544
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9249794931247474e-05,
+      "loss": 0.3073,
+      "step": 10545
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9243426682455856e-05,
+      "loss": 0.326,
+      "step": 10546
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9237059197914063e-05,
+      "loss": 0.2977,
+      "step": 10547
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9230692478211475e-05,
+      "loss": 0.2935,
+      "step": 10548
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.922432652393742e-05,
+      "loss": 0.3284,
+      "step": 10549
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9217961335681134e-05,
+      "loss": 0.3073,
+      "step": 10550
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.92115969140318e-05,
+      "loss": 0.3057,
+      "step": 10551
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9205233259578533e-05,
+      "loss": 0.3091,
+      "step": 10552
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.919887037291035e-05,
+      "loss": 0.3052,
+      "step": 10553
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.919250825461624e-05,
+      "loss": 0.2918,
+      "step": 10554
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9186146905285092e-05,
+      "loss": 0.3083,
+      "step": 10555
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9179786325505713e-05,
+      "loss": 0.3093,
+      "step": 10556
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9173426515866866e-05,
+      "loss": 0.3053,
+      "step": 10557
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9167067476957234e-05,
+      "loss": 0.3099,
+      "step": 10558
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9160709209365425e-05,
+      "loss": 0.2943,
+      "step": 10559
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.915435171367997e-05,
+      "loss": 0.3067,
+      "step": 10560
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9147994990489345e-05,
+      "loss": 0.3228,
+      "step": 10561
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9141639040381936e-05,
+      "loss": 0.3127,
+      "step": 10562
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9135283863946067e-05,
+      "loss": 0.297,
+      "step": 10563
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.912892946177e-05,
+      "loss": 0.2957,
+      "step": 10564
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.91225758344419e-05,
+      "loss": 0.2771,
+      "step": 10565
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9116222982549893e-05,
+      "loss": 0.313,
+      "step": 10566
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9109870906682003e-05,
+      "loss": 0.2996,
+      "step": 10567
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.910351960742619e-05,
+      "loss": 0.299,
+      "step": 10568
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.909716908537036e-05,
+      "loss": 0.3076,
+      "step": 10569
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9090819341102318e-05,
+      "loss": 0.2991,
+      "step": 10570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.908447037520983e-05,
+      "loss": 0.2881,
+      "step": 10571
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9078122188280567e-05,
+      "loss": 0.2901,
+      "step": 10572
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9071774780902125e-05,
+      "loss": 0.3066,
+      "step": 10573
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9065428153662045e-05,
+      "loss": 0.3229,
+      "step": 10574
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.905908230714778e-05,
+      "loss": 0.3116,
+      "step": 10575
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9052737241946734e-05,
+      "loss": 0.3205,
+      "step": 10576
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.904639295864621e-05,
+      "loss": 0.3165,
+      "step": 10577
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9040049457833442e-05,
+      "loss": 0.2984,
+      "step": 10578
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.903370674009562e-05,
+      "loss": 0.2917,
+      "step": 10579
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.902736480601983e-05,
+      "loss": 0.3124,
+      "step": 10580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.902102365619311e-05,
+      "loss": 0.2956,
+      "step": 10581
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9014683291202402e-05,
+      "loss": 0.2863,
+      "step": 10582
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9008343711634585e-05,
+      "loss": 0.3342,
+      "step": 10583
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9002004918076477e-05,
+      "loss": 0.2893,
+      "step": 10584
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8995666911114807e-05,
+      "loss": 0.2892,
+      "step": 10585
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8989329691336237e-05,
+      "loss": 0.3146,
+      "step": 10586
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.898299325932736e-05,
+      "loss": 0.3155,
+      "step": 10587
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8976657615674693e-05,
+      "loss": 0.298,
+      "step": 10588
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.897032276096467e-05,
+      "loss": 0.2931,
+      "step": 10589
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.896398869578367e-05,
+      "loss": 0.3077,
+      "step": 10590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8957655420717995e-05,
+      "loss": 0.2935,
+      "step": 10591
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8951322936353862e-05,
+      "loss": 0.3007,
+      "step": 10592
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8944991243277415e-05,
+      "loss": 0.3034,
+      "step": 10593
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.893866034207475e-05,
+      "loss": 0.2931,
+      "step": 10594
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8932330233331858e-05,
+      "loss": 0.3128,
+      "step": 10595
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8926000917634673e-05,
+      "loss": 0.3017,
+      "step": 10596
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.891967239556906e-05,
+      "loss": 0.3028,
+      "step": 10597
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8913344667720796e-05,
+      "loss": 0.298,
+      "step": 10598
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8907017734675592e-05,
+      "loss": 0.3097,
+      "step": 10599
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8900691597019083e-05,
+      "loss": 0.3197,
+      "step": 10600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8894366255336842e-05,
+      "loss": 0.2979,
+      "step": 10601
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8888041710214356e-05,
+      "loss": 0.3247,
+      "step": 10602
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8881717962237027e-05,
+      "loss": 0.297,
+      "step": 10603
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.887539501199022e-05,
+      "loss": 0.3132,
+      "step": 10604
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8869072860059188e-05,
+      "loss": 0.3156,
+      "step": 10605
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8862751507029136e-05,
+      "loss": 0.3081,
+      "step": 10606
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.885643095348517e-05,
+      "loss": 0.3098,
+      "step": 10607
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8850111200012354e-05,
+      "loss": 0.2985,
+      "step": 10608
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8843792247195658e-05,
+      "loss": 0.3029,
+      "step": 10609
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8837474095619967e-05,
+      "loss": 0.3045,
+      "step": 10610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8831156745870115e-05,
+      "loss": 0.3067,
+      "step": 10611
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.882484019853085e-05,
+      "loss": 0.3086,
+      "step": 10612
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8818524454186857e-05,
+      "loss": 0.2966,
+      "step": 10613
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.881220951342273e-05,
+      "loss": 0.3259,
+      "step": 10614
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.880589537682299e-05,
+      "loss": 0.2992,
+      "step": 10615
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8799582044972096e-05,
+      "loss": 0.3095,
+      "step": 10616
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.879326951845443e-05,
+      "loss": 0.308,
+      "step": 10617
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8786957797854293e-05,
+      "loss": 0.3163,
+      "step": 10618
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.878064688375591e-05,
+      "loss": 0.3989,
+      "step": 10619
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.877433677674344e-05,
+      "loss": 0.3093,
+      "step": 10620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8768027477400967e-05,
+      "loss": 0.3303,
+      "step": 10621
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8761718986312482e-05,
+      "loss": 0.3136,
+      "step": 10622
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.875541130406193e-05,
+      "loss": 0.2979,
+      "step": 10623
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8749104431233168e-05,
+      "loss": 0.328,
+      "step": 10624
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8742798368409958e-05,
+      "loss": 0.3066,
+      "step": 10625
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8736493116176017e-05,
+      "loss": 0.305,
+      "step": 10626
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.873018867511498e-05,
+      "loss": 0.3028,
+      "step": 10627
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8723885045810396e-05,
+      "loss": 0.3085,
+      "step": 10628
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8717582228845754e-05,
+      "loss": 0.3092,
+      "step": 10629
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.871128022480444e-05,
+      "loss": 0.3204,
+      "step": 10630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8704979034269805e-05,
+      "loss": 0.2899,
+      "step": 10631
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8698678657825087e-05,
+      "loss": 0.3178,
+      "step": 10632
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8692379096053475e-05,
+      "loss": 0.31,
+      "step": 10633
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8686080349538082e-05,
+      "loss": 0.3097,
+      "step": 10634
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8679782418861913e-05,
+      "loss": 0.3668,
+      "step": 10635
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8673485304607934e-05,
+      "loss": 0.2983,
+      "step": 10636
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8667189007359015e-05,
+      "loss": 0.2998,
+      "step": 10637
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8660893527697976e-05,
+      "loss": 0.3214,
+      "step": 10638
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8654598866207517e-05,
+      "loss": 0.3222,
+      "step": 10639
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.864830502347031e-05,
+      "loss": 0.2994,
+      "step": 10640
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8642012000068916e-05,
+      "loss": 0.3216,
+      "step": 10641
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8635719796585837e-05,
+      "loss": 0.3029,
+      "step": 10642
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8629428413603496e-05,
+      "loss": 0.3087,
+      "step": 10643
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.862313785170425e-05,
+      "loss": 0.3081,
+      "step": 10644
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.861684811147035e-05,
+      "loss": 0.2996,
+      "step": 10645
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8610559193484e-05,
+      "loss": 0.3094,
+      "step": 10646
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8604271098327326e-05,
+      "loss": 0.3204,
+      "step": 10647
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.859798382658236e-05,
+      "loss": 0.3018,
+      "step": 10648
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8591697378831073e-05,
+      "loss": 0.3031,
+      "step": 10649
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8585411755655358e-05,
+      "loss": 0.3058,
+      "step": 10650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8579126957637023e-05,
+      "loss": 0.3033,
+      "step": 10651
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8572842985357803e-05,
+      "loss": 0.3198,
+      "step": 10652
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.856655983939936e-05,
+      "loss": 0.3203,
+      "step": 10653
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8560277520343283e-05,
+      "loss": 0.3022,
+      "step": 10654
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.855399602877108e-05,
+      "loss": 0.3136,
+      "step": 10655
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8547715365264178e-05,
+      "loss": 0.3122,
+      "step": 10656
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8541435530403935e-05,
+      "loss": 0.2972,
+      "step": 10657
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8535156524771624e-05,
+      "loss": 0.317,
+      "step": 10658
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8528878348948448e-05,
+      "loss": 0.2989,
+      "step": 10659
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.852260100351554e-05,
+      "loss": 0.3018,
+      "step": 10660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8516324489053927e-05,
+      "loss": 0.3105,
+      "step": 10661
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.85100488061446e-05,
+      "loss": 0.2877,
+      "step": 10662
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8503773955368438e-05,
+      "loss": 0.2908,
+      "step": 10663
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8497499937306267e-05,
+      "loss": 0.3035,
+      "step": 10664
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8491226752538827e-05,
+      "loss": 0.3134,
+      "step": 10665
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.848495440164677e-05,
+      "loss": 0.3201,
+      "step": 10666
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.847868288521069e-05,
+      "loss": 0.2986,
+      "step": 10667
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.847241220381109e-05,
+      "loss": 0.3265,
+      "step": 10668
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8466142358028403e-05,
+      "loss": 0.3106,
+      "step": 10669
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.845987334844299e-05,
+      "loss": 0.3089,
+      "step": 10670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8453605175635107e-05,
+      "loss": 0.3067,
+      "step": 10671
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8447337840184974e-05,
+      "loss": 0.2856,
+      "step": 10672
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8441071342672687e-05,
+      "loss": 0.3184,
+      "step": 10673
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.843480568367832e-05,
+      "loss": 0.3045,
+      "step": 10674
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.842854086378182e-05,
+      "loss": 0.2962,
+      "step": 10675
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8422276883563072e-05,
+      "loss": 0.3094,
+      "step": 10676
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8416013743601896e-05,
+      "loss": 0.3024,
+      "step": 10677
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8409751444478012e-05,
+      "loss": 0.2931,
+      "step": 10678
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8403489986771087e-05,
+      "loss": 0.3217,
+      "step": 10679
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.83972293710607e-05,
+      "loss": 0.3211,
+      "step": 10680
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8390969597926333e-05,
+      "loss": 0.3013,
+      "step": 10681
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.838471066794743e-05,
+      "loss": 0.2988,
+      "step": 10682
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8378452581703308e-05,
+      "loss": 0.3016,
+      "step": 10683
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8372195339773253e-05,
+      "loss": 0.3083,
+      "step": 10684
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.836593894273644e-05,
+      "loss": 0.2974,
+      "step": 10685
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8359683391171985e-05,
+      "loss": 0.2829,
+      "step": 10686
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8353428685658906e-05,
+      "loss": 0.2985,
+      "step": 10687
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8347174826776166e-05,
+      "loss": 0.321,
+      "step": 10688
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8340921815102634e-05,
+      "loss": 0.3223,
+      "step": 10689
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.83346696512171e-05,
+      "loss": 0.3027,
+      "step": 10690
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8328418335698298e-05,
+      "loss": 0.3256,
+      "step": 10691
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8322167869124853e-05,
+      "loss": 0.3119,
+      "step": 10692
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8315918252075316e-05,
+      "loss": 0.3225,
+      "step": 10693
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8309669485128184e-05,
+      "loss": 0.3156,
+      "step": 10694
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8303421568861844e-05,
+      "loss": 0.274,
+      "step": 10695
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8297174503854642e-05,
+      "loss": 0.3102,
+      "step": 10696
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.82909282906848e-05,
+      "loss": 0.3098,
+      "step": 10697
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8284682929930492e-05,
+      "loss": 0.3197,
+      "step": 10698
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.82784384221698e-05,
+      "loss": 0.2828,
+      "step": 10699
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.827219476798074e-05,
+      "loss": 0.319,
+      "step": 10700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8265951967941243e-05,
+      "loss": 0.321,
+      "step": 10701
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8259710022629144e-05,
+      "loss": 0.2904,
+      "step": 10702
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8253468932622225e-05,
+      "loss": 0.3139,
+      "step": 10703
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.824722869849818e-05,
+      "loss": 0.2952,
+      "step": 10704
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8240989320834607e-05,
+      "loss": 0.3027,
+      "step": 10705
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.823475080020906e-05,
+      "loss": 0.323,
+      "step": 10706
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8228513137198977e-05,
+      "loss": 0.3073,
+      "step": 10707
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8222276332381727e-05,
+      "loss": 0.2956,
+      "step": 10708
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.821604038633462e-05,
+      "loss": 0.3315,
+      "step": 10709
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.820980529963487e-05,
+      "loss": 0.3294,
+      "step": 10710
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8203571072859603e-05,
+      "loss": 0.3037,
+      "step": 10711
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8197337706585883e-05,
+      "loss": 0.3307,
+      "step": 10712
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.819110520139068e-05,
+      "loss": 0.3009,
+      "step": 10713
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8184873557850897e-05,
+      "loss": 0.3098,
+      "step": 10714
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.817864277654334e-05,
+      "loss": 0.3134,
+      "step": 10715
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8172412858044767e-05,
+      "loss": 0.3065,
+      "step": 10716
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8166183802931824e-05,
+      "loss": 0.2897,
+      "step": 10717
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8159955611781075e-05,
+      "loss": 0.3146,
+      "step": 10718
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8153728285169043e-05,
+      "loss": 0.3035,
+      "step": 10719
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8147501823672126e-05,
+      "loss": 0.3045,
+      "step": 10720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8141276227866672e-05,
+      "loss": 0.3224,
+      "step": 10721
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8135051498328925e-05,
+      "loss": 0.3095,
+      "step": 10722
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.812882763563508e-05,
+      "loss": 0.3139,
+      "step": 10723
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.812260464036123e-05,
+      "loss": 0.3095,
+      "step": 10724
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.811638251308338e-05,
+      "loss": 0.2987,
+      "step": 10725
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8110161254377467e-05,
+      "loss": 0.3189,
+      "step": 10726
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8103940864819363e-05,
+      "loss": 0.3323,
+      "step": 10727
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.809772134498483e-05,
+      "loss": 0.2975,
+      "step": 10728
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.809150269544957e-05,
+      "loss": 0.2977,
+      "step": 10729
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.808528491678919e-05,
+      "loss": 0.3013,
+      "step": 10730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.807906800957922e-05,
+      "loss": 0.2864,
+      "step": 10731
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8072851974395132e-05,
+      "loss": 0.2934,
+      "step": 10732
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8066636811812283e-05,
+      "loss": 0.3082,
+      "step": 10733
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.806042252240597e-05,
+      "loss": 0.3102,
+      "step": 10734
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8054209106751395e-05,
+      "loss": 0.3157,
+      "step": 10735
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.804799656542369e-05,
+      "loss": 0.3179,
+      "step": 10736
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.804178489899792e-05,
+      "loss": 0.3197,
+      "step": 10737
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.803557410804904e-05,
+      "loss": 0.3105,
+      "step": 10738
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8029364193151944e-05,
+      "loss": 0.3121,
+      "step": 10739
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8023155154881422e-05,
+      "loss": 0.3046,
+      "step": 10740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8016946993812206e-05,
+      "loss": 0.3029,
+      "step": 10741
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8010739710518948e-05,
+      "loss": 0.3042,
+      "step": 10742
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8004533305576203e-05,
+      "loss": 0.3019,
+      "step": 10743
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.799832777955846e-05,
+      "loss": 0.3315,
+      "step": 10744
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.799212313304011e-05,
+      "loss": 0.3075,
+      "step": 10745
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.7985919366595465e-05,
+      "loss": 0.3008,
+      "step": 10746
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.797971648079878e-05,
+      "loss": 0.2895,
+      "step": 10747
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.7973514476224195e-05,
+      "loss": 0.3104,
+      "step": 10748
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.796731335344579e-05,
+      "loss": 0.3101,
+      "step": 10749
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.796111311303756e-05,
+      "loss": 0.3032,
+      "step": 10750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.7954913755573398e-05,
+      "loss": 0.3012,
+      "step": 10751
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.7948715281627153e-05,
+      "loss": 0.3026,
+      "step": 10752
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.794251769177257e-05,
+      "loss": 0.3195,
+      "step": 10753
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.7936320986583304e-05,
+      "loss": 0.3077,
+      "step": 10754
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.7930125166632944e-05,
+      "loss": 0.2905,
+      "step": 10755
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.7923930232494977e-05,
+      "loss": 0.3034,
+      "step": 10756
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.791773618474285e-05,
+      "loss": 0.3197,
+      "step": 10757
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7911543023949884e-05,
+      "loss": 0.3051,
+      "step": 10758
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7905350750689332e-05,
+      "loss": 0.2864,
+      "step": 10759
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7899159365534372e-05,
+      "loss": 0.3118,
+      "step": 10760
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.789296886905809e-05,
+      "loss": 0.3095,
+      "step": 10761
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.78867792618335e-05,
+      "loss": 0.3094,
+      "step": 10762
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.788059054443352e-05,
+      "loss": 0.3169,
+      "step": 10763
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7874402717431006e-05,
+      "loss": 0.3275,
+      "step": 10764
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7868215781398696e-05,
+      "loss": 0.3011,
+      "step": 10765
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.786202973690929e-05,
+      "loss": 0.3121,
+      "step": 10766
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7855844584535387e-05,
+      "loss": 0.3068,
+      "step": 10767
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7849660324849492e-05,
+      "loss": 0.3026,
+      "step": 10768
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7843476958424038e-05,
+      "loss": 0.3307,
+      "step": 10769
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7837294485831356e-05,
+      "loss": 0.3087,
+      "step": 10770
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7831112907643744e-05,
+      "loss": 0.2804,
+      "step": 10771
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7824932224433367e-05,
+      "loss": 0.3144,
+      "step": 10772
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7818752436772324e-05,
+      "loss": 0.299,
+      "step": 10773
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7812573545232637e-05,
+      "loss": 0.2878,
+      "step": 10774
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.780639555038623e-05,
+      "loss": 0.3024,
+      "step": 10775
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7800218452804967e-05,
+      "loss": 0.3044,
+      "step": 10776
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.779404225306062e-05,
+      "loss": 0.314,
+      "step": 10777
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7787866951724857e-05,
+      "loss": 0.3245,
+      "step": 10778
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7781692549369294e-05,
+      "loss": 0.2907,
+      "step": 10779
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.777551904656543e-05,
+      "loss": 0.2995,
+      "step": 10780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.776934644388473e-05,
+      "loss": 0.3146,
+      "step": 10781
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7763174741898532e-05,
+      "loss": 0.2969,
+      "step": 10782
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7757003941178096e-05,
+      "loss": 0.3092,
+      "step": 10783
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7750834042294618e-05,
+      "loss": 0.3016,
+      "step": 10784
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7744665045819187e-05,
+      "loss": 0.3116,
+      "step": 10785
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.773849695232284e-05,
+      "loss": 0.3077,
+      "step": 10786
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7732329762376503e-05,
+      "loss": 0.3074,
+      "step": 10787
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7726163476551028e-05,
+      "loss": 0.3008,
+      "step": 10788
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7719998095417178e-05,
+      "loss": 0.3201,
+      "step": 10789
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7713833619545634e-05,
+      "loss": 0.2963,
+      "step": 10790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7707670049507007e-05,
+      "loss": 0.3024,
+      "step": 10791
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7701507385871808e-05,
+      "loss": 0.3238,
+      "step": 10792
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7695345629210476e-05,
+      "loss": 0.3107,
+      "step": 10793
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7689184780093332e-05,
+      "loss": 0.3148,
+      "step": 10794
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7683024839090674e-05,
+      "loss": 0.3269,
+      "step": 10795
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.767686580677267e-05,
+      "loss": 0.3061,
+      "step": 10796
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.767070768370941e-05,
+      "loss": 0.3112,
+      "step": 10797
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.766455047047091e-05,
+      "loss": 0.3229,
+      "step": 10798
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.765839416762709e-05,
+      "loss": 0.3356,
+      "step": 10799
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7652238775747814e-05,
+      "loss": 0.3077,
+      "step": 10800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.764608429540282e-05,
+      "loss": 0.2959,
+      "step": 10801
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7639930727161796e-05,
+      "loss": 0.307,
+      "step": 10802
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7633778071594325e-05,
+      "loss": 0.2898,
+      "step": 10803
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7627626329269904e-05,
+      "loss": 0.2914,
+      "step": 10804
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7621475500757973e-05,
+      "loss": 0.3101,
+      "step": 10805
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7615325586627862e-05,
+      "loss": 0.3086,
+      "step": 10806
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7609176587448823e-05,
+      "loss": 0.3147,
+      "step": 10807
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.760302850379002e-05,
+      "loss": 0.3092,
+      "step": 10808
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.759688133622053e-05,
+      "loss": 0.2913,
+      "step": 10809
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7590735085309364e-05,
+      "loss": 0.3397,
+      "step": 10810
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.758458975162543e-05,
+      "loss": 0.3232,
+      "step": 10811
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.757844533573755e-05,
+      "loss": 0.3182,
+      "step": 10812
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7572301838214478e-05,
+      "loss": 0.3052,
+      "step": 10813
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7566159259624854e-05,
+      "loss": 0.3237,
+      "step": 10814
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.756001760053727e-05,
+      "loss": 0.3076,
+      "step": 10815
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.755387686152021e-05,
+      "loss": 0.3123,
+      "step": 10816
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7547737043142077e-05,
+      "loss": 0.3197,
+      "step": 10817
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7541598145971176e-05,
+      "loss": 0.3017,
+      "step": 10818
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7535460170575745e-05,
+      "loss": 0.3109,
+      "step": 10819
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7529323117523942e-05,
+      "loss": 0.3018,
+      "step": 10820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7523186987383823e-05,
+      "loss": 0.3121,
+      "step": 10821
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.751705178072336e-05,
+      "loss": 0.2948,
+      "step": 10822
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7510917498110446e-05,
+      "loss": 0.3115,
+      "step": 10823
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7504784140112868e-05,
+      "loss": 0.3105,
+      "step": 10824
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.749865170729838e-05,
+      "loss": 0.3017,
+      "step": 10825
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7492520200234598e-05,
+      "loss": 0.2973,
+      "step": 10826
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.748638961948907e-05,
+      "loss": 0.3243,
+      "step": 10827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7480259965629263e-05,
+      "loss": 0.3003,
+      "step": 10828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7474131239222536e-05,
+      "loss": 0.2853,
+      "step": 10829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7468003440836205e-05,
+      "loss": 0.3183,
+      "step": 10830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.746187657103746e-05,
+      "loss": 0.297,
+      "step": 10831
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7455750630393432e-05,
+      "loss": 0.3123,
+      "step": 10832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7449625619471137e-05,
+      "loss": 0.3095,
+      "step": 10833
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7443501538837525e-05,
+      "loss": 0.3123,
+      "step": 10834
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.743737838905947e-05,
+      "loss": 0.3009,
+      "step": 10835
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7431256170703742e-05,
+      "loss": 0.3121,
+      "step": 10836
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7425134884337024e-05,
+      "loss": 0.3193,
+      "step": 10837
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.741901453052592e-05,
+      "loss": 0.3114,
+      "step": 10838
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7412895109836934e-05,
+      "loss": 0.3282,
+      "step": 10839
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7406776622836525e-05,
+      "loss": 0.308,
+      "step": 10840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7400659070091013e-05,
+      "loss": 0.3046,
+      "step": 10841
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.739454245216666e-05,
+      "loss": 0.3026,
+      "step": 10842
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.738842676962964e-05,
+      "loss": 0.321,
+      "step": 10843
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7382312023046016e-05,
+      "loss": 0.2964,
+      "step": 10844
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7376198212981816e-05,
+      "loss": 0.3196,
+      "step": 10845
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7370085340002936e-05,
+      "loss": 0.3029,
+      "step": 10846
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7363973404675193e-05,
+      "loss": 0.304,
+      "step": 10847
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7357862407564334e-05,
+      "loss": 0.3202,
+      "step": 10848
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7351752349235995e-05,
+      "loss": 0.311,
+      "step": 10849
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7345643230255752e-05,
+      "loss": 0.3381,
+      "step": 10850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.733953505118909e-05,
+      "loss": 0.294,
+      "step": 10851
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7333427812601373e-05,
+      "loss": 0.319,
+      "step": 10852
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7327321515057916e-05,
+      "loss": 0.3231,
+      "step": 10853
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7321216159123924e-05,
+      "loss": 0.3142,
+      "step": 10854
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.731511174536454e-05,
+      "loss": 0.3046,
+      "step": 10855
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7309008274344798e-05,
+      "loss": 0.2846,
+      "step": 10856
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7302905746629653e-05,
+      "loss": 0.2905,
+      "step": 10857
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7296804162783965e-05,
+      "loss": 0.3144,
+      "step": 10858
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.729070352337251e-05,
+      "loss": 0.2966,
+      "step": 10859
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.728460382895999e-05,
+      "loss": 0.2948,
+      "step": 10860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7278505080111e-05,
+      "loss": 0.3304,
+      "step": 10861
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7272407277390066e-05,
+      "loss": 0.3087,
+      "step": 10862
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7266310421361603e-05,
+      "loss": 0.3432,
+      "step": 10863
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7260214512589956e-05,
+      "loss": 0.2923,
+      "step": 10864
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7254119551639382e-05,
+      "loss": 0.3178,
+      "step": 10865
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7248025539074048e-05,
+      "loss": 0.3132,
+      "step": 10866
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7241932475458027e-05,
+      "loss": 0.3147,
+      "step": 10867
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7235840361355312e-05,
+      "loss": 0.2936,
+      "step": 10868
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.722974919732979e-05,
+      "loss": 0.3082,
+      "step": 10869
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.72236589839453e-05,
+      "loss": 0.3196,
+      "step": 10870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7217569721765554e-05,
+      "loss": 0.3069,
+      "step": 10871
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7211481411354193e-05,
+      "loss": 0.2987,
+      "step": 10872
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7205394053274764e-05,
+      "loss": 0.2967,
+      "step": 10873
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.719930764809072e-05,
+      "loss": 0.3114,
+      "step": 10874
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7193222196365456e-05,
+      "loss": 0.3161,
+      "step": 10875
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7187137698662246e-05,
+      "loss": 0.2995,
+      "step": 10876
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7181054155544283e-05,
+      "loss": 0.3108,
+      "step": 10877
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7174971567574682e-05,
+      "loss": 0.3051,
+      "step": 10878
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7168889935316444e-05,
+      "loss": 0.2699,
+      "step": 10879
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7162809259332536e-05,
+      "loss": 0.3021,
+      "step": 10880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7156729540185773e-05,
+      "loss": 0.3143,
+      "step": 10881
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7150650778438926e-05,
+      "loss": 0.3047,
+      "step": 10882
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.714457297465464e-05,
+      "loss": 0.3145,
+      "step": 10883
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7138496129395515e-05,
+      "loss": 0.2731,
+      "step": 10884
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.713242024322403e-05,
+      "loss": 0.3147,
+      "step": 10885
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7126345316702592e-05,
+      "loss": 0.3115,
+      "step": 10886
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7120271350393488e-05,
+      "loss": 0.3003,
+      "step": 10887
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7114198344858965e-05,
+      "loss": 0.3046,
+      "step": 10888
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7108126300661152e-05,
+      "loss": 0.3001,
+      "step": 10889
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7102055218362086e-05,
+      "loss": 0.2945,
+      "step": 10890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7095985098523724e-05,
+      "loss": 0.3029,
+      "step": 10891
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7089915941707922e-05,
+      "loss": 0.3152,
+      "step": 10892
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7083847748476482e-05,
+      "loss": 0.2857,
+      "step": 10893
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.707778051939108e-05,
+      "loss": 0.3035,
+      "step": 10894
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7071714255013305e-05,
+      "loss": 0.315,
+      "step": 10895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.706564895590468e-05,
+      "loss": 0.3084,
+      "step": 10896
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7059584622626602e-05,
+      "loss": 0.285,
+      "step": 10897
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7053521255740432e-05,
+      "loss": 0.3108,
+      "step": 10898
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.704745885580739e-05,
+      "loss": 0.2949,
+      "step": 10899
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.704139742338864e-05,
+      "loss": 0.3236,
+      "step": 10900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.703533695904524e-05,
+      "loss": 0.2906,
+      "step": 10901
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7029277463338146e-05,
+      "loss": 0.3235,
+      "step": 10902
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7023218936828268e-05,
+      "loss": 0.2889,
+      "step": 10903
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7017161380076382e-05,
+      "loss": 0.3176,
+      "step": 10904
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7011104793643195e-05,
+      "loss": 0.3048,
+      "step": 10905
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.7005049178089322e-05,
+      "loss": 0.2849,
+      "step": 10906
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6998994533975273e-05,
+      "loss": 0.3149,
+      "step": 10907
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.69929408618615e-05,
+      "loss": 0.2871,
+      "step": 10908
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6986888162308344e-05,
+      "loss": 0.2907,
+      "step": 10909
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6980836435876053e-05,
+      "loss": 0.317,
+      "step": 10910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6974785683124795e-05,
+      "loss": 0.2932,
+      "step": 10911
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6968735904614628e-05,
+      "loss": 0.3097,
+      "step": 10912
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6962687100905555e-05,
+      "loss": 0.3029,
+      "step": 10913
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6956639272557462e-05,
+      "loss": 0.3058,
+      "step": 10914
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6950592420130153e-05,
+      "loss": 0.301,
+      "step": 10915
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6944546544183333e-05,
+      "loss": 0.3086,
+      "step": 10916
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6938501645276624e-05,
+      "loss": 0.3073,
+      "step": 10917
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6932457723969573e-05,
+      "loss": 0.3032,
+      "step": 10918
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6926414780821608e-05,
+      "loss": 0.3115,
+      "step": 10919
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6920372816392084e-05,
+      "loss": 0.3162,
+      "step": 10920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6914331831240255e-05,
+      "loss": 0.2877,
+      "step": 10921
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6908291825925287e-05,
+      "loss": 0.3156,
+      "step": 10922
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6902252801006276e-05,
+      "loss": 0.332,
+      "step": 10923
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.68962147570422e-05,
+      "loss": 0.3132,
+      "step": 10924
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6890177694591953e-05,
+      "loss": 0.2988,
+      "step": 10925
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6884141614214343e-05,
+      "loss": 0.3189,
+      "step": 10926
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6878106516468084e-05,
+      "loss": 0.3058,
+      "step": 10927
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6872072401911805e-05,
+      "loss": 0.3173,
+      "step": 10928
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6866039271104036e-05,
+      "loss": 0.3185,
+      "step": 10929
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6860007124603223e-05,
+      "loss": 0.3194,
+      "step": 10930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6853975962967712e-05,
+      "loss": 0.309,
+      "step": 10931
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6847945786755753e-05,
+      "loss": 0.3002,
+      "step": 10932
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6841916596525537e-05,
+      "loss": 0.3124,
+      "step": 10933
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6835888392835134e-05,
+      "loss": 0.3032,
+      "step": 10934
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6829861176242524e-05,
+      "loss": 0.2996,
+      "step": 10935
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6823834947305608e-05,
+      "loss": 0.3165,
+      "step": 10936
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.681780970658217e-05,
+      "loss": 0.3146,
+      "step": 10937
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.681178545462996e-05,
+      "loss": 0.3105,
+      "step": 10938
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6805762192006564e-05,
+      "loss": 0.2975,
+      "step": 10939
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6799739919269527e-05,
+      "loss": 0.2934,
+      "step": 10940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6793718636976282e-05,
+      "loss": 0.3038,
+      "step": 10941
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6787698345684168e-05,
+      "loss": 0.3025,
+      "step": 10942
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6781679045950455e-05,
+      "loss": 0.3139,
+      "step": 10943
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6775660738332294e-05,
+      "loss": 0.2972,
+      "step": 10944
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6769643423386756e-05,
+      "loss": 0.2971,
+      "step": 10945
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.676362710167082e-05,
+      "loss": 0.3091,
+      "step": 10946
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6757611773741365e-05,
+      "loss": 0.3047,
+      "step": 10947
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6751597440155202e-05,
+      "loss": 0.3079,
+      "step": 10948
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6745584101469025e-05,
+      "loss": 0.3237,
+      "step": 10949
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6739571758239438e-05,
+      "loss": 0.3207,
+      "step": 10950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.673356041102297e-05,
+      "loss": 0.3019,
+      "step": 10951
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6727550060376026e-05,
+      "loss": 0.3047,
+      "step": 10952
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.672154070685497e-05,
+      "loss": 0.3127,
+      "step": 10953
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6715532351016024e-05,
+      "loss": 0.3135,
+      "step": 10954
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6709524993415344e-05,
+      "loss": 0.3135,
+      "step": 10955
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.670351863460898e-05,
+      "loss": 0.2971,
+      "step": 10956
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6697513275152893e-05,
+      "loss": 0.3072,
+      "step": 10957
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.669150891560297e-05,
+      "loss": 0.3276,
+      "step": 10958
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6685505556514987e-05,
+      "loss": 0.3252,
+      "step": 10959
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6679503198444617e-05,
+      "loss": 0.2881,
+      "step": 10960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6673501841947465e-05,
+      "loss": 0.2988,
+      "step": 10961
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6667501487579017e-05,
+      "loss": 0.3107,
+      "step": 10962
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6661502135894705e-05,
+      "loss": 0.3099,
+      "step": 10963
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6655503787449832e-05,
+      "loss": 0.3026,
+      "step": 10964
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6649506442799623e-05,
+      "loss": 0.2914,
+      "step": 10965
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6643510102499198e-05,
+      "loss": 0.3123,
+      "step": 10966
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6637514767103608e-05,
+      "loss": 0.3058,
+      "step": 10967
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.663152043716779e-05,
+      "loss": 0.2934,
+      "step": 10968
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6625527113246597e-05,
+      "loss": 0.317,
+      "step": 10969
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6619534795894785e-05,
+      "loss": 0.3152,
+      "step": 10970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.661354348566701e-05,
+      "loss": 0.3067,
+      "step": 10971
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.660755318311786e-05,
+      "loss": 0.3078,
+      "step": 10972
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6601563888801805e-05,
+      "loss": 0.3394,
+      "step": 10973
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.659557560327323e-05,
+      "loss": 0.2877,
+      "step": 10974
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6589588327086423e-05,
+      "loss": 0.2996,
+      "step": 10975
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6583602060795577e-05,
+      "loss": 0.3165,
+      "step": 10976
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6577616804954813e-05,
+      "loss": 0.2996,
+      "step": 10977
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6571632560118127e-05,
+      "loss": 0.3062,
+      "step": 10978
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.656564932683945e-05,
+      "loss": 0.3101,
+      "step": 10979
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6559667105672588e-05,
+      "loss": 0.3212,
+      "step": 10980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.655368589717128e-05,
+      "loss": 0.2845,
+      "step": 10981
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6547705701889167e-05,
+      "loss": 0.287,
+      "step": 10982
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6541726520379786e-05,
+      "loss": 0.2971,
+      "step": 10983
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6535748353196584e-05,
+      "loss": 0.3002,
+      "step": 10984
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.652977120089292e-05,
+      "loss": 0.2981,
+      "step": 10985
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6523795064022046e-05,
+      "loss": 0.2952,
+      "step": 10986
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6517819943137138e-05,
+      "loss": 0.3273,
+      "step": 10987
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.651184583879127e-05,
+      "loss": 0.3261,
+      "step": 10988
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6505872751537413e-05,
+      "loss": 0.3124,
+      "step": 10989
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.649990068192846e-05,
+      "loss": 0.3127,
+      "step": 10990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.649392963051718e-05,
+      "loss": 0.3065,
+      "step": 10991
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.64879595978563e-05,
+      "loss": 0.3084,
+      "step": 10992
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6481990584498404e-05,
+      "loss": 0.3031,
+      "step": 10993
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6476022590996002e-05,
+      "loss": 0.295,
+      "step": 10994
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.647005561790151e-05,
+      "loss": 0.3196,
+      "step": 10995
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.646408966576723e-05,
+      "loss": 0.3031,
+      "step": 10996
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6458124735145407e-05,
+      "loss": 0.2976,
+      "step": 10997
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6452160826588166e-05,
+      "loss": 0.3097,
+      "step": 10998
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.644619794064754e-05,
+      "loss": 0.3284,
+      "step": 10999
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6440236077875468e-05,
+      "loss": 0.3248,
+      "step": 11000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.355400696864112e-08,
+      "loss": 0.2828,
+      "step": 11001
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.710801393728224e-08,
+      "loss": 0.3042,
+      "step": 11002
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3066202090592336e-07,
+      "loss": 0.2842,
+      "step": 11003
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7421602787456448e-07,
+      "loss": 0.3053,
+      "step": 11004
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.1777003484320556e-07,
+      "loss": 0.2967,
+      "step": 11005
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6132404181184673e-07,
+      "loss": 0.2839,
+      "step": 11006
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0487804878048784e-07,
+      "loss": 0.2914,
+      "step": 11007
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.4843205574912896e-07,
+      "loss": 0.2774,
+      "step": 11008
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.9198606271777007e-07,
+      "loss": 0.2887,
+      "step": 11009
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.3554006968641113e-07,
+      "loss": 0.2751,
+      "step": 11010
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.790940766550523e-07,
+      "loss": 0.2868,
+      "step": 11011
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.226480836236935e-07,
+      "loss": 0.3007,
+      "step": 11012
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.662020905923345e-07,
+      "loss": 0.2728,
+      "step": 11013
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.097560975609757e-07,
+      "loss": 0.2763,
+      "step": 11014
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.533101045296167e-07,
+      "loss": 0.292,
+      "step": 11015
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.968641114982579e-07,
+      "loss": 0.2776,
+      "step": 11016
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.40418118466899e-07,
+      "loss": 0.2819,
+      "step": 11017
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.839721254355401e-07,
+      "loss": 0.2943,
+      "step": 11018
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.275261324041812e-07,
+      "loss": 0.2968,
+      "step": 11019
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.710801393728223e-07,
+      "loss": 0.287,
+      "step": 11020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.146341463414634e-07,
+      "loss": 0.3149,
+      "step": 11021
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.581881533101046e-07,
+      "loss": 0.2869,
+      "step": 11022
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0017421602787455e-06,
+      "loss": 0.2828,
+      "step": 11023
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.045296167247387e-06,
+      "loss": 0.2754,
+      "step": 11024
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.088850174216028e-06,
+      "loss": 0.2963,
+      "step": 11025
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.132404181184669e-06,
+      "loss": 0.2956,
+      "step": 11026
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1759581881533102e-06,
+      "loss": 0.3167,
+      "step": 11027
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 0.3023,
+      "step": 11028
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2630662020905923e-06,
+      "loss": 0.3199,
+      "step": 11029
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3066202090592335e-06,
+      "loss": 0.3046,
+      "step": 11030
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3501742160278747e-06,
+      "loss": 0.2892,
+      "step": 11031
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3937282229965158e-06,
+      "loss": 0.2948,
+      "step": 11032
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4372822299651568e-06,
+      "loss": 0.2857,
+      "step": 11033
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.480836236933798e-06,
+      "loss": 0.2808,
+      "step": 11034
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5243902439024391e-06,
+      "loss": 0.3025,
+      "step": 11035
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5679442508710803e-06,
+      "loss": 0.309,
+      "step": 11036
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6114982578397212e-06,
+      "loss": 0.2935,
+      "step": 11037
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6550522648083624e-06,
+      "loss": 0.2517,
+      "step": 11038
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6986062717770036e-06,
+      "loss": 0.3052,
+      "step": 11039
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7421602787456445e-06,
+      "loss": 0.3025,
+      "step": 11040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7857142857142857e-06,
+      "loss": 0.2924,
+      "step": 11041
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8292682926829268e-06,
+      "loss": 0.2884,
+      "step": 11042
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.872822299651568e-06,
+      "loss": 0.2724,
+      "step": 11043
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.916376306620209e-06,
+      "loss": 0.3134,
+      "step": 11044
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.95993031358885e-06,
+      "loss": 0.2909,
+      "step": 11045
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.003484320557491e-06,
+      "loss": 0.3028,
+      "step": 11046
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.0470383275261322e-06,
+      "loss": 0.2994,
+      "step": 11047
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.090592334494774e-06,
+      "loss": 0.2723,
+      "step": 11048
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.134146341463415e-06,
+      "loss": 0.302,
+      "step": 11049
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.177700348432056e-06,
+      "loss": 0.2844,
+      "step": 11050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.221254355400697e-06,
+      "loss": 0.3138,
+      "step": 11051
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.264808362369338e-06,
+      "loss": 0.301,
+      "step": 11052
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3083623693379792e-06,
+      "loss": 0.2973,
+      "step": 11053
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3519163763066204e-06,
+      "loss": 0.281,
+      "step": 11054
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3954703832752616e-06,
+      "loss": 0.2845,
+      "step": 11055
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 0.3025,
+      "step": 11056
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.482578397212544e-06,
+      "loss": 0.2998,
+      "step": 11057
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.5261324041811846e-06,
+      "loss": 0.2796,
+      "step": 11058
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.569686411149826e-06,
+      "loss": 0.283,
+      "step": 11059
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.613240418118467e-06,
+      "loss": 0.309,
+      "step": 11060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.656794425087108e-06,
+      "loss": 0.2882,
+      "step": 11061
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.7003484320557493e-06,
+      "loss": 0.2719,
+      "step": 11062
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.7439024390243905e-06,
+      "loss": 0.307,
+      "step": 11063
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.7874564459930316e-06,
+      "loss": 0.2976,
+      "step": 11064
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8310104529616724e-06,
+      "loss": 0.2718,
+      "step": 11065
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8745644599303136e-06,
+      "loss": 0.2917,
+      "step": 11066
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.9181184668989547e-06,
+      "loss": 0.2811,
+      "step": 11067
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.961672473867596e-06,
+      "loss": 0.2971,
+      "step": 11068
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.005226480836237e-06,
+      "loss": 0.2948,
+      "step": 11069
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0487804878048782e-06,
+      "loss": 0.2922,
+      "step": 11070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0923344947735194e-06,
+      "loss": 0.3054,
+      "step": 11071
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.1358885017421605e-06,
+      "loss": 0.3176,
+      "step": 11072
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.1794425087108017e-06,
+      "loss": 0.3136,
+      "step": 11073
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.2229965156794425e-06,
+      "loss": 0.31,
+      "step": 11074
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.266550522648084e-06,
+      "loss": 0.2836,
+      "step": 11075
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.3101045296167248e-06,
+      "loss": 0.3048,
+      "step": 11076
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.3536585365853664e-06,
+      "loss": 0.3001,
+      "step": 11077
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.397212543554007e-06,
+      "loss": 0.2697,
+      "step": 11078
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.4407665505226483e-06,
+      "loss": 0.2916,
+      "step": 11079
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.484320557491289e-06,
+      "loss": 0.283,
+      "step": 11080
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5278745644599306e-06,
+      "loss": 0.2997,
+      "step": 11081
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 0.2952,
+      "step": 11082
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.614982578397213e-06,
+      "loss": 0.2781,
+      "step": 11083
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 0.2837,
+      "step": 11084
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7020905923344953e-06,
+      "loss": 0.3074,
+      "step": 11085
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.745644599303136e-06,
+      "loss": 0.2745,
+      "step": 11086
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.789198606271777e-06,
+      "loss": 0.2999,
+      "step": 11087
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.832752613240418e-06,
+      "loss": 0.3013,
+      "step": 11088
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.876306620209059e-06,
+      "loss": 0.2984,
+      "step": 11089
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.9198606271777e-06,
+      "loss": 0.2928,
+      "step": 11090
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.9634146341463414e-06,
+      "loss": 0.2789,
+      "step": 11091
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.006968641114982e-06,
+      "loss": 0.2886,
+      "step": 11092
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.050522648083624e-06,
+      "loss": 0.2998,
+      "step": 11093
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0940766550522645e-06,
+      "loss": 0.2898,
+      "step": 11094
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.137630662020906e-06,
+      "loss": 0.301,
+      "step": 11095
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.181184668989548e-06,
+      "loss": 0.2745,
+      "step": 11096
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.224738675958188e-06,
+      "loss": 0.3013,
+      "step": 11097
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.26829268292683e-06,
+      "loss": 0.2843,
+      "step": 11098
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.311846689895471e-06,
+      "loss": 0.2844,
+      "step": 11099
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.355400696864112e-06,
+      "loss": 0.3135,
+      "step": 11100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.398954703832753e-06,
+      "loss": 0.2873,
+      "step": 11101
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.442508710801394e-06,
+      "loss": 0.2768,
+      "step": 11102
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.4860627177700346e-06,
+      "loss": 0.3011,
+      "step": 11103
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.529616724738676e-06,
+      "loss": 0.3087,
+      "step": 11104
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.573170731707317e-06,
+      "loss": 0.2944,
+      "step": 11105
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6167247386759585e-06,
+      "loss": 0.2743,
+      "step": 11106
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.660278745644599e-06,
+      "loss": 0.3049,
+      "step": 11107
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.703832752613241e-06,
+      "loss": 0.3038,
+      "step": 11108
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.7473867595818816e-06,
+      "loss": 0.2963,
+      "step": 11109
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.790940766550523e-06,
+      "loss": 0.2846,
+      "step": 11110
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.834494773519164e-06,
+      "loss": 0.2896,
+      "step": 11111
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.2924,
+      "step": 11112
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.921602787456446e-06,
+      "loss": 0.313,
+      "step": 11113
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.965156794425088e-06,
+      "loss": 0.2854,
+      "step": 11114
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.0087108013937286e-06,
+      "loss": 0.2882,
+      "step": 11115
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.052264808362369e-06,
+      "loss": 0.2912,
+      "step": 11116
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.09581881533101e-06,
+      "loss": 0.2714,
+      "step": 11117
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.139372822299652e-06,
+      "loss": 0.3016,
+      "step": 11118
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.182926829268292e-06,
+      "loss": 0.3097,
+      "step": 11119
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.226480836236934e-06,
+      "loss": 0.3014,
+      "step": 11120
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.270034843205575e-06,
+      "loss": 0.2885,
+      "step": 11121
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.313588850174216e-06,
+      "loss": 0.2995,
+      "step": 11122
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.2871,
+      "step": 11123
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.400696864111499e-06,
+      "loss": 0.3076,
+      "step": 11124
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.444250871080139e-06,
+      "loss": 0.3002,
+      "step": 11125
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.487804878048781e-06,
+      "loss": 0.303,
+      "step": 11126
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.531358885017422e-06,
+      "loss": 0.2981,
+      "step": 11127
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.574912891986063e-06,
+      "loss": 0.2893,
+      "step": 11128
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.618466898954704e-06,
+      "loss": 0.2799,
+      "step": 11129
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.662020905923345e-06,
+      "loss": 0.2909,
+      "step": 11130
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.705574912891986e-06,
+      "loss": 0.2832,
+      "step": 11131
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.749128919860627e-06,
+      "loss": 0.299,
+      "step": 11132
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.792682926829269e-06,
+      "loss": 0.2974,
+      "step": 11133
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8362369337979094e-06,
+      "loss": 0.3154,
+      "step": 11134
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.879790940766551e-06,
+      "loss": 0.2793,
+      "step": 11135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.923344947735192e-06,
+      "loss": 0.292,
+      "step": 11136
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.966898954703833e-06,
+      "loss": 0.302,
+      "step": 11137
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.010452961672474e-06,
+      "loss": 0.2803,
+      "step": 11138
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.054006968641116e-06,
+      "loss": 0.2941,
+      "step": 11139
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 0.3027,
+      "step": 11140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.141114982578398e-06,
+      "loss": 0.2862,
+      "step": 11141
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.184668989547039e-06,
+      "loss": 0.2887,
+      "step": 11142
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.2282229965156795e-06,
+      "loss": 0.2894,
+      "step": 11143
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.271777003484321e-06,
+      "loss": 0.2875,
+      "step": 11144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.315331010452962e-06,
+      "loss": 0.2966,
+      "step": 11145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.3588850174216034e-06,
+      "loss": 0.2819,
+      "step": 11146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.402439024390244e-06,
+      "loss": 0.2799,
+      "step": 11147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.445993031358885e-06,
+      "loss": 0.3062,
+      "step": 11148
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.489547038327526e-06,
+      "loss": 0.2995,
+      "step": 11149
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.533101045296168e-06,
+      "loss": 0.3002,
+      "step": 11150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.576655052264809e-06,
+      "loss": 0.2932,
+      "step": 11151
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.6202090592334496e-06,
+      "loss": 0.2931,
+      "step": 11152
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.66376306620209e-06,
+      "loss": 0.3175,
+      "step": 11153
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.707317073170733e-06,
+      "loss": 0.3132,
+      "step": 11154
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.7508710801393735e-06,
+      "loss": 0.2847,
+      "step": 11155
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.794425087108014e-06,
+      "loss": 0.2952,
+      "step": 11156
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.837979094076655e-06,
+      "loss": 0.2876,
+      "step": 11157
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.8815331010452966e-06,
+      "loss": 0.2897,
+      "step": 11158
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.925087108013937e-06,
+      "loss": 0.2927,
+      "step": 11159
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.968641114982578e-06,
+      "loss": 0.3052,
+      "step": 11160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.0121951219512205e-06,
+      "loss": 0.3069,
+      "step": 11161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.055749128919861e-06,
+      "loss": 0.291,
+      "step": 11162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.099303135888502e-06,
+      "loss": 0.2831,
+      "step": 11163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.269,
+      "step": 11164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.186411149825785e-06,
+      "loss": 0.2984,
+      "step": 11165
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.229965156794426e-06,
+      "loss": 0.3021,
+      "step": 11166
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.273519163763067e-06,
+      "loss": 0.2905,
+      "step": 11167
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 0.2998,
+      "step": 11168
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.360627177700349e-06,
+      "loss": 0.2841,
+      "step": 11169
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.4041811846689906e-06,
+      "loss": 0.3126,
+      "step": 11170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.447735191637631e-06,
+      "loss": 0.3237,
+      "step": 11171
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.491289198606272e-06,
+      "loss": 0.3126,
+      "step": 11172
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.534843205574914e-06,
+      "loss": 0.2814,
+      "step": 11173
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.578397212543554e-06,
+      "loss": 0.3014,
+      "step": 11174
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.621951219512195e-06,
+      "loss": 0.2953,
+      "step": 11175
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.665505226480837e-06,
+      "loss": 0.3039,
+      "step": 11176
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.709059233449478e-06,
+      "loss": 0.2929,
+      "step": 11177
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.752613240418118e-06,
+      "loss": 0.3599,
+      "step": 11178
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.79616724738676e-06,
+      "loss": 0.3077,
+      "step": 11179
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.8397212543554e-06,
+      "loss": 0.2985,
+      "step": 11180
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.883275261324043e-06,
+      "loss": 0.3128,
+      "step": 11181
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.926829268292683e-06,
+      "loss": 0.2809,
+      "step": 11182
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.970383275261324e-06,
+      "loss": 0.2961,
+      "step": 11183
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.013937282229964e-06,
+      "loss": 0.2986,
+      "step": 11184
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.057491289198608e-06,
+      "loss": 0.2903,
+      "step": 11185
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.101045296167248e-06,
+      "loss": 0.3037,
+      "step": 11186
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.144599303135889e-06,
+      "loss": 0.2939,
+      "step": 11187
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.188153310104529e-06,
+      "loss": 0.3047,
+      "step": 11188
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.231707317073172e-06,
+      "loss": 0.2753,
+      "step": 11189
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.275261324041812e-06,
+      "loss": 0.2962,
+      "step": 11190
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.318815331010454e-06,
+      "loss": 0.2889,
+      "step": 11191
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.362369337979095e-06,
+      "loss": 0.2996,
+      "step": 11192
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.405923344947735e-06,
+      "loss": 0.2929,
+      "step": 11193
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.449477351916377e-06,
+      "loss": 0.2827,
+      "step": 11194
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.493031358885017e-06,
+      "loss": 0.2927,
+      "step": 11195
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.53658536585366e-06,
+      "loss": 0.3041,
+      "step": 11196
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.5801393728223e-06,
+      "loss": 0.2853,
+      "step": 11197
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.623693379790942e-06,
+      "loss": 0.283,
+      "step": 11198
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.667247386759581e-06,
+      "loss": 0.2809,
+      "step": 11199
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.710801393728225e-06,
+      "loss": 0.2873,
+      "step": 11200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.754355400696865e-06,
+      "loss": 0.3033,
+      "step": 11201
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.797909407665506e-06,
+      "loss": 0.2984,
+      "step": 11202
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.841463414634146e-06,
+      "loss": 0.2789,
+      "step": 11203
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.885017421602788e-06,
+      "loss": 0.2849,
+      "step": 11204
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.2994,
+      "step": 11205
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.972125435540069e-06,
+      "loss": 0.3058,
+      "step": 11206
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.01567944250871e-06,
+      "loss": 0.2978,
+      "step": 11207
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.059233449477352e-06,
+      "loss": 0.2948,
+      "step": 11208
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.102787456445994e-06,
+      "loss": 0.2989,
+      "step": 11209
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.146341463414634e-06,
+      "loss": 0.2824,
+      "step": 11210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.189895470383275e-06,
+      "loss": 0.2925,
+      "step": 11211
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.233449477351917e-06,
+      "loss": 0.2933,
+      "step": 11212
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.277003484320559e-06,
+      "loss": 0.2859,
+      "step": 11213
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.320557491289198e-06,
+      "loss": 0.2963,
+      "step": 11214
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.36411149825784e-06,
+      "loss": 0.2892,
+      "step": 11215
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.407665505226482e-06,
+      "loss": 0.2968,
+      "step": 11216
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.451219512195123e-06,
+      "loss": 0.3104,
+      "step": 11217
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.494773519163763e-06,
+      "loss": 0.2814,
+      "step": 11218
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.538327526132405e-06,
+      "loss": 0.299,
+      "step": 11219
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.581881533101046e-06,
+      "loss": 0.2992,
+      "step": 11220
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.625435540069686e-06,
+      "loss": 0.3031,
+      "step": 11221
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.668989547038328e-06,
+      "loss": 0.3016,
+      "step": 11222
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.712543554006968e-06,
+      "loss": 0.2881,
+      "step": 11223
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.2984,
+      "step": 11224
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.799651567944251e-06,
+      "loss": 0.3124,
+      "step": 11225
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.843205574912892e-06,
+      "loss": 0.2902,
+      "step": 11226
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.886759581881534e-06,
+      "loss": 0.2796,
+      "step": 11227
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.930313588850176e-06,
+      "loss": 0.2962,
+      "step": 11228
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.973867595818816e-06,
+      "loss": 0.3013,
+      "step": 11229
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0017421602787457e-05,
+      "loss": 0.3004,
+      "step": 11230
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0060975609756099e-05,
+      "loss": 0.3002,
+      "step": 11231
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0104529616724739e-05,
+      "loss": 0.3059,
+      "step": 11232
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.014808362369338e-05,
+      "loss": 0.3052,
+      "step": 11233
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.019163763066202e-05,
+      "loss": 0.3038,
+      "step": 11234
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0235191637630663e-05,
+      "loss": 0.2926,
+      "step": 11235
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0278745644599303e-05,
+      "loss": 0.3032,
+      "step": 11236
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0322299651567945e-05,
+      "loss": 0.2875,
+      "step": 11237
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0365853658536585e-05,
+      "loss": 0.2954,
+      "step": 11238
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0409407665505228e-05,
+      "loss": 0.285,
+      "step": 11239
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0452961672473868e-05,
+      "loss": 0.2842,
+      "step": 11240
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.049651567944251e-05,
+      "loss": 0.3079,
+      "step": 11241
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.054006968641115e-05,
+      "loss": 0.2964,
+      "step": 11242
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0583623693379793e-05,
+      "loss": 0.2992,
+      "step": 11243
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0627177700348433e-05,
+      "loss": 0.3049,
+      "step": 11244
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0670731707317074e-05,
+      "loss": 0.3014,
+      "step": 11245
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.3018,
+      "step": 11246
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0757839721254356e-05,
+      "loss": 0.3039,
+      "step": 11247
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0801393728222997e-05,
+      "loss": 0.2936,
+      "step": 11248
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0844947735191637e-05,
+      "loss": 0.3043,
+      "step": 11249
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0888501742160279e-05,
+      "loss": 0.3005,
+      "step": 11250
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.093205574912892e-05,
+      "loss": 0.2886,
+      "step": 11251
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0975609756097562e-05,
+      "loss": 0.2895,
+      "step": 11252
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1019163763066202e-05,
+      "loss": 0.2795,
+      "step": 11253
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1062717770034843e-05,
+      "loss": 0.3038,
+      "step": 11254
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1106271777003485e-05,
+      "loss": 0.296,
+      "step": 11255
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1149825783972127e-05,
+      "loss": 0.3074,
+      "step": 11256
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1193379790940766e-05,
+      "loss": 0.3104,
+      "step": 11257
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1236933797909408e-05,
+      "loss": 0.2843,
+      "step": 11258
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.128048780487805e-05,
+      "loss": 0.2962,
+      "step": 11259
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.132404181184669e-05,
+      "loss": 0.2756,
+      "step": 11260
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1367595818815331e-05,
+      "loss": 0.2939,
+      "step": 11261
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1411149825783973e-05,
+      "loss": 0.2906,
+      "step": 11262
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1454703832752614e-05,
+      "loss": 0.3276,
+      "step": 11263
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1498257839721254e-05,
+      "loss": 0.2906,
+      "step": 11264
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1541811846689896e-05,
+      "loss": 0.2961,
+      "step": 11265
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1585365853658537e-05,
+      "loss": 0.2674,
+      "step": 11266
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1628919860627179e-05,
+      "loss": 0.2885,
+      "step": 11267
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1672473867595819e-05,
+      "loss": 0.3098,
+      "step": 11268
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.171602787456446e-05,
+      "loss": 0.2829,
+      "step": 11269
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1759581881533102e-05,
+      "loss": 0.291,
+      "step": 11270
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1803135888501744e-05,
+      "loss": 0.281,
+      "step": 11271
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1846689895470384e-05,
+      "loss": 0.2826,
+      "step": 11272
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1890243902439025e-05,
+      "loss": 0.3047,
+      "step": 11273
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1933797909407667e-05,
+      "loss": 0.2867,
+      "step": 11274
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1977351916376307e-05,
+      "loss": 0.2908,
+      "step": 11275
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2020905923344948e-05,
+      "loss": 0.2849,
+      "step": 11276
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2064459930313588e-05,
+      "loss": 0.2974,
+      "step": 11277
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2108013937282231e-05,
+      "loss": 0.2825,
+      "step": 11278
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2151567944250871e-05,
+      "loss": 0.2893,
+      "step": 11279
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 0.2887,
+      "step": 11280
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2238675958188153e-05,
+      "loss": 0.3093,
+      "step": 11281
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2282229965156796e-05,
+      "loss": 0.293,
+      "step": 11282
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2325783972125436e-05,
+      "loss": 0.2954,
+      "step": 11283
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2369337979094078e-05,
+      "loss": 0.3062,
+      "step": 11284
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2412891986062717e-05,
+      "loss": 0.3159,
+      "step": 11285
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2456445993031359e-05,
+      "loss": 0.2873,
+      "step": 11286
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3034,
+      "step": 11287
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2543554006968642e-05,
+      "loss": 0.2876,
+      "step": 11288
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2587108013937282e-05,
+      "loss": 0.2919,
+      "step": 11289
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2630662020905924e-05,
+      "loss": 0.2838,
+      "step": 11290
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2674216027874567e-05,
+      "loss": 0.2904,
+      "step": 11291
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2717770034843207e-05,
+      "loss": 0.2806,
+      "step": 11292
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2761324041811848e-05,
+      "loss": 0.293,
+      "step": 11293
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2804878048780488e-05,
+      "loss": 0.294,
+      "step": 11294
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.284843205574913e-05,
+      "loss": 0.2896,
+      "step": 11295
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.289198606271777e-05,
+      "loss": 0.2786,
+      "step": 11296
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2935540069686411e-05,
+      "loss": 0.3011,
+      "step": 11297
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2979094076655051e-05,
+      "loss": 0.2821,
+      "step": 11298
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3022648083623695e-05,
+      "loss": 0.3006,
+      "step": 11299
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3066202090592336e-05,
+      "loss": 0.2985,
+      "step": 11300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3109756097560976e-05,
+      "loss": 0.2913,
+      "step": 11301
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3153310104529618e-05,
+      "loss": 0.2964,
+      "step": 11302
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3196864111498258e-05,
+      "loss": 0.3071,
+      "step": 11303
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3240418118466899e-05,
+      "loss": 0.2964,
+      "step": 11304
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3283972125435539e-05,
+      "loss": 0.302,
+      "step": 11305
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.332752613240418e-05,
+      "loss": 0.2913,
+      "step": 11306
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3371080139372824e-05,
+      "loss": 0.2787,
+      "step": 11307
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3414634146341466e-05,
+      "loss": 0.2976,
+      "step": 11308
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3458188153310105e-05,
+      "loss": 0.2893,
+      "step": 11309
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3501742160278747e-05,
+      "loss": 0.2968,
+      "step": 11310
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3545296167247387e-05,
+      "loss": 0.2931,
+      "step": 11311
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3588850174216028e-05,
+      "loss": 0.3025,
+      "step": 11312
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3632404181184668e-05,
+      "loss": 0.2856,
+      "step": 11313
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.367595818815331e-05,
+      "loss": 0.3138,
+      "step": 11314
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3719512195121953e-05,
+      "loss": 0.2995,
+      "step": 11315
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3763066202090593e-05,
+      "loss": 0.302,
+      "step": 11316
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3806620209059235e-05,
+      "loss": 0.2792,
+      "step": 11317
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3850174216027875e-05,
+      "loss": 0.2865,
+      "step": 11318
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3893728222996516e-05,
+      "loss": 0.2815,
+      "step": 11319
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3937282229965156e-05,
+      "loss": 0.3034,
+      "step": 11320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3980836236933798e-05,
+      "loss": 0.2725,
+      "step": 11321
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4024390243902441e-05,
+      "loss": 0.2779,
+      "step": 11322
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4067944250871083e-05,
+      "loss": 0.2857,
+      "step": 11323
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4111498257839722e-05,
+      "loss": 0.2798,
+      "step": 11324
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4155052264808364e-05,
+      "loss": 0.2959,
+      "step": 11325
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4198606271777004e-05,
+      "loss": 0.3053,
+      "step": 11326
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4242160278745646e-05,
+      "loss": 0.3064,
+      "step": 11327
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.301,
+      "step": 11328
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4329268292682927e-05,
+      "loss": 0.2949,
+      "step": 11329
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.437282229965157e-05,
+      "loss": 0.2895,
+      "step": 11330
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.441637630662021e-05,
+      "loss": 0.295,
+      "step": 11331
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4459930313588852e-05,
+      "loss": 0.284,
+      "step": 11332
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4503484320557492e-05,
+      "loss": 0.2894,
+      "step": 11333
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4547038327526133e-05,
+      "loss": 0.2944,
+      "step": 11334
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4590592334494773e-05,
+      "loss": 0.3027,
+      "step": 11335
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.3113,
+      "step": 11336
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4677700348432055e-05,
+      "loss": 0.3355,
+      "step": 11337
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4721254355400698e-05,
+      "loss": 0.2861,
+      "step": 11338
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.476480836236934e-05,
+      "loss": 0.3023,
+      "step": 11339
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4808362369337981e-05,
+      "loss": 0.2915,
+      "step": 11340
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4851916376306621e-05,
+      "loss": 0.3047,
+      "step": 11341
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4895470383275263e-05,
+      "loss": 0.2792,
+      "step": 11342
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4939024390243902e-05,
+      "loss": 0.2784,
+      "step": 11343
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4982578397212544e-05,
+      "loss": 0.3071,
+      "step": 11344
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5026132404181184e-05,
+      "loss": 0.2952,
+      "step": 11345
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5069686411149827e-05,
+      "loss": 0.2922,
+      "step": 11346
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5113240418118469e-05,
+      "loss": 0.3076,
+      "step": 11347
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5156794425087109e-05,
+      "loss": 0.3177,
+      "step": 11348
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.520034843205575e-05,
+      "loss": 0.2911,
+      "step": 11349
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.524390243902439e-05,
+      "loss": 0.3143,
+      "step": 11350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.528745644599303e-05,
+      "loss": 0.3028,
+      "step": 11351
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5331010452961673e-05,
+      "loss": 0.2759,
+      "step": 11352
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5374564459930317e-05,
+      "loss": 0.2782,
+      "step": 11353
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5418118466898957e-05,
+      "loss": 0.3022,
+      "step": 11354
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5461672473867596e-05,
+      "loss": 0.2759,
+      "step": 11355
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5505226480836236e-05,
+      "loss": 0.2947,
+      "step": 11356
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.554878048780488e-05,
+      "loss": 0.2922,
+      "step": 11357
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.559233449477352e-05,
+      "loss": 0.2871,
+      "step": 11358
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.563588850174216e-05,
+      "loss": 0.304,
+      "step": 11359
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.56794425087108e-05,
+      "loss": 0.3064,
+      "step": 11360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5722996515679446e-05,
+      "loss": 0.3137,
+      "step": 11361
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5766550522648086e-05,
+      "loss": 0.2857,
+      "step": 11362
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5810104529616726e-05,
+      "loss": 0.3062,
+      "step": 11363
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5853658536585366e-05,
+      "loss": 0.3013,
+      "step": 11364
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.589721254355401e-05,
+      "loss": 0.2981,
+      "step": 11365
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.594076655052265e-05,
+      "loss": 0.2845,
+      "step": 11366
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.598432055749129e-05,
+      "loss": 0.288,
+      "step": 11367
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.602787456445993e-05,
+      "loss": 0.3079,
+      "step": 11368
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.308,
+      "step": 11369
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6114982578397215e-05,
+      "loss": 0.2818,
+      "step": 11370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6158536585365855e-05,
+      "loss": 0.3046,
+      "step": 11371
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6202090592334495e-05,
+      "loss": 0.2812,
+      "step": 11372
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6245644599303135e-05,
+      "loss": 0.2953,
+      "step": 11373
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6289198606271778e-05,
+      "loss": 0.293,
+      "step": 11374
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6332752613240418e-05,
+      "loss": 0.2795,
+      "step": 11375
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6376306620209058e-05,
+      "loss": 0.2894,
+      "step": 11376
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.64198606271777e-05,
+      "loss": 0.2809,
+      "step": 11377
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6463414634146345e-05,
+      "loss": 0.2986,
+      "step": 11378
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6506968641114984e-05,
+      "loss": 0.2846,
+      "step": 11379
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6550522648083624e-05,
+      "loss": 0.2839,
+      "step": 11380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6594076655052264e-05,
+      "loss": 0.288,
+      "step": 11381
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6637630662020908e-05,
+      "loss": 0.276,
+      "step": 11382
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6681184668989547e-05,
+      "loss": 0.2863,
+      "step": 11383
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.672473867595819e-05,
+      "loss": 0.2891,
+      "step": 11384
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.676829268292683e-05,
+      "loss": 0.3507,
+      "step": 11385
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.681184668989547e-05,
+      "loss": 0.2952,
+      "step": 11386
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6855400696864114e-05,
+      "loss": 0.2819,
+      "step": 11387
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6898954703832754e-05,
+      "loss": 0.2931,
+      "step": 11388
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6942508710801394e-05,
+      "loss": 0.295,
+      "step": 11389
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6986062717770033e-05,
+      "loss": 0.3058,
+      "step": 11390
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7029616724738677e-05,
+      "loss": 0.3033,
+      "step": 11391
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.707317073170732e-05,
+      "loss": 0.3146,
+      "step": 11392
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.711672473867596e-05,
+      "loss": 0.311,
+      "step": 11393
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.71602787456446e-05,
+      "loss": 0.3078,
+      "step": 11394
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7203832752613243e-05,
+      "loss": 0.322,
+      "step": 11395
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7247386759581883e-05,
+      "loss": 0.2885,
+      "step": 11396
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7290940766550523e-05,
+      "loss": 0.2793,
+      "step": 11397
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7334494773519163e-05,
+      "loss": 0.28,
+      "step": 11398
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7378048780487806e-05,
+      "loss": 0.2805,
+      "step": 11399
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.742160278745645e-05,
+      "loss": 0.279,
+      "step": 11400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.746515679442509e-05,
+      "loss": 0.3083,
+      "step": 11401
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.750871080139373e-05,
+      "loss": 0.2741,
+      "step": 11402
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.755226480836237e-05,
+      "loss": 0.2786,
+      "step": 11403
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7595818815331012e-05,
+      "loss": 0.3038,
+      "step": 11404
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7639372822299652e-05,
+      "loss": 0.2906,
+      "step": 11405
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7682926829268292e-05,
+      "loss": 0.3023,
+      "step": 11406
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7726480836236932e-05,
+      "loss": 0.2977,
+      "step": 11407
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7770034843205575e-05,
+      "loss": 0.2905,
+      "step": 11408
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.781358885017422e-05,
+      "loss": 0.294,
+      "step": 11409
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.2844,
+      "step": 11410
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.79006968641115e-05,
+      "loss": 0.2854,
+      "step": 11411
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7944250871080138e-05,
+      "loss": 0.2846,
+      "step": 11412
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.798780487804878e-05,
+      "loss": 0.3019,
+      "step": 11413
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.803135888501742e-05,
+      "loss": 0.2816,
+      "step": 11414
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8074912891986065e-05,
+      "loss": 0.2874,
+      "step": 11415
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8118466898954705e-05,
+      "loss": 0.2998,
+      "step": 11416
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8162020905923348e-05,
+      "loss": 0.2893,
+      "step": 11417
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8205574912891988e-05,
+      "loss": 0.2854,
+      "step": 11418
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8249128919860628e-05,
+      "loss": 0.3063,
+      "step": 11419
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8292682926829268e-05,
+      "loss": 0.2832,
+      "step": 11420
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.833623693379791e-05,
+      "loss": 0.279,
+      "step": 11421
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.837979094076655e-05,
+      "loss": 0.2862,
+      "step": 11422
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8423344947735194e-05,
+      "loss": 0.2696,
+      "step": 11423
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8466898954703834e-05,
+      "loss": 0.2862,
+      "step": 11424
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8510452961672474e-05,
+      "loss": 0.2799,
+      "step": 11425
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8554006968641117e-05,
+      "loss": 0.3083,
+      "step": 11426
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8597560975609757e-05,
+      "loss": 0.2918,
+      "step": 11427
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8641114982578397e-05,
+      "loss": 0.2955,
+      "step": 11428
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8684668989547037e-05,
+      "loss": 0.2783,
+      "step": 11429
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.872822299651568e-05,
+      "loss": 0.2939,
+      "step": 11430
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8771777003484323e-05,
+      "loss": 0.3008,
+      "step": 11431
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8815331010452963e-05,
+      "loss": 0.3147,
+      "step": 11432
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8858885017421603e-05,
+      "loss": 0.2817,
+      "step": 11433
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8902439024390246e-05,
+      "loss": 0.2881,
+      "step": 11434
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8945993031358886e-05,
+      "loss": 0.3115,
+      "step": 11435
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8989547038327526e-05,
+      "loss": 0.283,
+      "step": 11436
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9033101045296166e-05,
+      "loss": 0.3012,
+      "step": 11437
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.907665505226481e-05,
+      "loss": 0.2938,
+      "step": 11438
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9120209059233453e-05,
+      "loss": 0.295,
+      "step": 11439
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9163763066202093e-05,
+      "loss": 0.2969,
+      "step": 11440
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9207317073170733e-05,
+      "loss": 0.3116,
+      "step": 11441
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9250871080139372e-05,
+      "loss": 0.2973,
+      "step": 11442
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9294425087108016e-05,
+      "loss": 0.2918,
+      "step": 11443
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9337979094076656e-05,
+      "loss": 0.2925,
+      "step": 11444
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9381533101045295e-05,
+      "loss": 0.2986,
+      "step": 11445
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9425087108013935e-05,
+      "loss": 0.3009,
+      "step": 11446
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9468641114982582e-05,
+      "loss": 0.3013,
+      "step": 11447
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.3038,
+      "step": 11448
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9555749128919862e-05,
+      "loss": 0.3011,
+      "step": 11449
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9599303135888502e-05,
+      "loss": 0.2911,
+      "step": 11450
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.3026,
+      "step": 11451
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9686411149825785e-05,
+      "loss": 0.2728,
+      "step": 11452
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9729965156794425e-05,
+      "loss": 0.2829,
+      "step": 11453
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9773519163763068e-05,
+      "loss": 0.2991,
+      "step": 11454
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9817073170731708e-05,
+      "loss": 0.3041,
+      "step": 11455
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.986062717770035e-05,
+      "loss": 0.2847,
+      "step": 11456
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.990418118466899e-05,
+      "loss": 0.2759,
+      "step": 11457
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.994773519163763e-05,
+      "loss": 0.3001,
+      "step": 11458
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.999128919860627e-05,
+      "loss": 0.2907,
+      "step": 11459
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0034843205574914e-05,
+      "loss": 0.3017,
+      "step": 11460
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0078397212543554e-05,
+      "loss": 0.2741,
+      "step": 11461
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0121951219512197e-05,
+      "loss": 0.2908,
+      "step": 11462
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0165505226480837e-05,
+      "loss": 0.2971,
+      "step": 11463
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0209059233449477e-05,
+      "loss": 0.2848,
+      "step": 11464
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.025261324041812e-05,
+      "loss": 0.296,
+      "step": 11465
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.029616724738676e-05,
+      "loss": 0.2912,
+      "step": 11466
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.03397212543554e-05,
+      "loss": 0.2911,
+      "step": 11467
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.038327526132404e-05,
+      "loss": 0.2901,
+      "step": 11468
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0426829268292683e-05,
+      "loss": 0.2955,
+      "step": 11469
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0470383275261327e-05,
+      "loss": 0.3011,
+      "step": 11470
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0513937282229967e-05,
+      "loss": 0.2951,
+      "step": 11471
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0557491289198607e-05,
+      "loss": 0.2952,
+      "step": 11472
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.060104529616725e-05,
+      "loss": 0.2903,
+      "step": 11473
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.064459930313589e-05,
+      "loss": 0.2918,
+      "step": 11474
+    },
+    {
+      "epoch": 2.0,
+      "step": 11474,
+      "total_flos": 1.258760766830999e+20,
+      "train_loss": 0.012149228684698973,
+      "train_runtime": 14546.155,
+      "train_samples_per_second": 201.94,
+      "train_steps_per_second": 0.789
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 11474,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "total_flos": 1.258760766830999e+20,
+  "trial_name": null,
+  "trial_params": null
+}