diff --git "a/.ipynb_checkpoints/trainer_state-checkpoint.json" "b/.ipynb_checkpoints/trainer_state-checkpoint.json"
new file mode 100644--- /dev/null
+++ "b/.ipynb_checkpoints/trainer_state-checkpoint.json"
@@ -0,0 +1,36016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.5717092337917484,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.9856793573174994e-09,
+      "loss": 1.4395,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3971358714634999e-08,
+      "loss": 1.378,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.09570380719525e-08,
+      "loss": 1.587,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7942717429269998e-08,
+      "loss": 1.477,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.49283967865875e-08,
+      "loss": 1.5244,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1914076143905e-08,
+      "loss": 1.3167,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8899755501222494e-08,
+      "loss": 1.3262,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.5885434858539996e-08,
+      "loss": 1.4078,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.28711142158575e-08,
+      "loss": 1.1713,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.9856793573175e-08,
+      "loss": 1.4081,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.684247293049249e-08,
+      "loss": 1.5938,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.382815228781e-08,
+      "loss": 1.5224,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.08138316451275e-08,
+      "loss": 1.3292,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.779951100244499e-08,
+      "loss": 1.5055,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.047851903597625e-07,
+      "loss": 1.4238,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1177086971707999e-07,
+      "loss": 1.7046,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1875654907439749e-07,
+      "loss": 1.2096,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25742228431715e-07,
+      "loss": 1.5781,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.327279077890325e-07,
+      "loss": 1.5714,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3971358714635e-07,
+      "loss": 1.4642,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.466992665036675e-07,
+      "loss": 1.4648,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5368494586098498e-07,
+      "loss": 1.4606,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.606706252183025e-07,
+      "loss": 1.2654,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6765630457562e-07,
+      "loss": 1.1986,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.746419839329375e-07,
+      "loss": 1.5095,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.81627663290255e-07,
+      "loss": 1.2356,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8861334264757248e-07,
+      "loss": 1.3246,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9559902200488998e-07,
+      "loss": 1.5105,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.025847013622075e-07,
+      "loss": 1.3697,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.09570380719525e-07,
+      "loss": 1.5698,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.165560600768425e-07,
+      "loss": 1.3552,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2354173943415998e-07,
+      "loss": 1.5244,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3052741879147748e-07,
+      "loss": 1.3645,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3751309814879497e-07,
+      "loss": 1.4106,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.444987775061125e-07,
+      "loss": 1.4133,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5148445686343e-07,
+      "loss": 1.6173,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.584701362207475e-07,
+      "loss": 1.5613,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.65455815578065e-07,
+      "loss": 1.3876,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.724414949353825e-07,
+      "loss": 1.5248,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.794271742927e-07,
+      "loss": 1.2577,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.864128536500175e-07,
+      "loss": 1.4868,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.93398533007335e-07,
+      "loss": 1.3306,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.003842123646525e-07,
+      "loss": 1.4708,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0736989172196997e-07,
+      "loss": 1.5021,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1435557107928746e-07,
+      "loss": 1.3023,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.21341250436605e-07,
+      "loss": 1.3778,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2832692979392245e-07,
+      "loss": 1.3115,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3531260915124e-07,
+      "loss": 1.3999,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4229828850855744e-07,
+      "loss": 1.4823,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.49283967865875e-07,
+      "loss": 1.4549,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.562696472231925e-07,
+      "loss": 1.3512,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6325532658051e-07,
+      "loss": 1.3154,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7024100593782747e-07,
+      "loss": 1.6101,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7722668529514497e-07,
+      "loss": 1.3029,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8421236465246246e-07,
+      "loss": 1.2815,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9119804400977996e-07,
+      "loss": 1.3324,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.981837233670975e-07,
+      "loss": 1.5183,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.05169402724415e-07,
+      "loss": 1.3837,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.121550820817325e-07,
+      "loss": 1.4974,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1914076143905e-07,
+      "loss": 1.5105,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.261264407963675e-07,
+      "loss": 1.3823,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.33112120153685e-07,
+      "loss": 1.616,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4009779951100247e-07,
+      "loss": 1.5144,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4708347886831996e-07,
+      "loss": 1.5093,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.540691582256375e-07,
+      "loss": 1.2671,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6105483758295495e-07,
+      "loss": 1.3624,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.680405169402725e-07,
+      "loss": 1.484,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7502619629758994e-07,
+      "loss": 1.5003,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.820118756549074e-07,
+      "loss": 1.4087,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88997555012225e-07,
+      "loss": 1.4172,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.959832343695424e-07,
+      "loss": 1.4386,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0296891372686e-07,
+      "loss": 1.2307,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.099545930841775e-07,
+      "loss": 1.3619,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.16940272441495e-07,
+      "loss": 1.3267,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.239259517988125e-07,
+      "loss": 1.3124,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3091163115613e-07,
+      "loss": 1.4446,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.378973105134475e-07,
+      "loss": 1.3285,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.44882989870765e-07,
+      "loss": 1.5004,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.518686692280825e-07,
+      "loss": 1.3193,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.588543485854e-07,
+      "loss": 1.3796,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.658400279427175e-07,
+      "loss": 1.3663,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.72825707300035e-07,
+      "loss": 1.5603,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.798113866573525e-07,
+      "loss": 1.4815,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8679706601467e-07,
+      "loss": 1.3132,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.937827453719875e-07,
+      "loss": 1.3939,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.00768424729305e-07,
+      "loss": 1.5233,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.077541040866225e-07,
+      "loss": 1.4247,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.147397834439399e-07,
+      "loss": 1.3964,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.217254628012575e-07,
+      "loss": 1.4369,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.287111421585749e-07,
+      "loss": 1.4345,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.356968215158925e-07,
+      "loss": 1.274,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4268250087321e-07,
+      "loss": 1.4396,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.496681802305274e-07,
+      "loss": 1.4857,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.566538595878449e-07,
+      "loss": 1.4514,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.636395389451625e-07,
+      "loss": 1.3358,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.7062521830248e-07,
+      "loss": 1.4524,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.776108976597975e-07,
+      "loss": 1.2834,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.845965770171149e-07,
+      "loss": 1.5804,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.915822563744325e-07,
+      "loss": 1.3692,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9856793573175e-07,
+      "loss": 1.2872,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.055536150890676e-07,
+      "loss": 1.4223,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.12539294446385e-07,
+      "loss": 1.2648,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.195249738037025e-07,
+      "loss": 1.4832,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2651065316102e-07,
+      "loss": 1.4123,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.334963325183376e-07,
+      "loss": 1.2506,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.404820118756549e-07,
+      "loss": 1.3025,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.474676912329724e-07,
+      "loss": 1.6453,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.544533705902899e-07,
+      "loss": 1.3191,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.614390499476075e-07,
+      "loss": 1.2377,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.684247293049249e-07,
+      "loss": 1.3758,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.754104086622424e-07,
+      "loss": 1.2395,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.823960880195599e-07,
+      "loss": 1.3322,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.893817673768775e-07,
+      "loss": 1.4124,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.96367446734195e-07,
+      "loss": 1.6356,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.033531260915124e-07,
+      "loss": 1.4947,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.1033880544883e-07,
+      "loss": 1.4247,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.173244848061475e-07,
+      "loss": 1.4601,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.24310164163465e-07,
+      "loss": 1.4397,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.312958435207824e-07,
+      "loss": 1.1851,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.382815228781e-07,
+      "loss": 1.4321,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.452672022354175e-07,
+      "loss": 1.2376,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.52252881592735e-07,
+      "loss": 1.1986,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.592385609500524e-07,
+      "loss": 1.3787,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.6622424030737e-07,
+      "loss": 1.3585,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.732099196646874e-07,
+      "loss": 1.2539,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.801955990220049e-07,
+      "loss": 1.6212,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.871812783793225e-07,
+      "loss": 1.398,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.941669577366399e-07,
+      "loss": 1.3806,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.011526370939574e-07,
+      "loss": 1.4171,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.08138316451275e-07,
+      "loss": 1.2667,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.151239958085925e-07,
+      "loss": 1.2492,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.221096751659099e-07,
+      "loss": 1.1939,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.290953545232274e-07,
+      "loss": 1.5278,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.36081033880545e-07,
+      "loss": 1.4402,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.430667132378625e-07,
+      "loss": 1.4659,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.500523925951799e-07,
+      "loss": 1.2736,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.570380719524975e-07,
+      "loss": 1.3346,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.640237513098149e-07,
+      "loss": 1.3891,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.710094306671325e-07,
+      "loss": 1.3055,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.7799511002445e-07,
+      "loss": 1.3461,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.849807893817675e-07,
+      "loss": 1.6012,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.919664687390849e-07,
+      "loss": 1.266,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989521480964025e-07,
+      "loss": 1.2874,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.00593782745372e-06,
+      "loss": 1.2581,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0129235068110374e-06,
+      "loss": 1.2244,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.019909186168355e-06,
+      "loss": 1.5638,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0268948655256724e-06,
+      "loss": 1.4074,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.03388054488299e-06,
+      "loss": 1.3958,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0408662242403074e-06,
+      "loss": 1.2429,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.047851903597625e-06,
+      "loss": 1.4675,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0548375829549424e-06,
+      "loss": 1.3431,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.06182326231226e-06,
+      "loss": 1.5155,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0688089416695776e-06,
+      "loss": 1.252,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.075794621026895e-06,
+      "loss": 1.3245,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0827803003842124e-06,
+      "loss": 1.4067,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.08976597974153e-06,
+      "loss": 1.3396,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0967516590988476e-06,
+      "loss": 1.2332,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.103737338456165e-06,
+      "loss": 1.2938,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1107230178134824e-06,
+      "loss": 1.3361,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1177086971708e-06,
+      "loss": 1.3835,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1246943765281176e-06,
+      "loss": 1.3198,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.131680055885435e-06,
+      "loss": 1.3486,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1386657352427523e-06,
+      "loss": 1.41,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.14565141460007e-06,
+      "loss": 1.2313,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1526370939573875e-06,
+      "loss": 1.3744,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.159622773314705e-06,
+      "loss": 1.3904,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1666084526720223e-06,
+      "loss": 1.4053,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.17359413202934e-06,
+      "loss": 1.2505,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1805798113866575e-06,
+      "loss": 1.4309,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.187565490743975e-06,
+      "loss": 1.4312,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1945511701012923e-06,
+      "loss": 1.405,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.20153684945861e-06,
+      "loss": 1.2493,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2085225288159275e-06,
+      "loss": 1.2524,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.215508208173245e-06,
+      "loss": 1.3255,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2224938875305625e-06,
+      "loss": 1.2766,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2294795668878799e-06,
+      "loss": 1.3791,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2364652462451975e-06,
+      "loss": 1.297,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.243450925602515e-06,
+      "loss": 1.2294,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2504366049598325e-06,
+      "loss": 1.4998,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2574222843171499e-06,
+      "loss": 1.2949,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2644079636744675e-06,
+      "loss": 1.1852,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.271393643031785e-06,
+      "loss": 1.2807,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2783793223891024e-06,
+      "loss": 1.6925,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.28536500174642e-06,
+      "loss": 1.3192,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2923506811037376e-06,
+      "loss": 1.3157,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2993363604610548e-06,
+      "loss": 1.5452,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3063220398183724e-06,
+      "loss": 1.3784,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3133077191756898e-06,
+      "loss": 1.3431,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3202933985330074e-06,
+      "loss": 1.3853,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.327279077890325e-06,
+      "loss": 1.015,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3342647572476424e-06,
+      "loss": 1.2661,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.34125043660496e-06,
+      "loss": 1.3743,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3482361159622776e-06,
+      "loss": 1.3612,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.355221795319595e-06,
+      "loss": 1.4087,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3622074746769124e-06,
+      "loss": 1.1743,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3691931540342298e-06,
+      "loss": 1.4002,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3761788333915474e-06,
+      "loss": 1.2383,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.383164512748865e-06,
+      "loss": 1.2123,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3901501921061826e-06,
+      "loss": 1.4371,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3971358714635e-06,
+      "loss": 1.3212,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4041215508208176e-06,
+      "loss": 1.4153,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4111072301781352e-06,
+      "loss": 1.2589,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4180929095354523e-06,
+      "loss": 1.2864,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.42507858889277e-06,
+      "loss": 1.3421,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4320642682500873e-06,
+      "loss": 1.461,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.439049947607405e-06,
+      "loss": 1.1337,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4460356269647225e-06,
+      "loss": 1.3503,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.45302130632204e-06,
+      "loss": 1.1948,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4600069856793575e-06,
+      "loss": 1.2287,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4669926650366751e-06,
+      "loss": 1.3381,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4739783443939925e-06,
+      "loss": 1.2314,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4809640237513099e-06,
+      "loss": 1.5321,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4879497031086273e-06,
+      "loss": 1.4746,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4949353824659449e-06,
+      "loss": 1.4683,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5019210618232625e-06,
+      "loss": 1.4426,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5089067411805799e-06,
+      "loss": 1.2754,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5158924205378975e-06,
+      "loss": 1.3198,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.522878099895215e-06,
+      "loss": 1.2375,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5298637792525325e-06,
+      "loss": 1.3457,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5368494586098498e-06,
+      "loss": 1.331,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5438351379671672e-06,
+      "loss": 1.2428,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5508208173244848e-06,
+      "loss": 1.2373,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5578064966818024e-06,
+      "loss": 1.3366,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5647921760391198e-06,
+      "loss": 1.4088,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5717778553964374e-06,
+      "loss": 1.244,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.578763534753755e-06,
+      "loss": 1.2542,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5857492141110726e-06,
+      "loss": 1.3669,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.59273489346839e-06,
+      "loss": 1.3214,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5997205728257074e-06,
+      "loss": 1.3409,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6067062521830248e-06,
+      "loss": 1.2829,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6136919315403424e-06,
+      "loss": 1.315,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.62067761089766e-06,
+      "loss": 1.5036,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6276632902549774e-06,
+      "loss": 1.3839,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.634648969612295e-06,
+      "loss": 1.3303,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6416346489696126e-06,
+      "loss": 1.4581,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.64862032832693e-06,
+      "loss": 1.1864,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6556060076842476e-06,
+      "loss": 1.3906,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6625916870415647e-06,
+      "loss": 1.1208,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6695773663988823e-06,
+      "loss": 1.388,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6765630457562e-06,
+      "loss": 1.2578,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6835487251135173e-06,
+      "loss": 1.5461,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.690534404470835e-06,
+      "loss": 1.323,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6975200838281525e-06,
+      "loss": 1.2719,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.70450576318547e-06,
+      "loss": 1.4094,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7114914425427875e-06,
+      "loss": 1.2282,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7184771219001047e-06,
+      "loss": 1.2858,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7254628012574223e-06,
+      "loss": 1.3895,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.73244848061474e-06,
+      "loss": 1.4129,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7394341599720573e-06,
+      "loss": 1.4079,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7464198393293749e-06,
+      "loss": 1.2564,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7534055186866925e-06,
+      "loss": 1.2452,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7603911980440099e-06,
+      "loss": 1.5228,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7673768774013275e-06,
+      "loss": 1.5038,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.774362556758645e-06,
+      "loss": 1.3066,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7813482361159623e-06,
+      "loss": 1.281,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7883339154732799e-06,
+      "loss": 1.3382,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7953195948305975e-06,
+      "loss": 1.4065,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8023052741879148e-06,
+      "loss": 1.1652,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8092909535452324e-06,
+      "loss": 1.1342,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.81627663290255e-06,
+      "loss": 1.5352,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8232623122598674e-06,
+      "loss": 1.2424,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.830247991617185e-06,
+      "loss": 1.2978,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8372336709745026e-06,
+      "loss": 1.3306,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8442193503318198e-06,
+      "loss": 1.3668,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8512050296891374e-06,
+      "loss": 1.3379,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8581907090464548e-06,
+      "loss": 1.3002,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8651763884037724e-06,
+      "loss": 1.376,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.87216206776109e-06,
+      "loss": 1.4188,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8791477471184074e-06,
+      "loss": 1.2456,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.886133426475725e-06,
+      "loss": 1.331,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8931191058330426e-06,
+      "loss": 1.4588,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9001047851903598e-06,
+      "loss": 1.5054,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9070904645476774e-06,
+      "loss": 1.3551,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.914076143904995e-06,
+      "loss": 1.2043,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9210618232623124e-06,
+      "loss": 1.2367,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9280475026196297e-06,
+      "loss": 1.33,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9350331819769476e-06,
+      "loss": 1.2815,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.942018861334265e-06,
+      "loss": 1.1488,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9490045406915823e-06,
+      "loss": 1.2798,
+      "step": 279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9559902200489e-06,
+      "loss": 1.415,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.962975899406217e-06,
+      "loss": 1.3643,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.969961578763535e-06,
+      "loss": 1.2221,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769472581208523e-06,
+      "loss": 1.1345,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9839329374781697e-06,
+      "loss": 1.2594,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909186168354875e-06,
+      "loss": 1.4332,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997904296192805e-06,
+      "loss": 1.2625,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0048899755501223e-06,
+      "loss": 1.1755,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.01187565490744e-06,
+      "loss": 1.3266,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.018861334264757e-06,
+      "loss": 1.3081,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.025847013622075e-06,
+      "loss": 1.256,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0328326929793923e-06,
+      "loss": 1.3574,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.03981837233671e-06,
+      "loss": 1.303,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0468040516940275e-06,
+      "loss": 1.4058,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.053789731051345e-06,
+      "loss": 1.2674,
+      "step": 294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0607754104086627e-06,
+      "loss": 1.4137,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.06776108976598e-06,
+      "loss": 1.3002,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0747467691232974e-06,
+      "loss": 1.5734,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.081732448480615e-06,
+      "loss": 1.3079,
+      "step": 298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0887181278379322e-06,
+      "loss": 1.3744,
+      "step": 299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.09570380719525e-06,
+      "loss": 1.3763,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1026894865525674e-06,
+      "loss": 1.2637,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.109675165909885e-06,
+      "loss": 1.4016,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1166608452672026e-06,
+      "loss": 1.458,
+      "step": 303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.12364652462452e-06,
+      "loss": 1.3302,
+      "step": 304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1306322039818374e-06,
+      "loss": 1.2406,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1376178833391552e-06,
+      "loss": 1.4229,
+      "step": 306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.144603562696472e-06,
+      "loss": 1.2181,
+      "step": 307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.15158924205379e-06,
+      "loss": 1.2442,
+      "step": 308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1585749214111074e-06,
+      "loss": 1.3766,
+      "step": 309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1655606007684248e-06,
+      "loss": 1.2702,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1725462801257426e-06,
+      "loss": 1.2899,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.17953195948306e-06,
+      "loss": 1.1227,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865176388403774e-06,
+      "loss": 1.3358,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.193503318197695e-06,
+      "loss": 1.3301,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.200488997555012e-06,
+      "loss": 1.3028,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.20747467691233e-06,
+      "loss": 1.2905,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2144603562696473e-06,
+      "loss": 1.3593,
+      "step": 317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2214460356269647e-06,
+      "loss": 1.2956,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2284317149842825e-06,
+      "loss": 1.4055,
+      "step": 319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2354173943416e-06,
+      "loss": 1.2658,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2424030736989173e-06,
+      "loss": 1.5361,
+      "step": 321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.249388753056235e-06,
+      "loss": 1.3089,
+      "step": 322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2563744324135525e-06,
+      "loss": 1.4515,
+      "step": 323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.26336011177087e-06,
+      "loss": 1.3313,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2703457911281873e-06,
+      "loss": 1.2721,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2773314704855047e-06,
+      "loss": 1.2477,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2843171498428225e-06,
+      "loss": 1.1197,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.29130282920014e-06,
+      "loss": 1.3109,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2982885085574573e-06,
+      "loss": 1.2679,
+      "step": 329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.305274187914775e-06,
+      "loss": 1.278,
+      "step": 330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3122598672720925e-06,
+      "loss": 1.4146,
+      "step": 331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.31924554662941e-06,
+      "loss": 1.2603,
+      "step": 332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3262312259867273e-06,
+      "loss": 1.187,
+      "step": 333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3332169053440446e-06,
+      "loss": 1.278,
+      "step": 334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3402025847013625e-06,
+      "loss": 1.1635,
+      "step": 335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.34718826405868e-06,
+      "loss": 1.3081,
+      "step": 336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3541739434159972e-06,
+      "loss": 1.1724,
+      "step": 337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.361159622773315e-06,
+      "loss": 1.1738,
+      "step": 338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3681453021306324e-06,
+      "loss": 1.1025,
+      "step": 339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.37513098148795e-06,
+      "loss": 1.2898,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.382116660845267e-06,
+      "loss": 1.374,
+      "step": 341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3891023402025846e-06,
+      "loss": 1.3478,
+      "step": 342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3960880195599024e-06,
+      "loss": 1.2961,
+      "step": 343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.40307369891722e-06,
+      "loss": 1.2492,
+      "step": 344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4100593782745376e-06,
+      "loss": 1.2347,
+      "step": 345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.417045057631855e-06,
+      "loss": 1.1512,
+      "step": 346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4240307369891724e-06,
+      "loss": 1.2998,
+      "step": 347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.43101641634649e-06,
+      "loss": 1.266,
+      "step": 348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4380020957038076e-06,
+      "loss": 1.2684,
+      "step": 349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.444987775061125e-06,
+      "loss": 1.2408,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4519734544184424e-06,
+      "loss": 1.3548,
+      "step": 351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4589591337757598e-06,
+      "loss": 1.3596,
+      "step": 352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4659448131330776e-06,
+      "loss": 1.516,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.472930492490395e-06,
+      "loss": 1.4587,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4799161718477123e-06,
+      "loss": 1.2567,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.48690185120503e-06,
+      "loss": 1.5524,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4938875305623475e-06,
+      "loss": 1.1951,
+      "step": 357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.500873209919665e-06,
+      "loss": 1.1797,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5078588892769823e-06,
+      "loss": 1.597,
+      "step": 359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5148445686342997e-06,
+      "loss": 1.1578,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5218302479916175e-06,
+      "loss": 1.1551,
+      "step": 361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.528815927348935e-06,
+      "loss": 1.2181,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5358016067062523e-06,
+      "loss": 1.1526,
+      "step": 363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.54278728606357e-06,
+      "loss": 1.5563,
+      "step": 364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5497729654208875e-06,
+      "loss": 1.204,
+      "step": 365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.556758644778205e-06,
+      "loss": 1.3086,
+      "step": 366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5637443241355227e-06,
+      "loss": 1.3867,
+      "step": 367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.57073000349284e-06,
+      "loss": 1.2053,
+      "step": 368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5777156828501575e-06,
+      "loss": 1.3874,
+      "step": 369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5847013622074753e-06,
+      "loss": 1.3688,
+      "step": 370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5916870415647923e-06,
+      "loss": 1.3529,
+      "step": 371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5986727209221096e-06,
+      "loss": 1.197,
+      "step": 372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.605658400279427e-06,
+      "loss": 1.0984,
+      "step": 373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.612644079636745e-06,
+      "loss": 1.2969,
+      "step": 374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6196297589940622e-06,
+      "loss": 1.2838,
+      "step": 375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6266154383513796e-06,
+      "loss": 1.2289,
+      "step": 376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6336011177086974e-06,
+      "loss": 1.2428,
+      "step": 377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.640586797066015e-06,
+      "loss": 1.2951,
+      "step": 378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.647572476423332e-06,
+      "loss": 1.3031,
+      "step": 379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.65455815578065e-06,
+      "loss": 1.3135,
+      "step": 380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6615438351379674e-06,
+      "loss": 1.1794,
+      "step": 381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.668529514495285e-06,
+      "loss": 1.1801,
+      "step": 382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6755151938526026e-06,
+      "loss": 1.2873,
+      "step": 383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.68250087320992e-06,
+      "loss": 1.3784,
+      "step": 384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6894865525672374e-06,
+      "loss": 1.1401,
+      "step": 385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.696472231924555e-06,
+      "loss": 1.088,
+      "step": 386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7034579112818726e-06,
+      "loss": 1.2689,
+      "step": 387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.71044359063919e-06,
+      "loss": 1.3174,
+      "step": 388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.717429269996507e-06,
+      "loss": 1.4157,
+      "step": 389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7244149493538248e-06,
+      "loss": 1.4755,
+      "step": 390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.731400628711142e-06,
+      "loss": 1.3091,
+      "step": 391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7383863080684595e-06,
+      "loss": 1.2798,
+      "step": 392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7453719874257773e-06,
+      "loss": 1.2112,
+      "step": 393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7523576667830947e-06,
+      "loss": 1.292,
+      "step": 394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.759343346140412e-06,
+      "loss": 1.378,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.76632902549773e-06,
+      "loss": 1.3564,
+      "step": 396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7733147048550473e-06,
+      "loss": 1.3591,
+      "step": 397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.780300384212365e-06,
+      "loss": 1.2549,
+      "step": 398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7872860635696825e-06,
+      "loss": 1.3777,
+      "step": 399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.794271742927e-06,
+      "loss": 1.2019,
+      "step": 400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8012574222843177e-06,
+      "loss": 1.128,
+      "step": 401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.808243101641635e-06,
+      "loss": 1.3522,
+      "step": 402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8152287809989525e-06,
+      "loss": 1.5424,
+      "step": 403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8222144603562703e-06,
+      "loss": 1.2855,
+      "step": 404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8292001397135877e-06,
+      "loss": 1.3095,
+      "step": 405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8361858190709047e-06,
+      "loss": 1.2101,
+      "step": 406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.843171498428222e-06,
+      "loss": 1.4208,
+      "step": 407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.85015717778554e-06,
+      "loss": 1.2215,
+      "step": 408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.2446,
+      "step": 409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8641285365001746e-06,
+      "loss": 1.4605,
+      "step": 410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8711142158574925e-06,
+      "loss": 1.3091,
+      "step": 411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.87809989521481e-06,
+      "loss": 1.3434,
+      "step": 412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8850855745721272e-06,
+      "loss": 1.3562,
+      "step": 413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.892071253929445e-06,
+      "loss": 1.2633,
+      "step": 414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8990569332867624e-06,
+      "loss": 1.4233,
+      "step": 415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.90604261264408e-06,
+      "loss": 1.1229,
+      "step": 416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9130282920013976e-06,
+      "loss": 1.3928,
+      "step": 417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.920013971358715e-06,
+      "loss": 1.5127,
+      "step": 418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9269996507160324e-06,
+      "loss": 1.1098,
+      "step": 419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9339853300733502e-06,
+      "loss": 1.3364,
+      "step": 420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9409710094306676e-06,
+      "loss": 1.4362,
+      "step": 421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.947956688787985e-06,
+      "loss": 1.2468,
+      "step": 422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.954942368145302e-06,
+      "loss": 1.2906,
+      "step": 423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9619280475026198e-06,
+      "loss": 1.3927,
+      "step": 424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.968913726859937e-06,
+      "loss": 1.0892,
+      "step": 425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9758994062172546e-06,
+      "loss": 1.3447,
+      "step": 426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9828850855745724e-06,
+      "loss": 1.2747,
+      "step": 427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9898707649318898e-06,
+      "loss": 1.3381,
+      "step": 428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.996856444289207e-06,
+      "loss": 1.3138,
+      "step": 429
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.003842123646525e-06,
+      "loss": 1.1448,
+      "step": 430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0108278030038423e-06,
+      "loss": 1.1203,
+      "step": 431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0178134823611597e-06,
+      "loss": 1.1146,
+      "step": 432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0247991617184775e-06,
+      "loss": 1.1614,
+      "step": 433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.031784841075795e-06,
+      "loss": 1.1545,
+      "step": 434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0387705204331123e-06,
+      "loss": 1.1324,
+      "step": 435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.04575619979043e-06,
+      "loss": 1.3561,
+      "step": 436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0527418791477475e-06,
+      "loss": 1.1336,
+      "step": 437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.059727558505065e-06,
+      "loss": 1.3058,
+      "step": 438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0667132378623827e-06,
+      "loss": 1.305,
+      "step": 439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0736989172196997e-06,
+      "loss": 1.2135,
+      "step": 440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.080684596577017e-06,
+      "loss": 1.236,
+      "step": 441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0876702759343345e-06,
+      "loss": 1.2182,
+      "step": 442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0946559552916523e-06,
+      "loss": 1.3429,
+      "step": 443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1016416346489697e-06,
+      "loss": 1.3023,
+      "step": 444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.108627314006287e-06,
+      "loss": 1.1586,
+      "step": 445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.115612993363605e-06,
+      "loss": 1.3803,
+      "step": 446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1225986727209223e-06,
+      "loss": 1.2364,
+      "step": 447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1295843520782396e-06,
+      "loss": 1.3148,
+      "step": 448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1365700314355575e-06,
+      "loss": 1.1631,
+      "step": 449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.143555710792875e-06,
+      "loss": 1.1828,
+      "step": 450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1505413901501922e-06,
+      "loss": 1.333,
+      "step": 451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.15752706950751e-06,
+      "loss": 1.3426,
+      "step": 452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1645127488648274e-06,
+      "loss": 1.5088,
+      "step": 453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1714984282221452e-06,
+      "loss": 1.1188,
+      "step": 454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1784841075794626e-06,
+      "loss": 1.5372,
+      "step": 455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.18546978693678e-06,
+      "loss": 1.3209,
+      "step": 456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.192455466294098e-06,
+      "loss": 1.0497,
+      "step": 457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.199441145651415e-06,
+      "loss": 1.1761,
+      "step": 458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.206426825008732e-06,
+      "loss": 1.4854,
+      "step": 459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2134125043660496e-06,
+      "loss": 1.4629,
+      "step": 460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2203981837233674e-06,
+      "loss": 1.1879,
+      "step": 461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2273838630806848e-06,
+      "loss": 1.1256,
+      "step": 462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.234369542438002e-06,
+      "loss": 1.3619,
+      "step": 463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.24135522179532e-06,
+      "loss": 1.2209,
+      "step": 464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2483409011526374e-06,
+      "loss": 1.2509,
+      "step": 465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2553265805099548e-06,
+      "loss": 1.2636,
+      "step": 466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2623122598672726e-06,
+      "loss": 1.321,
+      "step": 467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.26929793922459e-06,
+      "loss": 1.3575,
+      "step": 468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2762836185819073e-06,
+      "loss": 1.3709,
+      "step": 469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.283269297939225e-06,
+      "loss": 1.338,
+      "step": 470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2902549772965425e-06,
+      "loss": 1.1507,
+      "step": 471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.29724065665386e-06,
+      "loss": 1.306,
+      "step": 472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3042263360111777e-06,
+      "loss": 1.4932,
+      "step": 473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.311212015368495e-06,
+      "loss": 1.1615,
+      "step": 474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.318197694725812e-06,
+      "loss": 1.4089,
+      "step": 475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3251833740831295e-06,
+      "loss": 1.1964,
+      "step": 476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3321690534404473e-06,
+      "loss": 1.3663,
+      "step": 477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3391547327977647e-06,
+      "loss": 1.348,
+      "step": 478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.346140412155082e-06,
+      "loss": 1.0565,
+      "step": 479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3531260915124e-06,
+      "loss": 1.2124,
+      "step": 480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3601117708697173e-06,
+      "loss": 1.1465,
+      "step": 481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3670974502270347e-06,
+      "loss": 1.3834,
+      "step": 482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3740831295843525e-06,
+      "loss": 1.3424,
+      "step": 483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.38106880894167e-06,
+      "loss": 1.1192,
+      "step": 484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3880544882989873e-06,
+      "loss": 1.2585,
+      "step": 485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.395040167656305e-06,
+      "loss": 1.2975,
+      "step": 486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4020258470136225e-06,
+      "loss": 1.3256,
+      "step": 487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.40901152637094e-06,
+      "loss": 1.1669,
+      "step": 488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4159972057282577e-06,
+      "loss": 1.1371,
+      "step": 489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.422982885085575e-06,
+      "loss": 1.1957,
+      "step": 490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4299685644428924e-06,
+      "loss": 1.1701,
+      "step": 491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4369542438002094e-06,
+      "loss": 1.3177,
+      "step": 492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4439399231575272e-06,
+      "loss": 1.1904,
+      "step": 493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4509256025148446e-06,
+      "loss": 1.2258,
+      "step": 494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.457911281872162e-06,
+      "loss": 1.4526,
+      "step": 495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.46489696122948e-06,
+      "loss": 1.1657,
+      "step": 496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.471882640586797e-06,
+      "loss": 1.2294,
+      "step": 497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4788683199441146e-06,
+      "loss": 1.2635,
+      "step": 498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4858539993014324e-06,
+      "loss": 1.2771,
+      "step": 499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4928396786587498e-06,
+      "loss": 1.2341,
+      "step": 500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.499825358016067e-06,
+      "loss": 1.2761,
+      "step": 501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.506811037373385e-06,
+      "loss": 1.0662,
+      "step": 502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5137967167307024e-06,
+      "loss": 1.1611,
+      "step": 503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5207823960880198e-06,
+      "loss": 1.2035,
+      "step": 504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5277680754453376e-06,
+      "loss": 1.2773,
+      "step": 505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.534753754802655e-06,
+      "loss": 1.0509,
+      "step": 506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5417394341599728e-06,
+      "loss": 1.2165,
+      "step": 507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.54872511351729e-06,
+      "loss": 1.2864,
+      "step": 508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.555710792874607e-06,
+      "loss": 1.198,
+      "step": 509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5626964722319245e-06,
+      "loss": 1.1488,
+      "step": 510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5696821515892423e-06,
+      "loss": 1.1574,
+      "step": 511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5766678309465597e-06,
+      "loss": 1.1598,
+      "step": 512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.583653510303877e-06,
+      "loss": 1.1942,
+      "step": 513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.590639189661195e-06,
+      "loss": 1.2725,
+      "step": 514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5976248690185123e-06,
+      "loss": 1.3625,
+      "step": 515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6046105483758297e-06,
+      "loss": 1.2417,
+      "step": 516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6115962277331475e-06,
+      "loss": 1.2305,
+      "step": 517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.618581907090465e-06,
+      "loss": 1.2266,
+      "step": 518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6255675864477823e-06,
+      "loss": 1.2846,
+      "step": 519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6325532658051e-06,
+      "loss": 1.1393,
+      "step": 520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6395389451624175e-06,
+      "loss": 1.1757,
+      "step": 521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.646524624519735e-06,
+      "loss": 1.213,
+      "step": 522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6535103038770527e-06,
+      "loss": 1.1595,
+      "step": 523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.66049598323437e-06,
+      "loss": 1.2596,
+      "step": 524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6674816625916875e-06,
+      "loss": 1.2848,
+      "step": 525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6744673419490053e-06,
+      "loss": 1.1491,
+      "step": 526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6814530213063222e-06,
+      "loss": 1.2241,
+      "step": 527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6884387006636396e-06,
+      "loss": 1.229,
+      "step": 528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.695424380020957e-06,
+      "loss": 1.3531,
+      "step": 529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.702410059378275e-06,
+      "loss": 1.2878,
+      "step": 530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7093957387355922e-06,
+      "loss": 1.1776,
+      "step": 531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7163814180929096e-06,
+      "loss": 1.2563,
+      "step": 532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7233670974502274e-06,
+      "loss": 0.9994,
+      "step": 533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.730352776807545e-06,
+      "loss": 1.1566,
+      "step": 534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.737338456164862e-06,
+      "loss": 1.132,
+      "step": 535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.74432413552218e-06,
+      "loss": 1.2025,
+      "step": 536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7513098148794974e-06,
+      "loss": 1.0815,
+      "step": 537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7582954942368148e-06,
+      "loss": 1.1862,
+      "step": 538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7652811735941326e-06,
+      "loss": 1.3755,
+      "step": 539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.77226685295145e-06,
+      "loss": 1.3642,
+      "step": 540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7792525323087674e-06,
+      "loss": 1.1859,
+      "step": 541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.786238211666085e-06,
+      "loss": 1.5,
+      "step": 542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7932238910234026e-06,
+      "loss": 1.1412,
+      "step": 543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8002095703807195e-06,
+      "loss": 1.4648,
+      "step": 544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.807195249738037e-06,
+      "loss": 1.1977,
+      "step": 545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8141809290953547e-06,
+      "loss": 1.1897,
+      "step": 546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.821166608452672e-06,
+      "loss": 1.3648,
+      "step": 547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.82815228780999e-06,
+      "loss": 1.3787,
+      "step": 548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.835137967167307e-06,
+      "loss": 1.2318,
+      "step": 549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.842123646524625e-06,
+      "loss": 1.2707,
+      "step": 550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8491093258819425e-06,
+      "loss": 1.2297,
+      "step": 551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8560950052392595e-06,
+      "loss": 1.2593,
+      "step": 552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863080684596577e-06,
+      "loss": 1.3106,
+      "step": 553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.870066363953895e-06,
+      "loss": 1.1775,
+      "step": 554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.877052043311212e-06,
+      "loss": 1.2461,
+      "step": 555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.88403772266853e-06,
+      "loss": 1.1632,
+      "step": 556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.891023402025848e-06,
+      "loss": 1.226,
+      "step": 557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.898009081383165e-06,
+      "loss": 1.419,
+      "step": 558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9049947607404825e-06,
+      "loss": 1.1306,
+      "step": 559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9119804400978e-06,
+      "loss": 1.4094,
+      "step": 560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.918966119455117e-06,
+      "loss": 1.0913,
+      "step": 561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.925951798812434e-06,
+      "loss": 1.1865,
+      "step": 562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.932937478169752e-06,
+      "loss": 1.0372,
+      "step": 563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.93992315752707e-06,
+      "loss": 1.2802,
+      "step": 564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.946908836884387e-06,
+      "loss": 1.1082,
+      "step": 565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.953894516241705e-06,
+      "loss": 1.1323,
+      "step": 566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9608801955990224e-06,
+      "loss": 1.2518,
+      "step": 567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.967865874956339e-06,
+      "loss": 1.2726,
+      "step": 568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.974851554313657e-06,
+      "loss": 1.1906,
+      "step": 569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.981837233670975e-06,
+      "loss": 1.1739,
+      "step": 570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.988822913028292e-06,
+      "loss": 1.1111,
+      "step": 571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.99580859238561e-06,
+      "loss": 1.3159,
+      "step": 572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.002794271742928e-06,
+      "loss": 1.2635,
+      "step": 573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.009779951100245e-06,
+      "loss": 1.2474,
+      "step": 574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.016765630457562e-06,
+      "loss": 1.1875,
+      "step": 575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.02375130981488e-06,
+      "loss": 1.4562,
+      "step": 576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.030736989172197e-06,
+      "loss": 1.379,
+      "step": 577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.037722668529514e-06,
+      "loss": 1.217,
+      "step": 578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.044708347886832e-06,
+      "loss": 1.2318,
+      "step": 579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.05169402724415e-06,
+      "loss": 1.4222,
+      "step": 580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.058679706601467e-06,
+      "loss": 1.0874,
+      "step": 581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.0656653859587845e-06,
+      "loss": 1.2626,
+      "step": 582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.072651065316102e-06,
+      "loss": 1.3209,
+      "step": 583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.07963674467342e-06,
+      "loss": 1.25,
+      "step": 584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.086622424030737e-06,
+      "loss": 1.419,
+      "step": 585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.093608103388055e-06,
+      "loss": 1.3351,
+      "step": 586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.100593782745373e-06,
+      "loss": 1.2165,
+      "step": 587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.10757946210269e-06,
+      "loss": 1.21,
+      "step": 588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.1145651414600075e-06,
+      "loss": 1.3108,
+      "step": 589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.121550820817325e-06,
+      "loss": 1.7316,
+      "step": 590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.128536500174642e-06,
+      "loss": 1.0726,
+      "step": 591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.13552217953196e-06,
+      "loss": 1.2311,
+      "step": 592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.142507858889278e-06,
+      "loss": 1.2378,
+      "step": 593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.149493538246595e-06,
+      "loss": 1.2767,
+      "step": 594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.156479217603913e-06,
+      "loss": 1.1814,
+      "step": 595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.16346489696123e-06,
+      "loss": 1.2572,
+      "step": 596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1704505763185475e-06,
+      "loss": 1.2031,
+      "step": 597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1774362556758645e-06,
+      "loss": 1.1836,
+      "step": 598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.184421935033182e-06,
+      "loss": 1.262,
+      "step": 599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1914076143905e-06,
+      "loss": 1.5331,
+      "step": 600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.198393293747817e-06,
+      "loss": 1.2047,
+      "step": 601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.205378973105135e-06,
+      "loss": 1.1074,
+      "step": 602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.212364652462453e-06,
+      "loss": 1.121,
+      "step": 603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.21935033181977e-06,
+      "loss": 1.1868,
+      "step": 604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2263360111770874e-06,
+      "loss": 1.1489,
+      "step": 605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.233321690534405e-06,
+      "loss": 1.2641,
+      "step": 606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.240307369891722e-06,
+      "loss": 1.2483,
+      "step": 607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.24729304924904e-06,
+      "loss": 1.2511,
+      "step": 608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.254278728606358e-06,
+      "loss": 1.3123,
+      "step": 609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.261264407963675e-06,
+      "loss": 1.1757,
+      "step": 610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.268250087320993e-06,
+      "loss": 1.2889,
+      "step": 611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2752357666783104e-06,
+      "loss": 1.288,
+      "step": 612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.282221446035627e-06,
+      "loss": 1.2685,
+      "step": 613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.289207125392944e-06,
+      "loss": 1.0206,
+      "step": 614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.296192804750262e-06,
+      "loss": 1.244,
+      "step": 615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.30317848410758e-06,
+      "loss": 1.1972,
+      "step": 616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.310164163464897e-06,
+      "loss": 1.1357,
+      "step": 617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.317149842822215e-06,
+      "loss": 1.284,
+      "step": 618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.324135522179533e-06,
+      "loss": 1.2751,
+      "step": 619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3311212015368495e-06,
+      "loss": 1.2848,
+      "step": 620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.338106880894167e-06,
+      "loss": 1.3089,
+      "step": 621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.345092560251485e-06,
+      "loss": 1.1811,
+      "step": 622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.352078239608802e-06,
+      "loss": 1.2697,
+      "step": 623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.35906391896612e-06,
+      "loss": 1.1228,
+      "step": 624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.366049598323438e-06,
+      "loss": 1.2088,
+      "step": 625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.373035277680755e-06,
+      "loss": 1.239,
+      "step": 626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3800209570380725e-06,
+      "loss": 1.3455,
+      "step": 627
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.38700663639539e-06,
+      "loss": 1.3332,
+      "step": 628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.393992315752707e-06,
+      "loss": 1.1497,
+      "step": 629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.400977995110024e-06,
+      "loss": 1.1978,
+      "step": 630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.407963674467342e-06,
+      "loss": 1.2721,
+      "step": 631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.41494935382466e-06,
+      "loss": 1.3043,
+      "step": 632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.421935033181977e-06,
+      "loss": 1.2734,
+      "step": 633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.428920712539295e-06,
+      "loss": 1.1762,
+      "step": 634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4359063918966125e-06,
+      "loss": 1.095,
+      "step": 635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4428920712539295e-06,
+      "loss": 1.4542,
+      "step": 636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.449877750611247e-06,
+      "loss": 1.1374,
+      "step": 637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.456863429968565e-06,
+      "loss": 1.0548,
+      "step": 638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.463849109325882e-06,
+      "loss": 0.9901,
+      "step": 639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4708347886832e-06,
+      "loss": 1.2161,
+      "step": 640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.477820468040518e-06,
+      "loss": 1.1694,
+      "step": 641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.484806147397835e-06,
+      "loss": 1.0528,
+      "step": 642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4917918267551524e-06,
+      "loss": 1.3245,
+      "step": 643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.49877750611247e-06,
+      "loss": 1.2465,
+      "step": 644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.505763185469787e-06,
+      "loss": 1.045,
+      "step": 645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.512748864827105e-06,
+      "loss": 1.1136,
+      "step": 646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.519734544184422e-06,
+      "loss": 1.1746,
+      "step": 647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.52672022354174e-06,
+      "loss": 1.1303,
+      "step": 648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.533705902899057e-06,
+      "loss": 1.2037,
+      "step": 649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.540691582256375e-06,
+      "loss": 1.0069,
+      "step": 650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.547677261613692e-06,
+      "loss": 1.3668,
+      "step": 651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.554662940971009e-06,
+      "loss": 1.1766,
+      "step": 652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.561648620328327e-06,
+      "loss": 1.1553,
+      "step": 653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.568634299685645e-06,
+      "loss": 1.3197,
+      "step": 654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.575619979042962e-06,
+      "loss": 1.1426,
+      "step": 655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.58260565840028e-06,
+      "loss": 1.1993,
+      "step": 656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.589591337757598e-06,
+      "loss": 1.2331,
+      "step": 657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5965770171149145e-06,
+      "loss": 1.1468,
+      "step": 658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.603562696472232e-06,
+      "loss": 1.2101,
+      "step": 659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.61054837582955e-06,
+      "loss": 1.0976,
+      "step": 660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.617534055186867e-06,
+      "loss": 1.264,
+      "step": 661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.624519734544185e-06,
+      "loss": 1.2393,
+      "step": 662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.631505413901503e-06,
+      "loss": 1.1253,
+      "step": 663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.63849109325882e-06,
+      "loss": 1.128,
+      "step": 664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.645476772616137e-06,
+      "loss": 1.2846,
+      "step": 665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6524624519734545e-06,
+      "loss": 1.1613,
+      "step": 666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.659448131330772e-06,
+      "loss": 1.0201,
+      "step": 667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.666433810688089e-06,
+      "loss": 1.3337,
+      "step": 668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.673419490045407e-06,
+      "loss": 1.3436,
+      "step": 669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.680405169402725e-06,
+      "loss": 1.1394,
+      "step": 670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.687390848760042e-06,
+      "loss": 1.1382,
+      "step": 671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.69437652811736e-06,
+      "loss": 1.2085,
+      "step": 672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7013622074746775e-06,
+      "loss": 1.0832,
+      "step": 673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7083478868319945e-06,
+      "loss": 1.1669,
+      "step": 674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715333566189312e-06,
+      "loss": 1.232,
+      "step": 675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.72231924554663e-06,
+      "loss": 1.2953,
+      "step": 676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.729304924903947e-06,
+      "loss": 1.1432,
+      "step": 677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.736290604261265e-06,
+      "loss": 1.0934,
+      "step": 678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.743276283618583e-06,
+      "loss": 1.3829,
+      "step": 679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7502619629759e-06,
+      "loss": 1.2318,
+      "step": 680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7572476423332174e-06,
+      "loss": 1.1194,
+      "step": 681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.764233321690534e-06,
+      "loss": 1.172,
+      "step": 682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.771219001047852e-06,
+      "loss": 1.187,
+      "step": 683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.778204680405169e-06,
+      "loss": 1.0818,
+      "step": 684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.785190359762487e-06,
+      "loss": 1.2906,
+      "step": 685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.792176039119805e-06,
+      "loss": 1.3967,
+      "step": 686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.799161718477122e-06,
+      "loss": 1.012,
+      "step": 687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.80614739783444e-06,
+      "loss": 1.4962,
+      "step": 688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.813133077191757e-06,
+      "loss": 1.3215,
+      "step": 689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.820118756549075e-06,
+      "loss": 1.3811,
+      "step": 690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.827104435906392e-06,
+      "loss": 1.1809,
+      "step": 691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.83409011526371e-06,
+      "loss": 1.2042,
+      "step": 692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.841075794621028e-06,
+      "loss": 1.076,
+      "step": 693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.848061473978345e-06,
+      "loss": 1.3096,
+      "step": 694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.855047153335663e-06,
+      "loss": 1.1219,
+      "step": 695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.86203283269298e-06,
+      "loss": 1.1854,
+      "step": 696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.869018512050297e-06,
+      "loss": 1.2434,
+      "step": 697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.876004191407615e-06,
+      "loss": 1.4487,
+      "step": 698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.882989870764932e-06,
+      "loss": 1.2769,
+      "step": 699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.88997555012225e-06,
+      "loss": 1.19,
+      "step": 700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.896961229479567e-06,
+      "loss": 1.1628,
+      "step": 701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.903946908836885e-06,
+      "loss": 1.2091,
+      "step": 702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9109325881942025e-06,
+      "loss": 1.2402,
+      "step": 703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9179182675515195e-06,
+      "loss": 1.0974,
+      "step": 704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.924903946908837e-06,
+      "loss": 1.0749,
+      "step": 705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.931889626266155e-06,
+      "loss": 1.4719,
+      "step": 706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.938875305623472e-06,
+      "loss": 1.1765,
+      "step": 707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.94586098498079e-06,
+      "loss": 1.2264,
+      "step": 708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.952846664338108e-06,
+      "loss": 1.2813,
+      "step": 709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.959832343695425e-06,
+      "loss": 1.2698,
+      "step": 710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9668180230527425e-06,
+      "loss": 1.202,
+      "step": 711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.97380370241006e-06,
+      "loss": 1.3508,
+      "step": 712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.980789381767377e-06,
+      "loss": 1.2025,
+      "step": 713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.987775061124695e-06,
+      "loss": 1.2171,
+      "step": 714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.994760740482013e-06,
+      "loss": 1.0995,
+      "step": 715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.00174641983933e-06,
+      "loss": 1.2508,
+      "step": 716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.008732099196647e-06,
+      "loss": 1.2724,
+      "step": 717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.015717778553965e-06,
+      "loss": 0.9617,
+      "step": 718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.0227034579112825e-06,
+      "loss": 1.169,
+      "step": 719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.029689137268599e-06,
+      "loss": 1.1758,
+      "step": 720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.036674816625917e-06,
+      "loss": 1.1767,
+      "step": 721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.043660495983235e-06,
+      "loss": 1.1896,
+      "step": 722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.050646175340552e-06,
+      "loss": 1.3176,
+      "step": 723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.05763185469787e-06,
+      "loss": 1.0993,
+      "step": 724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.064617534055188e-06,
+      "loss": 1.2964,
+      "step": 725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.071603213412505e-06,
+      "loss": 1.2392,
+      "step": 726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.078588892769822e-06,
+      "loss": 1.0638,
+      "step": 727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.08557457212714e-06,
+      "loss": 1.1725,
+      "step": 728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.092560251484457e-06,
+      "loss": 1.0798,
+      "step": 729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.099545930841775e-06,
+      "loss": 1.1761,
+      "step": 730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.106531610199093e-06,
+      "loss": 1.1268,
+      "step": 731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.11351728955641e-06,
+      "loss": 1.1714,
+      "step": 732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.120502968913728e-06,
+      "loss": 1.2402,
+      "step": 733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.127488648271045e-06,
+      "loss": 1.154,
+      "step": 734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.134474327628362e-06,
+      "loss": 1.2271,
+      "step": 735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.14146000698568e-06,
+      "loss": 0.9843,
+      "step": 736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.148445686342998e-06,
+      "loss": 1.3356,
+      "step": 737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.155431365700315e-06,
+      "loss": 1.1345,
+      "step": 738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.162417045057633e-06,
+      "loss": 1.1855,
+      "step": 739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.169402724414951e-06,
+      "loss": 1.2417,
+      "step": 740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1763884037722675e-06,
+      "loss": 1.1992,
+      "step": 741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1833740831295845e-06,
+      "loss": 1.1606,
+      "step": 742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1903597624869015e-06,
+      "loss": 1.21,
+      "step": 743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.197345441844219e-06,
+      "loss": 1.2498,
+      "step": 744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.204331121201537e-06,
+      "loss": 1.0316,
+      "step": 745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.211316800558854e-06,
+      "loss": 1.2076,
+      "step": 746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.218302479916172e-06,
+      "loss": 1.2454,
+      "step": 747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.22528815927349e-06,
+      "loss": 1.262,
+      "step": 748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.232273838630807e-06,
+      "loss": 1.1437,
+      "step": 749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.2392595179881245e-06,
+      "loss": 1.1901,
+      "step": 750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.246245197345442e-06,
+      "loss": 1.2697,
+      "step": 751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.253230876702759e-06,
+      "loss": 1.162,
+      "step": 752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.260216556060077e-06,
+      "loss": 1.0647,
+      "step": 753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.267202235417395e-06,
+      "loss": 1.3005,
+      "step": 754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.274187914774712e-06,
+      "loss": 1.0271,
+      "step": 755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.28117359413203e-06,
+      "loss": 1.3314,
+      "step": 756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.2881592734893475e-06,
+      "loss": 1.1942,
+      "step": 757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.295144952846664e-06,
+      "loss": 1.1465,
+      "step": 758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.302130632203982e-06,
+      "loss": 1.1743,
+      "step": 759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.3091163115613e-06,
+      "loss": 1.332,
+      "step": 760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.316101990918617e-06,
+      "loss": 1.0209,
+      "step": 761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.323087670275935e-06,
+      "loss": 1.1561,
+      "step": 762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.330073349633253e-06,
+      "loss": 1.1927,
+      "step": 763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.33705902899057e-06,
+      "loss": 1.2162,
+      "step": 764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.344044708347887e-06,
+      "loss": 1.1595,
+      "step": 765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.351030387705205e-06,
+      "loss": 1.1858,
+      "step": 766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.358016067062522e-06,
+      "loss": 1.1638,
+      "step": 767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.36500174641984e-06,
+      "loss": 1.3354,
+      "step": 768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.371987425777158e-06,
+      "loss": 1.1311,
+      "step": 769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.378973105134475e-06,
+      "loss": 1.1942,
+      "step": 770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.385958784491793e-06,
+      "loss": 1.3184,
+      "step": 771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.39294446384911e-06,
+      "loss": 1.0974,
+      "step": 772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.399930143206427e-06,
+      "loss": 1.218,
+      "step": 773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.406915822563745e-06,
+      "loss": 1.0977,
+      "step": 774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.413901501921063e-06,
+      "loss": 1.1124,
+      "step": 775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.42088718127838e-06,
+      "loss": 1.0862,
+      "step": 776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.427872860635697e-06,
+      "loss": 1.2856,
+      "step": 777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.434858539993014e-06,
+      "loss": 1.3335,
+      "step": 778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.441844219350332e-06,
+      "loss": 1.1935,
+      "step": 779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4488298987076495e-06,
+      "loss": 1.1436,
+      "step": 780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4558155780649665e-06,
+      "loss": 1.1119,
+      "step": 781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.462801257422284e-06,
+      "loss": 0.9242,
+      "step": 782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.469786936779602e-06,
+      "loss": 1.3825,
+      "step": 783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.476772616136919e-06,
+      "loss": 1.0698,
+      "step": 784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.483758295494237e-06,
+      "loss": 1.4055,
+      "step": 785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.490743974851555e-06,
+      "loss": 1.2031,
+      "step": 786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.497729654208872e-06,
+      "loss": 1.0746,
+      "step": 787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.5047153335661895e-06,
+      "loss": 1.1294,
+      "step": 788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.511701012923507e-06,
+      "loss": 1.1484,
+      "step": 789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.518686692280824e-06,
+      "loss": 1.1047,
+      "step": 790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.525672371638142e-06,
+      "loss": 1.036,
+      "step": 791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.53265805099546e-06,
+      "loss": 1.1136,
+      "step": 792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.539643730352777e-06,
+      "loss": 1.2354,
+      "step": 793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.546629409710095e-06,
+      "loss": 1.184,
+      "step": 794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.5536150890674125e-06,
+      "loss": 1.2569,
+      "step": 795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.56060076842473e-06,
+      "loss": 1.0893,
+      "step": 796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.567586447782047e-06,
+      "loss": 1.1557,
+      "step": 797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.574572127139365e-06,
+      "loss": 1.1109,
+      "step": 798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.581557806496683e-06,
+      "loss": 1.1977,
+      "step": 799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.588543485854e-06,
+      "loss": 1.2609,
+      "step": 800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.595529165211318e-06,
+      "loss": 1.0304,
+      "step": 801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.6025148445686354e-06,
+      "loss": 1.2636,
+      "step": 802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.609500523925952e-06,
+      "loss": 1.1125,
+      "step": 803
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.61648620328327e-06,
+      "loss": 1.1817,
+      "step": 804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.623471882640588e-06,
+      "loss": 1.2215,
+      "step": 805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.630457561997905e-06,
+      "loss": 1.4473,
+      "step": 806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.637443241355223e-06,
+      "loss": 1.0963,
+      "step": 807
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.644428920712541e-06,
+      "loss": 1.1696,
+      "step": 808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.651414600069858e-06,
+      "loss": 1.1266,
+      "step": 809
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.658400279427175e-06,
+      "loss": 1.1113,
+      "step": 810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.6653859587844915e-06,
+      "loss": 1.4099,
+      "step": 811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.672371638141809e-06,
+      "loss": 1.2315,
+      "step": 812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.679357317499127e-06,
+      "loss": 1.3051,
+      "step": 813
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.686342996856444e-06,
+      "loss": 1.0832,
+      "step": 814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.693328676213762e-06,
+      "loss": 1.1912,
+      "step": 815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.70031435557108e-06,
+      "loss": 1.0668,
+      "step": 816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.707300034928397e-06,
+      "loss": 1.1328,
+      "step": 817
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.181,
+      "step": 818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.721271393643032e-06,
+      "loss": 1.1391,
+      "step": 819
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.728257073000349e-06,
+      "loss": 1.0239,
+      "step": 820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.735242752357667e-06,
+      "loss": 1.3575,
+      "step": 821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.742228431714985e-06,
+      "loss": 1.1818,
+      "step": 822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.749214111072302e-06,
+      "loss": 1.3121,
+      "step": 823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.75619979042962e-06,
+      "loss": 0.9908,
+      "step": 824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.7631854697869375e-06,
+      "loss": 1.0762,
+      "step": 825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.7701711491442545e-06,
+      "loss": 1.3779,
+      "step": 826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.777156828501572e-06,
+      "loss": 1.0789,
+      "step": 827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.78414250785889e-06,
+      "loss": 1.3539,
+      "step": 828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.791128187216207e-06,
+      "loss": 1.1408,
+      "step": 829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.798113866573525e-06,
+      "loss": 1.3983,
+      "step": 830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.805099545930843e-06,
+      "loss": 1.1543,
+      "step": 831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.81208522528816e-06,
+      "loss": 1.1138,
+      "step": 832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8190709046454775e-06,
+      "loss": 1.1353,
+      "step": 833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.826056584002795e-06,
+      "loss": 1.1939,
+      "step": 834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.833042263360112e-06,
+      "loss": 1.0814,
+      "step": 835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.84002794271743e-06,
+      "loss": 1.4223,
+      "step": 836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.847013622074748e-06,
+      "loss": 1.0728,
+      "step": 837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.853999301432065e-06,
+      "loss": 1.1781,
+      "step": 838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.860984980789383e-06,
+      "loss": 1.1844,
+      "step": 839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8679706601467004e-06,
+      "loss": 1.2201,
+      "step": 840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.874956339504017e-06,
+      "loss": 1.182,
+      "step": 841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.881942018861335e-06,
+      "loss": 1.0265,
+      "step": 842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.888927698218653e-06,
+      "loss": 1.2627,
+      "step": 843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.89591337757597e-06,
+      "loss": 1.2351,
+      "step": 844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.902899056933288e-06,
+      "loss": 1.2215,
+      "step": 845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.909884736290604e-06,
+      "loss": 1.3724,
+      "step": 846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.916870415647922e-06,
+      "loss": 1.3495,
+      "step": 847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9238560950052396e-06,
+      "loss": 1.1752,
+      "step": 848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9308417743625565e-06,
+      "loss": 1.1963,
+      "step": 849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.937827453719874e-06,
+      "loss": 1.1892,
+      "step": 850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.944813133077192e-06,
+      "loss": 1.1968,
+      "step": 851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.951798812434509e-06,
+      "loss": 1.1076,
+      "step": 852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958784491791827e-06,
+      "loss": 1.0992,
+      "step": 853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.965770171149145e-06,
+      "loss": 1.0583,
+      "step": 854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.972755850506462e-06,
+      "loss": 1.1023,
+      "step": 855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9797415298637795e-06,
+      "loss": 1.0997,
+      "step": 856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.986727209221097e-06,
+      "loss": 1.1353,
+      "step": 857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.993712888578414e-06,
+      "loss": 1.1499,
+      "step": 858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.000698567935732e-06,
+      "loss": 1.149,
+      "step": 859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.00768424729305e-06,
+      "loss": 0.9947,
+      "step": 860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.014669926650367e-06,
+      "loss": 1.2686,
+      "step": 861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.021655606007685e-06,
+      "loss": 1.2689,
+      "step": 862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.0286412853650025e-06,
+      "loss": 1.1053,
+      "step": 863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.0356269647223195e-06,
+      "loss": 1.1129,
+      "step": 864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.042612644079637e-06,
+      "loss": 1.1899,
+      "step": 865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.049598323436955e-06,
+      "loss": 1.1462,
+      "step": 866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.056584002794272e-06,
+      "loss": 1.0946,
+      "step": 867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.06356968215159e-06,
+      "loss": 1.2186,
+      "step": 868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.070555361508908e-06,
+      "loss": 1.2943,
+      "step": 869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.077541040866225e-06,
+      "loss": 1.1691,
+      "step": 870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.0845267202235425e-06,
+      "loss": 1.0016,
+      "step": 871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.09151239958086e-06,
+      "loss": 1.2695,
+      "step": 872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.098498078938177e-06,
+      "loss": 1.0929,
+      "step": 873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.105483758295495e-06,
+      "loss": 1.1535,
+      "step": 874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.112469437652813e-06,
+      "loss": 1.1391,
+      "step": 875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.11945511701013e-06,
+      "loss": 1.1697,
+      "step": 876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.126440796367448e-06,
+      "loss": 1.2933,
+      "step": 877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1334264757247655e-06,
+      "loss": 1.1638,
+      "step": 878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.140412155082082e-06,
+      "loss": 1.0749,
+      "step": 879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.147397834439399e-06,
+      "loss": 1.0159,
+      "step": 880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.154383513796716e-06,
+      "loss": 1.1831,
+      "step": 881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.161369193154034e-06,
+      "loss": 1.1173,
+      "step": 882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.168354872511352e-06,
+      "loss": 1.4833,
+      "step": 883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.175340551868669e-06,
+      "loss": 1.206,
+      "step": 884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.182326231225987e-06,
+      "loss": 0.9882,
+      "step": 885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1893119105833046e-06,
+      "loss": 1.2003,
+      "step": 886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1962975899406215e-06,
+      "loss": 1.0946,
+      "step": 887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.203283269297939e-06,
+      "loss": 1.3339,
+      "step": 888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.210268948655257e-06,
+      "loss": 1.1321,
+      "step": 889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.217254628012574e-06,
+      "loss": 1.1421,
+      "step": 890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.224240307369892e-06,
+      "loss": 1.1981,
+      "step": 891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.23122598672721e-06,
+      "loss": 1.1013,
+      "step": 892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.238211666084527e-06,
+      "loss": 1.0668,
+      "step": 893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.2451973454418445e-06,
+      "loss": 1.3106,
+      "step": 894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.252183024799162e-06,
+      "loss": 1.1113,
+      "step": 895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.259168704156479e-06,
+      "loss": 1.2092,
+      "step": 896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.266154383513797e-06,
+      "loss": 1.1292,
+      "step": 897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.273140062871115e-06,
+      "loss": 1.4579,
+      "step": 898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.280125742228432e-06,
+      "loss": 1.0749,
+      "step": 899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.28711142158575e-06,
+      "loss": 1.3221,
+      "step": 900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.2940971009430675e-06,
+      "loss": 1.1341,
+      "step": 901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.3010827803003845e-06,
+      "loss": 1.108,
+      "step": 902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.308068459657702e-06,
+      "loss": 1.0477,
+      "step": 903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.31505413901502e-06,
+      "loss": 1.0473,
+      "step": 904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.322039818372338e-06,
+      "loss": 1.0066,
+      "step": 905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.329025497729655e-06,
+      "loss": 1.3251,
+      "step": 906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.336011177086973e-06,
+      "loss": 1.1933,
+      "step": 907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.3429968564442905e-06,
+      "loss": 1.4009,
+      "step": 908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.3499825358016075e-06,
+      "loss": 1.1332,
+      "step": 909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.356968215158925e-06,
+      "loss": 1.1764,
+      "step": 910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.363953894516243e-06,
+      "loss": 0.9786,
+      "step": 911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.37093957387356e-06,
+      "loss": 1.0951,
+      "step": 912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.377925253230878e-06,
+      "loss": 1.1594,
+      "step": 913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.384910932588196e-06,
+      "loss": 1.3696,
+      "step": 914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.391896611945512e-06,
+      "loss": 1.1063,
+      "step": 915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.39888229130283e-06,
+      "loss": 1.2411,
+      "step": 916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.4058679706601466e-06,
+      "loss": 1.0628,
+      "step": 917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.412853650017464e-06,
+      "loss": 1.0523,
+      "step": 918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.419839329374782e-06,
+      "loss": 1.1779,
+      "step": 919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.426825008732099e-06,
+      "loss": 1.2368,
+      "step": 920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.433810688089417e-06,
+      "loss": 1.0511,
+      "step": 921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.440796367446735e-06,
+      "loss": 1.0662,
+      "step": 922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.447782046804052e-06,
+      "loss": 0.9902,
+      "step": 923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.4547677261613696e-06,
+      "loss": 1.105,
+      "step": 924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.461753405518687e-06,
+      "loss": 1.1856,
+      "step": 925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.468739084876004e-06,
+      "loss": 1.3325,
+      "step": 926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.475724764233322e-06,
+      "loss": 1.0856,
+      "step": 927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.48271044359064e-06,
+      "loss": 1.1201,
+      "step": 928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.489696122947957e-06,
+      "loss": 0.9884,
+      "step": 929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.496681802305275e-06,
+      "loss": 1.1077,
+      "step": 930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.5036674816625926e-06,
+      "loss": 1.3583,
+      "step": 931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.5106531610199095e-06,
+      "loss": 1.1164,
+      "step": 932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.517638840377227e-06,
+      "loss": 1.1808,
+      "step": 933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.524624519734545e-06,
+      "loss": 1.2655,
+      "step": 934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.531610199091862e-06,
+      "loss": 1.0057,
+      "step": 935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.53859587844918e-06,
+      "loss": 1.2452,
+      "step": 936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.545581557806498e-06,
+      "loss": 1.1205,
+      "step": 937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.552567237163815e-06,
+      "loss": 1.1138,
+      "step": 938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.5595529165211325e-06,
+      "loss": 1.1962,
+      "step": 939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.56653859587845e-06,
+      "loss": 1.0647,
+      "step": 940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.573524275235767e-06,
+      "loss": 1.0679,
+      "step": 941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.580509954593085e-06,
+      "loss": 1.1137,
+      "step": 942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.587495633950403e-06,
+      "loss": 1.2917,
+      "step": 943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.59448131330772e-06,
+      "loss": 1.1805,
+      "step": 944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.601466992665038e-06,
+      "loss": 1.2402,
+      "step": 945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.6084526720223555e-06,
+      "loss": 1.0912,
+      "step": 946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.6154383513796725e-06,
+      "loss": 1.1727,
+      "step": 947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.62242403073699e-06,
+      "loss": 1.3146,
+      "step": 948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.629409710094306e-06,
+      "loss": 1.0267,
+      "step": 949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.636395389451624e-06,
+      "loss": 1.1265,
+      "step": 950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.643381068808942e-06,
+      "loss": 1.1449,
+      "step": 951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.650366748166259e-06,
+      "loss": 1.142,
+      "step": 952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.657352427523577e-06,
+      "loss": 1.2089,
+      "step": 953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.664338106880895e-06,
+      "loss": 1.088,
+      "step": 954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.6713237862382116e-06,
+      "loss": 1.2173,
+      "step": 955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.678309465595529e-06,
+      "loss": 1.1874,
+      "step": 956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.685295144952847e-06,
+      "loss": 1.1247,
+      "step": 957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.692280824310164e-06,
+      "loss": 1.1741,
+      "step": 958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.699266503667482e-06,
+      "loss": 1.2065,
+      "step": 959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7062521830248e-06,
+      "loss": 1.1447,
+      "step": 960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.713237862382117e-06,
+      "loss": 1.2755,
+      "step": 961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7202235417394346e-06,
+      "loss": 1.2663,
+      "step": 962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.727209221096752e-06,
+      "loss": 1.3535,
+      "step": 963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.734194900454069e-06,
+      "loss": 1.1639,
+      "step": 964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.741180579811387e-06,
+      "loss": 1.2464,
+      "step": 965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.748166259168705e-06,
+      "loss": 1.2848,
+      "step": 966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.755151938526022e-06,
+      "loss": 1.141,
+      "step": 967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.76213761788334e-06,
+      "loss": 1.2462,
+      "step": 968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7691232972406576e-06,
+      "loss": 0.9378,
+      "step": 969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7761089765979745e-06,
+      "loss": 0.9355,
+      "step": 970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.783094655955292e-06,
+      "loss": 1.1788,
+      "step": 971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.79008033531261e-06,
+      "loss": 1.2673,
+      "step": 972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.797066014669927e-06,
+      "loss": 1.0909,
+      "step": 973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.804051694027245e-06,
+      "loss": 1.1653,
+      "step": 974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.811037373384563e-06,
+      "loss": 1.1691,
+      "step": 975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.81802305274188e-06,
+      "loss": 1.2072,
+      "step": 976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.8250087320991975e-06,
+      "loss": 1.2989,
+      "step": 977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.831994411456515e-06,
+      "loss": 1.1174,
+      "step": 978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.838980090813832e-06,
+      "loss": 0.9299,
+      "step": 979
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.84596577017115e-06,
+      "loss": 1.0226,
+      "step": 980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.852951449528468e-06,
+      "loss": 1.1709,
+      "step": 981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.859937128885785e-06,
+      "loss": 0.9945,
+      "step": 982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.866922808243103e-06,
+      "loss": 0.9831,
+      "step": 983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.873908487600419e-06,
+      "loss": 1.052,
+      "step": 984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.880894166957737e-06,
+      "loss": 1.0082,
+      "step": 985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.8878798463150544e-06,
+      "loss": 1.0372,
+      "step": 986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.894865525672371e-06,
+      "loss": 1.205,
+      "step": 987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.901851205029689e-06,
+      "loss": 0.9915,
+      "step": 988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.908836884387007e-06,
+      "loss": 1.2041,
+      "step": 989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.915822563744324e-06,
+      "loss": 1.0686,
+      "step": 990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.922808243101642e-06,
+      "loss": 1.0592,
+      "step": 991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.92979392245896e-06,
+      "loss": 1.0525,
+      "step": 992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.9367796018162766e-06,
+      "loss": 1.0723,
+      "step": 993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.943765281173594e-06,
+      "loss": 1.0668,
+      "step": 994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.950750960530912e-06,
+      "loss": 1.0838,
+      "step": 995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.957736639888229e-06,
+      "loss": 1.1156,
+      "step": 996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.964722319245547e-06,
+      "loss": 1.1774,
+      "step": 997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.971707998602865e-06,
+      "loss": 1.0824,
+      "step": 998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.978693677960182e-06,
+      "loss": 1.2013,
+      "step": 999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.9856793573174996e-06,
+      "loss": 1.1249,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.992665036674817e-06,
+      "loss": 1.2026,
+      "step": 1001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.999650716032134e-06,
+      "loss": 1.1063,
+      "step": 1002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.006636395389452e-06,
+      "loss": 1.1961,
+      "step": 1003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.01362207474677e-06,
+      "loss": 1.0332,
+      "step": 1004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.020607754104087e-06,
+      "loss": 1.0794,
+      "step": 1005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.027593433461405e-06,
+      "loss": 1.1263,
+      "step": 1006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.0345791128187226e-06,
+      "loss": 1.2194,
+      "step": 1007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.0415647921760395e-06,
+      "loss": 1.0177,
+      "step": 1008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.048550471533357e-06,
+      "loss": 1.1275,
+      "step": 1009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.055536150890675e-06,
+      "loss": 1.3614,
+      "step": 1010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.062521830247993e-06,
+      "loss": 1.2917,
+      "step": 1011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.06950750960531e-06,
+      "loss": 1.1242,
+      "step": 1012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.076493188962628e-06,
+      "loss": 1.3068,
+      "step": 1013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.0834788683199455e-06,
+      "loss": 1.166,
+      "step": 1014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.0904645476772625e-06,
+      "loss": 1.1178,
+      "step": 1015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.09745022703458e-06,
+      "loss": 1.0325,
+      "step": 1016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.104435906391898e-06,
+      "loss": 1.2209,
+      "step": 1017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.111421585749214e-06,
+      "loss": 1.1005,
+      "step": 1018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.118407265106532e-06,
+      "loss": 1.0585,
+      "step": 1019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.125392944463849e-06,
+      "loss": 1.1264,
+      "step": 1020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.132378623821167e-06,
+      "loss": 1.1732,
+      "step": 1021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.139364303178485e-06,
+      "loss": 1.0686,
+      "step": 1022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.146349982535802e-06,
+      "loss": 1.2896,
+      "step": 1023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.1533356618931194e-06,
+      "loss": 1.1112,
+      "step": 1024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.160321341250437e-06,
+      "loss": 0.993,
+      "step": 1025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.167307020607754e-06,
+      "loss": 1.1689,
+      "step": 1026
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.174292699965072e-06,
+      "loss": 1.1818,
+      "step": 1027
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.18127837932239e-06,
+      "loss": 1.1,
+      "step": 1028
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.188264058679707e-06,
+      "loss": 1.0705,
+      "step": 1029
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.195249738037025e-06,
+      "loss": 1.0829,
+      "step": 1030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2022354173943424e-06,
+      "loss": 1.0234,
+      "step": 1031
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.209221096751659e-06,
+      "loss": 1.1229,
+      "step": 1032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.216206776108977e-06,
+      "loss": 1.1893,
+      "step": 1033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.223192455466295e-06,
+      "loss": 1.2186,
+      "step": 1034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.230178134823612e-06,
+      "loss": 1.1591,
+      "step": 1035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.23716381418093e-06,
+      "loss": 1.0311,
+      "step": 1036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.244149493538248e-06,
+      "loss": 1.1166,
+      "step": 1037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2511351728955646e-06,
+      "loss": 1.1902,
+      "step": 1038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.258120852252882e-06,
+      "loss": 1.244,
+      "step": 1039
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2651065316102e-06,
+      "loss": 1.069,
+      "step": 1040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.272092210967517e-06,
+      "loss": 1.0669,
+      "step": 1041
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.279077890324835e-06,
+      "loss": 1.1441,
+      "step": 1042
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.286063569682153e-06,
+      "loss": 1.2316,
+      "step": 1043
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.29304924903947e-06,
+      "loss": 1.2607,
+      "step": 1044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.3000349283967876e-06,
+      "loss": 1.0504,
+      "step": 1045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.307020607754105e-06,
+      "loss": 1.1299,
+      "step": 1046
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.314006287111422e-06,
+      "loss": 1.0015,
+      "step": 1047
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.32099196646874e-06,
+      "loss": 1.1391,
+      "step": 1048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.327977645826058e-06,
+      "loss": 1.0767,
+      "step": 1049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.334963325183375e-06,
+      "loss": 1.3686,
+      "step": 1050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.341949004540693e-06,
+      "loss": 1.1962,
+      "step": 1051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.3489346838980105e-06,
+      "loss": 1.2094,
+      "step": 1052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.355920363255327e-06,
+      "loss": 1.1653,
+      "step": 1053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.3629060426126445e-06,
+      "loss": 1.1511,
+      "step": 1054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.3698917219699614e-06,
+      "loss": 1.0565,
+      "step": 1055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.376877401327279e-06,
+      "loss": 1.1702,
+      "step": 1056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.383863080684597e-06,
+      "loss": 1.0898,
+      "step": 1057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.390848760041914e-06,
+      "loss": 1.0467,
+      "step": 1058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.397834439399232e-06,
+      "loss": 1.1014,
+      "step": 1059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.40482011875655e-06,
+      "loss": 1.1278,
+      "step": 1060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.411805798113867e-06,
+      "loss": 1.1274,
+      "step": 1061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.4187914774711844e-06,
+      "loss": 1.1136,
+      "step": 1062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.425777156828502e-06,
+      "loss": 1.2585,
+      "step": 1063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.432762836185819e-06,
+      "loss": 1.0951,
+      "step": 1064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.439748515543137e-06,
+      "loss": 1.1729,
+      "step": 1065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.446734194900455e-06,
+      "loss": 1.0821,
+      "step": 1066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.453719874257772e-06,
+      "loss": 1.0998,
+      "step": 1067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.46070555361509e-06,
+      "loss": 1.1339,
+      "step": 1068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.4676912329724074e-06,
+      "loss": 1.1858,
+      "step": 1069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.474676912329724e-06,
+      "loss": 1.1244,
+      "step": 1070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.481662591687042e-06,
+      "loss": 1.3606,
+      "step": 1071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.48864827104436e-06,
+      "loss": 1.0949,
+      "step": 1072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.495633950401677e-06,
+      "loss": 1.3719,
+      "step": 1073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.502619629758995e-06,
+      "loss": 1.093,
+      "step": 1074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.509605309116313e-06,
+      "loss": 1.0662,
+      "step": 1075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.5165909884736296e-06,
+      "loss": 1.109,
+      "step": 1076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.523576667830947e-06,
+      "loss": 1.1648,
+      "step": 1077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.530562347188265e-06,
+      "loss": 1.0067,
+      "step": 1078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.537548026545582e-06,
+      "loss": 1.0886,
+      "step": 1079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.5445337059029e-06,
+      "loss": 1.3001,
+      "step": 1080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.551519385260218e-06,
+      "loss": 0.9982,
+      "step": 1081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.558505064617535e-06,
+      "loss": 1.1937,
+      "step": 1082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.5654907439748526e-06,
+      "loss": 1.0505,
+      "step": 1083
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.57247642333217e-06,
+      "loss": 1.158,
+      "step": 1084
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.579462102689487e-06,
+      "loss": 1.1947,
+      "step": 1085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.586447782046805e-06,
+      "loss": 1.0351,
+      "step": 1086
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.593433461404121e-06,
+      "loss": 0.9456,
+      "step": 1087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.600419140761439e-06,
+      "loss": 1.1926,
+      "step": 1088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.607404820118757e-06,
+      "loss": 1.0148,
+      "step": 1089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.614390499476074e-06,
+      "loss": 1.0586,
+      "step": 1090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.621376178833392e-06,
+      "loss": 0.9762,
+      "step": 1091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.6283618581907095e-06,
+      "loss": 1.1221,
+      "step": 1092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.635347537548026e-06,
+      "loss": 1.1833,
+      "step": 1093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.642333216905344e-06,
+      "loss": 1.113,
+      "step": 1094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.649318896262662e-06,
+      "loss": 1.1647,
+      "step": 1095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.65630457561998e-06,
+      "loss": 1.2043,
+      "step": 1096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.663290254977298e-06,
+      "loss": 0.9713,
+      "step": 1097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.670275934334614e-06,
+      "loss": 0.9674,
+      "step": 1098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.677261613691932e-06,
+      "loss": 1.0311,
+      "step": 1099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.68424729304925e-06,
+      "loss": 1.1766,
+      "step": 1100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.691232972406567e-06,
+      "loss": 1.0189,
+      "step": 1101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.698218651763885e-06,
+      "loss": 0.99,
+      "step": 1102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.705204331121203e-06,
+      "loss": 1.1309,
+      "step": 1103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.712190010478519e-06,
+      "loss": 1.014,
+      "step": 1104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.719175689835837e-06,
+      "loss": 0.9649,
+      "step": 1105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.726161369193155e-06,
+      "loss": 1.0691,
+      "step": 1106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.733147048550472e-06,
+      "loss": 1.1396,
+      "step": 1107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.74013272790779e-06,
+      "loss": 1.38,
+      "step": 1108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.747118407265108e-06,
+      "loss": 1.491,
+      "step": 1109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.754104086622424e-06,
+      "loss": 1.1332,
+      "step": 1110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.761089765979742e-06,
+      "loss": 1.1721,
+      "step": 1111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.76807544533706e-06,
+      "loss": 1.177,
+      "step": 1112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.775061124694378e-06,
+      "loss": 1.045,
+      "step": 1113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.782046804051695e-06,
+      "loss": 1.1726,
+      "step": 1114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.789032483409013e-06,
+      "loss": 1.0496,
+      "step": 1115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.79601816276633e-06,
+      "loss": 1.0953,
+      "step": 1116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.803003842123647e-06,
+      "loss": 1.137,
+      "step": 1117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.809989521480965e-06,
+      "loss": 1.1734,
+      "step": 1118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.816975200838283e-06,
+      "loss": 1.1564,
+      "step": 1119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.8239608801956e-06,
+      "loss": 1.2342,
+      "step": 1120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.830946559552918e-06,
+      "loss": 1.0247,
+      "step": 1121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.837932238910235e-06,
+      "loss": 1.1545,
+      "step": 1122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.844917918267552e-06,
+      "loss": 1.0117,
+      "step": 1123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.851903597624868e-06,
+      "loss": 0.9944,
+      "step": 1124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.858889276982186e-06,
+      "loss": 1.1168,
+      "step": 1125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.865874956339504e-06,
+      "loss": 1.173,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.872860635696822e-06,
+      "loss": 1.0494,
+      "step": 1127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.87984631505414e-06,
+      "loss": 1.216,
+      "step": 1128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.886831994411458e-06,
+      "loss": 1.0822,
+      "step": 1129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.893817673768774e-06,
+      "loss": 1.0659,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.900803353126091e-06,
+      "loss": 1.2653,
+      "step": 1131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.90778903248341e-06,
+      "loss": 1.0779,
+      "step": 1132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.914774711840727e-06,
+      "loss": 1.2177,
+      "step": 1133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.921760391198045e-06,
+      "loss": 1.0716,
+      "step": 1134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.928746070555363e-06,
+      "loss": 1.0118,
+      "step": 1135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.935731749912679e-06,
+      "loss": 1.2374,
+      "step": 1136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.942717429269997e-06,
+      "loss": 1.2059,
+      "step": 1137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.949703108627314e-06,
+      "loss": 1.1668,
+      "step": 1138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.956688787984632e-06,
+      "loss": 1.0031,
+      "step": 1139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.96367446734195e-06,
+      "loss": 0.9859,
+      "step": 1140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.970660146699268e-06,
+      "loss": 1.0618,
+      "step": 1141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.977645826056584e-06,
+      "loss": 1.0964,
+      "step": 1142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.984631505413902e-06,
+      "loss": 1.1604,
+      "step": 1143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.99161718477122e-06,
+      "loss": 1.053,
+      "step": 1144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.998602864128537e-06,
+      "loss": 1.4088,
+      "step": 1145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.005588543485855e-06,
+      "loss": 1.0513,
+      "step": 1146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.012574222843173e-06,
+      "loss": 1.0278,
+      "step": 1147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.01955990220049e-06,
+      "loss": 0.9826,
+      "step": 1148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.026545581557807e-06,
+      "loss": 1.0264,
+      "step": 1149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.033531260915125e-06,
+      "loss": 1.0922,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.040516940272443e-06,
+      "loss": 1.2956,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.04750261962976e-06,
+      "loss": 1.141,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.054488298987078e-06,
+      "loss": 0.9719,
+      "step": 1153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.061473978344394e-06,
+      "loss": 1.288,
+      "step": 1154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.068459657701712e-06,
+      "loss": 0.972,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.075445337059028e-06,
+      "loss": 1.1767,
+      "step": 1156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.082431016416346e-06,
+      "loss": 1.126,
+      "step": 1157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.089416695773664e-06,
+      "loss": 1.1823,
+      "step": 1158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.096402375130982e-06,
+      "loss": 0.9922,
+      "step": 1159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.1033880544883e-06,
+      "loss": 1.0735,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.110373733845617e-06,
+      "loss": 1.2451,
+      "step": 1161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.117359413202933e-06,
+      "loss": 1.1223,
+      "step": 1162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.124345092560251e-06,
+      "loss": 1.3777,
+      "step": 1163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.131330771917569e-06,
+      "loss": 0.9809,
+      "step": 1164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138316451274887e-06,
+      "loss": 0.9895,
+      "step": 1165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.145302130632205e-06,
+      "loss": 0.994,
+      "step": 1166
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.152287809989523e-06,
+      "loss": 1.0897,
+      "step": 1167
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.15927348934684e-06,
+      "loss": 0.9562,
+      "step": 1168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.166259168704156e-06,
+      "loss": 0.9956,
+      "step": 1169
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.173244848061474e-06,
+      "loss": 1.1233,
+      "step": 1170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.180230527418792e-06,
+      "loss": 1.1622,
+      "step": 1171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.18721620677611e-06,
+      "loss": 1.0891,
+      "step": 1172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.194201886133428e-06,
+      "loss": 1.2218,
+      "step": 1173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.201187565490746e-06,
+      "loss": 0.9867,
+      "step": 1174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.208173244848062e-06,
+      "loss": 1.1316,
+      "step": 1175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.21515892420538e-06,
+      "loss": 1.0695,
+      "step": 1176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.222144603562697e-06,
+      "loss": 1.1156,
+      "step": 1177
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.229130282920015e-06,
+      "loss": 0.8624,
+      "step": 1178
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.236115962277333e-06,
+      "loss": 1.153,
+      "step": 1179
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.24310164163465e-06,
+      "loss": 1.2592,
+      "step": 1180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.250087320991967e-06,
+      "loss": 0.965,
+      "step": 1181
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.257073000349285e-06,
+      "loss": 1.1933,
+      "step": 1182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.264058679706602e-06,
+      "loss": 1.0527,
+      "step": 1183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.27104435906392e-06,
+      "loss": 1.1073,
+      "step": 1184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.278030038421238e-06,
+      "loss": 1.0596,
+      "step": 1185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.285015717778556e-06,
+      "loss": 0.9647,
+      "step": 1186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.292001397135872e-06,
+      "loss": 0.9955,
+      "step": 1187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.29898707649319e-06,
+      "loss": 1.2591,
+      "step": 1188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.305972755850508e-06,
+      "loss": 1.0047,
+      "step": 1189
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.312958435207825e-06,
+      "loss": 1.207,
+      "step": 1190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.319944114565142e-06,
+      "loss": 1.2602,
+      "step": 1191
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.32692979392246e-06,
+      "loss": 1.0984,
+      "step": 1192
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.333915473279777e-06,
+      "loss": 1.1516,
+      "step": 1193
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.340901152637095e-06,
+      "loss": 1.0703,
+      "step": 1194
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.347886831994411e-06,
+      "loss": 1.1238,
+      "step": 1195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.354872511351729e-06,
+      "loss": 1.0136,
+      "step": 1196
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.361858190709047e-06,
+      "loss": 1.3173,
+      "step": 1197
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.368843870066365e-06,
+      "loss": 1.0438,
+      "step": 1198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.375829549423682e-06,
+      "loss": 1.0867,
+      "step": 1199
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.382815228781e-06,
+      "loss": 1.0268,
+      "step": 1200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.389800908138316e-06,
+      "loss": 1.2407,
+      "step": 1201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.396786587495634e-06,
+      "loss": 1.0216,
+      "step": 1202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.403772266852952e-06,
+      "loss": 1.1782,
+      "step": 1203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.41075794621027e-06,
+      "loss": 1.1898,
+      "step": 1204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.417743625567588e-06,
+      "loss": 1.0449,
+      "step": 1205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.424729304924905e-06,
+      "loss": 1.1025,
+      "step": 1206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.431714984282221e-06,
+      "loss": 1.1548,
+      "step": 1207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.43870066363954e-06,
+      "loss": 0.9652,
+      "step": 1208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.445686342996857e-06,
+      "loss": 0.9479,
+      "step": 1209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.452672022354175e-06,
+      "loss": 0.9658,
+      "step": 1210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.459657701711493e-06,
+      "loss": 1.1873,
+      "step": 1211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.46664338106881e-06,
+      "loss": 1.1882,
+      "step": 1212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.473629060426127e-06,
+      "loss": 1.0687,
+      "step": 1213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.480614739783444e-06,
+      "loss": 1.0143,
+      "step": 1214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.487600419140762e-06,
+      "loss": 1.2559,
+      "step": 1215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.49458609849808e-06,
+      "loss": 1.0636,
+      "step": 1216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.501571777855398e-06,
+      "loss": 1.0072,
+      "step": 1217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.508557457212716e-06,
+      "loss": 1.1395,
+      "step": 1218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.515543136570032e-06,
+      "loss": 1.0788,
+      "step": 1219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.52252881592735e-06,
+      "loss": 1.0328,
+      "step": 1220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.529514495284667e-06,
+      "loss": 1.0337,
+      "step": 1221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.536500174641985e-06,
+      "loss": 1.0089,
+      "step": 1222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.543485853999303e-06,
+      "loss": 1.2692,
+      "step": 1223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.550471533356621e-06,
+      "loss": 0.9779,
+      "step": 1224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.557457212713937e-06,
+      "loss": 1.1229,
+      "step": 1225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.564442892071255e-06,
+      "loss": 0.968,
+      "step": 1226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.1355,
+      "step": 1227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.578414250785889e-06,
+      "loss": 1.1167,
+      "step": 1228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.585399930143207e-06,
+      "loss": 1.1482,
+      "step": 1229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.592385609500524e-06,
+      "loss": 0.9757,
+      "step": 1230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.599371288857842e-06,
+      "loss": 0.9273,
+      "step": 1231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.60635696821516e-06,
+      "loss": 0.9766,
+      "step": 1232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.613342647572476e-06,
+      "loss": 1.2285,
+      "step": 1233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.620328326929794e-06,
+      "loss": 1.337,
+      "step": 1234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.627314006287112e-06,
+      "loss": 1.2258,
+      "step": 1235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.63429968564443e-06,
+      "loss": 1.1253,
+      "step": 1236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.641285365001747e-06,
+      "loss": 1.1331,
+      "step": 1237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.648271044359065e-06,
+      "loss": 1.1082,
+      "step": 1238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.655256723716381e-06,
+      "loss": 1.1977,
+      "step": 1239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.662242403073699e-06,
+      "loss": 1.2983,
+      "step": 1240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.669228082431017e-06,
+      "loss": 0.9319,
+      "step": 1241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.676213761788335e-06,
+      "loss": 1.0618,
+      "step": 1242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.683199441145653e-06,
+      "loss": 1.0123,
+      "step": 1243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.69018512050297e-06,
+      "loss": 1.127,
+      "step": 1244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.697170799860286e-06,
+      "loss": 0.9231,
+      "step": 1245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.704156479217604e-06,
+      "loss": 1.0739,
+      "step": 1246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.711142158574922e-06,
+      "loss": 1.2418,
+      "step": 1247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.71812783793224e-06,
+      "loss": 1.1432,
+      "step": 1248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.725113517289558e-06,
+      "loss": 0.9105,
+      "step": 1249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.732099196646876e-06,
+      "loss": 1.0745,
+      "step": 1250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.739084876004192e-06,
+      "loss": 1.1295,
+      "step": 1251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.74607055536151e-06,
+      "loss": 1.1029,
+      "step": 1252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.753056234718827e-06,
+      "loss": 1.0508,
+      "step": 1253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.760041914076145e-06,
+      "loss": 0.9513,
+      "step": 1254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.767027593433463e-06,
+      "loss": 0.9839,
+      "step": 1255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.77401327279078e-06,
+      "loss": 1.158,
+      "step": 1256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.780998952148097e-06,
+      "loss": 1.1099,
+      "step": 1257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.787984631505415e-06,
+      "loss": 1.0387,
+      "step": 1258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.794970310862732e-06,
+      "loss": 1.1678,
+      "step": 1259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.801955990220049e-06,
+      "loss": 1.057,
+      "step": 1260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.808941669577366e-06,
+      "loss": 1.0382,
+      "step": 1261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.815927348934684e-06,
+      "loss": 1.1657,
+      "step": 1262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.822913028292002e-06,
+      "loss": 1.063,
+      "step": 1263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.82989870764932e-06,
+      "loss": 1.0099,
+      "step": 1264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.836884387006636e-06,
+      "loss": 1.2619,
+      "step": 1265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.843870066363954e-06,
+      "loss": 1.1516,
+      "step": 1266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.850855745721272e-06,
+      "loss": 1.0919,
+      "step": 1267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.85784142507859e-06,
+      "loss": 1.1035,
+      "step": 1268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.864827104435907e-06,
+      "loss": 0.939,
+      "step": 1269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.871812783793225e-06,
+      "loss": 1.161,
+      "step": 1270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.878798463150541e-06,
+      "loss": 0.9641,
+      "step": 1271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.885784142507859e-06,
+      "loss": 1.0655,
+      "step": 1272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.892769821865177e-06,
+      "loss": 1.1281,
+      "step": 1273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.899755501222495e-06,
+      "loss": 1.233,
+      "step": 1274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.906741180579812e-06,
+      "loss": 0.9749,
+      "step": 1275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.91372685993713e-06,
+      "loss": 1.0217,
+      "step": 1276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.920712539294448e-06,
+      "loss": 1.1812,
+      "step": 1277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.927698218651764e-06,
+      "loss": 1.0588,
+      "step": 1278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.934683898009082e-06,
+      "loss": 0.9533,
+      "step": 1279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.9416695773664e-06,
+      "loss": 1.0227,
+      "step": 1280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.948655256723718e-06,
+      "loss": 1.2081,
+      "step": 1281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.955640936081035e-06,
+      "loss": 1.0741,
+      "step": 1282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.962626615438353e-06,
+      "loss": 1.1146,
+      "step": 1283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.96961229479567e-06,
+      "loss": 1.1611,
+      "step": 1284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.976597974152987e-06,
+      "loss": 1.1562,
+      "step": 1285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.983583653510305e-06,
+      "loss": 0.939,
+      "step": 1286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.990569332867623e-06,
+      "loss": 1.0433,
+      "step": 1287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.99755501222494e-06,
+      "loss": 0.9477,
+      "step": 1288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.004540691582258e-06,
+      "loss": 1.0798,
+      "step": 1289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.011526370939574e-06,
+      "loss": 1.0917,
+      "step": 1290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.018512050296892e-06,
+      "loss": 1.1489,
+      "step": 1291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.02549772965421e-06,
+      "loss": 1.1059,
+      "step": 1292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.032483409011528e-06,
+      "loss": 1.0204,
+      "step": 1293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.039469088368844e-06,
+      "loss": 0.9371,
+      "step": 1294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.046454767726162e-06,
+      "loss": 1.0264,
+      "step": 1295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.05344044708348e-06,
+      "loss": 1.0838,
+      "step": 1296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.060426126440797e-06,
+      "loss": 0.8561,
+      "step": 1297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.067411805798114e-06,
+      "loss": 1.0952,
+      "step": 1298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.074397485155431e-06,
+      "loss": 1.0056,
+      "step": 1299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.08138316451275e-06,
+      "loss": 0.9554,
+      "step": 1300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.088368843870067e-06,
+      "loss": 0.8875,
+      "step": 1301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.095354523227385e-06,
+      "loss": 0.9998,
+      "step": 1302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.102340202584703e-06,
+      "loss": 0.973,
+      "step": 1303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.109325881942019e-06,
+      "loss": 0.9976,
+      "step": 1304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.116311561299337e-06,
+      "loss": 1.2535,
+      "step": 1305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.123297240656654e-06,
+      "loss": 0.9656,
+      "step": 1306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.130282920013972e-06,
+      "loss": 1.1636,
+      "step": 1307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.13726859937129e-06,
+      "loss": 1.1264,
+      "step": 1308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.144254278728608e-06,
+      "loss": 1.0612,
+      "step": 1309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.151239958085924e-06,
+      "loss": 0.9304,
+      "step": 1310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.158225637443242e-06,
+      "loss": 1.099,
+      "step": 1311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.16521131680056e-06,
+      "loss": 0.8749,
+      "step": 1312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.172196996157877e-06,
+      "loss": 0.9988,
+      "step": 1313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.179182675515195e-06,
+      "loss": 1.0548,
+      "step": 1314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.186168354872513e-06,
+      "loss": 1.4301,
+      "step": 1315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.193154034229829e-06,
+      "loss": 1.2622,
+      "step": 1316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.200139713587147e-06,
+      "loss": 1.2534,
+      "step": 1317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.207125392944465e-06,
+      "loss": 1.1139,
+      "step": 1318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.214111072301783e-06,
+      "loss": 1.1818,
+      "step": 1319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.2210967516591e-06,
+      "loss": 1.062,
+      "step": 1320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.228082431016418e-06,
+      "loss": 1.0844,
+      "step": 1321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.235068110373734e-06,
+      "loss": 1.0216,
+      "step": 1322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.242053789731052e-06,
+      "loss": 0.9635,
+      "step": 1323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.24903946908837e-06,
+      "loss": 1.0439,
+      "step": 1324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.256025148445688e-06,
+      "loss": 1.0045,
+      "step": 1325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.263010827803006e-06,
+      "loss": 1.0508,
+      "step": 1326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.269996507160323e-06,
+      "loss": 0.9816,
+      "step": 1327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.27698218651764e-06,
+      "loss": 1.0544,
+      "step": 1328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.283967865874957e-06,
+      "loss": 0.9438,
+      "step": 1329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.290953545232273e-06,
+      "loss": 1.0216,
+      "step": 1330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.297939224589591e-06,
+      "loss": 1.0519,
+      "step": 1331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.304924903946909e-06,
+      "loss": 1.0149,
+      "step": 1332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.311910583304227e-06,
+      "loss": 0.9157,
+      "step": 1333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.318896262661545e-06,
+      "loss": 1.2704,
+      "step": 1334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.325881942018862e-06,
+      "loss": 0.8968,
+      "step": 1335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.332867621376179e-06,
+      "loss": 1.123,
+      "step": 1336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.339853300733496e-06,
+      "loss": 1.0815,
+      "step": 1337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.346838980090814e-06,
+      "loss": 1.0884,
+      "step": 1338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.353824659448132e-06,
+      "loss": 1.0399,
+      "step": 1339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.36081033880545e-06,
+      "loss": 1.1665,
+      "step": 1340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.367796018162768e-06,
+      "loss": 1.0052,
+      "step": 1341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.374781697520084e-06,
+      "loss": 1.0273,
+      "step": 1342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.381767376877402e-06,
+      "loss": 1.0306,
+      "step": 1343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.38875305623472e-06,
+      "loss": 1.172,
+      "step": 1344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.395738735592037e-06,
+      "loss": 1.1502,
+      "step": 1345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.402724414949355e-06,
+      "loss": 1.1449,
+      "step": 1346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.409710094306673e-06,
+      "loss": 1.2357,
+      "step": 1347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.416695773663989e-06,
+      "loss": 0.8855,
+      "step": 1348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.423681453021307e-06,
+      "loss": 1.1163,
+      "step": 1349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.430667132378625e-06,
+      "loss": 1.0401,
+      "step": 1350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.437652811735942e-06,
+      "loss": 0.9353,
+      "step": 1351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.44463849109326e-06,
+      "loss": 0.8153,
+      "step": 1352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.451624170450578e-06,
+      "loss": 1.1113,
+      "step": 1353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.458609849807894e-06,
+      "loss": 1.0312,
+      "step": 1354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.465595529165212e-06,
+      "loss": 1.1165,
+      "step": 1355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.47258120852253e-06,
+      "loss": 0.9646,
+      "step": 1356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.479566887879848e-06,
+      "loss": 1.0864,
+      "step": 1357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.486552567237165e-06,
+      "loss": 0.9317,
+      "step": 1358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.493538246594483e-06,
+      "loss": 1.2162,
+      "step": 1359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.5005239259518e-06,
+      "loss": 1.0982,
+      "step": 1360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.507509605309117e-06,
+      "loss": 1.2154,
+      "step": 1361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.514495284666435e-06,
+      "loss": 1.0134,
+      "step": 1362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.521480964023751e-06,
+      "loss": 0.84,
+      "step": 1363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.528466643381069e-06,
+      "loss": 1.0956,
+      "step": 1364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.535452322738387e-06,
+      "loss": 1.1525,
+      "step": 1365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.542438002095704e-06,
+      "loss": 0.8488,
+      "step": 1366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.549423681453022e-06,
+      "loss": 1.078,
+      "step": 1367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.556409360810338e-06,
+      "loss": 1.2113,
+      "step": 1368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.563395040167656e-06,
+      "loss": 1.1155,
+      "step": 1369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.570380719524974e-06,
+      "loss": 1.0689,
+      "step": 1370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.577366398882292e-06,
+      "loss": 1.0115,
+      "step": 1371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.58435207823961e-06,
+      "loss": 1.0813,
+      "step": 1372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.591337757596927e-06,
+      "loss": 0.9866,
+      "step": 1373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.598323436954244e-06,
+      "loss": 1.0498,
+      "step": 1374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.605309116311561e-06,
+      "loss": 0.9373,
+      "step": 1375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.61229479566888e-06,
+      "loss": 0.9695,
+      "step": 1376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.619280475026197e-06,
+      "loss": 1.1033,
+      "step": 1377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.626266154383515e-06,
+      "loss": 1.0195,
+      "step": 1378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.633251833740833e-06,
+      "loss": 0.9394,
+      "step": 1379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.64023751309815e-06,
+      "loss": 1.2196,
+      "step": 1380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.647223192455467e-06,
+      "loss": 1.0463,
+      "step": 1381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.654208871812784e-06,
+      "loss": 1.1831,
+      "step": 1382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.661194551170102e-06,
+      "loss": 1.1994,
+      "step": 1383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.66818023052742e-06,
+      "loss": 1.1979,
+      "step": 1384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.675165909884738e-06,
+      "loss": 1.2288,
+      "step": 1385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.682151589242056e-06,
+      "loss": 0.9792,
+      "step": 1386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.689137268599372e-06,
+      "loss": 0.9927,
+      "step": 1387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.69612294795669e-06,
+      "loss": 1.0831,
+      "step": 1388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.703108627314007e-06,
+      "loss": 0.996,
+      "step": 1389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.710094306671325e-06,
+      "loss": 1.1698,
+      "step": 1390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.717079986028643e-06,
+      "loss": 0.9889,
+      "step": 1391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.72406566538596e-06,
+      "loss": 0.8795,
+      "step": 1392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.731051344743277e-06,
+      "loss": 1.0013,
+      "step": 1393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.738037024100595e-06,
+      "loss": 1.159,
+      "step": 1394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.745022703457913e-06,
+      "loss": 1.1905,
+      "step": 1395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.75200838281523e-06,
+      "loss": 1.447,
+      "step": 1396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.758994062172548e-06,
+      "loss": 1.1375,
+      "step": 1397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.765979741529864e-06,
+      "loss": 1.0666,
+      "step": 1398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.772965420887182e-06,
+      "loss": 1.3094,
+      "step": 1399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.7799511002445e-06,
+      "loss": 0.978,
+      "step": 1400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.786936779601816e-06,
+      "loss": 1.1219,
+      "step": 1401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.793922458959134e-06,
+      "loss": 0.8758,
+      "step": 1402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.800908138316452e-06,
+      "loss": 1.0848,
+      "step": 1403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.80789381767377e-06,
+      "loss": 1.1483,
+      "step": 1404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.814879497031087e-06,
+      "loss": 1.0213,
+      "step": 1405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.821865176388405e-06,
+      "loss": 1.0195,
+      "step": 1406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.828850855745721e-06,
+      "loss": 1.099,
+      "step": 1407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.835836535103039e-06,
+      "loss": 1.0105,
+      "step": 1408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.842822214460357e-06,
+      "loss": 0.9089,
+      "step": 1409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.849807893817675e-06,
+      "loss": 1.1503,
+      "step": 1410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.856793573174992e-06,
+      "loss": 1.0661,
+      "step": 1411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.86377925253231e-06,
+      "loss": 0.8991,
+      "step": 1412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.870764931889626e-06,
+      "loss": 1.0755,
+      "step": 1413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.877750611246944e-06,
+      "loss": 1.0177,
+      "step": 1414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.884736290604262e-06,
+      "loss": 1.0759,
+      "step": 1415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.89172196996158e-06,
+      "loss": 1.2507,
+      "step": 1416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.898707649318898e-06,
+      "loss": 1.0452,
+      "step": 1417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.905693328676215e-06,
+      "loss": 1.0108,
+      "step": 1418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.912679008033532e-06,
+      "loss": 1.1044,
+      "step": 1419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.91966468739085e-06,
+      "loss": 1.145,
+      "step": 1420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.926650366748167e-06,
+      "loss": 0.9409,
+      "step": 1421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.933636046105485e-06,
+      "loss": 0.9584,
+      "step": 1422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.940621725462803e-06,
+      "loss": 1.1168,
+      "step": 1423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.94760740482012e-06,
+      "loss": 1.2371,
+      "step": 1424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.954593084177437e-06,
+      "loss": 1.1532,
+      "step": 1425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.961578763534755e-06,
+      "loss": 0.9478,
+      "step": 1426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.968564442892072e-06,
+      "loss": 1.0417,
+      "step": 1427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.97555012224939e-06,
+      "loss": 1.2498,
+      "step": 1428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.982535801606708e-06,
+      "loss": 0.9267,
+      "step": 1429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.989521480964026e-06,
+      "loss": 1.0495,
+      "step": 1430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.996507160321342e-06,
+      "loss": 0.9942,
+      "step": 1431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.000349283967866e-05,
+      "loss": 1.0648,
+      "step": 1432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0010478519035978e-05,
+      "loss": 1.0832,
+      "step": 1433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0017464198393294e-05,
+      "loss": 0.9488,
+      "step": 1434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0024449877750613e-05,
+      "loss": 1.0661,
+      "step": 1435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.003143555710793e-05,
+      "loss": 1.2035,
+      "step": 1436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0038421236465247e-05,
+      "loss": 1.0578,
+      "step": 1437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0045406915822565e-05,
+      "loss": 0.9555,
+      "step": 1438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0052392595179883e-05,
+      "loss": 0.9185,
+      "step": 1439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0059378274537199e-05,
+      "loss": 0.9555,
+      "step": 1440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0066363953894518e-05,
+      "loss": 0.8988,
+      "step": 1441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0073349633251834e-05,
+      "loss": 1.1019,
+      "step": 1442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0080335312609152e-05,
+      "loss": 0.9927,
+      "step": 1443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.008732099196647e-05,
+      "loss": 1.1244,
+      "step": 1444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0094306671323788e-05,
+      "loss": 1.1463,
+      "step": 1445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0101292350681104e-05,
+      "loss": 1.082,
+      "step": 1446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0108278030038424e-05,
+      "loss": 1.1542,
+      "step": 1447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.011526370939574e-05,
+      "loss": 0.9663,
+      "step": 1448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0122249388753056e-05,
+      "loss": 0.9325,
+      "step": 1449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0129235068110375e-05,
+      "loss": 0.9802,
+      "step": 1450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0136220747467691e-05,
+      "loss": 1.0799,
+      "step": 1451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.014320642682501e-05,
+      "loss": 1.0222,
+      "step": 1452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0150192106182325e-05,
+      "loss": 1.16,
+      "step": 1453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0157177785539645e-05,
+      "loss": 1.0262,
+      "step": 1454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0164163464896961e-05,
+      "loss": 1.0124,
+      "step": 1455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.017114914425428e-05,
+      "loss": 1.0276,
+      "step": 1456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0178134823611597e-05,
+      "loss": 1.0962,
+      "step": 1457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0185120502968914e-05,
+      "loss": 0.8556,
+      "step": 1458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.019210618232623e-05,
+      "loss": 1.0801,
+      "step": 1459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.019909186168355e-05,
+      "loss": 1.3106,
+      "step": 1460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0206077541040866e-05,
+      "loss": 0.9575,
+      "step": 1461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0213063220398186e-05,
+      "loss": 0.9193,
+      "step": 1462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0220048899755502e-05,
+      "loss": 1.0867,
+      "step": 1463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.022703457911282e-05,
+      "loss": 1.2414,
+      "step": 1464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0234020258470136e-05,
+      "loss": 1.2133,
+      "step": 1465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0241005937827455e-05,
+      "loss": 1.0062,
+      "step": 1466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0247991617184771e-05,
+      "loss": 0.9314,
+      "step": 1467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.025497729654209e-05,
+      "loss": 1.1397,
+      "step": 1468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0261962975899407e-05,
+      "loss": 1.1417,
+      "step": 1469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0268948655256725e-05,
+      "loss": 0.9726,
+      "step": 1470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.027593433461404e-05,
+      "loss": 1.1652,
+      "step": 1471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.028292001397136e-05,
+      "loss": 0.9249,
+      "step": 1472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0289905693328676e-05,
+      "loss": 1.0828,
+      "step": 1473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0296891372685996e-05,
+      "loss": 1.1481,
+      "step": 1474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0303877052043312e-05,
+      "loss": 1.1262,
+      "step": 1475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.031086273140063e-05,
+      "loss": 1.0115,
+      "step": 1476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0317848410757946e-05,
+      "loss": 1.1965,
+      "step": 1477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0324834090115266e-05,
+      "loss": 0.9826,
+      "step": 1478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0331819769472582e-05,
+      "loss": 1.0624,
+      "step": 1479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0338805448829901e-05,
+      "loss": 1.0813,
+      "step": 1480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0345791128187217e-05,
+      "loss": 1.2576,
+      "step": 1481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0352776807544535e-05,
+      "loss": 1.04,
+      "step": 1482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0359762486901851e-05,
+      "loss": 1.1134,
+      "step": 1483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0366748166259169e-05,
+      "loss": 0.9966,
+      "step": 1484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0373733845616487e-05,
+      "loss": 0.9973,
+      "step": 1485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0380719524973803e-05,
+      "loss": 1.0072,
+      "step": 1486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0387705204331122e-05,
+      "loss": 0.9508,
+      "step": 1487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0394690883688439e-05,
+      "loss": 0.9979,
+      "step": 1488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0401676563045758e-05,
+      "loss": 1.0371,
+      "step": 1489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0408662242403074e-05,
+      "loss": 0.9825,
+      "step": 1490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0415647921760392e-05,
+      "loss": 1.0424,
+      "step": 1491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0422633601117708e-05,
+      "loss": 1.1731,
+      "step": 1492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0429619280475028e-05,
+      "loss": 0.9085,
+      "step": 1493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0436604959832344e-05,
+      "loss": 0.9615,
+      "step": 1494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0443590639189663e-05,
+      "loss": 1.2075,
+      "step": 1495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.045057631854698e-05,
+      "loss": 1.0462,
+      "step": 1496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0457561997904297e-05,
+      "loss": 0.9117,
+      "step": 1497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0464547677261613e-05,
+      "loss": 0.9083,
+      "step": 1498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0471533356618933e-05,
+      "loss": 1.124,
+      "step": 1499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0478519035976249e-05,
+      "loss": 1.0294,
+      "step": 1500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0485504715333568e-05,
+      "loss": 1.0891,
+      "step": 1501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0492490394690885e-05,
+      "loss": 1.2051,
+      "step": 1502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0499476074048202e-05,
+      "loss": 1.0911,
+      "step": 1503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0506461753405518e-05,
+      "loss": 1.1757,
+      "step": 1504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0513447432762838e-05,
+      "loss": 1.0035,
+      "step": 1505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0520433112120154e-05,
+      "loss": 1.197,
+      "step": 1506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0527418791477474e-05,
+      "loss": 1.0552,
+      "step": 1507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.053440447083479e-05,
+      "loss": 0.8929,
+      "step": 1508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0541390150192108e-05,
+      "loss": 1.1345,
+      "step": 1509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0548375829549424e-05,
+      "loss": 1.1865,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0555361508906743e-05,
+      "loss": 1.0741,
+      "step": 1511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.056234718826406e-05,
+      "loss": 1.3126,
+      "step": 1512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0569332867621379e-05,
+      "loss": 0.9801,
+      "step": 1513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0576318546978695e-05,
+      "loss": 1.0937,
+      "step": 1514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0583304226336013e-05,
+      "loss": 1.1558,
+      "step": 1515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0590289905693329e-05,
+      "loss": 1.0705,
+      "step": 1516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0597275585050648e-05,
+      "loss": 1.1737,
+      "step": 1517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0604261264407964e-05,
+      "loss": 0.9133,
+      "step": 1518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.061124694376528e-05,
+      "loss": 0.9452,
+      "step": 1519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.06182326231226e-05,
+      "loss": 1.1432,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0625218302479916e-05,
+      "loss": 1.2111,
+      "step": 1521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0632203981837234e-05,
+      "loss": 1.0035,
+      "step": 1522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0639189661194552e-05,
+      "loss": 0.9852,
+      "step": 1523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.064617534055187e-05,
+      "loss": 1.0266,
+      "step": 1524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0653161019909186e-05,
+      "loss": 1.0294,
+      "step": 1525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0660146699266505e-05,
+      "loss": 1.1696,
+      "step": 1526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0667132378623821e-05,
+      "loss": 1.0428,
+      "step": 1527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.067411805798114e-05,
+      "loss": 0.9221,
+      "step": 1528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0681103737338457e-05,
+      "loss": 1.0619,
+      "step": 1529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0688089416695775e-05,
+      "loss": 1.1713,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0695075096053091e-05,
+      "loss": 1.1092,
+      "step": 1531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.070206077541041e-05,
+      "loss": 1.2173,
+      "step": 1532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0709046454767727e-05,
+      "loss": 0.8596,
+      "step": 1533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0716032134125044e-05,
+      "loss": 1.1521,
+      "step": 1534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0723017813482362e-05,
+      "loss": 1.0895,
+      "step": 1535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.073000349283968e-05,
+      "loss": 1.067,
+      "step": 1536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0736989172196996e-05,
+      "loss": 1.1432,
+      "step": 1537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0743974851554316e-05,
+      "loss": 0.9149,
+      "step": 1538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0750960530911632e-05,
+      "loss": 0.8209,
+      "step": 1539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.075794621026895e-05,
+      "loss": 1.0524,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0764931889626267e-05,
+      "loss": 1.0937,
+      "step": 1541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0771917568983585e-05,
+      "loss": 1.0254,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0778903248340901e-05,
+      "loss": 1.0709,
+      "step": 1543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.078588892769822e-05,
+      "loss": 0.9354,
+      "step": 1544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0792874607055537e-05,
+      "loss": 1.0455,
+      "step": 1545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0799860286412855e-05,
+      "loss": 1.028,
+      "step": 1546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0806845965770173e-05,
+      "loss": 1.1741,
+      "step": 1547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.081383164512749e-05,
+      "loss": 1.0566,
+      "step": 1548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0820817324484806e-05,
+      "loss": 1.0289,
+      "step": 1549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0827803003842126e-05,
+      "loss": 1.0569,
+      "step": 1550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0834788683199442e-05,
+      "loss": 0.9154,
+      "step": 1551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.084177436255676e-05,
+      "loss": 0.8884,
+      "step": 1552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0848760041914078e-05,
+      "loss": 1.1301,
+      "step": 1553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0855745721271394e-05,
+      "loss": 0.9181,
+      "step": 1554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0862731400628712e-05,
+      "loss": 1.0243,
+      "step": 1555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0869717079986028e-05,
+      "loss": 0.9102,
+      "step": 1556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0876702759343347e-05,
+      "loss": 1.047,
+      "step": 1557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0883688438700663e-05,
+      "loss": 0.8831,
+      "step": 1558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0890674118057983e-05,
+      "loss": 1.2262,
+      "step": 1559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0897659797415299e-05,
+      "loss": 1.0481,
+      "step": 1560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0904645476772617e-05,
+      "loss": 1.1375,
+      "step": 1561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0911631156129933e-05,
+      "loss": 1.1803,
+      "step": 1562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0918616835487252e-05,
+      "loss": 1.0091,
+      "step": 1563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0925602514844569e-05,
+      "loss": 1.1861,
+      "step": 1564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0932588194201888e-05,
+      "loss": 0.9802,
+      "step": 1565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0939573873559204e-05,
+      "loss": 0.9628,
+      "step": 1566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0946559552916522e-05,
+      "loss": 1.0327,
+      "step": 1567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0953545232273838e-05,
+      "loss": 1.1398,
+      "step": 1568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0960530911631158e-05,
+      "loss": 0.9527,
+      "step": 1569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0967516590988474e-05,
+      "loss": 0.9652,
+      "step": 1570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0974502270345793e-05,
+      "loss": 1.1273,
+      "step": 1571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.098148794970311e-05,
+      "loss": 1.0163,
+      "step": 1572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0988473629060427e-05,
+      "loss": 0.9356,
+      "step": 1573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0995459308417743e-05,
+      "loss": 0.9179,
+      "step": 1574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1002444987775063e-05,
+      "loss": 1.0205,
+      "step": 1575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1009430667132379e-05,
+      "loss": 1.036,
+      "step": 1576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1016416346489698e-05,
+      "loss": 1.2432,
+      "step": 1577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1023402025847015e-05,
+      "loss": 1.3637,
+      "step": 1578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1030387705204332e-05,
+      "loss": 1.0137,
+      "step": 1579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1037373384561648e-05,
+      "loss": 1.0958,
+      "step": 1580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1044359063918968e-05,
+      "loss": 1.0443,
+      "step": 1581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1051344743276284e-05,
+      "loss": 1.1383,
+      "step": 1582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1058330422633604e-05,
+      "loss": 0.9534,
+      "step": 1583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.106531610199092e-05,
+      "loss": 1.0473,
+      "step": 1584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1072301781348238e-05,
+      "loss": 1.1406,
+      "step": 1585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1079287460705554e-05,
+      "loss": 0.8799,
+      "step": 1586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1086273140062871e-05,
+      "loss": 1.1746,
+      "step": 1587
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.109325881942019e-05,
+      "loss": 1.0887,
+      "step": 1588
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1100244498777505e-05,
+      "loss": 0.9775,
+      "step": 1589
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1107230178134825e-05,
+      "loss": 1.0745,
+      "step": 1590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1114215857492141e-05,
+      "loss": 0.8276,
+      "step": 1591
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.112120153684946e-05,
+      "loss": 1.2351,
+      "step": 1592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1128187216206777e-05,
+      "loss": 1.0249,
+      "step": 1593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1135172895564094e-05,
+      "loss": 1.1063,
+      "step": 1594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.114215857492141e-05,
+      "loss": 0.9237,
+      "step": 1595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.114914425427873e-05,
+      "loss": 0.8371,
+      "step": 1596
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1156129933636046e-05,
+      "loss": 0.9733,
+      "step": 1597
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1163115612993366e-05,
+      "loss": 1.1912,
+      "step": 1598
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1170101292350682e-05,
+      "loss": 1.0499,
+      "step": 1599
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1177086971708e-05,
+      "loss": 1.1213,
+      "step": 1600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1184072651065316e-05,
+      "loss": 0.9443,
+      "step": 1601
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1191058330422635e-05,
+      "loss": 0.9728,
+      "step": 1602
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1198044009779951e-05,
+      "loss": 1.0187,
+      "step": 1603
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1205029689137271e-05,
+      "loss": 0.9766,
+      "step": 1604
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1212015368494587e-05,
+      "loss": 0.8431,
+      "step": 1605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1219001047851905e-05,
+      "loss": 1.0535,
+      "step": 1606
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1225986727209221e-05,
+      "loss": 0.8765,
+      "step": 1607
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.123297240656654e-05,
+      "loss": 1.0742,
+      "step": 1608
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1239958085923857e-05,
+      "loss": 0.9631,
+      "step": 1609
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1246943765281176e-05,
+      "loss": 0.9616,
+      "step": 1610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1253929444638492e-05,
+      "loss": 1.2589,
+      "step": 1611
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.126091512399581e-05,
+      "loss": 1.3351,
+      "step": 1612
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1267900803353126e-05,
+      "loss": 1.1576,
+      "step": 1613
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1274886482710446e-05,
+      "loss": 1.1071,
+      "step": 1614
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1281872162067762e-05,
+      "loss": 1.0065,
+      "step": 1615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1288857841425081e-05,
+      "loss": 1.0056,
+      "step": 1616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1295843520782397e-05,
+      "loss": 0.9825,
+      "step": 1617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1302829200139715e-05,
+      "loss": 0.9191,
+      "step": 1618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1309814879497031e-05,
+      "loss": 1.1053,
+      "step": 1619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.131680055885435e-05,
+      "loss": 0.9773,
+      "step": 1620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1323786238211667e-05,
+      "loss": 0.8838,
+      "step": 1621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1330771917568983e-05,
+      "loss": 1.1892,
+      "step": 1622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1337757596926303e-05,
+      "loss": 1.0263,
+      "step": 1623
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1344743276283619e-05,
+      "loss": 1.0162,
+      "step": 1624
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1351728955640936e-05,
+      "loss": 1.0391,
+      "step": 1625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1358714634998254e-05,
+      "loss": 1.0437,
+      "step": 1626
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1365700314355572e-05,
+      "loss": 1.0177,
+      "step": 1627
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1372685993712888e-05,
+      "loss": 0.9257,
+      "step": 1628
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1379671673070208e-05,
+      "loss": 1.0404,
+      "step": 1629
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1386657352427524e-05,
+      "loss": 1.0946,
+      "step": 1630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1393643031784842e-05,
+      "loss": 0.9673,
+      "step": 1631
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.140062871114216e-05,
+      "loss": 0.9782,
+      "step": 1632
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1407614390499477e-05,
+      "loss": 0.9124,
+      "step": 1633
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1414600069856793e-05,
+      "loss": 0.9968,
+      "step": 1634
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1421585749214113e-05,
+      "loss": 0.8866,
+      "step": 1635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.9485,
+      "step": 1636
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1435557107928747e-05,
+      "loss": 0.9387,
+      "step": 1637
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1442542787286065e-05,
+      "loss": 1.0258,
+      "step": 1638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1449528466643382e-05,
+      "loss": 1.0167,
+      "step": 1639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1456514146000699e-05,
+      "loss": 0.9922,
+      "step": 1640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1463499825358018e-05,
+      "loss": 1.2024,
+      "step": 1641
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1470485504715334e-05,
+      "loss": 0.9489,
+      "step": 1642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1477471184072652e-05,
+      "loss": 1.0257,
+      "step": 1643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.148445686342997e-05,
+      "loss": 1.1372,
+      "step": 1644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1491442542787288e-05,
+      "loss": 1.1057,
+      "step": 1645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1498428222144604e-05,
+      "loss": 1.0334,
+      "step": 1646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1505413901501923e-05,
+      "loss": 0.8415,
+      "step": 1647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.151239958085924e-05,
+      "loss": 1.146,
+      "step": 1648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1519385260216557e-05,
+      "loss": 1.0466,
+      "step": 1649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1526370939573875e-05,
+      "loss": 0.9968,
+      "step": 1650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1533356618931193e-05,
+      "loss": 1.0005,
+      "step": 1651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1540342298288509e-05,
+      "loss": 1.2151,
+      "step": 1652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1547327977645828e-05,
+      "loss": 1.1683,
+      "step": 1653
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1554313657003145e-05,
+      "loss": 0.8831,
+      "step": 1654
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1561299336360462e-05,
+      "loss": 0.8404,
+      "step": 1655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.156828501571778e-05,
+      "loss": 0.9004,
+      "step": 1656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1575270695075096e-05,
+      "loss": 0.9177,
+      "step": 1657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1582256374432414e-05,
+      "loss": 1.0456,
+      "step": 1658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.158924205378973e-05,
+      "loss": 1.0378,
+      "step": 1659
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.159622773314705e-05,
+      "loss": 1.0109,
+      "step": 1660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1603213412504366e-05,
+      "loss": 1.1062,
+      "step": 1661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1610199091861685e-05,
+      "loss": 1.0466,
+      "step": 1662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1617184771219001e-05,
+      "loss": 1.1016,
+      "step": 1663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.162417045057632e-05,
+      "loss": 0.9013,
+      "step": 1664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1631156129933635e-05,
+      "loss": 0.8254,
+      "step": 1665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1638141809290955e-05,
+      "loss": 1.0784,
+      "step": 1666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1645127488648271e-05,
+      "loss": 1.0615,
+      "step": 1667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.165211316800559e-05,
+      "loss": 0.9426,
+      "step": 1668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1659098847362907e-05,
+      "loss": 0.916,
+      "step": 1669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1666084526720224e-05,
+      "loss": 1.0233,
+      "step": 1670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.167307020607754e-05,
+      "loss": 0.9796,
+      "step": 1671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.168005588543486e-05,
+      "loss": 1.0708,
+      "step": 1672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1687041564792176e-05,
+      "loss": 1.196,
+      "step": 1673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1694027244149496e-05,
+      "loss": 1.0616,
+      "step": 1674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1701012923506812e-05,
+      "loss": 1.0533,
+      "step": 1675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.170799860286413e-05,
+      "loss": 0.9226,
+      "step": 1676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1714984282221446e-05,
+      "loss": 1.1261,
+      "step": 1677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1721969961578765e-05,
+      "loss": 1.0012,
+      "step": 1678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1728955640936081e-05,
+      "loss": 1.0141,
+      "step": 1679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1735941320293401e-05,
+      "loss": 0.9148,
+      "step": 1680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1742926999650717e-05,
+      "loss": 0.8515,
+      "step": 1681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1749912679008035e-05,
+      "loss": 0.931,
+      "step": 1682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1756898358365351e-05,
+      "loss": 1.0334,
+      "step": 1683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.176388403772267e-05,
+      "loss": 0.9755,
+      "step": 1684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1770869717079987e-05,
+      "loss": 1.1368,
+      "step": 1685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1777855396437306e-05,
+      "loss": 0.8307,
+      "step": 1686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1784841075794622e-05,
+      "loss": 0.8417,
+      "step": 1687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.179182675515194e-05,
+      "loss": 0.9648,
+      "step": 1688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1798812434509256e-05,
+      "loss": 1.0115,
+      "step": 1689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1805798113866576e-05,
+      "loss": 1.0071,
+      "step": 1690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1812783793223892e-05,
+      "loss": 0.9325,
+      "step": 1691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1819769472581208e-05,
+      "loss": 0.9741,
+      "step": 1692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1826755151938527e-05,
+      "loss": 0.925,
+      "step": 1693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1833740831295843e-05,
+      "loss": 1.2803,
+      "step": 1694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1840726510653161e-05,
+      "loss": 0.9506,
+      "step": 1695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1847712190010479e-05,
+      "loss": 1.0372,
+      "step": 1696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1854697869367797e-05,
+      "loss": 1.0642,
+      "step": 1697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1861683548725113e-05,
+      "loss": 1.1371,
+      "step": 1698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1868669228082433e-05,
+      "loss": 0.931,
+      "step": 1699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1875654907439749e-05,
+      "loss": 0.9544,
+      "step": 1700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1882640586797068e-05,
+      "loss": 1.0006,
+      "step": 1701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1889626266154384e-05,
+      "loss": 1.0936,
+      "step": 1702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1896611945511702e-05,
+      "loss": 0.917,
+      "step": 1703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1903597624869018e-05,
+      "loss": 1.0462,
+      "step": 1704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1910583304226338e-05,
+      "loss": 1.0009,
+      "step": 1705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1917568983583654e-05,
+      "loss": 0.9962,
+      "step": 1706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1924554662940973e-05,
+      "loss": 0.9667,
+      "step": 1707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193154034229829e-05,
+      "loss": 1.0539,
+      "step": 1708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938526021655607e-05,
+      "loss": 0.9752,
+      "step": 1709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1945511701012923e-05,
+      "loss": 1.1546,
+      "step": 1710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952497380370243e-05,
+      "loss": 0.834,
+      "step": 1711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959483059727559e-05,
+      "loss": 1.0482,
+      "step": 1712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966468739084879e-05,
+      "loss": 0.9855,
+      "step": 1713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973454418442195e-05,
+      "loss": 1.2674,
+      "step": 1714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1980440097799512e-05,
+      "loss": 0.8866,
+      "step": 1715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987425777156829e-05,
+      "loss": 0.9224,
+      "step": 1716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994411456514148e-05,
+      "loss": 1.3131,
+      "step": 1717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001397135871464e-05,
+      "loss": 0.9632,
+      "step": 1718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2008382815228784e-05,
+      "loss": 0.977,
+      "step": 1719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20153684945861e-05,
+      "loss": 1.0196,
+      "step": 1720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022354173943418e-05,
+      "loss": 1.0851,
+      "step": 1721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029339853300734e-05,
+      "loss": 1.0192,
+      "step": 1722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036325532658053e-05,
+      "loss": 0.9258,
+      "step": 1723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204331121201537e-05,
+      "loss": 1.1219,
+      "step": 1724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050296891372686e-05,
+      "loss": 1.0534,
+      "step": 1725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2057282570730005e-05,
+      "loss": 1.0963,
+      "step": 1726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064268250087321e-05,
+      "loss": 0.9521,
+      "step": 1727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071253929444639e-05,
+      "loss": 0.8933,
+      "step": 1728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078239608801957e-05,
+      "loss": 0.9934,
+      "step": 1729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085225288159275e-05,
+      "loss": 1.0857,
+      "step": 1730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209221096751659e-05,
+      "loss": 0.8403,
+      "step": 1731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209919664687391e-05,
+      "loss": 0.8773,
+      "step": 1732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106182326231226e-05,
+      "loss": 0.8374,
+      "step": 1733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2113168005588544e-05,
+      "loss": 1.0755,
+      "step": 1734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120153684945862e-05,
+      "loss": 1.0762,
+      "step": 1735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212713936430318e-05,
+      "loss": 0.9919,
+      "step": 1736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2134125043660496e-05,
+      "loss": 0.9924,
+      "step": 1737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2141110723017815e-05,
+      "loss": 0.9421,
+      "step": 1738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2148096402375131e-05,
+      "loss": 1.0615,
+      "step": 1739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.215508208173245e-05,
+      "loss": 1.152,
+      "step": 1740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2162067761089767e-05,
+      "loss": 1.0399,
+      "step": 1741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2169053440447085e-05,
+      "loss": 1.0605,
+      "step": 1742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2176039119804401e-05,
+      "loss": 0.9015,
+      "step": 1743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.218302479916172e-05,
+      "loss": 1.0649,
+      "step": 1744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2190010478519037e-05,
+      "loss": 1.0296,
+      "step": 1745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2196996157876354e-05,
+      "loss": 0.9058,
+      "step": 1746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2203981837233672e-05,
+      "loss": 0.9266,
+      "step": 1747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.221096751659099e-05,
+      "loss": 0.9423,
+      "step": 1748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2217953195948306e-05,
+      "loss": 0.9674,
+      "step": 1749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2224938875305626e-05,
+      "loss": 0.9865,
+      "step": 1750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2231924554662942e-05,
+      "loss": 1.0922,
+      "step": 1751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.223891023402026e-05,
+      "loss": 1.0017,
+      "step": 1752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2245895913377577e-05,
+      "loss": 1.0481,
+      "step": 1753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2252881592734895e-05,
+      "loss": 0.984,
+      "step": 1754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2259867272092211e-05,
+      "loss": 1.0967,
+      "step": 1755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2266852951449531e-05,
+      "loss": 1.0252,
+      "step": 1756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2273838630806847e-05,
+      "loss": 0.9277,
+      "step": 1757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2280824310164165e-05,
+      "loss": 0.9558,
+      "step": 1758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2287809989521483e-05,
+      "loss": 0.9917,
+      "step": 1759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2294795668878799e-05,
+      "loss": 0.8842,
+      "step": 1760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2301781348236117e-05,
+      "loss": 0.8569,
+      "step": 1761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2308767027593433e-05,
+      "loss": 0.9539,
+      "step": 1762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2315752706950752e-05,
+      "loss": 1.0694,
+      "step": 1763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2322738386308068e-05,
+      "loss": 1.029,
+      "step": 1764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2329724065665388e-05,
+      "loss": 0.9311,
+      "step": 1765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2336709745022704e-05,
+      "loss": 1.1165,
+      "step": 1766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2343695424380022e-05,
+      "loss": 0.9167,
+      "step": 1767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2350681103737338e-05,
+      "loss": 1.0697,
+      "step": 1768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2357666783094657e-05,
+      "loss": 1.0626,
+      "step": 1769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2364652462451973e-05,
+      "loss": 1.0554,
+      "step": 1770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2371638141809293e-05,
+      "loss": 1.0579,
+      "step": 1771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2378623821166609e-05,
+      "loss": 1.0021,
+      "step": 1772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2385609500523927e-05,
+      "loss": 1.0229,
+      "step": 1773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2392595179881243e-05,
+      "loss": 1.0207,
+      "step": 1774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2399580859238563e-05,
+      "loss": 0.956,
+      "step": 1775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2406566538595879e-05,
+      "loss": 1.1538,
+      "step": 1776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2413552217953198e-05,
+      "loss": 1.0681,
+      "step": 1777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2420537897310514e-05,
+      "loss": 1.0392,
+      "step": 1778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2427523576667832e-05,
+      "loss": 0.8126,
+      "step": 1779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2434509256025148e-05,
+      "loss": 1.0304,
+      "step": 1780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2441494935382468e-05,
+      "loss": 0.9665,
+      "step": 1781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2448480614739784e-05,
+      "loss": 0.8955,
+      "step": 1782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2455466294097103e-05,
+      "loss": 1.1409,
+      "step": 1783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.246245197345442e-05,
+      "loss": 0.9863,
+      "step": 1784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2469437652811737e-05,
+      "loss": 1.1632,
+      "step": 1785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2476423332169053e-05,
+      "loss": 0.9573,
+      "step": 1786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2483409011526373e-05,
+      "loss": 1.1974,
+      "step": 1787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2490394690883689e-05,
+      "loss": 0.8698,
+      "step": 1788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2497380370241009e-05,
+      "loss": 1.0114,
+      "step": 1789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2504366049598325e-05,
+      "loss": 0.9939,
+      "step": 1790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2511351728955642e-05,
+      "loss": 0.8554,
+      "step": 1791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2518337408312959e-05,
+      "loss": 0.932,
+      "step": 1792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2525323087670278e-05,
+      "loss": 0.9952,
+      "step": 1793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2532308767027594e-05,
+      "loss": 0.9365,
+      "step": 1794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.253929444638491e-05,
+      "loss": 0.9542,
+      "step": 1795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.254628012574223e-05,
+      "loss": 0.8327,
+      "step": 1796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2553265805099546e-05,
+      "loss": 1.0,
+      "step": 1797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2560251484456864e-05,
+      "loss": 1.0246,
+      "step": 1798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2567237163814182e-05,
+      "loss": 0.8119,
+      "step": 1799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.25742228431715e-05,
+      "loss": 0.9851,
+      "step": 1800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2581208522528816e-05,
+      "loss": 0.879,
+      "step": 1801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2588194201886135e-05,
+      "loss": 1.1616,
+      "step": 1802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2595179881243451e-05,
+      "loss": 0.9473,
+      "step": 1803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2602165560600769e-05,
+      "loss": 1.2467,
+      "step": 1804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2609151239958087e-05,
+      "loss": 0.9614,
+      "step": 1805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2616136919315405e-05,
+      "loss": 1.0628,
+      "step": 1806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.262312259867272e-05,
+      "loss": 0.9146,
+      "step": 1807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.263010827803004e-05,
+      "loss": 0.9902,
+      "step": 1808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2637093957387356e-05,
+      "loss": 0.927,
+      "step": 1809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2644079636744676e-05,
+      "loss": 1.0164,
+      "step": 1810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2651065316101992e-05,
+      "loss": 0.9393,
+      "step": 1811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.265805099545931e-05,
+      "loss": 1.1561,
+      "step": 1812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2665036674816626e-05,
+      "loss": 1.0285,
+      "step": 1813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2672022354173945e-05,
+      "loss": 1.0378,
+      "step": 1814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2679008033531261e-05,
+      "loss": 1.0979,
+      "step": 1815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2685993712888581e-05,
+      "loss": 1.077,
+      "step": 1816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2692979392245897e-05,
+      "loss": 1.1119,
+      "step": 1817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2699965071603215e-05,
+      "loss": 1.1158,
+      "step": 1818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2706950750960531e-05,
+      "loss": 0.9184,
+      "step": 1819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.271393643031785e-05,
+      "loss": 0.9937,
+      "step": 1820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2720922109675167e-05,
+      "loss": 1.133,
+      "step": 1821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2727907789032486e-05,
+      "loss": 1.0531,
+      "step": 1822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2734893468389802e-05,
+      "loss": 0.9213,
+      "step": 1823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.274187914774712e-05,
+      "loss": 0.951,
+      "step": 1824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2748864827104436e-05,
+      "loss": 1.1019,
+      "step": 1825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2755850506461756e-05,
+      "loss": 0.9136,
+      "step": 1826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2762836185819072e-05,
+      "loss": 0.9535,
+      "step": 1827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2769821865176391e-05,
+      "loss": 1.0383,
+      "step": 1828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2776807544533707e-05,
+      "loss": 1.1178,
+      "step": 1829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2783793223891024e-05,
+      "loss": 1.0637,
+      "step": 1830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2790778903248341e-05,
+      "loss": 1.0042,
+      "step": 1831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.279776458260566e-05,
+      "loss": 1.022,
+      "step": 1832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2804750261962977e-05,
+      "loss": 1.0014,
+      "step": 1833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2811735941320293e-05,
+      "loss": 0.8859,
+      "step": 1834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2818721620677613e-05,
+      "loss": 1.0973,
+      "step": 1835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2825707300034929e-05,
+      "loss": 0.884,
+      "step": 1836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2832692979392247e-05,
+      "loss": 0.87,
+      "step": 1837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2839678658749564e-05,
+      "loss": 0.9807,
+      "step": 1838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2846664338106882e-05,
+      "loss": 0.9611,
+      "step": 1839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2853650017464198e-05,
+      "loss": 0.913,
+      "step": 1840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2860635696821518e-05,
+      "loss": 1.0382,
+      "step": 1841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2867621376178834e-05,
+      "loss": 1.0124,
+      "step": 1842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2874607055536152e-05,
+      "loss": 0.7551,
+      "step": 1843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.288159273489347e-05,
+      "loss": 0.8788,
+      "step": 1844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2888578414250787e-05,
+      "loss": 1.1806,
+      "step": 1845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2895564093608103e-05,
+      "loss": 1.1318,
+      "step": 1846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2902549772965423e-05,
+      "loss": 0.9722,
+      "step": 1847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2909535452322739e-05,
+      "loss": 0.9693,
+      "step": 1848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2916521131680057e-05,
+      "loss": 1.0693,
+      "step": 1849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2923506811037375e-05,
+      "loss": 0.9471,
+      "step": 1850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2930492490394693e-05,
+      "loss": 0.877,
+      "step": 1851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2937478169752009e-05,
+      "loss": 1.2471,
+      "step": 1852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2944463849109328e-05,
+      "loss": 0.9482,
+      "step": 1853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2951449528466644e-05,
+      "loss": 1.0007,
+      "step": 1854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2958435207823962e-05,
+      "loss": 1.0593,
+      "step": 1855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.296542088718128e-05,
+      "loss": 1.109,
+      "step": 1856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2972406566538598e-05,
+      "loss": 1.0252,
+      "step": 1857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2979392245895914e-05,
+      "loss": 0.9803,
+      "step": 1858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2986377925253233e-05,
+      "loss": 0.8407,
+      "step": 1859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.299336360461055e-05,
+      "loss": 1.0107,
+      "step": 1860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3000349283967867e-05,
+      "loss": 0.9062,
+      "step": 1861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3007334963325185e-05,
+      "loss": 0.9486,
+      "step": 1862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3014320642682501e-05,
+      "loss": 1.1293,
+      "step": 1863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3021306322039819e-05,
+      "loss": 0.9828,
+      "step": 1864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3028292001397135e-05,
+      "loss": 0.7753,
+      "step": 1865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3035277680754455e-05,
+      "loss": 0.8831,
+      "step": 1866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.304226336011177e-05,
+      "loss": 1.132,
+      "step": 1867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.304924903946909e-05,
+      "loss": 0.932,
+      "step": 1868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3056234718826406e-05,
+      "loss": 1.023,
+      "step": 1869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3063220398183724e-05,
+      "loss": 1.1083,
+      "step": 1870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.307020607754104e-05,
+      "loss": 1.1702,
+      "step": 1871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.307719175689836e-05,
+      "loss": 1.0056,
+      "step": 1872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3084177436255676e-05,
+      "loss": 0.9907,
+      "step": 1873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3091163115612995e-05,
+      "loss": 0.9508,
+      "step": 1874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3098148794970312e-05,
+      "loss": 1.1291,
+      "step": 1875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.310513447432763e-05,
+      "loss": 0.9355,
+      "step": 1876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3112120153684946e-05,
+      "loss": 1.0655,
+      "step": 1877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3119105833042265e-05,
+      "loss": 0.9892,
+      "step": 1878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3126091512399581e-05,
+      "loss": 1.0874,
+      "step": 1879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.31330771917569e-05,
+      "loss": 0.9841,
+      "step": 1880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3140062871114217e-05,
+      "loss": 0.9561,
+      "step": 1881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3147048550471535e-05,
+      "loss": 0.9849,
+      "step": 1882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.315403422982885e-05,
+      "loss": 1.1744,
+      "step": 1883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.316101990918617e-05,
+      "loss": 0.9358,
+      "step": 1884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3168005588543486e-05,
+      "loss": 0.9476,
+      "step": 1885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3174991267900806e-05,
+      "loss": 1.037,
+      "step": 1886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3181976947258122e-05,
+      "loss": 1.1312,
+      "step": 1887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.318896262661544e-05,
+      "loss": 0.9369,
+      "step": 1888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3195948305972756e-05,
+      "loss": 0.8973,
+      "step": 1889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3202933985330075e-05,
+      "loss": 0.9864,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3209919664687391e-05,
+      "loss": 1.1475,
+      "step": 1891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3216905344044711e-05,
+      "loss": 1.1853,
+      "step": 1892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3223891023402027e-05,
+      "loss": 1.0106,
+      "step": 1893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3230876702759345e-05,
+      "loss": 0.9312,
+      "step": 1894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3237862382116661e-05,
+      "loss": 0.9994,
+      "step": 1895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.324484806147398e-05,
+      "loss": 0.9966,
+      "step": 1896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3251833740831297e-05,
+      "loss": 0.9211,
+      "step": 1897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3258819420188613e-05,
+      "loss": 0.828,
+      "step": 1898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3265805099545932e-05,
+      "loss": 0.9892,
+      "step": 1899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3272790778903248e-05,
+      "loss": 0.9314,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3279776458260566e-05,
+      "loss": 0.8668,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3286762137617884e-05,
+      "loss": 0.9788,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3293747816975202e-05,
+      "loss": 1.2685,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3300733496332518e-05,
+      "loss": 0.9859,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3307719175689837e-05,
+      "loss": 0.8621,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3314704855047154e-05,
+      "loss": 0.9502,
+      "step": 1906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3321690534404471e-05,
+      "loss": 1.1123,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.332867621376179e-05,
+      "loss": 0.9382,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3335661893119107e-05,
+      "loss": 0.9141,
+      "step": 1909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3342647572476423e-05,
+      "loss": 1.0648,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3349633251833743e-05,
+      "loss": 0.795,
+      "step": 1911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3356618931191059e-05,
+      "loss": 1.0119,
+      "step": 1912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3363604610548378e-05,
+      "loss": 0.9578,
+      "step": 1913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3370590289905694e-05,
+      "loss": 0.8244,
+      "step": 1914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3377575969263012e-05,
+      "loss": 1.0499,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3384561648620328e-05,
+      "loss": 0.9045,
+      "step": 1916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3391547327977648e-05,
+      "loss": 1.0938,
+      "step": 1917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3398533007334964e-05,
+      "loss": 0.9677,
+      "step": 1918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3405518686692283e-05,
+      "loss": 1.0079,
+      "step": 1919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.34125043660496e-05,
+      "loss": 1.1218,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3419490045406917e-05,
+      "loss": 0.9935,
+      "step": 1921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3426475724764234e-05,
+      "loss": 0.858,
+      "step": 1922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3433461404121553e-05,
+      "loss": 1.012,
+      "step": 1923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3440447083478869e-05,
+      "loss": 0.9148,
+      "step": 1924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3447432762836189e-05,
+      "loss": 1.0259,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3454418442193505e-05,
+      "loss": 1.102,
+      "step": 1926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3461404121550823e-05,
+      "loss": 1.1035,
+      "step": 1927
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3468389800908139e-05,
+      "loss": 1.1308,
+      "step": 1928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3475375480265458e-05,
+      "loss": 0.9161,
+      "step": 1929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3482361159622774e-05,
+      "loss": 1.0974,
+      "step": 1930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3489346838980094e-05,
+      "loss": 0.9657,
+      "step": 1931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.349633251833741e-05,
+      "loss": 0.9867,
+      "step": 1932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3503318197694726e-05,
+      "loss": 0.8825,
+      "step": 1933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3510303877052044e-05,
+      "loss": 1.083,
+      "step": 1934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3517289556409362e-05,
+      "loss": 1.0107,
+      "step": 1935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.352427523576668e-05,
+      "loss": 0.841,
+      "step": 1936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3531260915123996e-05,
+      "loss": 0.7762,
+      "step": 1937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3538246594481315e-05,
+      "loss": 0.9161,
+      "step": 1938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3545232273838631e-05,
+      "loss": 1.0341,
+      "step": 1939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3552217953195949e-05,
+      "loss": 0.9411,
+      "step": 1940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3559203632553267e-05,
+      "loss": 1.0295,
+      "step": 1941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3566189311910585e-05,
+      "loss": 0.9683,
+      "step": 1942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.35731749912679e-05,
+      "loss": 1.0734,
+      "step": 1943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.358016067062522e-05,
+      "loss": 0.9305,
+      "step": 1944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3587146349982536e-05,
+      "loss": 1.0578,
+      "step": 1945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3594132029339854e-05,
+      "loss": 0.9491,
+      "step": 1946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3601117708697172e-05,
+      "loss": 0.9039,
+      "step": 1947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.360810338805449e-05,
+      "loss": 1.0064,
+      "step": 1948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3615089067411806e-05,
+      "loss": 0.8726,
+      "step": 1949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3622074746769125e-05,
+      "loss": 1.0858,
+      "step": 1950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3629060426126442e-05,
+      "loss": 1.0694,
+      "step": 1951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.363604610548376e-05,
+      "loss": 0.9084,
+      "step": 1952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3643031784841077e-05,
+      "loss": 0.8934,
+      "step": 1953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3650017464198395e-05,
+      "loss": 1.0208,
+      "step": 1954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3657003143555711e-05,
+      "loss": 0.7647,
+      "step": 1955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.366398882291303e-05,
+      "loss": 0.8775,
+      "step": 1956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3670974502270347e-05,
+      "loss": 0.9784,
+      "step": 1957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3677960181627665e-05,
+      "loss": 1.224,
+      "step": 1958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3684945860984982e-05,
+      "loss": 0.9599,
+      "step": 1959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.36919315403423e-05,
+      "loss": 1.0296,
+      "step": 1960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3698917219699616e-05,
+      "loss": 0.9326,
+      "step": 1961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3705902899056936e-05,
+      "loss": 1.1234,
+      "step": 1962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3712888578414252e-05,
+      "loss": 1.013,
+      "step": 1963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.371987425777157e-05,
+      "loss": 0.7232,
+      "step": 1964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3726859937128888e-05,
+      "loss": 0.7965,
+      "step": 1965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3733845616486205e-05,
+      "loss": 0.8808,
+      "step": 1966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3740831295843521e-05,
+      "loss": 1.1448,
+      "step": 1967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3747816975200838e-05,
+      "loss": 1.0642,
+      "step": 1968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3754802654558157e-05,
+      "loss": 1.4013,
+      "step": 1969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3761788333915473e-05,
+      "loss": 1.0885,
+      "step": 1970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3768774013272793e-05,
+      "loss": 0.9373,
+      "step": 1971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3775759692630109e-05,
+      "loss": 0.9862,
+      "step": 1972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3782745371987427e-05,
+      "loss": 0.9734,
+      "step": 1973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3789731051344743e-05,
+      "loss": 1.1843,
+      "step": 1974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3796716730702062e-05,
+      "loss": 1.0321,
+      "step": 1975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3803702410059378e-05,
+      "loss": 0.916,
+      "step": 1976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3810688089416698e-05,
+      "loss": 1.125,
+      "step": 1977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3817673768774014e-05,
+      "loss": 0.9165,
+      "step": 1978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3824659448131332e-05,
+      "loss": 1.1786,
+      "step": 1979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3831645127488648e-05,
+      "loss": 0.9729,
+      "step": 1980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3838630806845967e-05,
+      "loss": 0.9777,
+      "step": 1981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3845616486203284e-05,
+      "loss": 0.9148,
+      "step": 1982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3852602165560603e-05,
+      "loss": 0.8159,
+      "step": 1983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.385958784491792e-05,
+      "loss": 1.0576,
+      "step": 1984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3866573524275237e-05,
+      "loss": 1.0464,
+      "step": 1985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3873559203632553e-05,
+      "loss": 0.9295,
+      "step": 1986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3880544882989873e-05,
+      "loss": 0.8609,
+      "step": 1987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3887530562347189e-05,
+      "loss": 0.9368,
+      "step": 1988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3894516241704508e-05,
+      "loss": 1.1466,
+      "step": 1989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3901501921061824e-05,
+      "loss": 0.878,
+      "step": 1990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3908487600419142e-05,
+      "loss": 1.0344,
+      "step": 1991
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3915473279776458e-05,
+      "loss": 0.9835,
+      "step": 1992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3922458959133778e-05,
+      "loss": 1.0778,
+      "step": 1993
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3929444638491094e-05,
+      "loss": 0.9282,
+      "step": 1994
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3936430317848413e-05,
+      "loss": 1.0273,
+      "step": 1995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.394341599720573e-05,
+      "loss": 0.9169,
+      "step": 1996
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3950401676563047e-05,
+      "loss": 0.8015,
+      "step": 1997
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3957387355920364e-05,
+      "loss": 0.924,
+      "step": 1998
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3964373035277683e-05,
+      "loss": 1.135,
+      "step": 1999
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3971358714634999e-05,
+      "loss": 0.9837,
+      "step": 2000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3978344393992315e-05,
+      "loss": 0.9969,
+      "step": 2001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3985330073349635e-05,
+      "loss": 0.8542,
+      "step": 2002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3992315752706951e-05,
+      "loss": 1.0389,
+      "step": 2003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3999301432064269e-05,
+      "loss": 0.9954,
+      "step": 2004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4006287111421586e-05,
+      "loss": 0.8848,
+      "step": 2005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4013272790778904e-05,
+      "loss": 0.8941,
+      "step": 2006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.402025847013622e-05,
+      "loss": 0.8736,
+      "step": 2007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.402724414949354e-05,
+      "loss": 1.0014,
+      "step": 2008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4034229828850856e-05,
+      "loss": 0.979,
+      "step": 2009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4041215508208174e-05,
+      "loss": 0.9609,
+      "step": 2010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4048201187565492e-05,
+      "loss": 0.9723,
+      "step": 2011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.405518686692281e-05,
+      "loss": 0.9558,
+      "step": 2012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4062172546280126e-05,
+      "loss": 1.0293,
+      "step": 2013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4069158225637445e-05,
+      "loss": 1.0784,
+      "step": 2014
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4076143904994761e-05,
+      "loss": 1.034,
+      "step": 2015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4083129584352079e-05,
+      "loss": 1.1925,
+      "step": 2016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4090115263709397e-05,
+      "loss": 0.89,
+      "step": 2017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4097100943066715e-05,
+      "loss": 0.8457,
+      "step": 2018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.410408662242403e-05,
+      "loss": 0.9714,
+      "step": 2019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.411107230178135e-05,
+      "loss": 0.8754,
+      "step": 2020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4118057981138666e-05,
+      "loss": 0.9813,
+      "step": 2021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4125043660495986e-05,
+      "loss": 0.9284,
+      "step": 2022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4132029339853302e-05,
+      "loss": 0.9669,
+      "step": 2023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.413901501921062e-05,
+      "loss": 1.0317,
+      "step": 2024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4146000698567936e-05,
+      "loss": 0.8313,
+      "step": 2025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4152986377925255e-05,
+      "loss": 0.9864,
+      "step": 2026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4159972057282572e-05,
+      "loss": 0.9057,
+      "step": 2027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4166957736639891e-05,
+      "loss": 0.8452,
+      "step": 2028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4173943415997207e-05,
+      "loss": 0.8752,
+      "step": 2029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4180929095354525e-05,
+      "loss": 0.8199,
+      "step": 2030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4187914774711841e-05,
+      "loss": 1.0703,
+      "step": 2031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.419490045406916e-05,
+      "loss": 0.8967,
+      "step": 2032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4201886133426477e-05,
+      "loss": 1.1318,
+      "step": 2033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4208871812783796e-05,
+      "loss": 1.0681,
+      "step": 2034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4215857492141112e-05,
+      "loss": 0.9702,
+      "step": 2035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4222843171498429e-05,
+      "loss": 0.999,
+      "step": 2036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4229828850855746e-05,
+      "loss": 1.0696,
+      "step": 2037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4236814530213064e-05,
+      "loss": 1.0397,
+      "step": 2038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4243800209570382e-05,
+      "loss": 0.9558,
+      "step": 2039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4250785888927698e-05,
+      "loss": 0.9222,
+      "step": 2040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4257771568285018e-05,
+      "loss": 1.1268,
+      "step": 2041
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4264757247642334e-05,
+      "loss": 1.0373,
+      "step": 2042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4271742926999652e-05,
+      "loss": 1.0148,
+      "step": 2043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.427872860635697e-05,
+      "loss": 1.1985,
+      "step": 2044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 1.0749,
+      "step": 2045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4292699965071603e-05,
+      "loss": 0.8968,
+      "step": 2046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4299685644428923e-05,
+      "loss": 0.8689,
+      "step": 2047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4306671323786239e-05,
+      "loss": 1.0486,
+      "step": 2048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4313657003143557e-05,
+      "loss": 0.8943,
+      "step": 2049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4320642682500874e-05,
+      "loss": 0.9144,
+      "step": 2050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4327628361858192e-05,
+      "loss": 0.9508,
+      "step": 2051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4334614041215508e-05,
+      "loss": 0.9617,
+      "step": 2052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4341599720572828e-05,
+      "loss": 1.0184,
+      "step": 2053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4348585399930144e-05,
+      "loss": 1.0441,
+      "step": 2054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4355571079287462e-05,
+      "loss": 0.8599,
+      "step": 2055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.436255675864478e-05,
+      "loss": 0.9089,
+      "step": 2056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4369542438002097e-05,
+      "loss": 0.9672,
+      "step": 2057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4376528117359414e-05,
+      "loss": 0.9939,
+      "step": 2058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4383513796716733e-05,
+      "loss": 0.9255,
+      "step": 2059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.439049947607405e-05,
+      "loss": 0.8739,
+      "step": 2060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4397485155431367e-05,
+      "loss": 0.8469,
+      "step": 2061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4404470834788685e-05,
+      "loss": 1.1228,
+      "step": 2062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4411456514146003e-05,
+      "loss": 1.1902,
+      "step": 2063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4418442193503319e-05,
+      "loss": 0.9889,
+      "step": 2064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4425427872860638e-05,
+      "loss": 1.0055,
+      "step": 2065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4432413552217954e-05,
+      "loss": 0.9034,
+      "step": 2066
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4439399231575272e-05,
+      "loss": 0.9601,
+      "step": 2067
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.444638491093259e-05,
+      "loss": 1.0674,
+      "step": 2068
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4453370590289908e-05,
+      "loss": 1.0422,
+      "step": 2069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4460356269647224e-05,
+      "loss": 0.9518,
+      "step": 2070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.446734194900454e-05,
+      "loss": 1.1396,
+      "step": 2071
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.447432762836186e-05,
+      "loss": 0.842,
+      "step": 2072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4481313307719176e-05,
+      "loss": 1.1409,
+      "step": 2073
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4488298987076495e-05,
+      "loss": 1.1957,
+      "step": 2074
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4495284666433811e-05,
+      "loss": 1.0312,
+      "step": 2075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4502270345791129e-05,
+      "loss": 0.9234,
+      "step": 2076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4509256025148445e-05,
+      "loss": 0.8946,
+      "step": 2077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4516241704505765e-05,
+      "loss": 0.8513,
+      "step": 2078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4523227383863081e-05,
+      "loss": 1.0337,
+      "step": 2079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.45302130632204e-05,
+      "loss": 0.885,
+      "step": 2080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4537198742577717e-05,
+      "loss": 1.027,
+      "step": 2081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4544184421935034e-05,
+      "loss": 0.9763,
+      "step": 2082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.455117010129235e-05,
+      "loss": 1.0716,
+      "step": 2083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.455815578064967e-05,
+      "loss": 0.9305,
+      "step": 2084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4565141460006986e-05,
+      "loss": 1.1248,
+      "step": 2085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4572127139364306e-05,
+      "loss": 0.9423,
+      "step": 2086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4579112818721622e-05,
+      "loss": 0.9544,
+      "step": 2087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.458609849807894e-05,
+      "loss": 0.905,
+      "step": 2088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4593084177436256e-05,
+      "loss": 0.866,
+      "step": 2089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4600069856793575e-05,
+      "loss": 0.8257,
+      "step": 2090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4607055536150891e-05,
+      "loss": 0.9341,
+      "step": 2091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.461404121550821e-05,
+      "loss": 1.1985,
+      "step": 2092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4621026894865527e-05,
+      "loss": 0.9968,
+      "step": 2093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4628012574222845e-05,
+      "loss": 1.0903,
+      "step": 2094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.463499825358016e-05,
+      "loss": 0.8995,
+      "step": 2095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.464198393293748e-05,
+      "loss": 0.7981,
+      "step": 2096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4648969612294796e-05,
+      "loss": 0.879,
+      "step": 2097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4655955291652116e-05,
+      "loss": 0.9336,
+      "step": 2098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4662940971009432e-05,
+      "loss": 0.8542,
+      "step": 2099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.466992665036675e-05,
+      "loss": 1.0579,
+      "step": 2100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4676912329724066e-05,
+      "loss": 1.1485,
+      "step": 2101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4683898009081385e-05,
+      "loss": 1.2083,
+      "step": 2102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4690883688438702e-05,
+      "loss": 1.11,
+      "step": 2103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4697869367796021e-05,
+      "loss": 1.0184,
+      "step": 2104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4704855047153337e-05,
+      "loss": 0.9128,
+      "step": 2105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4711840726510653e-05,
+      "loss": 0.8946,
+      "step": 2106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4718826405867971e-05,
+      "loss": 0.8277,
+      "step": 2107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4725812085225289e-05,
+      "loss": 0.8955,
+      "step": 2108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4732797764582607e-05,
+      "loss": 1.1943,
+      "step": 2109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4739783443939923e-05,
+      "loss": 0.9256,
+      "step": 2110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4746769123297242e-05,
+      "loss": 0.9221,
+      "step": 2111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4753754802654559e-05,
+      "loss": 0.8712,
+      "step": 2112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4760740482011876e-05,
+      "loss": 0.9128,
+      "step": 2113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4767726161369194e-05,
+      "loss": 1.1117,
+      "step": 2114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4774711840726512e-05,
+      "loss": 0.9187,
+      "step": 2115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4781697520083828e-05,
+      "loss": 1.2087,
+      "step": 2116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4788683199441148e-05,
+      "loss": 0.9684,
+      "step": 2117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4795668878798464e-05,
+      "loss": 0.9231,
+      "step": 2118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4802654558155782e-05,
+      "loss": 0.9238,
+      "step": 2119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.48096402375131e-05,
+      "loss": 0.9337,
+      "step": 2120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4816625916870417e-05,
+      "loss": 1.0465,
+      "step": 2121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4823611596227733e-05,
+      "loss": 0.8199,
+      "step": 2122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4830597275585053e-05,
+      "loss": 1.2414,
+      "step": 2123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4837582954942369e-05,
+      "loss": 0.8186,
+      "step": 2124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4844568634299687e-05,
+      "loss": 0.9632,
+      "step": 2125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4851554313657004e-05,
+      "loss": 0.979,
+      "step": 2126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4858539993014322e-05,
+      "loss": 0.9152,
+      "step": 2127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4865525672371638e-05,
+      "loss": 1.0891,
+      "step": 2128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4872511351728958e-05,
+      "loss": 0.8688,
+      "step": 2129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4879497031086274e-05,
+      "loss": 0.899,
+      "step": 2130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4886482710443594e-05,
+      "loss": 0.8952,
+      "step": 2131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.489346838980091e-05,
+      "loss": 0.8569,
+      "step": 2132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4900454069158227e-05,
+      "loss": 1.0557,
+      "step": 2133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4907439748515544e-05,
+      "loss": 1.0579,
+      "step": 2134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4914425427872863e-05,
+      "loss": 1.0294,
+      "step": 2135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.492141110723018e-05,
+      "loss": 0.9976,
+      "step": 2136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4928396786587499e-05,
+      "loss": 0.9712,
+      "step": 2137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4935382465944815e-05,
+      "loss": 0.8307,
+      "step": 2138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4942368145302131e-05,
+      "loss": 0.8014,
+      "step": 2139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4949353824659449e-05,
+      "loss": 1.0643,
+      "step": 2140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4956339504016767e-05,
+      "loss": 1.0043,
+      "step": 2141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4963325183374084e-05,
+      "loss": 1.0577,
+      "step": 2142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.49703108627314e-05,
+      "loss": 0.8675,
+      "step": 2143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.497729654208872e-05,
+      "loss": 1.1369,
+      "step": 2144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4984282221446036e-05,
+      "loss": 0.8572,
+      "step": 2145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4991267900803354e-05,
+      "loss": 1.1583,
+      "step": 2146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4998253580160672e-05,
+      "loss": 1.0841,
+      "step": 2147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.500523925951799e-05,
+      "loss": 0.9588,
+      "step": 2148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5012224938875306e-05,
+      "loss": 1.1178,
+      "step": 2149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5019210618232625e-05,
+      "loss": 0.8983,
+      "step": 2150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5026196297589941e-05,
+      "loss": 0.8855,
+      "step": 2151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5033181976947259e-05,
+      "loss": 1.0041,
+      "step": 2152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5040167656304577e-05,
+      "loss": 1.0097,
+      "step": 2153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5047153335661895e-05,
+      "loss": 0.8864,
+      "step": 2154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5054139015019211e-05,
+      "loss": 0.9101,
+      "step": 2155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.506112469437653e-05,
+      "loss": 1.0554,
+      "step": 2156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5068110373733847e-05,
+      "loss": 1.0881,
+      "step": 2157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5075096053091164e-05,
+      "loss": 1.1569,
+      "step": 2158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5082081732448482e-05,
+      "loss": 0.9259,
+      "step": 2159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.50890674118058e-05,
+      "loss": 0.9098,
+      "step": 2160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5096053091163116e-05,
+      "loss": 0.9246,
+      "step": 2161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5103038770520436e-05,
+      "loss": 1.0138,
+      "step": 2162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5110024449877752e-05,
+      "loss": 0.9591,
+      "step": 2163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.511701012923507e-05,
+      "loss": 0.8716,
+      "step": 2164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5123995808592387e-05,
+      "loss": 0.9208,
+      "step": 2165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5130981487949705e-05,
+      "loss": 0.9863,
+      "step": 2166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5137967167307021e-05,
+      "loss": 0.9858,
+      "step": 2167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.514495284666434e-05,
+      "loss": 1.0493,
+      "step": 2168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5151938526021657e-05,
+      "loss": 0.8372,
+      "step": 2169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5158924205378975e-05,
+      "loss": 0.8965,
+      "step": 2170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5165909884736292e-05,
+      "loss": 0.9455,
+      "step": 2171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.517289556409361e-05,
+      "loss": 0.9331,
+      "step": 2172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5179881243450926e-05,
+      "loss": 1.0299,
+      "step": 2173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5186866922808243e-05,
+      "loss": 0.9444,
+      "step": 2174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5193852602165562e-05,
+      "loss": 1.0641,
+      "step": 2175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5200838281522878e-05,
+      "loss": 0.9713,
+      "step": 2176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5207823960880198e-05,
+      "loss": 1.1086,
+      "step": 2177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5214809640237514e-05,
+      "loss": 0.9929,
+      "step": 2178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5221795319594832e-05,
+      "loss": 1.0381,
+      "step": 2179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5228780998952148e-05,
+      "loss": 1.1502,
+      "step": 2180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5235766678309467e-05,
+      "loss": 0.8429,
+      "step": 2181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5242752357666783e-05,
+      "loss": 0.9165,
+      "step": 2182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5249738037024103e-05,
+      "loss": 1.0833,
+      "step": 2183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5256723716381419e-05,
+      "loss": 0.9267,
+      "step": 2184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.526370939573874e-05,
+      "loss": 0.8544,
+      "step": 2185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5270695075096053e-05,
+      "loss": 1.1834,
+      "step": 2186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.527768075445337e-05,
+      "loss": 0.9554,
+      "step": 2187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.528466643381069e-05,
+      "loss": 0.9071,
+      "step": 2188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5291652113168006e-05,
+      "loss": 0.9453,
+      "step": 2189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5298637792525324e-05,
+      "loss": 1.0378,
+      "step": 2190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5305623471882642e-05,
+      "loss": 0.8982,
+      "step": 2191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.531260915123996e-05,
+      "loss": 1.1432,
+      "step": 2192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5319594830597278e-05,
+      "loss": 0.8178,
+      "step": 2193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5326580509954595e-05,
+      "loss": 0.8993,
+      "step": 2194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5333566189311913e-05,
+      "loss": 1.1376,
+      "step": 2195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5340551868669228e-05,
+      "loss": 0.9901,
+      "step": 2196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.534753754802655e-05,
+      "loss": 1.0286,
+      "step": 2197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5354523227383863e-05,
+      "loss": 0.9971,
+      "step": 2198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.536150890674118e-05,
+      "loss": 0.8826,
+      "step": 2199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.53684945860985e-05,
+      "loss": 1.0454,
+      "step": 2200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5375480265455817e-05,
+      "loss": 1.1235,
+      "step": 2201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5382465944813135e-05,
+      "loss": 0.9722,
+      "step": 2202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5389451624170452e-05,
+      "loss": 0.9414,
+      "step": 2203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.539643730352777e-05,
+      "loss": 0.8865,
+      "step": 2204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5403422982885088e-05,
+      "loss": 0.9133,
+      "step": 2205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5410408662242406e-05,
+      "loss": 0.8084,
+      "step": 2206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5417394341599724e-05,
+      "loss": 0.9517,
+      "step": 2207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5424380020957038e-05,
+      "loss": 0.9914,
+      "step": 2208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5431365700314356e-05,
+      "loss": 0.8923,
+      "step": 2209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5438351379671674e-05,
+      "loss": 1.0487,
+      "step": 2210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.544533705902899e-05,
+      "loss": 0.9963,
+      "step": 2211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.545232273838631e-05,
+      "loss": 0.9111,
+      "step": 2212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5459308417743627e-05,
+      "loss": 0.8357,
+      "step": 2213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5466294097100945e-05,
+      "loss": 0.7329,
+      "step": 2214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.547327977645826e-05,
+      "loss": 0.9833,
+      "step": 2215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.548026545581558e-05,
+      "loss": 0.8965,
+      "step": 2216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5487251135172895e-05,
+      "loss": 0.9067,
+      "step": 2217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5494236814530216e-05,
+      "loss": 0.9744,
+      "step": 2218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.550122249388753e-05,
+      "loss": 0.9051,
+      "step": 2219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.550820817324485e-05,
+      "loss": 0.9217,
+      "step": 2220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5515193852602166e-05,
+      "loss": 1.0308,
+      "step": 2221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5522179531959484e-05,
+      "loss": 0.8746,
+      "step": 2222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5529165211316802e-05,
+      "loss": 0.9553,
+      "step": 2223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.553615089067412e-05,
+      "loss": 0.8369,
+      "step": 2224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5543136570031437e-05,
+      "loss": 0.9373,
+      "step": 2225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5550122249388755e-05,
+      "loss": 1.0121,
+      "step": 2226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5557107928746073e-05,
+      "loss": 1.0615,
+      "step": 2227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.556409360810339e-05,
+      "loss": 1.1449,
+      "step": 2228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5571079287460705e-05,
+      "loss": 1.0102,
+      "step": 2229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5578064966818026e-05,
+      "loss": 0.7906,
+      "step": 2230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.558505064617534e-05,
+      "loss": 0.8568,
+      "step": 2231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.559203632553266e-05,
+      "loss": 0.8775,
+      "step": 2232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5599022004889977e-05,
+      "loss": 0.8794,
+      "step": 2233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5606007684247294e-05,
+      "loss": 1.1121,
+      "step": 2234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5612993363604612e-05,
+      "loss": 0.9232,
+      "step": 2235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.561997904296193e-05,
+      "loss": 1.2208,
+      "step": 2236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5626964722319248e-05,
+      "loss": 0.9262,
+      "step": 2237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5633950401676566e-05,
+      "loss": 0.942,
+      "step": 2238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5640936081033883e-05,
+      "loss": 0.9577,
+      "step": 2239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.56479217603912e-05,
+      "loss": 0.9463,
+      "step": 2240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5654907439748516e-05,
+      "loss": 0.9545,
+      "step": 2241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5661893119105837e-05,
+      "loss": 0.9796,
+      "step": 2242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.566887879846315e-05,
+      "loss": 0.6732,
+      "step": 2243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.567586447782047e-05,
+      "loss": 0.9146,
+      "step": 2244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5682850157177787e-05,
+      "loss": 0.7231,
+      "step": 2245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5689835836535105e-05,
+      "loss": 0.9027,
+      "step": 2246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5696821515892422e-05,
+      "loss": 1.1755,
+      "step": 2247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5703807195249737e-05,
+      "loss": 0.7979,
+      "step": 2248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5710792874607058e-05,
+      "loss": 0.93,
+      "step": 2249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5717778553964373e-05,
+      "loss": 1.0984,
+      "step": 2250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5724764233321694e-05,
+      "loss": 1.0229,
+      "step": 2251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5731749912679008e-05,
+      "loss": 0.893,
+      "step": 2252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5738735592036326e-05,
+      "loss": 0.7087,
+      "step": 2253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5745721271393644e-05,
+      "loss": 0.8513,
+      "step": 2254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.575270695075096e-05,
+      "loss": 0.925,
+      "step": 2255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.575969263010828e-05,
+      "loss": 0.9624,
+      "step": 2256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5766678309465597e-05,
+      "loss": 0.9589,
+      "step": 2257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5773663988822915e-05,
+      "loss": 0.9465,
+      "step": 2258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5780649668180233e-05,
+      "loss": 1.1483,
+      "step": 2259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5787635347537547e-05,
+      "loss": 1.0445,
+      "step": 2260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.579462102689487e-05,
+      "loss": 1.1133,
+      "step": 2261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5801606706252183e-05,
+      "loss": 1.0566,
+      "step": 2262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5808592385609504e-05,
+      "loss": 0.8572,
+      "step": 2263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.581557806496682e-05,
+      "loss": 0.9675,
+      "step": 2264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5822563744324136e-05,
+      "loss": 0.9895,
+      "step": 2265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5829549423681454e-05,
+      "loss": 0.7586,
+      "step": 2266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5836535103038772e-05,
+      "loss": 0.7759,
+      "step": 2267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.584352078239609e-05,
+      "loss": 0.9685,
+      "step": 2268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5850506461753408e-05,
+      "loss": 0.9071,
+      "step": 2269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5857492141110725e-05,
+      "loss": 0.9376,
+      "step": 2270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5864477820468043e-05,
+      "loss": 0.9119,
+      "step": 2271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5871463499825358e-05,
+      "loss": 0.935,
+      "step": 2272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.587844917918268e-05,
+      "loss": 0.8421,
+      "step": 2273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5885434858539993e-05,
+      "loss": 1.1682,
+      "step": 2274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5892420537897314e-05,
+      "loss": 0.9709,
+      "step": 2275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.589940621725463e-05,
+      "loss": 0.9942,
+      "step": 2276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5906391896611947e-05,
+      "loss": 0.7222,
+      "step": 2277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5913377575969265e-05,
+      "loss": 1.0394,
+      "step": 2278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5920363255326582e-05,
+      "loss": 0.7971,
+      "step": 2279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.59273489346839e-05,
+      "loss": 0.8561,
+      "step": 2280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5934334614041215e-05,
+      "loss": 0.9095,
+      "step": 2281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5941320293398536e-05,
+      "loss": 0.8574,
+      "step": 2282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.594830597275585e-05,
+      "loss": 0.9399,
+      "step": 2283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5955291652113168e-05,
+      "loss": 0.9443,
+      "step": 2284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5962277331470486e-05,
+      "loss": 0.8905,
+      "step": 2285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5969263010827804e-05,
+      "loss": 0.9823,
+      "step": 2286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.597624869018512e-05,
+      "loss": 0.9692,
+      "step": 2287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.598323436954244e-05,
+      "loss": 0.9101,
+      "step": 2288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5990220048899757e-05,
+      "loss": 1.1678,
+      "step": 2289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5997205728257075e-05,
+      "loss": 0.9841,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6004191407614393e-05,
+      "loss": 0.9559,
+      "step": 2291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.601117708697171e-05,
+      "loss": 0.9314,
+      "step": 2292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6018162766329025e-05,
+      "loss": 0.9803,
+      "step": 2293
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6025148445686346e-05,
+      "loss": 1.1957,
+      "step": 2294
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.603213412504366e-05,
+      "loss": 0.8612,
+      "step": 2295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.603911980440098e-05,
+      "loss": 1.0196,
+      "step": 2296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6046105483758296e-05,
+      "loss": 0.8599,
+      "step": 2297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6053091163115614e-05,
+      "loss": 0.7823,
+      "step": 2298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6060076842472932e-05,
+      "loss": 0.8309,
+      "step": 2299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.606706252183025e-05,
+      "loss": 0.9899,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6074048201187567e-05,
+      "loss": 0.7189,
+      "step": 2301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6081033880544885e-05,
+      "loss": 1.1128,
+      "step": 2302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6088019559902203e-05,
+      "loss": 0.8944,
+      "step": 2303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.609500523925952e-05,
+      "loss": 0.8903,
+      "step": 2304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6101990918616835e-05,
+      "loss": 1.0213,
+      "step": 2305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6108976597974156e-05,
+      "loss": 0.9912,
+      "step": 2306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.611596227733147e-05,
+      "loss": 0.7911,
+      "step": 2307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.612294795668879e-05,
+      "loss": 0.8554,
+      "step": 2308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6129933636046107e-05,
+      "loss": 1.038,
+      "step": 2309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6136919315403424e-05,
+      "loss": 0.8943,
+      "step": 2310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6143904994760742e-05,
+      "loss": 0.9143,
+      "step": 2311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6150890674118057e-05,
+      "loss": 0.9615,
+      "step": 2312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6157876353475378e-05,
+      "loss": 0.9902,
+      "step": 2313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6164862032832692e-05,
+      "loss": 0.8709,
+      "step": 2314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6171847712190013e-05,
+      "loss": 0.9678,
+      "step": 2315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6178833391547328e-05,
+      "loss": 1.0283,
+      "step": 2316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6185819070904646e-05,
+      "loss": 1.0853,
+      "step": 2317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6192804750261963e-05,
+      "loss": 0.8989,
+      "step": 2318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.619979042961928e-05,
+      "loss": 0.9293,
+      "step": 2319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.62067761089766e-05,
+      "loss": 1.0675,
+      "step": 2320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6213761788333917e-05,
+      "loss": 0.9879,
+      "step": 2321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6220747467691235e-05,
+      "loss": 0.9475,
+      "step": 2322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6227733147048552e-05,
+      "loss": 0.9612,
+      "step": 2323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6234718826405867e-05,
+      "loss": 1.0126,
+      "step": 2324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6241704505763188e-05,
+      "loss": 0.8049,
+      "step": 2325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6248690185120503e-05,
+      "loss": 0.8346,
+      "step": 2326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6255675864477824e-05,
+      "loss": 0.8947,
+      "step": 2327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6262661543835138e-05,
+      "loss": 0.8839,
+      "step": 2328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6269647223192456e-05,
+      "loss": 0.9778,
+      "step": 2329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6276632902549774e-05,
+      "loss": 0.7869,
+      "step": 2330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628361858190709e-05,
+      "loss": 1.1236,
+      "step": 2331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629060426126441e-05,
+      "loss": 0.9191,
+      "step": 2332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6297589940621727e-05,
+      "loss": 0.9158,
+      "step": 2333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6304575619979045e-05,
+      "loss": 0.9217,
+      "step": 2334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6311561299336363e-05,
+      "loss": 1.0133,
+      "step": 2335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.631854697869368e-05,
+      "loss": 1.0124,
+      "step": 2336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6325532658051e-05,
+      "loss": 1.0104,
+      "step": 2337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6332518337408313e-05,
+      "loss": 0.8105,
+      "step": 2338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6339504016765634e-05,
+      "loss": 0.9603,
+      "step": 2339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634648969612295e-05,
+      "loss": 1.0469,
+      "step": 2340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6353475375480266e-05,
+      "loss": 1.0001,
+      "step": 2341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6360461054837584e-05,
+      "loss": 0.9254,
+      "step": 2342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6367446734194902e-05,
+      "loss": 1.0518,
+      "step": 2343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.637443241355222e-05,
+      "loss": 0.8959,
+      "step": 2344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6381418092909538e-05,
+      "loss": 0.8607,
+      "step": 2345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6388403772266855e-05,
+      "loss": 1.0398,
+      "step": 2346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.639538945162417e-05,
+      "loss": 0.6518,
+      "step": 2347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.640237513098149e-05,
+      "loss": 0.8976,
+      "step": 2348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6409360810338805e-05,
+      "loss": 0.7893,
+      "step": 2349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6416346489696123e-05,
+      "loss": 1.0693,
+      "step": 2350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.642333216905344e-05,
+      "loss": 1.03,
+      "step": 2351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.643031784841076e-05,
+      "loss": 0.9278,
+      "step": 2352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6437303527768077e-05,
+      "loss": 0.9158,
+      "step": 2353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6444289207125395e-05,
+      "loss": 0.9108,
+      "step": 2354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6451274886482712e-05,
+      "loss": 0.8077,
+      "step": 2355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.645826056584003e-05,
+      "loss": 1.018,
+      "step": 2356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6465246245197345e-05,
+      "loss": 0.785,
+      "step": 2357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6472231924554666e-05,
+      "loss": 0.8332,
+      "step": 2358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.647921760391198e-05,
+      "loss": 1.1018,
+      "step": 2359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.64862032832693e-05,
+      "loss": 0.8847,
+      "step": 2360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6493188962626616e-05,
+      "loss": 0.9506,
+      "step": 2361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6500174641983934e-05,
+      "loss": 0.8018,
+      "step": 2362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.650716032134125e-05,
+      "loss": 0.8754,
+      "step": 2363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.651414600069857e-05,
+      "loss": 1.0249,
+      "step": 2364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6521131680055887e-05,
+      "loss": 0.9444,
+      "step": 2365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6528117359413205e-05,
+      "loss": 0.957,
+      "step": 2366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6535103038770523e-05,
+      "loss": 0.9312,
+      "step": 2367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.654208871812784e-05,
+      "loss": 0.9256,
+      "step": 2368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6549074397485155e-05,
+      "loss": 0.9338,
+      "step": 2369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6556060076842476e-05,
+      "loss": 0.8055,
+      "step": 2370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.656304575619979e-05,
+      "loss": 1.0491,
+      "step": 2371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6570031435557112e-05,
+      "loss": 0.8816,
+      "step": 2372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6577017114914426e-05,
+      "loss": 1.1273,
+      "step": 2373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6584002794271744e-05,
+      "loss": 0.989,
+      "step": 2374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6590988473629062e-05,
+      "loss": 0.8948,
+      "step": 2375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.659797415298638e-05,
+      "loss": 0.8866,
+      "step": 2376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6604959832343697e-05,
+      "loss": 0.848,
+      "step": 2377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6611945511701015e-05,
+      "loss": 1.038,
+      "step": 2378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6618931191058333e-05,
+      "loss": 0.8838,
+      "step": 2379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.662591687041565e-05,
+      "loss": 0.9579,
+      "step": 2380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6632902549772965e-05,
+      "loss": 0.8364,
+      "step": 2381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6639888229130283e-05,
+      "loss": 0.9802,
+      "step": 2382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.66468739084876e-05,
+      "loss": 0.8458,
+      "step": 2383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.665385958784492e-05,
+      "loss": 1.0139,
+      "step": 2384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6660845267202237e-05,
+      "loss": 0.9641,
+      "step": 2385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6667830946559554e-05,
+      "loss": 0.7917,
+      "step": 2386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6674816625916872e-05,
+      "loss": 1.0625,
+      "step": 2387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.668180230527419e-05,
+      "loss": 0.9993,
+      "step": 2388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6688787984631508e-05,
+      "loss": 1.1218,
+      "step": 2389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6695773663988822e-05,
+      "loss": 0.8218,
+      "step": 2390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6702759343346143e-05,
+      "loss": 0.8396,
+      "step": 2391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6709745022703458e-05,
+      "loss": 0.9124,
+      "step": 2392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6716730702060776e-05,
+      "loss": 0.8545,
+      "step": 2393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6723716381418093e-05,
+      "loss": 0.9439,
+      "step": 2394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.673070206077541e-05,
+      "loss": 0.9207,
+      "step": 2395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.673768774013273e-05,
+      "loss": 1.2031,
+      "step": 2396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6744673419490047e-05,
+      "loss": 0.8091,
+      "step": 2397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6751659098847365e-05,
+      "loss": 0.8525,
+      "step": 2398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6758644778204683e-05,
+      "loss": 0.9296,
+      "step": 2399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6765630457562e-05,
+      "loss": 0.9896,
+      "step": 2400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6772616136919318e-05,
+      "loss": 0.8983,
+      "step": 2401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6779601816276633e-05,
+      "loss": 0.9161,
+      "step": 2402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6786587495633954e-05,
+      "loss": 0.9212,
+      "step": 2403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6793573174991268e-05,
+      "loss": 1.0386,
+      "step": 2404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6800558854348586e-05,
+      "loss": 0.8673,
+      "step": 2405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6807544533705904e-05,
+      "loss": 0.9443,
+      "step": 2406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.681453021306322e-05,
+      "loss": 0.9882,
+      "step": 2407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.682151589242054e-05,
+      "loss": 1.1505,
+      "step": 2408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6828501571777857e-05,
+      "loss": 0.8284,
+      "step": 2409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6835487251135175e-05,
+      "loss": 0.8997,
+      "step": 2410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6842472930492493e-05,
+      "loss": 0.9183,
+      "step": 2411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.684945860984981e-05,
+      "loss": 1.0002,
+      "step": 2412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.685644428920713e-05,
+      "loss": 1.0393,
+      "step": 2413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6863429968564443e-05,
+      "loss": 1.1409,
+      "step": 2414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.687041564792176e-05,
+      "loss": 0.8033,
+      "step": 2415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.687740132727908e-05,
+      "loss": 0.9675,
+      "step": 2416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6884387006636396e-05,
+      "loss": 1.0372,
+      "step": 2417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6891372685993714e-05,
+      "loss": 0.9475,
+      "step": 2418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6898358365351032e-05,
+      "loss": 0.9193,
+      "step": 2419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.690534404470835e-05,
+      "loss": 1.2905,
+      "step": 2420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6912329724065664e-05,
+      "loss": 0.8044,
+      "step": 2421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6919315403422985e-05,
+      "loss": 0.9124,
+      "step": 2422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.69263010827803e-05,
+      "loss": 1.0947,
+      "step": 2423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.693328676213762e-05,
+      "loss": 0.8911,
+      "step": 2424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6940272441494935e-05,
+      "loss": 0.8456,
+      "step": 2425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6947258120852253e-05,
+      "loss": 1.109,
+      "step": 2426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.695424380020957e-05,
+      "loss": 0.907,
+      "step": 2427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.696122947956689e-05,
+      "loss": 0.8436,
+      "step": 2428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6968215158924207e-05,
+      "loss": 1.0603,
+      "step": 2429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6975200838281525e-05,
+      "loss": 0.9141,
+      "step": 2430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6982186517638842e-05,
+      "loss": 0.7396,
+      "step": 2431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.698917219699616e-05,
+      "loss": 0.917,
+      "step": 2432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6996157876353475e-05,
+      "loss": 0.721,
+      "step": 2433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7003143555710796e-05,
+      "loss": 0.9354,
+      "step": 2434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.701012923506811e-05,
+      "loss": 1.0451,
+      "step": 2435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.701711491442543e-05,
+      "loss": 0.9672,
+      "step": 2436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7024100593782746e-05,
+      "loss": 0.8881,
+      "step": 2437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7031086273140064e-05,
+      "loss": 1.1153,
+      "step": 2438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.703807195249738e-05,
+      "loss": 0.9816,
+      "step": 2439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.70450576318547e-05,
+      "loss": 1.0192,
+      "step": 2440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7052043311212017e-05,
+      "loss": 1.0079,
+      "step": 2441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7059028990569335e-05,
+      "loss": 0.7962,
+      "step": 2442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7066014669926653e-05,
+      "loss": 0.9166,
+      "step": 2443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.707300034928397e-05,
+      "loss": 0.8443,
+      "step": 2444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.707998602864129e-05,
+      "loss": 1.0897,
+      "step": 2445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7086971707998606e-05,
+      "loss": 0.9498,
+      "step": 2446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.709395738735592e-05,
+      "loss": 1.0753,
+      "step": 2447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7100943066713242e-05,
+      "loss": 0.9658,
+      "step": 2448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7107928746070556e-05,
+      "loss": 0.8695,
+      "step": 2449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7114914425427874e-05,
+      "loss": 0.8666,
+      "step": 2450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7121900104785192e-05,
+      "loss": 1.0092,
+      "step": 2451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.712888578414251e-05,
+      "loss": 0.9134,
+      "step": 2452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7135871463499827e-05,
+      "loss": 0.7828,
+      "step": 2453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.9141,
+      "step": 2454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7149842822214463e-05,
+      "loss": 0.9441,
+      "step": 2455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7156828501571777e-05,
+      "loss": 1.1522,
+      "step": 2456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.71638141809291e-05,
+      "loss": 0.8607,
+      "step": 2457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7170799860286413e-05,
+      "loss": 0.7988,
+      "step": 2458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.717778553964373e-05,
+      "loss": 0.9208,
+      "step": 2459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.718477121900105e-05,
+      "loss": 0.876,
+      "step": 2460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7191756898358367e-05,
+      "loss": 1.0195,
+      "step": 2461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7198742577715684e-05,
+      "loss": 0.6568,
+      "step": 2462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7205728257073002e-05,
+      "loss": 0.9372,
+      "step": 2463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.721271393643032e-05,
+      "loss": 0.9944,
+      "step": 2464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7219699615787638e-05,
+      "loss": 0.9451,
+      "step": 2465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7226685295144952e-05,
+      "loss": 0.9319,
+      "step": 2466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7233670974502273e-05,
+      "loss": 0.8163,
+      "step": 2467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7240656653859588e-05,
+      "loss": 0.8014,
+      "step": 2468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.724764233321691e-05,
+      "loss": 0.8418,
+      "step": 2469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7254628012574223e-05,
+      "loss": 1.2509,
+      "step": 2470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.726161369193154e-05,
+      "loss": 0.8441,
+      "step": 2471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.726859937128886e-05,
+      "loss": 0.7974,
+      "step": 2472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7275585050646177e-05,
+      "loss": 0.8591,
+      "step": 2473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7282570730003495e-05,
+      "loss": 0.9237,
+      "step": 2474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7289556409360813e-05,
+      "loss": 0.9989,
+      "step": 2475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.729654208871813e-05,
+      "loss": 0.7852,
+      "step": 2476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7303527768075448e-05,
+      "loss": 0.8205,
+      "step": 2477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7310513447432763e-05,
+      "loss": 1.0168,
+      "step": 2478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7317499126790084e-05,
+      "loss": 0.8136,
+      "step": 2479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7324484806147398e-05,
+      "loss": 0.9734,
+      "step": 2480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.733147048550472e-05,
+      "loss": 0.9124,
+      "step": 2481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7338456164862034e-05,
+      "loss": 0.849,
+      "step": 2482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.734544184421935e-05,
+      "loss": 1.2122,
+      "step": 2483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.735242752357667e-05,
+      "loss": 1.0366,
+      "step": 2484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7359413202933987e-05,
+      "loss": 1.0695,
+      "step": 2485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7366398882291305e-05,
+      "loss": 0.8647,
+      "step": 2486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.737338456164862e-05,
+      "loss": 0.958,
+      "step": 2487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.738037024100594e-05,
+      "loss": 0.8228,
+      "step": 2488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7387355920363255e-05,
+      "loss": 0.9784,
+      "step": 2489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7394341599720573e-05,
+      "loss": 0.9016,
+      "step": 2490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.740132727907789e-05,
+      "loss": 0.9331,
+      "step": 2491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.740831295843521e-05,
+      "loss": 0.9425,
+      "step": 2492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7415298637792526e-05,
+      "loss": 0.8829,
+      "step": 2493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7422284317149844e-05,
+      "loss": 1.047,
+      "step": 2494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7429269996507162e-05,
+      "loss": 0.9702,
+      "step": 2495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.743625567586448e-05,
+      "loss": 0.8964,
+      "step": 2496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7443241355221798e-05,
+      "loss": 1.1405,
+      "step": 2497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7450227034579115e-05,
+      "loss": 1.0497,
+      "step": 2498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.745721271393643e-05,
+      "loss": 0.7559,
+      "step": 2499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.746419839329375e-05,
+      "loss": 0.9726,
+      "step": 2500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7471184072651065e-05,
+      "loss": 0.9841,
+      "step": 2501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7478169752008383e-05,
+      "loss": 1.0558,
+      "step": 2502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.74851554313657e-05,
+      "loss": 0.9004,
+      "step": 2503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.749214111072302e-05,
+      "loss": 0.974,
+      "step": 2504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7499126790080337e-05,
+      "loss": 0.7103,
+      "step": 2505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7506112469437655e-05,
+      "loss": 0.9553,
+      "step": 2506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7513098148794972e-05,
+      "loss": 0.9042,
+      "step": 2507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.752008382815229e-05,
+      "loss": 0.7365,
+      "step": 2508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7527069507509608e-05,
+      "loss": 0.9136,
+      "step": 2509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7534055186866926e-05,
+      "loss": 0.953,
+      "step": 2510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.754104086622424e-05,
+      "loss": 1.0103,
+      "step": 2511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.754802654558156e-05,
+      "loss": 1.1098,
+      "step": 2512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7555012224938876e-05,
+      "loss": 1.0258,
+      "step": 2513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7561997904296194e-05,
+      "loss": 0.8989,
+      "step": 2514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.756898358365351e-05,
+      "loss": 1.0059,
+      "step": 2515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.757596926301083e-05,
+      "loss": 1.0014,
+      "step": 2516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7582954942368147e-05,
+      "loss": 0.8503,
+      "step": 2517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7589940621725465e-05,
+      "loss": 0.8516,
+      "step": 2518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7596926301082783e-05,
+      "loss": 0.744,
+      "step": 2519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7603911980440097e-05,
+      "loss": 0.8298,
+      "step": 2520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.761089765979742e-05,
+      "loss": 0.8761,
+      "step": 2521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7617883339154733e-05,
+      "loss": 0.8601,
+      "step": 2522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.762486901851205e-05,
+      "loss": 0.9648,
+      "step": 2523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.763185469786937e-05,
+      "loss": 1.2971,
+      "step": 2524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7638840377226686e-05,
+      "loss": 0.8134,
+      "step": 2525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7645826056584004e-05,
+      "loss": 1.02,
+      "step": 2526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7652811735941322e-05,
+      "loss": 1.1908,
+      "step": 2527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.765979741529864e-05,
+      "loss": 0.887,
+      "step": 2528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7666783094655957e-05,
+      "loss": 0.931,
+      "step": 2529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7673768774013272e-05,
+      "loss": 0.8405,
+      "step": 2530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7680754453370593e-05,
+      "loss": 0.945,
+      "step": 2531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7687740132727907e-05,
+      "loss": 0.9403,
+      "step": 2532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.769472581208523e-05,
+      "loss": 0.9577,
+      "step": 2533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7701711491442543e-05,
+      "loss": 1.0043,
+      "step": 2534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.770869717079986e-05,
+      "loss": 1.257,
+      "step": 2535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.771568285015718e-05,
+      "loss": 0.9345,
+      "step": 2536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7722668529514497e-05,
+      "loss": 0.8254,
+      "step": 2537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7729654208871814e-05,
+      "loss": 0.9997,
+      "step": 2538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7736639888229132e-05,
+      "loss": 0.7122,
+      "step": 2539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.774362556758645e-05,
+      "loss": 0.8062,
+      "step": 2540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7750611246943768e-05,
+      "loss": 0.9328,
+      "step": 2541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7757596926301082e-05,
+      "loss": 1.0081,
+      "step": 2542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7764582605658403e-05,
+      "loss": 1.0134,
+      "step": 2543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7771568285015718e-05,
+      "loss": 0.9765,
+      "step": 2544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.777855396437304e-05,
+      "loss": 0.8905,
+      "step": 2545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7785539643730353e-05,
+      "loss": 1.0939,
+      "step": 2546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.779252532308767e-05,
+      "loss": 0.7575,
+      "step": 2547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.779951100244499e-05,
+      "loss": 1.0929,
+      "step": 2548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7806496681802307e-05,
+      "loss": 0.8689,
+      "step": 2549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7813482361159625e-05,
+      "loss": 1.0038,
+      "step": 2550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7820468040516943e-05,
+      "loss": 0.9786,
+      "step": 2551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.782745371987426e-05,
+      "loss": 0.7902,
+      "step": 2552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7834439399231575e-05,
+      "loss": 0.9392,
+      "step": 2553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7841425078588896e-05,
+      "loss": 0.7647,
+      "step": 2554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.784841075794621e-05,
+      "loss": 0.8846,
+      "step": 2555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7855396437303528e-05,
+      "loss": 0.8237,
+      "step": 2556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7862382116660846e-05,
+      "loss": 0.9746,
+      "step": 2557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7869367796018164e-05,
+      "loss": 0.8829,
+      "step": 2558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.787635347537548e-05,
+      "loss": 1.0753,
+      "step": 2559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.78833391547328e-05,
+      "loss": 1.1609,
+      "step": 2560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7890324834090117e-05,
+      "loss": 0.836,
+      "step": 2561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7897310513447435e-05,
+      "loss": 1.0712,
+      "step": 2562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.790429619280475e-05,
+      "loss": 0.8608,
+      "step": 2563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.791128187216207e-05,
+      "loss": 0.7564,
+      "step": 2564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7918267551519385e-05,
+      "loss": 0.8994,
+      "step": 2565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7925253230876706e-05,
+      "loss": 0.8491,
+      "step": 2566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.793223891023402e-05,
+      "loss": 0.9804,
+      "step": 2567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.793922458959134e-05,
+      "loss": 0.873,
+      "step": 2568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7946210268948656e-05,
+      "loss": 0.999,
+      "step": 2569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7953195948305974e-05,
+      "loss": 0.8908,
+      "step": 2570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7960181627663292e-05,
+      "loss": 0.9985,
+      "step": 2571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.796716730702061e-05,
+      "loss": 0.9113,
+      "step": 2572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7974152986377928e-05,
+      "loss": 0.8193,
+      "step": 2573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7981138665735245e-05,
+      "loss": 0.8476,
+      "step": 2574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.798812434509256e-05,
+      "loss": 0.85,
+      "step": 2575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.799511002444988e-05,
+      "loss": 0.8121,
+      "step": 2576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8002095703807195e-05,
+      "loss": 0.8367,
+      "step": 2577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8009081383164517e-05,
+      "loss": 0.866,
+      "step": 2578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801606706252183e-05,
+      "loss": 0.8312,
+      "step": 2579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802305274187915e-05,
+      "loss": 0.9736,
+      "step": 2580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030038421236467e-05,
+      "loss": 0.9664,
+      "step": 2581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037024100593785e-05,
+      "loss": 0.8667,
+      "step": 2582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8044009779951102e-05,
+      "loss": 0.8719,
+      "step": 2583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805099545930842e-05,
+      "loss": 1.0157,
+      "step": 2584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8057981138665738e-05,
+      "loss": 0.8922,
+      "step": 2585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8064966818023056e-05,
+      "loss": 0.8374,
+      "step": 2586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.807195249738037e-05,
+      "loss": 0.8917,
+      "step": 2587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8078938176737688e-05,
+      "loss": 0.8587,
+      "step": 2588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8085923856095006e-05,
+      "loss": 0.7796,
+      "step": 2589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8092909535452324e-05,
+      "loss": 0.7786,
+      "step": 2590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.809989521480964e-05,
+      "loss": 0.8496,
+      "step": 2591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.810688089416696e-05,
+      "loss": 0.9464,
+      "step": 2592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8113866573524277e-05,
+      "loss": 1.0406,
+      "step": 2593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8120852252881595e-05,
+      "loss": 1.031,
+      "step": 2594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8127837932238913e-05,
+      "loss": 0.7873,
+      "step": 2595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8134823611596227e-05,
+      "loss": 1.0256,
+      "step": 2596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.814180929095355e-05,
+      "loss": 0.7608,
+      "step": 2597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8148794970310863e-05,
+      "loss": 0.8322,
+      "step": 2598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.815578064966818e-05,
+      "loss": 0.8935,
+      "step": 2599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.81627663290255e-05,
+      "loss": 0.9545,
+      "step": 2600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8169752008382816e-05,
+      "loss": 0.9542,
+      "step": 2601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8176737687740134e-05,
+      "loss": 0.9325,
+      "step": 2602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8183723367097452e-05,
+      "loss": 0.9344,
+      "step": 2603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.819070904645477e-05,
+      "loss": 0.7275,
+      "step": 2604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8197694725812087e-05,
+      "loss": 0.7673,
+      "step": 2605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8204680405169405e-05,
+      "loss": 0.8682,
+      "step": 2606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8211666084526723e-05,
+      "loss": 0.9895,
+      "step": 2607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8218651763884037e-05,
+      "loss": 0.8291,
+      "step": 2608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.822563744324136e-05,
+      "loss": 0.9603,
+      "step": 2609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8232623122598673e-05,
+      "loss": 0.9517,
+      "step": 2610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.823960880195599e-05,
+      "loss": 0.9838,
+      "step": 2611
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.824659448131331e-05,
+      "loss": 0.7107,
+      "step": 2612
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8253580160670627e-05,
+      "loss": 0.8315,
+      "step": 2613
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8260565840027944e-05,
+      "loss": 0.9254,
+      "step": 2614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8267551519385262e-05,
+      "loss": 1.0074,
+      "step": 2615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.827453719874258e-05,
+      "loss": 0.9729,
+      "step": 2616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8281522878099898e-05,
+      "loss": 0.8193,
+      "step": 2617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8288508557457216e-05,
+      "loss": 0.7222,
+      "step": 2618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8295494236814533e-05,
+      "loss": 0.9538,
+      "step": 2619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8302479916171848e-05,
+      "loss": 1.0389,
+      "step": 2620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.830946559552917e-05,
+      "loss": 0.9608,
+      "step": 2621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8316451274886483e-05,
+      "loss": 0.9788,
+      "step": 2622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.83234369542438e-05,
+      "loss": 0.8732,
+      "step": 2623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.833042263360112e-05,
+      "loss": 0.8933,
+      "step": 2624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8337408312958437e-05,
+      "loss": 1.0387,
+      "step": 2625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8344393992315755e-05,
+      "loss": 0.8681,
+      "step": 2626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.835137967167307e-05,
+      "loss": 0.9288,
+      "step": 2627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.835836535103039e-05,
+      "loss": 0.8898,
+      "step": 2628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8365351030387705e-05,
+      "loss": 0.9083,
+      "step": 2629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8372336709745026e-05,
+      "loss": 0.8826,
+      "step": 2630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.837932238910234e-05,
+      "loss": 0.9513,
+      "step": 2631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8386308068459658e-05,
+      "loss": 0.7785,
+      "step": 2632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8393293747816976e-05,
+      "loss": 0.8187,
+      "step": 2633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8400279427174294e-05,
+      "loss": 0.8217,
+      "step": 2634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.840726510653161e-05,
+      "loss": 1.1253,
+      "step": 2635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.841425078588893e-05,
+      "loss": 0.759,
+      "step": 2636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8421236465246247e-05,
+      "loss": 1.0305,
+      "step": 2637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8428222144603565e-05,
+      "loss": 0.7421,
+      "step": 2638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.843520782396088e-05,
+      "loss": 0.9293,
+      "step": 2639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.84421935033182e-05,
+      "loss": 0.6587,
+      "step": 2640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8449179182675515e-05,
+      "loss": 0.826,
+      "step": 2641
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8456164862032836e-05,
+      "loss": 0.9407,
+      "step": 2642
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.846315054139015e-05,
+      "loss": 0.8789,
+      "step": 2643
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.847013622074747e-05,
+      "loss": 0.8372,
+      "step": 2644
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8477121900104786e-05,
+      "loss": 1.007,
+      "step": 2645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8484107579462104e-05,
+      "loss": 0.9428,
+      "step": 2646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8491093258819422e-05,
+      "loss": 0.784,
+      "step": 2647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.849807893817674e-05,
+      "loss": 1.0504,
+      "step": 2648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8505064617534058e-05,
+      "loss": 0.8113,
+      "step": 2649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8512050296891375e-05,
+      "loss": 0.7734,
+      "step": 2650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8519035976248693e-05,
+      "loss": 0.8592,
+      "step": 2651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.852602165560601e-05,
+      "loss": 0.9181,
+      "step": 2652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8533007334963325e-05,
+      "loss": 0.7936,
+      "step": 2653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8539993014320647e-05,
+      "loss": 0.7898,
+      "step": 2654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.854697869367796e-05,
+      "loss": 0.7256,
+      "step": 2655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.855396437303528e-05,
+      "loss": 0.9396,
+      "step": 2656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8560950052392597e-05,
+      "loss": 1.0459,
+      "step": 2657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8567935731749915e-05,
+      "loss": 0.8495,
+      "step": 2658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8574921411107232e-05,
+      "loss": 0.7849,
+      "step": 2659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8581907090464547e-05,
+      "loss": 0.8418,
+      "step": 2660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8588892769821868e-05,
+      "loss": 0.8431,
+      "step": 2661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8595878449179182e-05,
+      "loss": 0.8823,
+      "step": 2662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8602864128536504e-05,
+      "loss": 0.8484,
+      "step": 2663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8609849807893818e-05,
+      "loss": 0.9379,
+      "step": 2664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8616835487251136e-05,
+      "loss": 0.8137,
+      "step": 2665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8623821166608454e-05,
+      "loss": 0.8271,
+      "step": 2666
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.863080684596577e-05,
+      "loss": 0.7238,
+      "step": 2667
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.863779252532309e-05,
+      "loss": 0.9286,
+      "step": 2668
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8644778204680407e-05,
+      "loss": 0.8361,
+      "step": 2669
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8651763884037725e-05,
+      "loss": 0.8819,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8658749563395043e-05,
+      "loss": 0.8923,
+      "step": 2671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8665735242752357e-05,
+      "loss": 1.1812,
+      "step": 2672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.867272092210968e-05,
+      "loss": 1.0171,
+      "step": 2673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8679706601466993e-05,
+      "loss": 0.9213,
+      "step": 2674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8686692280824314e-05,
+      "loss": 1.2219,
+      "step": 2675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.869367796018163e-05,
+      "loss": 0.8744,
+      "step": 2676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8700663639538946e-05,
+      "loss": 0.9625,
+      "step": 2677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8707649318896264e-05,
+      "loss": 0.8417,
+      "step": 2678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8714634998253582e-05,
+      "loss": 0.8391,
+      "step": 2679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.87216206776109e-05,
+      "loss": 0.8763,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8728606356968217e-05,
+      "loss": 0.7948,
+      "step": 2681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8735592036325535e-05,
+      "loss": 0.9,
+      "step": 2682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8742577715682853e-05,
+      "loss": 1.0221,
+      "step": 2683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8749563395040167e-05,
+      "loss": 0.8967,
+      "step": 2684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.875654907439749e-05,
+      "loss": 0.9053,
+      "step": 2685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8763534753754803e-05,
+      "loss": 1.1107,
+      "step": 2686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8770520433112124e-05,
+      "loss": 0.9668,
+      "step": 2687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.877750611246944e-05,
+      "loss": 0.894,
+      "step": 2688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8784491791826757e-05,
+      "loss": 0.936,
+      "step": 2689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8791477471184074e-05,
+      "loss": 0.8672,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8798463150541392e-05,
+      "loss": 0.7134,
+      "step": 2691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.880544882989871e-05,
+      "loss": 0.9701,
+      "step": 2692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8812434509256024e-05,
+      "loss": 0.9069,
+      "step": 2693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8819420188613346e-05,
+      "loss": 0.9043,
+      "step": 2694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.882640586797066e-05,
+      "loss": 1.0907,
+      "step": 2695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8833391547327978e-05,
+      "loss": 0.99,
+      "step": 2696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8840377226685296e-05,
+      "loss": 0.8459,
+      "step": 2697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8847362906042613e-05,
+      "loss": 1.0868,
+      "step": 2698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.885434858539993e-05,
+      "loss": 0.7584,
+      "step": 2699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.886133426475725e-05,
+      "loss": 0.8516,
+      "step": 2700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8868319944114567e-05,
+      "loss": 0.9301,
+      "step": 2701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8875305623471885e-05,
+      "loss": 0.9975,
+      "step": 2702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8882291302829203e-05,
+      "loss": 0.8486,
+      "step": 2703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.888927698218652e-05,
+      "loss": 0.8305,
+      "step": 2704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8896262661543835e-05,
+      "loss": 1.1364,
+      "step": 2705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8903248340901156e-05,
+      "loss": 0.8401,
+      "step": 2706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.891023402025847e-05,
+      "loss": 0.8813,
+      "step": 2707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8917219699615788e-05,
+      "loss": 0.9249,
+      "step": 2708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8924205378973106e-05,
+      "loss": 0.8602,
+      "step": 2709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8931191058330424e-05,
+      "loss": 0.8413,
+      "step": 2710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.893817673768774e-05,
+      "loss": 0.8734,
+      "step": 2711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.894516241704506e-05,
+      "loss": 0.6723,
+      "step": 2712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8952148096402377e-05,
+      "loss": 0.822,
+      "step": 2713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8959133775759695e-05,
+      "loss": 0.9356,
+      "step": 2714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8966119455117013e-05,
+      "loss": 0.9448,
+      "step": 2715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.897310513447433e-05,
+      "loss": 0.8497,
+      "step": 2716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8980090813831645e-05,
+      "loss": 0.8443,
+      "step": 2717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8987076493188966e-05,
+      "loss": 1.0001,
+      "step": 2718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.899406217254628e-05,
+      "loss": 0.798,
+      "step": 2719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.90010478519036e-05,
+      "loss": 0.7786,
+      "step": 2720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9008033531260916e-05,
+      "loss": 1.1532,
+      "step": 2721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9015019210618234e-05,
+      "loss": 0.8448,
+      "step": 2722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9022004889975552e-05,
+      "loss": 0.9118,
+      "step": 2723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.902899056933287e-05,
+      "loss": 0.8587,
+      "step": 2724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9035976248690188e-05,
+      "loss": 0.9603,
+      "step": 2725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9042961928047502e-05,
+      "loss": 0.8375,
+      "step": 2726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9049947607404823e-05,
+      "loss": 0.9179,
+      "step": 2727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9056933286762138e-05,
+      "loss": 1.0042,
+      "step": 2728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9063918966119455e-05,
+      "loss": 1.0237,
+      "step": 2729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9070904645476773e-05,
+      "loss": 1.0467,
+      "step": 2730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.907789032483409e-05,
+      "loss": 0.9742,
+      "step": 2731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.908487600419141e-05,
+      "loss": 0.9982,
+      "step": 2732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9091861683548727e-05,
+      "loss": 0.9217,
+      "step": 2733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9098847362906045e-05,
+      "loss": 0.8167,
+      "step": 2734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9105833042263362e-05,
+      "loss": 0.8505,
+      "step": 2735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9112818721620677e-05,
+      "loss": 0.9368,
+      "step": 2736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9119804400977998e-05,
+      "loss": 1.028,
+      "step": 2737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9126790080335312e-05,
+      "loss": 0.9855,
+      "step": 2738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9133775759692634e-05,
+      "loss": 0.6985,
+      "step": 2739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9140761439049948e-05,
+      "loss": 0.9402,
+      "step": 2740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9147747118407266e-05,
+      "loss": 0.7885,
+      "step": 2741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9154732797764584e-05,
+      "loss": 0.9504,
+      "step": 2742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.91617184771219e-05,
+      "loss": 0.9076,
+      "step": 2743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.916870415647922e-05,
+      "loss": 0.8917,
+      "step": 2744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9175689835836537e-05,
+      "loss": 0.9434,
+      "step": 2745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9182675515193855e-05,
+      "loss": 0.756,
+      "step": 2746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9189661194551173e-05,
+      "loss": 0.8557,
+      "step": 2747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9196646873908487e-05,
+      "loss": 1.0243,
+      "step": 2748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.920363255326581e-05,
+      "loss": 0.756,
+      "step": 2749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9210618232623123e-05,
+      "loss": 0.7825,
+      "step": 2750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9217603911980444e-05,
+      "loss": 0.9209,
+      "step": 2751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.922458959133776e-05,
+      "loss": 0.722,
+      "step": 2752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9231575270695076e-05,
+      "loss": 0.9341,
+      "step": 2753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9238560950052394e-05,
+      "loss": 0.9718,
+      "step": 2754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9245546629409712e-05,
+      "loss": 1.0149,
+      "step": 2755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.925253230876703e-05,
+      "loss": 0.8747,
+      "step": 2756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9259517988124347e-05,
+      "loss": 1.1145,
+      "step": 2757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9266503667481665e-05,
+      "loss": 0.9497,
+      "step": 2758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9273489346838983e-05,
+      "loss": 0.8919,
+      "step": 2759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.92804750261963e-05,
+      "loss": 1.0226,
+      "step": 2760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9287460705553615e-05,
+      "loss": 0.8852,
+      "step": 2761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9294446384910933e-05,
+      "loss": 0.8251,
+      "step": 2762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.930143206426825e-05,
+      "loss": 0.7083,
+      "step": 2763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.930841774362557e-05,
+      "loss": 1.0907,
+      "step": 2764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315403422982887e-05,
+      "loss": 0.8219,
+      "step": 2765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9322389102340204e-05,
+      "loss": 0.8435,
+      "step": 2766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329374781697522e-05,
+      "loss": 0.9056,
+      "step": 2767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.933636046105484e-05,
+      "loss": 0.8545,
+      "step": 2768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9343346140412154e-05,
+      "loss": 0.7598,
+      "step": 2769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9350331819769476e-05,
+      "loss": 0.8648,
+      "step": 2770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.935731749912679e-05,
+      "loss": 0.9574,
+      "step": 2771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.936430317848411e-05,
+      "loss": 0.9193,
+      "step": 2772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9371288857841426e-05,
+      "loss": 0.7608,
+      "step": 2773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9378274537198743e-05,
+      "loss": 0.9479,
+      "step": 2774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.938526021655606e-05,
+      "loss": 1.0517,
+      "step": 2775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.939224589591338e-05,
+      "loss": 0.8464,
+      "step": 2776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9399231575270697e-05,
+      "loss": 0.8445,
+      "step": 2777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9406217254628015e-05,
+      "loss": 0.958,
+      "step": 2778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9413202933985333e-05,
+      "loss": 0.8113,
+      "step": 2779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.942018861334265e-05,
+      "loss": 0.9576,
+      "step": 2780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9427174292699965e-05,
+      "loss": 0.9036,
+      "step": 2781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9434159972057286e-05,
+      "loss": 0.8628,
+      "step": 2782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.94411456514146e-05,
+      "loss": 0.9093,
+      "step": 2783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.944813133077192e-05,
+      "loss": 0.8242,
+      "step": 2784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9455117010129236e-05,
+      "loss": 0.9248,
+      "step": 2785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9462102689486554e-05,
+      "loss": 0.6811,
+      "step": 2786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.946908836884387e-05,
+      "loss": 0.7633,
+      "step": 2787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.947607404820119e-05,
+      "loss": 0.9027,
+      "step": 2788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9483059727558507e-05,
+      "loss": 0.8938,
+      "step": 2789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9490045406915825e-05,
+      "loss": 0.9531,
+      "step": 2790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9497031086273143e-05,
+      "loss": 0.8209,
+      "step": 2791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.950401676563046e-05,
+      "loss": 0.8595,
+      "step": 2792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9511002444987775e-05,
+      "loss": 0.7591,
+      "step": 2793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9517988124345096e-05,
+      "loss": 1.0327,
+      "step": 2794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.952497380370241e-05,
+      "loss": 0.7643,
+      "step": 2795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.953195948305973e-05,
+      "loss": 0.7925,
+      "step": 2796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9538945162417046e-05,
+      "loss": 0.7798,
+      "step": 2797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9545930841774364e-05,
+      "loss": 0.7856,
+      "step": 2798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9552916521131682e-05,
+      "loss": 0.9514,
+      "step": 2799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9559902200489e-05,
+      "loss": 0.8151,
+      "step": 2800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9566887879846318e-05,
+      "loss": 0.9479,
+      "step": 2801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9573873559203632e-05,
+      "loss": 0.8355,
+      "step": 2802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9580859238560953e-05,
+      "loss": 0.906,
+      "step": 2803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9587844917918268e-05,
+      "loss": 0.8575,
+      "step": 2804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9594830597275585e-05,
+      "loss": 0.7653,
+      "step": 2805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9601816276632903e-05,
+      "loss": 0.9879,
+      "step": 2806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.960880195599022e-05,
+      "loss": 1.0,
+      "step": 2807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.961578763534754e-05,
+      "loss": 0.818,
+      "step": 2808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9622773314704857e-05,
+      "loss": 0.8111,
+      "step": 2809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9629758994062175e-05,
+      "loss": 0.967,
+      "step": 2810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9636744673419492e-05,
+      "loss": 0.8543,
+      "step": 2811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.964373035277681e-05,
+      "loss": 0.8419,
+      "step": 2812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9650716032134128e-05,
+      "loss": 0.7228,
+      "step": 2813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9657701711491442e-05,
+      "loss": 0.8093,
+      "step": 2814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9664687390848764e-05,
+      "loss": 0.9412,
+      "step": 2815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9671673070206078e-05,
+      "loss": 0.9106,
+      "step": 2816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9678658749563396e-05,
+      "loss": 0.7721,
+      "step": 2817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9685644428920714e-05,
+      "loss": 0.8408,
+      "step": 2818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.969263010827803e-05,
+      "loss": 0.8882,
+      "step": 2819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.969961578763535e-05,
+      "loss": 0.903,
+      "step": 2820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9706601466992667e-05,
+      "loss": 0.9404,
+      "step": 2821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9713587146349985e-05,
+      "loss": 0.8791,
+      "step": 2822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9720572825707303e-05,
+      "loss": 0.855,
+      "step": 2823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.972755850506462e-05,
+      "loss": 0.9824,
+      "step": 2824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.973454418442194e-05,
+      "loss": 1.0503,
+      "step": 2825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9741529863779253e-05,
+      "loss": 1.0127,
+      "step": 2826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9748515543136574e-05,
+      "loss": 1.0268,
+      "step": 2827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.975550122249389e-05,
+      "loss": 1.0205,
+      "step": 2828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9762486901851206e-05,
+      "loss": 0.6864,
+      "step": 2829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9769472581208524e-05,
+      "loss": 0.726,
+      "step": 2830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9776458260565842e-05,
+      "loss": 0.8573,
+      "step": 2831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.978344393992316e-05,
+      "loss": 1.0642,
+      "step": 2832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9790429619280474e-05,
+      "loss": 0.7985,
+      "step": 2833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9797415298637795e-05,
+      "loss": 0.699,
+      "step": 2834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.980440097799511e-05,
+      "loss": 0.9945,
+      "step": 2835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.981138665735243e-05,
+      "loss": 0.9485,
+      "step": 2836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9818372336709745e-05,
+      "loss": 1.0347,
+      "step": 2837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9825358016067063e-05,
+      "loss": 0.855,
+      "step": 2838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.983234369542438e-05,
+      "loss": 1.1487,
+      "step": 2839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.98393293747817e-05,
+      "loss": 0.7844,
+      "step": 2840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9846315054139017e-05,
+      "loss": 0.9339,
+      "step": 2841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9853300733496334e-05,
+      "loss": 0.8661,
+      "step": 2842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9860286412853652e-05,
+      "loss": 1.0204,
+      "step": 2843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.986727209221097e-05,
+      "loss": 0.8238,
+      "step": 2844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9874257771568284e-05,
+      "loss": 1.0459,
+      "step": 2845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9881243450925606e-05,
+      "loss": 0.9648,
+      "step": 2846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.988822913028292e-05,
+      "loss": 0.6355,
+      "step": 2847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.989521480964024e-05,
+      "loss": 0.6523,
+      "step": 2848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9902200488997556e-05,
+      "loss": 0.9531,
+      "step": 2849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9909186168354873e-05,
+      "loss": 0.9089,
+      "step": 2850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.991617184771219e-05,
+      "loss": 0.906,
+      "step": 2851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.992315752706951e-05,
+      "loss": 0.8568,
+      "step": 2852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9930143206426827e-05,
+      "loss": 0.8547,
+      "step": 2853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9937128885784145e-05,
+      "loss": 0.9632,
+      "step": 2854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9944114565141463e-05,
+      "loss": 0.8129,
+      "step": 2855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.995110024449878e-05,
+      "loss": 0.8811,
+      "step": 2856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9958085923856095e-05,
+      "loss": 0.8769,
+      "step": 2857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9965071603213416e-05,
+      "loss": 0.8693,
+      "step": 2858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.997205728257073e-05,
+      "loss": 0.7416,
+      "step": 2859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.997904296192805e-05,
+      "loss": 0.8671,
+      "step": 2860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9986028641285366e-05,
+      "loss": 0.9267,
+      "step": 2861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9993014320642684e-05,
+      "loss": 0.8566,
+      "step": 2862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.8281,
+      "step": 2863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999999994240245e-05,
+      "loss": 0.7701,
+      "step": 2864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999999976960966e-05,
+      "loss": 0.7463,
+      "step": 2865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999999948162175e-05,
+      "loss": 0.9199,
+      "step": 2866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999999990784386e-05,
+      "loss": 0.8614,
+      "step": 2867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999999856006034e-05,
+      "loss": 0.9113,
+      "step": 2868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999999792648688e-05,
+      "loss": 0.8545,
+      "step": 2869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999999971777183e-05,
+      "loss": 0.9585,
+      "step": 2870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999999631375447e-05,
+      "loss": 0.7884,
+      "step": 2871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999999953345955e-05,
+      "loss": 0.8202,
+      "step": 2872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999999424024137e-05,
+      "loss": 0.8345,
+      "step": 2873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999999303069207e-05,
+      "loss": 0.7527,
+      "step": 2874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999999917059476e-05,
+      "loss": 0.871,
+      "step": 2875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999999026600798e-05,
+      "loss": 0.8887,
+      "step": 2876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999998871087317e-05,
+      "loss": 1.053,
+      "step": 2877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999999870405432e-05,
+      "loss": 0.8391,
+      "step": 2878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999998525501808e-05,
+      "loss": 0.8467,
+      "step": 2879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999998335429783e-05,
+      "loss": 0.8868,
+      "step": 2880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.999999813383824e-05,
+      "loss": 0.7661,
+      "step": 2881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9999997920727182e-05,
+      "loss": 0.7187,
+      "step": 2882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999999769609661e-05,
+      "loss": 1.0012,
+      "step": 2883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999997459946517e-05,
+      "loss": 0.9841,
+      "step": 2884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999999721227692e-05,
+      "loss": 0.9589,
+      "step": 2885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.99999969530878e-05,
+      "loss": 1.2283,
+      "step": 2886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999999668237917e-05,
+      "loss": 1.1199,
+      "step": 2887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999996400151026e-05,
+      "loss": 0.8228,
+      "step": 2888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999999610640337e-05,
+      "loss": 0.8539,
+      "step": 2889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999995801136197e-05,
+      "loss": 0.6324,
+      "step": 2890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999995484349517e-05,
+      "loss": 0.8044,
+      "step": 2891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999999515604332e-05,
+      "loss": 0.8048,
+      "step": 2892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999994816217613e-05,
+      "loss": 0.8428,
+      "step": 2893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999994464872395e-05,
+      "loss": 1.1157,
+      "step": 2894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999994102007668e-05,
+      "loss": 0.7931,
+      "step": 2895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999993727623425e-05,
+      "loss": 1.0402,
+      "step": 2896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999993341719673e-05,
+      "loss": 0.8566,
+      "step": 2897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999992944296416e-05,
+      "loss": 0.8349,
+      "step": 2898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999999253535365e-05,
+      "loss": 0.8626,
+      "step": 2899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999999211489137e-05,
+      "loss": 1.0878,
+      "step": 2900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999999168290958e-05,
+      "loss": 0.9839,
+      "step": 2901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999991239408286e-05,
+      "loss": 1.0484,
+      "step": 2902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999990784387486e-05,
+      "loss": 0.9722,
+      "step": 2903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999999031784718e-05,
+      "loss": 0.9822,
+      "step": 2904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999989839787365e-05,
+      "loss": 0.8332,
+      "step": 2905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999989350208044e-05,
+      "loss": 0.8035,
+      "step": 2906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999988849109217e-05,
+      "loss": 0.8864,
+      "step": 2907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999988336490888e-05,
+      "loss": 0.9687,
+      "step": 2908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999987812353053e-05,
+      "loss": 0.7308,
+      "step": 2909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999987276695716e-05,
+      "loss": 0.9108,
+      "step": 2910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999986729518876e-05,
+      "loss": 0.9037,
+      "step": 2911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999986170822537e-05,
+      "loss": 0.7308,
+      "step": 2912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999998560060669e-05,
+      "loss": 0.9052,
+      "step": 2913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999985018871346e-05,
+      "loss": 0.8786,
+      "step": 2914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999984425616503e-05,
+      "loss": 0.8457,
+      "step": 2915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.999998382084216e-05,
+      "loss": 0.8604,
+      "step": 2916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999983204548314e-05,
+      "loss": 0.8922,
+      "step": 2917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999982576734975e-05,
+      "loss": 0.8977,
+      "step": 2918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999981937402136e-05,
+      "loss": 0.911,
+      "step": 2919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9999981286549802e-05,
+      "loss": 0.9564,
+      "step": 2920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999998062417797e-05,
+      "loss": 0.9225,
+      "step": 2921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999979950286644e-05,
+      "loss": 0.75,
+      "step": 2922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999979264875823e-05,
+      "loss": 0.8869,
+      "step": 2923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999997856794551e-05,
+      "loss": 0.8093,
+      "step": 2924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999977859495704e-05,
+      "loss": 0.7039,
+      "step": 2925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999977139526407e-05,
+      "loss": 0.8069,
+      "step": 2926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999976408037617e-05,
+      "loss": 0.9909,
+      "step": 2927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999975665029338e-05,
+      "loss": 0.5941,
+      "step": 2928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999974910501567e-05,
+      "loss": 0.8476,
+      "step": 2929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999997414445431e-05,
+      "loss": 0.9902,
+      "step": 2930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999973366887565e-05,
+      "loss": 0.7208,
+      "step": 2931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999972577801335e-05,
+      "loss": 0.8603,
+      "step": 2932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999971777195615e-05,
+      "loss": 0.8675,
+      "step": 2933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999970965070413e-05,
+      "loss": 0.853,
+      "step": 2934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999970141425726e-05,
+      "loss": 1.0775,
+      "step": 2935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999969306261557e-05,
+      "loss": 1.1076,
+      "step": 2936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999968459577906e-05,
+      "loss": 0.9111,
+      "step": 2937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999967601374773e-05,
+      "loss": 0.8802,
+      "step": 2938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999996673165216e-05,
+      "loss": 0.726,
+      "step": 2939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999965850410067e-05,
+      "loss": 0.8194,
+      "step": 2940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999964957648498e-05,
+      "loss": 0.8313,
+      "step": 2941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999996405336745e-05,
+      "loss": 0.7755,
+      "step": 2942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999963137566926e-05,
+      "loss": 0.8838,
+      "step": 2943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999962210246928e-05,
+      "loss": 0.8484,
+      "step": 2944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999961271407457e-05,
+      "loss": 0.9019,
+      "step": 2945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999996032104851e-05,
+      "loss": 1.0027,
+      "step": 2946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999959359170097e-05,
+      "loss": 0.7433,
+      "step": 2947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999958385772208e-05,
+      "loss": 0.8958,
+      "step": 2948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999957400854853e-05,
+      "loss": 0.8297,
+      "step": 2949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999995640441803e-05,
+      "loss": 0.8263,
+      "step": 2950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999955396461735e-05,
+      "loss": 0.7128,
+      "step": 2951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999954376985978e-05,
+      "loss": 0.8974,
+      "step": 2952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9999953345990756e-05,
+      "loss": 0.672,
+      "step": 2953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999995230347607e-05,
+      "loss": 0.7636,
+      "step": 2954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999995124944192e-05,
+      "loss": 0.8376,
+      "step": 2955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999995018388831e-05,
+      "loss": 0.87,
+      "step": 2956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999994910681524e-05,
+      "loss": 0.9995,
+      "step": 2957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.999994801822271e-05,
+      "loss": 0.7207,
+      "step": 2958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999946918110725e-05,
+      "loss": 0.9621,
+      "step": 2959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999994580647928e-05,
+      "loss": 0.9048,
+      "step": 2960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999944683328382e-05,
+      "loss": 0.9654,
+      "step": 2961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999994354865803e-05,
+      "loss": 0.9452,
+      "step": 2962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999942402468226e-05,
+      "loss": 0.7221,
+      "step": 2963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999941244758973e-05,
+      "loss": 0.9749,
+      "step": 2964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999940075530268e-05,
+      "loss": 0.8427,
+      "step": 2965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999938894782118e-05,
+      "loss": 0.8552,
+      "step": 2966
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999937702514516e-05,
+      "loss": 0.7816,
+      "step": 2967
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999936498727466e-05,
+      "loss": 1.056,
+      "step": 2968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999993528342098e-05,
+      "loss": 0.9242,
+      "step": 2969
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999934056595046e-05,
+      "loss": 0.8683,
+      "step": 2970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999932818249672e-05,
+      "loss": 0.7184,
+      "step": 2971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999931568384857e-05,
+      "loss": 0.8074,
+      "step": 2972
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999930307000604e-05,
+      "loss": 0.7173,
+      "step": 2973
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999929034096916e-05,
+      "loss": 0.9264,
+      "step": 2974
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999992774967379e-05,
+      "loss": 0.8819,
+      "step": 2975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999992645373123e-05,
+      "loss": 0.7812,
+      "step": 2976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999925146269237e-05,
+      "loss": 0.772,
+      "step": 2977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999923827287814e-05,
+      "loss": 1.0115,
+      "step": 2978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999992249678696e-05,
+      "loss": 0.729,
+      "step": 2979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999921154766677e-05,
+      "loss": 0.8227,
+      "step": 2980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999991980122697e-05,
+      "loss": 1.0685,
+      "step": 2981
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999918436167836e-05,
+      "loss": 0.7168,
+      "step": 2982
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999991705958928e-05,
+      "loss": 0.8549,
+      "step": 2983
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.99999156714913e-05,
+      "loss": 0.9166,
+      "step": 2984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999914271873903e-05,
+      "loss": 0.8926,
+      "step": 2985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999912860737084e-05,
+      "loss": 1.0688,
+      "step": 2986
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999991143808085e-05,
+      "loss": 0.9157,
+      "step": 2987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.99999100039052e-05,
+      "loss": 0.9293,
+      "step": 2988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999990855821014e-05,
+      "loss": 0.7151,
+      "step": 2989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999907100995662e-05,
+      "loss": 0.9106,
+      "step": 2990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999905632261775e-05,
+      "loss": 0.9219,
+      "step": 2991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999990415200848e-05,
+      "loss": 0.6498,
+      "step": 2992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999990266023578e-05,
+      "loss": 0.798,
+      "step": 2993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.999990115694367e-05,
+      "loss": 0.7789,
+      "step": 2994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999899642132162e-05,
+      "loss": 0.7264,
+      "step": 2995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9999898115801248e-05,
+      "loss": 0.9054,
+      "step": 2996
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999896577950936e-05,
+      "loss": 0.856,
+      "step": 2997
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999895028581227e-05,
+      "loss": 0.8354,
+      "step": 2998
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999989346769212e-05,
+      "loss": 0.9042,
+      "step": 2999
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999989189528362e-05,
+      "loss": 0.8506,
+      "step": 3000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999890311355722e-05,
+      "loss": 0.7594,
+      "step": 3001
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999888715908437e-05,
+      "loss": 0.8478,
+      "step": 3002
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999988710894176e-05,
+      "loss": 0.8299,
+      "step": 3003
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999885490455695e-05,
+      "loss": 0.7582,
+      "step": 3004
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999883860450252e-05,
+      "loss": 0.8897,
+      "step": 3005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999988221892542e-05,
+      "loss": 0.8439,
+      "step": 3006
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999880565881208e-05,
+      "loss": 0.7869,
+      "step": 3007
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999878901317613e-05,
+      "loss": 1.1752,
+      "step": 3008
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999987722523464e-05,
+      "loss": 0.8942,
+      "step": 3009
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999875537632296e-05,
+      "loss": 0.9083,
+      "step": 3010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999873838510574e-05,
+      "loss": 0.7138,
+      "step": 3011
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999987212786948e-05,
+      "loss": 0.7377,
+      "step": 3012
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999870405709018e-05,
+      "loss": 0.8441,
+      "step": 3013
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999868672029185e-05,
+      "loss": 0.6648,
+      "step": 3014
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999986692682999e-05,
+      "loss": 0.7849,
+      "step": 3015
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999865170111428e-05,
+      "loss": 0.7906,
+      "step": 3016
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.99998634018735e-05,
+      "loss": 0.9101,
+      "step": 3017
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999861622116218e-05,
+      "loss": 0.628,
+      "step": 3018
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999859830839577e-05,
+      "loss": 0.9126,
+      "step": 3019
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999985802804358e-05,
+      "loss": 0.8371,
+      "step": 3020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999985621372823e-05,
+      "loss": 0.9651,
+      "step": 3021
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999854387893526e-05,
+      "loss": 0.9221,
+      "step": 3022
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999852550539473e-05,
+      "loss": 0.9914,
+      "step": 3023
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999985070166607e-05,
+      "loss": 0.8642,
+      "step": 3024
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999848841273326e-05,
+      "loss": 0.7696,
+      "step": 3025
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999846969361236e-05,
+      "loss": 0.7693,
+      "step": 3026
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999845085929806e-05,
+      "loss": 1.0268,
+      "step": 3027
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999843190979037e-05,
+      "loss": 0.8408,
+      "step": 3028
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999984128450893e-05,
+      "loss": 1.0227,
+      "step": 3029
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999983936651949e-05,
+      "loss": 0.6931,
+      "step": 3030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999837437010717e-05,
+      "loss": 0.7202,
+      "step": 3031
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999835495982612e-05,
+      "loss": 0.8149,
+      "step": 3032
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.999983354343518e-05,
+      "loss": 0.8184,
+      "step": 3033
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9999831579368424e-05,
+      "loss": 1.092,
+      "step": 3034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999829603782344e-05,
+      "loss": 0.7828,
+      "step": 3035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999827616676945e-05,
+      "loss": 0.8057,
+      "step": 3036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999825618052223e-05,
+      "loss": 0.9406,
+      "step": 3037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999823607908188e-05,
+      "loss": 0.9561,
+      "step": 3038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999821586244837e-05,
+      "loss": 0.9932,
+      "step": 3039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999819553062174e-05,
+      "loss": 0.8494,
+      "step": 3040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99998175083602e-05,
+      "loss": 0.8141,
+      "step": 3041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999981545213892e-05,
+      "loss": 0.9813,
+      "step": 3042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999813384398337e-05,
+      "loss": 0.9489,
+      "step": 3043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999981130513845e-05,
+      "loss": 0.9235,
+      "step": 3044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999809214359264e-05,
+      "loss": 0.9052,
+      "step": 3045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999980711206078e-05,
+      "loss": 0.951,
+      "step": 3046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999804998242998e-05,
+      "loss": 0.8823,
+      "step": 3047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999802872905928e-05,
+      "loss": 0.7336,
+      "step": 3048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999800736049565e-05,
+      "loss": 0.9271,
+      "step": 3049
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999798587673914e-05,
+      "loss": 0.8032,
+      "step": 3050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999796427778977e-05,
+      "loss": 0.8588,
+      "step": 3051
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999979425636476e-05,
+      "loss": 0.8118,
+      "step": 3052
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999792073431257e-05,
+      "loss": 0.8206,
+      "step": 3053
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999978987897848e-05,
+      "loss": 0.9727,
+      "step": 3054
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999978767300643e-05,
+      "loss": 0.9417,
+      "step": 3055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999785455515105e-05,
+      "loss": 0.8921,
+      "step": 3056
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999783226504506e-05,
+      "loss": 0.8615,
+      "step": 3057
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999780985974645e-05,
+      "loss": 0.9065,
+      "step": 3058
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999977873392552e-05,
+      "loss": 0.8996,
+      "step": 3059
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999776470357123e-05,
+      "loss": 0.9525,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999774195269473e-05,
+      "loss": 0.6649,
+      "step": 3061
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999771908662565e-05,
+      "loss": 0.9567,
+      "step": 3062
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99997696105364e-05,
+      "loss": 0.8665,
+      "step": 3063
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999767300890986e-05,
+      "loss": 0.8286,
+      "step": 3064
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999764979726323e-05,
+      "loss": 1.0307,
+      "step": 3065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999976264704241e-05,
+      "loss": 0.8701,
+      "step": 3066
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999760302839254e-05,
+      "loss": 0.7657,
+      "step": 3067
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999975794711686e-05,
+      "loss": 0.923,
+      "step": 3068
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999755579875227e-05,
+      "loss": 0.6983,
+      "step": 3069
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999753201114352e-05,
+      "loss": 0.9515,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999975081083425e-05,
+      "loss": 1.0082,
+      "step": 3071
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999974840903492e-05,
+      "loss": 1.037,
+      "step": 3072
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9999745995716355e-05,
+      "loss": 0.864,
+      "step": 3073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999974357087857e-05,
+      "loss": 0.8855,
+      "step": 3074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999974113452156e-05,
+      "loss": 0.8045,
+      "step": 3075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999738686645332e-05,
+      "loss": 0.7715,
+      "step": 3076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999973622724989e-05,
+      "loss": 0.8002,
+      "step": 3077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999973375633523e-05,
+      "loss": 0.9447,
+      "step": 3078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999973127390136e-05,
+      "loss": 0.8119,
+      "step": 3079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999728779948288e-05,
+      "loss": 0.8198,
+      "step": 3080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999726274476004e-05,
+      "loss": 0.813,
+      "step": 3081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999723757484523e-05,
+      "loss": 1.0415,
+      "step": 3082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999721228973838e-05,
+      "loss": 0.9826,
+      "step": 3083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999718688943958e-05,
+      "loss": 0.6768,
+      "step": 3084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999716137394888e-05,
+      "loss": 0.7943,
+      "step": 3085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999713574326624e-05,
+      "loss": 0.7991,
+      "step": 3086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999710999739173e-05,
+      "loss": 0.6975,
+      "step": 3087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999970841363254e-05,
+      "loss": 0.7921,
+      "step": 3088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999970581600672e-05,
+      "loss": 0.9067,
+      "step": 3089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999703206861728e-05,
+      "loss": 0.9053,
+      "step": 3090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999700586197558e-05,
+      "loss": 0.8232,
+      "step": 3091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999697954014214e-05,
+      "loss": 0.9099,
+      "step": 3092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999695310311703e-05,
+      "loss": 0.6373,
+      "step": 3093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999692655090025e-05,
+      "loss": 0.8105,
+      "step": 3094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999689988349184e-05,
+      "loss": 0.6597,
+      "step": 3095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999968731008918e-05,
+      "loss": 0.8968,
+      "step": 3096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999968462031002e-05,
+      "loss": 0.9601,
+      "step": 3097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999681919011705e-05,
+      "loss": 0.9069,
+      "step": 3098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999679206194244e-05,
+      "loss": 0.8422,
+      "step": 3099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999676481857632e-05,
+      "loss": 0.7728,
+      "step": 3100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999673746001873e-05,
+      "loss": 0.8146,
+      "step": 3101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999670998626975e-05,
+      "loss": 0.7127,
+      "step": 3102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999966823973294e-05,
+      "loss": 0.895,
+      "step": 3103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999665469319765e-05,
+      "loss": 0.9569,
+      "step": 3104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999966268738746e-05,
+      "loss": 0.7203,
+      "step": 3105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999659893936027e-05,
+      "loss": 0.9722,
+      "step": 3106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999965708896547e-05,
+      "loss": 1.0524,
+      "step": 3107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999965427247579e-05,
+      "loss": 0.8238,
+      "step": 3108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9999651444466987e-05,
+      "loss": 0.9739,
+      "step": 3109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999964860493907e-05,
+      "loss": 0.9295,
+      "step": 3110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.999964575389204e-05,
+      "loss": 0.8801,
+      "step": 3111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999642891325905e-05,
+      "loss": 0.8404,
+      "step": 3112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999640017240658e-05,
+      "loss": 0.6453,
+      "step": 3113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999963713163631e-05,
+      "loss": 0.7682,
+      "step": 3114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999634234512863e-05,
+      "loss": 0.8571,
+      "step": 3115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999631325870325e-05,
+      "loss": 0.7013,
+      "step": 3116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999628405708687e-05,
+      "loss": 0.9109,
+      "step": 3117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999962547402796e-05,
+      "loss": 0.7527,
+      "step": 3118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999622530828153e-05,
+      "loss": 0.6656,
+      "step": 3119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999961957610926e-05,
+      "loss": 0.9851,
+      "step": 3120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999616609871286e-05,
+      "loss": 0.8778,
+      "step": 3121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999613632114237e-05,
+      "loss": 0.8504,
+      "step": 3122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999610642838116e-05,
+      "loss": 0.8974,
+      "step": 3123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999607642042927e-05,
+      "loss": 0.8127,
+      "step": 3124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999604629728676e-05,
+      "loss": 0.8331,
+      "step": 3125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999960160589536e-05,
+      "loss": 0.7648,
+      "step": 3126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999959857054298e-05,
+      "loss": 0.6673,
+      "step": 3127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999959552367155e-05,
+      "loss": 0.9415,
+      "step": 3128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999592465281072e-05,
+      "loss": 0.7647,
+      "step": 3129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999589395371538e-05,
+      "loss": 0.862,
+      "step": 3130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999586313942966e-05,
+      "loss": 0.9198,
+      "step": 3131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999583220995352e-05,
+      "loss": 0.8737,
+      "step": 3132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999580116528697e-05,
+      "loss": 1.1232,
+      "step": 3133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999957700054301e-05,
+      "loss": 0.7875,
+      "step": 3134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.99995738730383e-05,
+      "loss": 0.8537,
+      "step": 3135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999570734014556e-05,
+      "loss": 0.8406,
+      "step": 3136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999956758347179e-05,
+      "loss": 1.0381,
+      "step": 3137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999564421410003e-05,
+      "loss": 0.9575,
+      "step": 3138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999561247829203e-05,
+      "loss": 0.7927,
+      "step": 3139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999558062729392e-05,
+      "loss": 0.7957,
+      "step": 3140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999955486611057e-05,
+      "loss": 0.8236,
+      "step": 3141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999551657972744e-05,
+      "loss": 0.7261,
+      "step": 3142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999548438315916e-05,
+      "loss": 0.7886,
+      "step": 3143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999545207140095e-05,
+      "loss": 0.7685,
+      "step": 3144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999541964445276e-05,
+      "loss": 0.9201,
+      "step": 3145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999538710231473e-05,
+      "loss": 0.8901,
+      "step": 3146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9999535444498675e-05,
+      "loss": 0.965,
+      "step": 3147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.99995321672469e-05,
+      "loss": 0.8248,
+      "step": 3148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.999952887847615e-05,
+      "loss": 0.8882,
+      "step": 3149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999952557818642e-05,
+      "loss": 0.8041,
+      "step": 3150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999952226637772e-05,
+      "loss": 0.7021,
+      "step": 3151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999518943050057e-05,
+      "loss": 0.7931,
+      "step": 3152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999515608203427e-05,
+      "loss": 0.7627,
+      "step": 3153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999512261837837e-05,
+      "loss": 0.7861,
+      "step": 3154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999950890395329e-05,
+      "loss": 0.8303,
+      "step": 3155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999505534549795e-05,
+      "loss": 0.9629,
+      "step": 3156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999950215362735e-05,
+      "loss": 0.8341,
+      "step": 3157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999498761185963e-05,
+      "loss": 0.8221,
+      "step": 3158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999495357225632e-05,
+      "loss": 0.8432,
+      "step": 3159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999491941746368e-05,
+      "loss": 0.843,
+      "step": 3160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999488514748174e-05,
+      "loss": 0.8294,
+      "step": 3161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999948507623105e-05,
+      "loss": 1.0966,
+      "step": 3162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999481626195e-05,
+      "loss": 1.0293,
+      "step": 3163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999947816464003e-05,
+      "loss": 0.6839,
+      "step": 3164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999474691566148e-05,
+      "loss": 0.9533,
+      "step": 3165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999947120697335e-05,
+      "loss": 0.8112,
+      "step": 3166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999467710861644e-05,
+      "loss": 0.8586,
+      "step": 3167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999464203231033e-05,
+      "loss": 0.962,
+      "step": 3168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999460684081522e-05,
+      "loss": 0.8956,
+      "step": 3169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999457153413116e-05,
+      "loss": 0.8785,
+      "step": 3170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999453611225817e-05,
+      "loss": 0.7238,
+      "step": 3171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999450057519633e-05,
+      "loss": 0.765,
+      "step": 3172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999446492294562e-05,
+      "loss": 0.9385,
+      "step": 3173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999944291555061e-05,
+      "loss": 0.9746,
+      "step": 3174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999439327287785e-05,
+      "loss": 0.8345,
+      "step": 3175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999435727506088e-05,
+      "loss": 0.759,
+      "step": 3176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999432116205525e-05,
+      "loss": 0.7172,
+      "step": 3177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999428493386094e-05,
+      "loss": 1.0086,
+      "step": 3178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999942485904781e-05,
+      "loss": 0.7971,
+      "step": 3179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999421213190666e-05,
+      "loss": 0.7986,
+      "step": 3180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999417555814672e-05,
+      "loss": 0.9431,
+      "step": 3181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999413886919833e-05,
+      "loss": 0.8634,
+      "step": 3182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999410206506152e-05,
+      "loss": 0.9839,
+      "step": 3183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999406514573634e-05,
+      "loss": 0.8285,
+      "step": 3184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.999940281112228e-05,
+      "loss": 0.803,
+      "step": 3185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999399096152098e-05,
+      "loss": 0.8582,
+      "step": 3186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9999395369663088e-05,
+      "loss": 0.8219,
+      "step": 3187
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999939163165526e-05,
+      "loss": 0.7952,
+      "step": 3188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999387882128617e-05,
+      "loss": 0.7636,
+      "step": 3189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999938412108316e-05,
+      "loss": 0.7289,
+      "step": 3190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999380348518895e-05,
+      "loss": 0.8613,
+      "step": 3191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999376564435825e-05,
+      "loss": 0.6922,
+      "step": 3192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999372768833955e-05,
+      "loss": 0.9451,
+      "step": 3193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999368961713293e-05,
+      "loss": 1.0957,
+      "step": 3194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999365143073837e-05,
+      "loss": 0.7812,
+      "step": 3195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.99993613129156e-05,
+      "loss": 0.698,
+      "step": 3196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999357471238577e-05,
+      "loss": 0.8594,
+      "step": 3197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999353618042776e-05,
+      "loss": 0.6963,
+      "step": 3198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999349753328207e-05,
+      "loss": 0.9728,
+      "step": 3199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999345877094868e-05,
+      "loss": 0.8455,
+      "step": 3200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999341989342762e-05,
+      "loss": 0.9267,
+      "step": 3201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.99993380900719e-05,
+      "loss": 0.8105,
+      "step": 3202
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999334179282282e-05,
+      "loss": 0.7343,
+      "step": 3203
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999933025697391e-05,
+      "loss": 0.8379,
+      "step": 3204
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999326323146798e-05,
+      "loss": 0.8401,
+      "step": 3205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999322377800944e-05,
+      "loss": 0.8189,
+      "step": 3206
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999931842093635e-05,
+      "loss": 0.8737,
+      "step": 3207
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999314452553026e-05,
+      "loss": 0.9019,
+      "step": 3208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999931047265097e-05,
+      "loss": 0.6692,
+      "step": 3209
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999306481230195e-05,
+      "loss": 0.9084,
+      "step": 3210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.99993024782907e-05,
+      "loss": 0.689,
+      "step": 3211
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999298463832492e-05,
+      "loss": 0.855,
+      "step": 3212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999294437855576e-05,
+      "loss": 0.9333,
+      "step": 3213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999290400359954e-05,
+      "loss": 0.9042,
+      "step": 3214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999928635134563e-05,
+      "loss": 0.7403,
+      "step": 3215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999282290812615e-05,
+      "loss": 0.8092,
+      "step": 3216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999278218760905e-05,
+      "loss": 0.7859,
+      "step": 3217
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999274135190513e-05,
+      "loss": 0.7079,
+      "step": 3218
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999270040101438e-05,
+      "loss": 0.7725,
+      "step": 3219
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999265933493684e-05,
+      "loss": 0.8719,
+      "step": 3220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999926181536726e-05,
+      "loss": 0.8255,
+      "step": 3221
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999257685722172e-05,
+      "loss": 0.8664,
+      "step": 3222
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999925354455842e-05,
+      "loss": 0.7982,
+      "step": 3223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.999924939187601e-05,
+      "loss": 0.765,
+      "step": 3224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9999245227674944e-05,
+      "loss": 0.9403,
+      "step": 3225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999241051955234e-05,
+      "loss": 0.8678,
+      "step": 3226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999923686471688e-05,
+      "loss": 0.8226,
+      "step": 3227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999232665959886e-05,
+      "loss": 0.7934,
+      "step": 3228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999922845568426e-05,
+      "loss": 0.7785,
+      "step": 3229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999224233890007e-05,
+      "loss": 0.9912,
+      "step": 3230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999922000057713e-05,
+      "loss": 0.8324,
+      "step": 3231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999921575574563e-05,
+      "loss": 0.7086,
+      "step": 3232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999211499395518e-05,
+      "loss": 0.7209,
+      "step": 3233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.99992072315268e-05,
+      "loss": 0.8444,
+      "step": 3234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999202952139472e-05,
+      "loss": 0.8944,
+      "step": 3235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999919866123355e-05,
+      "loss": 0.7631,
+      "step": 3236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999919435880903e-05,
+      "loss": 0.768,
+      "step": 3237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999190044865925e-05,
+      "loss": 0.705,
+      "step": 3238
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999185719404234e-05,
+      "loss": 0.6005,
+      "step": 3239
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999918138242396e-05,
+      "loss": 0.8125,
+      "step": 3240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999177033925113e-05,
+      "loss": 0.8483,
+      "step": 3241
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.99991726739077e-05,
+      "loss": 0.7236,
+      "step": 3242
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999916830237172e-05,
+      "loss": 0.7702,
+      "step": 3243
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999163919317182e-05,
+      "loss": 0.7582,
+      "step": 3244
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999159524744088e-05,
+      "loss": 0.6884,
+      "step": 3245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999155118652447e-05,
+      "loss": 0.9782,
+      "step": 3246
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999150701042258e-05,
+      "loss": 0.7728,
+      "step": 3247
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999146271913533e-05,
+      "loss": 0.7751,
+      "step": 3248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999141831266274e-05,
+      "loss": 0.8392,
+      "step": 3249
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999137379100486e-05,
+      "loss": 0.6982,
+      "step": 3250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999132915416174e-05,
+      "loss": 0.6694,
+      "step": 3251
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999128440213343e-05,
+      "loss": 1.0347,
+      "step": 3252
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999123953491998e-05,
+      "loss": 0.8462,
+      "step": 3253
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999119455252147e-05,
+      "loss": 0.9003,
+      "step": 3254
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999911494549379e-05,
+      "loss": 0.9214,
+      "step": 3255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999911042421694e-05,
+      "loss": 0.85,
+      "step": 3256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999910589142159e-05,
+      "loss": 0.7737,
+      "step": 3257
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999101347107758e-05,
+      "loss": 0.8395,
+      "step": 3258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999096791275444e-05,
+      "loss": 0.7231,
+      "step": 3259
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.999909222392465e-05,
+      "loss": 1.0129,
+      "step": 3260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999087645055387e-05,
+      "loss": 0.7458,
+      "step": 3261
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999083054667655e-05,
+      "loss": 0.7709,
+      "step": 3262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9999078452761465e-05,
+      "loss": 1.0026,
+      "step": 3263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999073839336815e-05,
+      "loss": 0.825,
+      "step": 3264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999069214393717e-05,
+      "loss": 0.7652,
+      "step": 3265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999064577932174e-05,
+      "loss": 0.6819,
+      "step": 3266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999905992995219e-05,
+      "loss": 0.8692,
+      "step": 3267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999055270453774e-05,
+      "loss": 0.7414,
+      "step": 3268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999050599436926e-05,
+      "loss": 0.7937,
+      "step": 3269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999045916901655e-05,
+      "loss": 0.9652,
+      "step": 3270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999041222847966e-05,
+      "loss": 0.835,
+      "step": 3271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999036517275866e-05,
+      "loss": 0.8759,
+      "step": 3272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999031800185355e-05,
+      "loss": 0.9298,
+      "step": 3273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999027071576446e-05,
+      "loss": 0.7532,
+      "step": 3274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999022331449135e-05,
+      "loss": 0.8797,
+      "step": 3275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999017579803437e-05,
+      "loss": 0.8458,
+      "step": 3276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999901281663935e-05,
+      "loss": 0.7116,
+      "step": 3277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999008041956887e-05,
+      "loss": 0.8132,
+      "step": 3278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9999003255756046e-05,
+      "loss": 0.7816,
+      "step": 3279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998998458036838e-05,
+      "loss": 0.7929,
+      "step": 3280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998993648799265e-05,
+      "loss": 0.8282,
+      "step": 3281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998988828043333e-05,
+      "loss": 1.0173,
+      "step": 3282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999898399576905e-05,
+      "loss": 0.9461,
+      "step": 3283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999897915197642e-05,
+      "loss": 0.7571,
+      "step": 3284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998974296665445e-05,
+      "loss": 1.0202,
+      "step": 3285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999896942983614e-05,
+      "loss": 0.8831,
+      "step": 3286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.99989645514885e-05,
+      "loss": 0.785,
+      "step": 3287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998959661622538e-05,
+      "loss": 0.9911,
+      "step": 3288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998954760238257e-05,
+      "loss": 0.9194,
+      "step": 3289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998949847335663e-05,
+      "loss": 0.9999,
+      "step": 3290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999894492291476e-05,
+      "loss": 0.7461,
+      "step": 3291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998939986975552e-05,
+      "loss": 0.9771,
+      "step": 3292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999893503951805e-05,
+      "loss": 0.8824,
+      "step": 3293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999893008054226e-05,
+      "loss": 0.7658,
+      "step": 3294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998925110048185e-05,
+      "loss": 0.9352,
+      "step": 3295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998920128035827e-05,
+      "loss": 0.9452,
+      "step": 3296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.99989151345052e-05,
+      "loss": 1.048,
+      "step": 3297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.99989101294563e-05,
+      "loss": 0.8817,
+      "step": 3298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998905112889142e-05,
+      "loss": 0.7178,
+      "step": 3299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999890008480373e-05,
+      "loss": 0.9139,
+      "step": 3300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9998895045200062e-05,
+      "loss": 0.6969,
+      "step": 3301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.999888999407815e-05,
+      "loss": 0.9698,
+      "step": 3302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998884931438004e-05,
+      "loss": 0.8024,
+      "step": 3303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998879857279622e-05,
+      "loss": 0.9125,
+      "step": 3304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999887477160301e-05,
+      "loss": 0.8014,
+      "step": 3305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998869674408183e-05,
+      "loss": 0.8749,
+      "step": 3306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998864565695136e-05,
+      "loss": 0.686,
+      "step": 3307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998859445463884e-05,
+      "loss": 0.8395,
+      "step": 3308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998854313714424e-05,
+      "loss": 0.7347,
+      "step": 3309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999884917044677e-05,
+      "loss": 0.9088,
+      "step": 3310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999884401566092e-05,
+      "loss": 0.8666,
+      "step": 3311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998838849356887e-05,
+      "loss": 0.8498,
+      "step": 3312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998833671534676e-05,
+      "loss": 0.7465,
+      "step": 3313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998828482194287e-05,
+      "loss": 1.003,
+      "step": 3314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998823281335734e-05,
+      "loss": 0.6932,
+      "step": 3315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998818068959017e-05,
+      "loss": 0.7363,
+      "step": 3316
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998812845064143e-05,
+      "loss": 0.8983,
+      "step": 3317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999880760965112e-05,
+      "loss": 0.7602,
+      "step": 3318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999880236271995e-05,
+      "loss": 0.8531,
+      "step": 3319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998797104270648e-05,
+      "loss": 0.8247,
+      "step": 3320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998791834303212e-05,
+      "loss": 0.8025,
+      "step": 3321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999878655281765e-05,
+      "loss": 0.851,
+      "step": 3322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998781259813967e-05,
+      "loss": 0.92,
+      "step": 3323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998775955292175e-05,
+      "loss": 0.7174,
+      "step": 3324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998770639252272e-05,
+      "loss": 0.663,
+      "step": 3325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998765311694267e-05,
+      "loss": 0.9165,
+      "step": 3326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998759972618172e-05,
+      "loss": 0.72,
+      "step": 3327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998754622023984e-05,
+      "loss": 0.9223,
+      "step": 3328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999874925991171e-05,
+      "loss": 0.7549,
+      "step": 3329
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998743886281367e-05,
+      "loss": 0.801,
+      "step": 3330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999873850113295e-05,
+      "loss": 0.8775,
+      "step": 3331
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998733104466466e-05,
+      "loss": 0.7363,
+      "step": 3332
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999872769628193e-05,
+      "loss": 0.8927,
+      "step": 3333
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998722276579336e-05,
+      "loss": 0.7989,
+      "step": 3334
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.99987168453587e-05,
+      "loss": 0.7157,
+      "step": 3335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998711402620024e-05,
+      "loss": 0.7842,
+      "step": 3336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998705948363315e-05,
+      "loss": 0.7257,
+      "step": 3337
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999870048258858e-05,
+      "loss": 0.9041,
+      "step": 3338
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.999869500529582e-05,
+      "loss": 0.7784,
+      "step": 3339
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9998689516485054e-05,
+      "loss": 0.8285,
+      "step": 3340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998684016156274e-05,
+      "loss": 0.972,
+      "step": 3341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998678504309494e-05,
+      "loss": 0.8949,
+      "step": 3342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999867298094472e-05,
+      "loss": 0.7766,
+      "step": 3343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998667446061958e-05,
+      "loss": 0.9416,
+      "step": 3344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999866189966121e-05,
+      "loss": 0.6703,
+      "step": 3345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998656341742488e-05,
+      "loss": 0.8657,
+      "step": 3346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998650772305795e-05,
+      "loss": 0.8249,
+      "step": 3347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999864519135114e-05,
+      "loss": 0.6849,
+      "step": 3348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998639598878533e-05,
+      "loss": 0.9099,
+      "step": 3349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998633994887968e-05,
+      "loss": 0.7385,
+      "step": 3350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998628379379466e-05,
+      "loss": 0.865,
+      "step": 3351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999862275235302e-05,
+      "loss": 0.9669,
+      "step": 3352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998617113808646e-05,
+      "loss": 0.7914,
+      "step": 3353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999861146374635e-05,
+      "loss": 0.7951,
+      "step": 3354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998605802166133e-05,
+      "loss": 0.8075,
+      "step": 3355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998600129068006e-05,
+      "loss": 0.7729,
+      "step": 3356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998594444451976e-05,
+      "loss": 0.9004,
+      "step": 3357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998588748318043e-05,
+      "loss": 0.9896,
+      "step": 3358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998583040666223e-05,
+      "loss": 0.9683,
+      "step": 3359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998577321496514e-05,
+      "loss": 0.9399,
+      "step": 3360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999857159080893e-05,
+      "loss": 0.8764,
+      "step": 3361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999856584860347e-05,
+      "loss": 0.8492,
+      "step": 3362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998560094880147e-05,
+      "loss": 0.7414,
+      "step": 3363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998554329638964e-05,
+      "loss": 0.8386,
+      "step": 3364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999854855287993e-05,
+      "loss": 0.7658,
+      "step": 3365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.999854276460305e-05,
+      "loss": 0.761,
+      "step": 3366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998536964808335e-05,
+      "loss": 0.8318,
+      "step": 3367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998531153495783e-05,
+      "loss": 0.7688,
+      "step": 3368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998525330665406e-05,
+      "loss": 0.901,
+      "step": 3369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998519496317214e-05,
+      "loss": 0.7846,
+      "step": 3370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998513650451207e-05,
+      "loss": 0.7255,
+      "step": 3371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998507793067395e-05,
+      "loss": 0.6962,
+      "step": 3372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998501924165788e-05,
+      "loss": 0.8459,
+      "step": 3373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998496043746386e-05,
+      "loss": 0.7732,
+      "step": 3374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998490151809196e-05,
+      "loss": 0.7652,
+      "step": 3375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998484248354236e-05,
+      "loss": 0.7676,
+      "step": 3376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998478333381497e-05,
+      "loss": 0.9063,
+      "step": 3377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9998472406890997e-05,
+      "loss": 0.8946,
+      "step": 3378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999846646888274e-05,
+      "loss": 0.7457,
+      "step": 3379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999846051935673e-05,
+      "loss": 0.8805,
+      "step": 3380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998454558312978e-05,
+      "loss": 0.6738,
+      "step": 3381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998448585751485e-05,
+      "loss": 0.8387,
+      "step": 3382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998442601672266e-05,
+      "loss": 0.8723,
+      "step": 3383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998436606075323e-05,
+      "loss": 0.7625,
+      "step": 3384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998430598960663e-05,
+      "loss": 0.6825,
+      "step": 3385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998424580328294e-05,
+      "loss": 0.9915,
+      "step": 3386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999841855017822e-05,
+      "loss": 0.7348,
+      "step": 3387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999841250851045e-05,
+      "loss": 0.9169,
+      "step": 3388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998406455324995e-05,
+      "loss": 0.8214,
+      "step": 3389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998400390621856e-05,
+      "loss": 0.9843,
+      "step": 3390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998394314401044e-05,
+      "loss": 1.0427,
+      "step": 3391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998388226662563e-05,
+      "loss": 0.9021,
+      "step": 3392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998382127406422e-05,
+      "loss": 0.7699,
+      "step": 3393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998376016632626e-05,
+      "loss": 0.7092,
+      "step": 3394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998369894341184e-05,
+      "loss": 0.7931,
+      "step": 3395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998363760532103e-05,
+      "loss": 0.7653,
+      "step": 3396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998357615205387e-05,
+      "loss": 0.8524,
+      "step": 3397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999835145836105e-05,
+      "loss": 0.8617,
+      "step": 3398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999834528999909e-05,
+      "loss": 0.7089,
+      "step": 3399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999833911011952e-05,
+      "loss": 0.714,
+      "step": 3400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999833291872235e-05,
+      "loss": 0.9192,
+      "step": 3401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998326715807577e-05,
+      "loss": 0.7697,
+      "step": 3402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998320501375216e-05,
+      "loss": 1.0592,
+      "step": 3403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999831427542527e-05,
+      "loss": 0.8605,
+      "step": 3404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998308037957755e-05,
+      "loss": 0.839,
+      "step": 3405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998301788972668e-05,
+      "loss": 0.9195,
+      "step": 3406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999829552847002e-05,
+      "loss": 0.7045,
+      "step": 3407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998289256449817e-05,
+      "loss": 0.7647,
+      "step": 3408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998282972912067e-05,
+      "loss": 0.695,
+      "step": 3409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999827667785678e-05,
+      "loss": 0.7087,
+      "step": 3410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999827037128396e-05,
+      "loss": 1.0906,
+      "step": 3411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998264053193614e-05,
+      "loss": 0.7795,
+      "step": 3412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.999825772358575e-05,
+      "loss": 0.8387,
+      "step": 3413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998251382460377e-05,
+      "loss": 0.7425,
+      "step": 3414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.99982450298175e-05,
+      "loss": 0.728,
+      "step": 3415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9998238665657127e-05,
+      "loss": 0.8018,
+      "step": 3416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998232289979268e-05,
+      "loss": 0.9343,
+      "step": 3417
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998225902783927e-05,
+      "loss": 1.0602,
+      "step": 3418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998219504071112e-05,
+      "loss": 0.7648,
+      "step": 3419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999821309384083e-05,
+      "loss": 0.7289,
+      "step": 3420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999820667209309e-05,
+      "loss": 0.7719,
+      "step": 3421
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998200238827893e-05,
+      "loss": 0.7908,
+      "step": 3422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998193794045258e-05,
+      "loss": 0.8736,
+      "step": 3423
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998187337745184e-05,
+      "loss": 0.7888,
+      "step": 3424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999818086992768e-05,
+      "loss": 0.8342,
+      "step": 3425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998174390592757e-05,
+      "loss": 0.7761,
+      "step": 3426
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999816789974042e-05,
+      "loss": 0.7426,
+      "step": 3427
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999816139737067e-05,
+      "loss": 0.8492,
+      "step": 3428
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998154883483528e-05,
+      "loss": 0.6631,
+      "step": 3429
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999814835807899e-05,
+      "loss": 0.7301,
+      "step": 3430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999814182115707e-05,
+      "loss": 0.8422,
+      "step": 3431
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998135272717768e-05,
+      "loss": 0.8329,
+      "step": 3432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.99981287127611e-05,
+      "loss": 0.8398,
+      "step": 3433
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998122141287072e-05,
+      "loss": 0.9701,
+      "step": 3434
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998115558295687e-05,
+      "loss": 0.9107,
+      "step": 3435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999810896378696e-05,
+      "loss": 0.6612,
+      "step": 3436
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998102357760885e-05,
+      "loss": 1.003,
+      "step": 3437
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998095740217485e-05,
+      "loss": 0.8017,
+      "step": 3438
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999808911115676e-05,
+      "loss": 0.882,
+      "step": 3439
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999808247057872e-05,
+      "loss": 0.7341,
+      "step": 3440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998075818483372e-05,
+      "loss": 0.9477,
+      "step": 3441
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999806915487072e-05,
+      "loss": 0.7061,
+      "step": 3442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999806247974078e-05,
+      "loss": 0.8169,
+      "step": 3443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998055793093548e-05,
+      "loss": 0.7766,
+      "step": 3444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999804909492904e-05,
+      "loss": 0.7983,
+      "step": 3445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998042385247266e-05,
+      "loss": 0.6904,
+      "step": 3446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998035664048226e-05,
+      "loss": 0.7508,
+      "step": 3447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998028931331933e-05,
+      "loss": 0.922,
+      "step": 3448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999802218709839e-05,
+      "loss": 1.0595,
+      "step": 3449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999801543134761e-05,
+      "loss": 0.8571,
+      "step": 3450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.99980086640796e-05,
+      "loss": 0.9199,
+      "step": 3451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9998001885294367e-05,
+      "loss": 0.7529,
+      "step": 3452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9997995094991916e-05,
+      "loss": 0.8018,
+      "step": 3453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999798829317226e-05,
+      "loss": 0.8429,
+      "step": 3454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.99979814798354e-05,
+      "loss": 0.9534,
+      "step": 3455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997974654981348e-05,
+      "loss": 0.8594,
+      "step": 3456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997967818610117e-05,
+      "loss": 0.7159,
+      "step": 3457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997960970721704e-05,
+      "loss": 0.7908,
+      "step": 3458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997954111316126e-05,
+      "loss": 0.8459,
+      "step": 3459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997947240393384e-05,
+      "loss": 0.7828,
+      "step": 3460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.999794035795349e-05,
+      "loss": 0.9666,
+      "step": 3461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997933463996457e-05,
+      "loss": 0.7734,
+      "step": 3462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997926558522282e-05,
+      "loss": 0.8127,
+      "step": 3463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997919641530976e-05,
+      "loss": 0.9315,
+      "step": 3464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.999791271302255e-05,
+      "loss": 0.7115,
+      "step": 3465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997905772997013e-05,
+      "loss": 0.9429,
+      "step": 3466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.999789882145437e-05,
+      "loss": 1.0446,
+      "step": 3467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997891858394633e-05,
+      "loss": 0.8927,
+      "step": 3468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997884883817806e-05,
+      "loss": 0.7963,
+      "step": 3469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997877897723896e-05,
+      "loss": 0.7423,
+      "step": 3470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997870900112916e-05,
+      "loss": 0.8664,
+      "step": 3471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997863890984866e-05,
+      "loss": 0.8211,
+      "step": 3472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997856870339764e-05,
+      "loss": 0.8459,
+      "step": 3473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997849838177612e-05,
+      "loss": 0.7525,
+      "step": 3474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997842794498417e-05,
+      "loss": 0.7646,
+      "step": 3475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997835739302194e-05,
+      "loss": 0.7866,
+      "step": 3476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997828672588945e-05,
+      "loss": 0.8174,
+      "step": 3477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.999782159435868e-05,
+      "loss": 0.8539,
+      "step": 3478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997814504611406e-05,
+      "loss": 0.6221,
+      "step": 3479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.999780740334713e-05,
+      "loss": 0.7786,
+      "step": 3480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997800290565867e-05,
+      "loss": 0.8319,
+      "step": 3481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997793166267622e-05,
+      "loss": 0.7159,
+      "step": 3482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997786030452395e-05,
+      "loss": 0.9491,
+      "step": 3483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997778883120203e-05,
+      "loss": 0.8041,
+      "step": 3484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997771724271057e-05,
+      "loss": 0.9083,
+      "step": 3485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997764553904956e-05,
+      "loss": 0.9607,
+      "step": 3486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997757372021915e-05,
+      "loss": 0.7485,
+      "step": 3487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.999775017862194e-05,
+      "loss": 0.8485,
+      "step": 3488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997742973705035e-05,
+      "loss": 0.7376,
+      "step": 3489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997735757271215e-05,
+      "loss": 0.654,
+      "step": 3490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997728529320487e-05,
+      "loss": 0.841,
+      "step": 3491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.999772128985286e-05,
+      "loss": 0.8107,
+      "step": 3492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9997714038868334e-05,
+      "loss": 0.94,
+      "step": 3493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.999770677636693e-05,
+      "loss": 0.7056,
+      "step": 3494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997699502348648e-05,
+      "loss": 0.6868,
+      "step": 3495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.99976922168135e-05,
+      "loss": 0.8607,
+      "step": 3496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997684919761493e-05,
+      "loss": 0.9329,
+      "step": 3497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997677611192632e-05,
+      "loss": 0.9555,
+      "step": 3498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997670291106933e-05,
+      "loss": 0.8159,
+      "step": 3499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997662959504397e-05,
+      "loss": 0.7914,
+      "step": 3500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.999765561638504e-05,
+      "loss": 0.9226,
+      "step": 3501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997648261748862e-05,
+      "loss": 0.7875,
+      "step": 3502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997640895595876e-05,
+      "loss": 0.7475,
+      "step": 3503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997633517926092e-05,
+      "loss": 0.7681,
+      "step": 3504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.999762612873952e-05,
+      "loss": 0.9332,
+      "step": 3505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997618728036157e-05,
+      "loss": 0.9116,
+      "step": 3506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997611315816024e-05,
+      "loss": 0.8692,
+      "step": 3507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997603892079126e-05,
+      "loss": 0.8009,
+      "step": 3508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.999759645682547e-05,
+      "loss": 0.9937,
+      "step": 3509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997589010055067e-05,
+      "loss": 0.7194,
+      "step": 3510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997581551767922e-05,
+      "loss": 0.63,
+      "step": 3511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997574081964046e-05,
+      "loss": 0.7128,
+      "step": 3512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997566600643447e-05,
+      "loss": 0.9095,
+      "step": 3513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997559107806133e-05,
+      "loss": 0.8244,
+      "step": 3514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997551603452115e-05,
+      "loss": 0.726,
+      "step": 3515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997544087581397e-05,
+      "loss": 0.6365,
+      "step": 3516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997536560193996e-05,
+      "loss": 0.9204,
+      "step": 3517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.999752902128991e-05,
+      "loss": 0.8917,
+      "step": 3518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997521470869156e-05,
+      "loss": 0.8087,
+      "step": 3519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.999751390893174e-05,
+      "loss": 0.7362,
+      "step": 3520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997506335477667e-05,
+      "loss": 0.8899,
+      "step": 3521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.999749875050695e-05,
+      "loss": 1.0043,
+      "step": 3522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.99974911540196e-05,
+      "loss": 0.9038,
+      "step": 3523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997483546015625e-05,
+      "loss": 0.9163,
+      "step": 3524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997475926495024e-05,
+      "loss": 0.7715,
+      "step": 3525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997468295457816e-05,
+      "loss": 0.9741,
+      "step": 3526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997460652904007e-05,
+      "loss": 0.9067,
+      "step": 3527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.999745299883361e-05,
+      "loss": 0.8171,
+      "step": 3528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997445333246627e-05,
+      "loss": 0.7539,
+      "step": 3529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997437656143065e-05,
+      "loss": 0.9517,
+      "step": 3530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9997429967522942e-05,
+      "loss": 0.83,
+      "step": 3531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999742226738626e-05,
+      "loss": 0.7412,
+      "step": 3532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997414555733032e-05,
+      "loss": 0.8547,
+      "step": 3533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997406832563265e-05,
+      "loss": 0.915,
+      "step": 3534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997399097876968e-05,
+      "loss": 0.8137,
+      "step": 3535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997391351674146e-05,
+      "loss": 0.8756,
+      "step": 3536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997383593954813e-05,
+      "loss": 0.824,
+      "step": 3537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999737582471898e-05,
+      "loss": 0.8918,
+      "step": 3538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999736804396665e-05,
+      "loss": 0.7109,
+      "step": 3539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997360251697832e-05,
+      "loss": 0.8261,
+      "step": 3540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997352447912542e-05,
+      "loss": 0.862,
+      "step": 3541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999734463261078e-05,
+      "loss": 0.7327,
+      "step": 3542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997336805792564e-05,
+      "loss": 0.8238,
+      "step": 3543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997328967457897e-05,
+      "loss": 0.9071,
+      "step": 3544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999732111760679e-05,
+      "loss": 0.8991,
+      "step": 3545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997313256239252e-05,
+      "loss": 0.7924,
+      "step": 3546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997305383355287e-05,
+      "loss": 0.7627,
+      "step": 3547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999729749895491e-05,
+      "loss": 0.6514,
+      "step": 3548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997289603038133e-05,
+      "loss": 0.6829,
+      "step": 3549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997281695604958e-05,
+      "loss": 0.9334,
+      "step": 3550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997273776655394e-05,
+      "loss": 1.0054,
+      "step": 3551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999726584618946e-05,
+      "loss": 0.8381,
+      "step": 3552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999725790420715e-05,
+      "loss": 0.7445,
+      "step": 3553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999724995070849e-05,
+      "loss": 0.7989,
+      "step": 3554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997241985693475e-05,
+      "loss": 0.7949,
+      "step": 3555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999723400916212e-05,
+      "loss": 0.8177,
+      "step": 3556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997226021114434e-05,
+      "loss": 0.8469,
+      "step": 3557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997218021550427e-05,
+      "loss": 0.714,
+      "step": 3558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997210010470106e-05,
+      "loss": 0.7294,
+      "step": 3559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997201987873484e-05,
+      "loss": 0.7219,
+      "step": 3560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999719395376057e-05,
+      "loss": 0.8347,
+      "step": 3561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997185908131366e-05,
+      "loss": 0.7597,
+      "step": 3562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997177850985887e-05,
+      "loss": 0.9572,
+      "step": 3563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999716978232414e-05,
+      "loss": 0.8718,
+      "step": 3564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999716170214614e-05,
+      "loss": 1.0264,
+      "step": 3565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997153610451892e-05,
+      "loss": 0.9309,
+      "step": 3566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997145507241403e-05,
+      "loss": 0.7358,
+      "step": 3567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997137392514685e-05,
+      "loss": 0.8251,
+      "step": 3568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9997129266271747e-05,
+      "loss": 0.9672,
+      "step": 3569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.99971211285126e-05,
+      "loss": 0.8002,
+      "step": 3570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999711297923725e-05,
+      "loss": 0.8599,
+      "step": 3571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999710481844571e-05,
+      "loss": 0.8987,
+      "step": 3572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999709664613799e-05,
+      "loss": 1.0715,
+      "step": 3573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9997088462314095e-05,
+      "loss": 0.6654,
+      "step": 3574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9997080266974035e-05,
+      "loss": 0.8741,
+      "step": 3575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999707206011782e-05,
+      "loss": 0.7477,
+      "step": 3576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9997063841745463e-05,
+      "loss": 0.7956,
+      "step": 3577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9997055611856972e-05,
+      "loss": 0.8637,
+      "step": 3578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999704737045235e-05,
+      "loss": 0.8856,
+      "step": 3579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9997039117531617e-05,
+      "loss": 0.924,
+      "step": 3580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9997030853094776e-05,
+      "loss": 0.8961,
+      "step": 3581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9997022577141842e-05,
+      "loss": 0.8133,
+      "step": 3582
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9997014289672813e-05,
+      "loss": 0.8816,
+      "step": 3583
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999700599068771e-05,
+      "loss": 0.7086,
+      "step": 3584
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999699768018654e-05,
+      "loss": 1.0529,
+      "step": 3585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996989358169307e-05,
+      "loss": 0.8491,
+      "step": 3586
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996981024636028e-05,
+      "loss": 0.7002,
+      "step": 3587
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996972679586706e-05,
+      "loss": 0.8526,
+      "step": 3588
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996964323021358e-05,
+      "loss": 0.9849,
+      "step": 3589
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996955954939987e-05,
+      "loss": 0.8055,
+      "step": 3590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996947575342603e-05,
+      "loss": 1.032,
+      "step": 3591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999693918422922e-05,
+      "loss": 0.7031,
+      "step": 3592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996930781599846e-05,
+      "loss": 0.7065,
+      "step": 3593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996922367454492e-05,
+      "loss": 0.8095,
+      "step": 3594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996913941793163e-05,
+      "loss": 0.9176,
+      "step": 3595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996905504615872e-05,
+      "loss": 0.7054,
+      "step": 3596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999689705592263e-05,
+      "loss": 1.0991,
+      "step": 3597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996888595713442e-05,
+      "loss": 0.7964,
+      "step": 3598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996880123988324e-05,
+      "loss": 0.6694,
+      "step": 3599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999687164074728e-05,
+      "loss": 0.7835,
+      "step": 3600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996863145990323e-05,
+      "loss": 0.5773,
+      "step": 3601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996854639717465e-05,
+      "loss": 0.9096,
+      "step": 3602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999684612192871e-05,
+      "loss": 1.1153,
+      "step": 3603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999683759262407e-05,
+      "loss": 0.7191,
+      "step": 3604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9996829051803557e-05,
+      "loss": 1.063,
+      "step": 3605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999682049946718e-05,
+      "loss": 0.822,
+      "step": 3606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999681193561495e-05,
+      "loss": 0.885,
+      "step": 3607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.999680336024687e-05,
+      "loss": 0.7472,
+      "step": 3608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996794773362956e-05,
+      "loss": 0.8571,
+      "step": 3609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.999678617496322e-05,
+      "loss": 0.6971,
+      "step": 3610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996777565047668e-05,
+      "loss": 0.8266,
+      "step": 3611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996768943616306e-05,
+      "loss": 0.9361,
+      "step": 3612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996760310669156e-05,
+      "loss": 0.9637,
+      "step": 3613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996751666206213e-05,
+      "loss": 0.9621,
+      "step": 3614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996743010227498e-05,
+      "loss": 0.8882,
+      "step": 3615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.999673434273302e-05,
+      "loss": 0.91,
+      "step": 3616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996725663722786e-05,
+      "loss": 0.7234,
+      "step": 3617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996716973196802e-05,
+      "loss": 0.6765,
+      "step": 3618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996708271155084e-05,
+      "loss": 0.8172,
+      "step": 3619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996699557597644e-05,
+      "loss": 0.7377,
+      "step": 3620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996690832524486e-05,
+      "loss": 0.9771,
+      "step": 3621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.999668209593562e-05,
+      "loss": 0.9349,
+      "step": 3622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996673347831065e-05,
+      "loss": 0.9014,
+      "step": 3623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996664588210818e-05,
+      "loss": 0.8632,
+      "step": 3624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.99966558170749e-05,
+      "loss": 0.652,
+      "step": 3625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996647034423317e-05,
+      "loss": 0.7157,
+      "step": 3626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.999663824025608e-05,
+      "loss": 0.7066,
+      "step": 3627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996629434573195e-05,
+      "loss": 0.8124,
+      "step": 3628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996620617374677e-05,
+      "loss": 0.9755,
+      "step": 3629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.999661178866053e-05,
+      "loss": 0.8656,
+      "step": 3630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996602948430773e-05,
+      "loss": 0.7739,
+      "step": 3631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996594096685414e-05,
+      "loss": 0.7963,
+      "step": 3632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.999658523342446e-05,
+      "loss": 0.7717,
+      "step": 3633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996576358647918e-05,
+      "loss": 0.7793,
+      "step": 3634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996567472355805e-05,
+      "loss": 0.7365,
+      "step": 3635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.999655857454813e-05,
+      "loss": 0.8757,
+      "step": 3636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.99965496652249e-05,
+      "loss": 0.6587,
+      "step": 3637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996540744386127e-05,
+      "loss": 0.7985,
+      "step": 3638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996531812031824e-05,
+      "loss": 0.8687,
+      "step": 3639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996522868161995e-05,
+      "loss": 0.8195,
+      "step": 3640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996513912776656e-05,
+      "loss": 0.867,
+      "step": 3641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996504945875815e-05,
+      "loss": 0.9413,
+      "step": 3642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996495967459488e-05,
+      "loss": 0.731,
+      "step": 3643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9996486977527672e-05,
+      "loss": 0.7162,
+      "step": 3644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.999647797608039e-05,
+      "loss": 0.8546,
+      "step": 3645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996468963117647e-05,
+      "loss": 0.5744,
+      "step": 3646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996459938639454e-05,
+      "loss": 0.6853,
+      "step": 3647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999645090264582e-05,
+      "loss": 0.7244,
+      "step": 3648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999644185513676e-05,
+      "loss": 0.9463,
+      "step": 3649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999643279611228e-05,
+      "loss": 0.8902,
+      "step": 3650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999642372557239e-05,
+      "loss": 0.8847,
+      "step": 3651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996414643517106e-05,
+      "loss": 0.8712,
+      "step": 3652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999640554994643e-05,
+      "loss": 0.7582,
+      "step": 3653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996396444860382e-05,
+      "loss": 0.9824,
+      "step": 3654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996387328258967e-05,
+      "loss": 0.7935,
+      "step": 3655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996378200142193e-05,
+      "loss": 0.712,
+      "step": 3656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996369060510074e-05,
+      "loss": 0.7115,
+      "step": 3657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996359909362624e-05,
+      "loss": 0.8448,
+      "step": 3658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996350746699845e-05,
+      "loss": 0.9445,
+      "step": 3659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999634157252176e-05,
+      "loss": 0.7537,
+      "step": 3660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996332386828364e-05,
+      "loss": 0.8267,
+      "step": 3661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999632318961968e-05,
+      "loss": 0.9162,
+      "step": 3662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996313980895713e-05,
+      "loss": 0.8019,
+      "step": 3663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996304760656472e-05,
+      "loss": 0.8067,
+      "step": 3664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999629552890197e-05,
+      "loss": 0.7899,
+      "step": 3665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996286285632224e-05,
+      "loss": 0.6787,
+      "step": 3666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996277030847233e-05,
+      "loss": 0.7441,
+      "step": 3667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999626776454702e-05,
+      "loss": 0.8575,
+      "step": 3668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999625848673158e-05,
+      "loss": 0.9,
+      "step": 3669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996249197400938e-05,
+      "loss": 0.7487,
+      "step": 3670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996239896555098e-05,
+      "loss": 0.7996,
+      "step": 3671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996230584194072e-05,
+      "loss": 0.8477,
+      "step": 3672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999622126031787e-05,
+      "loss": 0.7109,
+      "step": 3673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996211924926504e-05,
+      "loss": 0.8749,
+      "step": 3674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996202578019988e-05,
+      "loss": 0.8628,
+      "step": 3675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996193219598324e-05,
+      "loss": 0.7163,
+      "step": 3676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999618384966153e-05,
+      "loss": 0.7454,
+      "step": 3677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996174468209614e-05,
+      "loss": 0.7088,
+      "step": 3678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996165075242588e-05,
+      "loss": 0.8683,
+      "step": 3679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.999615567076046e-05,
+      "loss": 0.7373,
+      "step": 3680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996146254763247e-05,
+      "loss": 0.9288,
+      "step": 3681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996136827250955e-05,
+      "loss": 0.9596,
+      "step": 3682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9996127388223593e-05,
+      "loss": 0.84,
+      "step": 3683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996117937681176e-05,
+      "loss": 0.7203,
+      "step": 3684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996108475623713e-05,
+      "loss": 0.6984,
+      "step": 3685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996099002051216e-05,
+      "loss": 0.7919,
+      "step": 3686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996089516963696e-05,
+      "loss": 0.6844,
+      "step": 3687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996080020361166e-05,
+      "loss": 0.8879,
+      "step": 3688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996070512243627e-05,
+      "loss": 1.0573,
+      "step": 3689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996060992611104e-05,
+      "loss": 0.752,
+      "step": 3690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.99960514614636e-05,
+      "loss": 0.7237,
+      "step": 3691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996041918801126e-05,
+      "loss": 0.6536,
+      "step": 3692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996032364623693e-05,
+      "loss": 1.0499,
+      "step": 3693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996022798931315e-05,
+      "loss": 0.7716,
+      "step": 3694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9996013221723998e-05,
+      "loss": 0.8606,
+      "step": 3695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.999600363300176e-05,
+      "loss": 0.8004,
+      "step": 3696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995994032764603e-05,
+      "loss": 0.9224,
+      "step": 3697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995984421012545e-05,
+      "loss": 0.8944,
+      "step": 3698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.99959747977456e-05,
+      "loss": 0.6653,
+      "step": 3699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.999596516296377e-05,
+      "loss": 0.7418,
+      "step": 3700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995955516667075e-05,
+      "loss": 0.7144,
+      "step": 3701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995945858855517e-05,
+      "loss": 0.7527,
+      "step": 3702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995936189529112e-05,
+      "loss": 0.748,
+      "step": 3703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995926508687873e-05,
+      "loss": 0.6442,
+      "step": 3704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995916816331808e-05,
+      "loss": 0.9109,
+      "step": 3705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.999590711246093e-05,
+      "loss": 0.884,
+      "step": 3706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995897397075247e-05,
+      "loss": 0.7732,
+      "step": 3707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995887670174776e-05,
+      "loss": 0.8157,
+      "step": 3708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.999587793175952e-05,
+      "loss": 0.7805,
+      "step": 3709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995868181829502e-05,
+      "loss": 0.8307,
+      "step": 3710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995858420384723e-05,
+      "loss": 0.813,
+      "step": 3711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995848647425195e-05,
+      "loss": 0.8865,
+      "step": 3712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995838862950932e-05,
+      "loss": 0.846,
+      "step": 3713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995829066961948e-05,
+      "loss": 0.9142,
+      "step": 3714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.999581925945825e-05,
+      "loss": 0.9247,
+      "step": 3715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995809440439854e-05,
+      "loss": 0.9329,
+      "step": 3716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995799609906765e-05,
+      "loss": 0.9856,
+      "step": 3717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995789767858995e-05,
+      "loss": 0.8592,
+      "step": 3718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995779914296557e-05,
+      "loss": 0.7139,
+      "step": 3719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9995770049219466e-05,
+      "loss": 0.7016,
+      "step": 3720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.999576017262773e-05,
+      "loss": 0.8224,
+      "step": 3721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.999575028452136e-05,
+      "loss": 0.8437,
+      "step": 3722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999574038490037e-05,
+      "loss": 0.6318,
+      "step": 3723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995730473764766e-05,
+      "loss": 0.7101,
+      "step": 3724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995720551114564e-05,
+      "loss": 0.4902,
+      "step": 3725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995710616949776e-05,
+      "loss": 0.7076,
+      "step": 3726
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999570067127041e-05,
+      "loss": 0.8466,
+      "step": 3727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999569071407648e-05,
+      "loss": 0.9808,
+      "step": 3728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995680745367995e-05,
+      "loss": 0.8824,
+      "step": 3729
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995670765144973e-05,
+      "loss": 0.8265,
+      "step": 3730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995660773407417e-05,
+      "loss": 0.9371,
+      "step": 3731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999565077015534e-05,
+      "loss": 0.801,
+      "step": 3732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999564075538876e-05,
+      "loss": 0.7167,
+      "step": 3733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995630729107683e-05,
+      "loss": 0.6287,
+      "step": 3734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999562069131212e-05,
+      "loss": 0.6819,
+      "step": 3735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999561064200209e-05,
+      "loss": 0.7448,
+      "step": 3736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999560058117759e-05,
+      "loss": 0.7411,
+      "step": 3737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995590508838646e-05,
+      "loss": 0.7375,
+      "step": 3738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999558042498526e-05,
+      "loss": 0.7484,
+      "step": 3739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995570329617454e-05,
+      "loss": 0.6792,
+      "step": 3740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995560222735228e-05,
+      "loss": 0.768,
+      "step": 3741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995550104338605e-05,
+      "loss": 0.6763,
+      "step": 3742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995539974427585e-05,
+      "loss": 0.7547,
+      "step": 3743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995529833002187e-05,
+      "loss": 0.8483,
+      "step": 3744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995519680062422e-05,
+      "loss": 0.8767,
+      "step": 3745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.99955095156083e-05,
+      "loss": 0.8215,
+      "step": 3746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999549933963983e-05,
+      "loss": 0.8674,
+      "step": 3747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995489152157033e-05,
+      "loss": 0.8746,
+      "step": 3748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995478953159913e-05,
+      "loss": 0.7189,
+      "step": 3749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995468742648484e-05,
+      "loss": 0.7997,
+      "step": 3750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995458520622754e-05,
+      "loss": 0.8065,
+      "step": 3751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999544828708274e-05,
+      "loss": 0.9693,
+      "step": 3752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995438042028456e-05,
+      "loss": 0.8222,
+      "step": 3753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995427785459905e-05,
+      "loss": 0.7079,
+      "step": 3754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995417517377104e-05,
+      "loss": 0.9701,
+      "step": 3755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995407237780066e-05,
+      "loss": 0.6465,
+      "step": 3756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995396946668797e-05,
+      "loss": 0.773,
+      "step": 3757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995386644043318e-05,
+      "loss": 0.6647,
+      "step": 3758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9995376329903635e-05,
+      "loss": 0.8283,
+      "step": 3759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999536600424976e-05,
+      "loss": 0.8096,
+      "step": 3760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995355667081708e-05,
+      "loss": 0.9129,
+      "step": 3761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995345318399483e-05,
+      "loss": 0.8811,
+      "step": 3762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999533495820311e-05,
+      "loss": 0.8106,
+      "step": 3763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995324586492586e-05,
+      "loss": 0.6355,
+      "step": 3764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995314203267934e-05,
+      "loss": 0.6251,
+      "step": 3765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995303808529164e-05,
+      "loss": 0.9238,
+      "step": 3766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995293402276285e-05,
+      "loss": 0.9221,
+      "step": 3767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999528298450931e-05,
+      "loss": 0.8728,
+      "step": 3768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999527255522825e-05,
+      "loss": 1.0485,
+      "step": 3769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995262114433122e-05,
+      "loss": 0.6635,
+      "step": 3770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999525166212393e-05,
+      "loss": 0.7538,
+      "step": 3771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995241198300693e-05,
+      "loss": 1.01,
+      "step": 3772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999523072296342e-05,
+      "loss": 0.7639,
+      "step": 3773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995220236112124e-05,
+      "loss": 0.7691,
+      "step": 3774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995209737746818e-05,
+      "loss": 0.8473,
+      "step": 3775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999519922786751e-05,
+      "loss": 0.9941,
+      "step": 3776
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995188706474215e-05,
+      "loss": 0.8331,
+      "step": 3777
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995178173566945e-05,
+      "loss": 0.7957,
+      "step": 3778
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995167629145717e-05,
+      "loss": 0.7237,
+      "step": 3779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995157073210532e-05,
+      "loss": 0.7309,
+      "step": 3780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999514650576141e-05,
+      "loss": 0.8185,
+      "step": 3781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995135926798362e-05,
+      "loss": 0.8204,
+      "step": 3782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.99951253363214e-05,
+      "loss": 0.6248,
+      "step": 3783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995114734330537e-05,
+      "loss": 0.6952,
+      "step": 3784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995104120825782e-05,
+      "loss": 0.8722,
+      "step": 3785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995093495807153e-05,
+      "loss": 0.9055,
+      "step": 3786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995082859274655e-05,
+      "loss": 0.7147,
+      "step": 3787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995072211228307e-05,
+      "loss": 0.8805,
+      "step": 3788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995061551668114e-05,
+      "loss": 0.8,
+      "step": 3789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9995050880594094e-05,
+      "loss": 0.7097,
+      "step": 3790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999504019800626e-05,
+      "loss": 0.6832,
+      "step": 3791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999502950390462e-05,
+      "loss": 0.7647,
+      "step": 3792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999501879828919e-05,
+      "loss": 0.8075,
+      "step": 3793
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999500808115998e-05,
+      "loss": 0.7763,
+      "step": 3794
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9994997352517002e-05,
+      "loss": 0.8018,
+      "step": 3795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999498661236027e-05,
+      "loss": 0.7989,
+      "step": 3796
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9994975860689797e-05,
+      "loss": 0.8296,
+      "step": 3797
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9994965097505593e-05,
+      "loss": 0.8448,
+      "step": 3798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999495432280767e-05,
+      "loss": 0.8423,
+      "step": 3799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994943536596045e-05,
+      "loss": 0.8286,
+      "step": 3800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994932738870724e-05,
+      "loss": 0.913,
+      "step": 3801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994921929631726e-05,
+      "loss": 0.685,
+      "step": 3802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999491110887906e-05,
+      "loss": 0.9005,
+      "step": 3803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994900276612738e-05,
+      "loss": 0.8113,
+      "step": 3804
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994889432832773e-05,
+      "loss": 0.8832,
+      "step": 3805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994878577539177e-05,
+      "loss": 0.6449,
+      "step": 3806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994867710731963e-05,
+      "loss": 0.6954,
+      "step": 3807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994856832411146e-05,
+      "loss": 0.8403,
+      "step": 3808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994845942576732e-05,
+      "loss": 0.58,
+      "step": 3809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994835041228742e-05,
+      "loss": 0.8037,
+      "step": 3810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994824128367182e-05,
+      "loss": 0.8683,
+      "step": 3811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999481320399207e-05,
+      "loss": 0.6816,
+      "step": 3812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994802268103413e-05,
+      "loss": 0.765,
+      "step": 3813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994791320701225e-05,
+      "loss": 0.7548,
+      "step": 3814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999478036178552e-05,
+      "loss": 0.7194,
+      "step": 3815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994769391356315e-05,
+      "loss": 0.8224,
+      "step": 3816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994758409413614e-05,
+      "loss": 0.9312,
+      "step": 3817
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994747415957435e-05,
+      "loss": 0.8979,
+      "step": 3818
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994736410987788e-05,
+      "loss": 0.7527,
+      "step": 3819
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999472539450469e-05,
+      "loss": 1.0855,
+      "step": 3820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994714366508146e-05,
+      "loss": 0.8183,
+      "step": 3821
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994703326998176e-05,
+      "loss": 0.6974,
+      "step": 3822
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999469227597479e-05,
+      "loss": 0.5947,
+      "step": 3823
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994681213438e-05,
+      "loss": 0.8285,
+      "step": 3824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999467013938782e-05,
+      "loss": 0.732,
+      "step": 3825
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994659053824263e-05,
+      "loss": 0.6669,
+      "step": 3826
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999464795674734e-05,
+      "loss": 0.6872,
+      "step": 3827
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994636848157067e-05,
+      "loss": 0.8763,
+      "step": 3828
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994625728053453e-05,
+      "loss": 0.6813,
+      "step": 3829
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994614596436513e-05,
+      "loss": 0.5293,
+      "step": 3830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999460345330626e-05,
+      "loss": 0.7923,
+      "step": 3831
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994592298662704e-05,
+      "loss": 0.9079,
+      "step": 3832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.999458113250586e-05,
+      "loss": 0.7625,
+      "step": 3833
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994569954835744e-05,
+      "loss": 0.9081,
+      "step": 3834
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994558765652365e-05,
+      "loss": 0.7833,
+      "step": 3835
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9994547564955734e-05,
+      "loss": 0.834,
+      "step": 3836
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999453635274587e-05,
+      "loss": 0.6591,
+      "step": 3837
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999452512902278e-05,
+      "loss": 0.7221,
+      "step": 3838
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994513893786478e-05,
+      "loss": 0.69,
+      "step": 3839
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999450264703698e-05,
+      "loss": 0.6819,
+      "step": 3840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994491388774298e-05,
+      "loss": 0.7264,
+      "step": 3841
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994480118998444e-05,
+      "loss": 0.5804,
+      "step": 3842
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999446883770943e-05,
+      "loss": 0.8809,
+      "step": 3843
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999445754490727e-05,
+      "loss": 0.7167,
+      "step": 3844
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999444624059198e-05,
+      "loss": 0.9152,
+      "step": 3845
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994434924763568e-05,
+      "loss": 0.833,
+      "step": 3846
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999442359742205e-05,
+      "loss": 0.9512,
+      "step": 3847
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994412258567438e-05,
+      "loss": 0.819,
+      "step": 3848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994400908199747e-05,
+      "loss": 0.7939,
+      "step": 3849
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994389546318988e-05,
+      "loss": 0.9623,
+      "step": 3850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999437817292517e-05,
+      "loss": 0.9727,
+      "step": 3851
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994366788018315e-05,
+      "loss": 0.703,
+      "step": 3852
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994355391598432e-05,
+      "loss": 0.6167,
+      "step": 3853
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994343983665532e-05,
+      "loss": 0.7539,
+      "step": 3854
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999433256421963e-05,
+      "loss": 0.8801,
+      "step": 3855
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994321133260743e-05,
+      "loss": 0.6654,
+      "step": 3856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994309690788874e-05,
+      "loss": 0.8583,
+      "step": 3857
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999429823680405e-05,
+      "loss": 0.6478,
+      "step": 3858
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999428677130627e-05,
+      "loss": 0.8394,
+      "step": 3859
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994275294295557e-05,
+      "loss": 0.7233,
+      "step": 3860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994263805771924e-05,
+      "loss": 0.7776,
+      "step": 3861
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994252305735378e-05,
+      "loss": 0.7617,
+      "step": 3862
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994240794185937e-05,
+      "loss": 0.8825,
+      "step": 3863
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999422927112361e-05,
+      "loss": 0.8188,
+      "step": 3864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994217736548415e-05,
+      "loss": 0.7816,
+      "step": 3865
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994206190460365e-05,
+      "loss": 0.888,
+      "step": 3866
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994194632859466e-05,
+      "loss": 0.8276,
+      "step": 3867
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994183063745744e-05,
+      "loss": 0.6227,
+      "step": 3868
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994171483119204e-05,
+      "loss": 0.8603,
+      "step": 3869
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999415989097986e-05,
+      "loss": 0.7538,
+      "step": 3870
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994148287327725e-05,
+      "loss": 0.9564,
+      "step": 3871
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994136672162813e-05,
+      "loss": 0.7391,
+      "step": 3872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.999412504548514e-05,
+      "loss": 0.7712,
+      "step": 3873
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9994113407294716e-05,
+      "loss": 1.0765,
+      "step": 3874
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9994101757591555e-05,
+      "loss": 0.8116,
+      "step": 3875
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9994090096375675e-05,
+      "loss": 0.923,
+      "step": 3876
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999407842364708e-05,
+      "loss": 0.8856,
+      "step": 3877
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999406673940579e-05,
+      "loss": 0.8299,
+      "step": 3878
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9994055043651822e-05,
+      "loss": 0.8291,
+      "step": 3879
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999404333638518e-05,
+      "loss": 0.6604,
+      "step": 3880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9994031617605888e-05,
+      "loss": 1.024,
+      "step": 3881
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999401988731395e-05,
+      "loss": 0.7648,
+      "step": 3882
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999400814550938e-05,
+      "loss": 0.8842,
+      "step": 3883
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.99939963921922e-05,
+      "loss": 0.6048,
+      "step": 3884
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993984627362418e-05,
+      "loss": 0.8009,
+      "step": 3885
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993972851020045e-05,
+      "loss": 0.9216,
+      "step": 3886
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.99939610631651e-05,
+      "loss": 0.5949,
+      "step": 3887
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993949263797596e-05,
+      "loss": 0.617,
+      "step": 3888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999393745291754e-05,
+      "loss": 0.7511,
+      "step": 3889
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993925630524953e-05,
+      "loss": 0.7268,
+      "step": 3890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993913796619844e-05,
+      "loss": 0.8632,
+      "step": 3891
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993901951202233e-05,
+      "loss": 0.6842,
+      "step": 3892
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993890094272122e-05,
+      "loss": 0.6688,
+      "step": 3893
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999387822582954e-05,
+      "loss": 0.7769,
+      "step": 3894
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993866345874488e-05,
+      "loss": 0.7345,
+      "step": 3895
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993854454406985e-05,
+      "loss": 0.7903,
+      "step": 3896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993842551427044e-05,
+      "loss": 0.6919,
+      "step": 3897
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999383063693468e-05,
+      "loss": 0.9316,
+      "step": 3898
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993818710929905e-05,
+      "loss": 0.7105,
+      "step": 3899
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999380677341273e-05,
+      "loss": 0.952,
+      "step": 3900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993794824383176e-05,
+      "loss": 0.7496,
+      "step": 3901
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999378286384125e-05,
+      "loss": 0.7736,
+      "step": 3902
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999377089178697e-05,
+      "loss": 0.759,
+      "step": 3903
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993758908220347e-05,
+      "loss": 0.8776,
+      "step": 3904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.99937469131414e-05,
+      "loss": 1.1241,
+      "step": 3905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999373490655013e-05,
+      "loss": 0.779,
+      "step": 3906
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993722888446568e-05,
+      "loss": 0.7676,
+      "step": 3907
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993710858830716e-05,
+      "loss": 0.7483,
+      "step": 3908
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993698817702594e-05,
+      "loss": 0.8149,
+      "step": 3909
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.999368676506221e-05,
+      "loss": 0.6952,
+      "step": 3910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993674700909583e-05,
+      "loss": 0.7692,
+      "step": 3911
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9993662625244723e-05,
+      "loss": 0.9388,
+      "step": 3912
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999365053806765e-05,
+      "loss": 0.6909,
+      "step": 3913
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999363843937837e-05,
+      "loss": 0.8208,
+      "step": 3914
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.99936263291769e-05,
+      "loss": 0.7695,
+      "step": 3915
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993614207463257e-05,
+      "loss": 0.735,
+      "step": 3916
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993602074237454e-05,
+      "loss": 1.052,
+      "step": 3917
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993589929499503e-05,
+      "loss": 0.7013,
+      "step": 3918
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993577773249417e-05,
+      "loss": 0.673,
+      "step": 3919
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993565605487213e-05,
+      "loss": 0.6873,
+      "step": 3920
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993553426212904e-05,
+      "loss": 0.7254,
+      "step": 3921
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993541235426502e-05,
+      "loss": 0.807,
+      "step": 3922
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993529033128025e-05,
+      "loss": 0.6756,
+      "step": 3923
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993516819317485e-05,
+      "loss": 0.7561,
+      "step": 3924
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993504593994894e-05,
+      "loss": 0.6852,
+      "step": 3925
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993492357160267e-05,
+      "loss": 0.7704,
+      "step": 3926
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999348010881362e-05,
+      "loss": 0.7801,
+      "step": 3927
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993467848954967e-05,
+      "loss": 0.7024,
+      "step": 3928
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999345557758432e-05,
+      "loss": 0.7446,
+      "step": 3929
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993443294701697e-05,
+      "loss": 0.6015,
+      "step": 3930
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999343100030711e-05,
+      "loss": 0.7548,
+      "step": 3931
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993418694400568e-05,
+      "loss": 0.5978,
+      "step": 3932
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993406376982093e-05,
+      "loss": 0.6511,
+      "step": 3933
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993394048051696e-05,
+      "loss": 0.6749,
+      "step": 3934
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999338170760939e-05,
+      "loss": 0.7413,
+      "step": 3935
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993369355655195e-05,
+      "loss": 0.703,
+      "step": 3936
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993356992189116e-05,
+      "loss": 0.7016,
+      "step": 3937
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993344617211175e-05,
+      "loss": 0.7645,
+      "step": 3938
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999333223072138e-05,
+      "loss": 0.7863,
+      "step": 3939
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999331983271975e-05,
+      "loss": 0.8319,
+      "step": 3940
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.99933074232063e-05,
+      "loss": 0.7837,
+      "step": 3941
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999329500218104e-05,
+      "loss": 0.6303,
+      "step": 3942
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993282569643987e-05,
+      "loss": 0.8997,
+      "step": 3943
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993270125595154e-05,
+      "loss": 0.8073,
+      "step": 3944
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993257670034557e-05,
+      "loss": 0.8009,
+      "step": 3945
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999324520296221e-05,
+      "loss": 0.7607,
+      "step": 3946
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993232724378126e-05,
+      "loss": 0.6074,
+      "step": 3947
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.999322023428232e-05,
+      "loss": 0.6459,
+      "step": 3948
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993207732674805e-05,
+      "loss": 0.6961,
+      "step": 3949
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993195219555595e-05,
+      "loss": 0.8114,
+      "step": 3950
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9993182694924713e-05,
+      "loss": 0.7063,
+      "step": 3951
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993170158782162e-05,
+      "loss": 0.8115,
+      "step": 3952
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993157611127965e-05,
+      "loss": 0.8621,
+      "step": 3953
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.999314505196213e-05,
+      "loss": 0.8568,
+      "step": 3954
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993132481284677e-05,
+      "loss": 0.9638,
+      "step": 3955
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993119899095613e-05,
+      "loss": 0.6443,
+      "step": 3956
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.999310730539496e-05,
+      "loss": 0.7019,
+      "step": 3957
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993094700182727e-05,
+      "loss": 0.9038,
+      "step": 3958
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993082083458933e-05,
+      "loss": 0.8837,
+      "step": 3959
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993069455223592e-05,
+      "loss": 0.6448,
+      "step": 3960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993056815476718e-05,
+      "loss": 0.8049,
+      "step": 3961
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993044164218323e-05,
+      "loss": 0.8014,
+      "step": 3962
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993031501448422e-05,
+      "loss": 0.8168,
+      "step": 3963
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993018827167032e-05,
+      "loss": 0.7292,
+      "step": 3964
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9993006141374166e-05,
+      "loss": 0.8145,
+      "step": 3965
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.999299344406984e-05,
+      "loss": 0.651,
+      "step": 3966
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.999298073525407e-05,
+      "loss": 0.8538,
+      "step": 3967
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992968014926864e-05,
+      "loss": 0.6576,
+      "step": 3968
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992955283088243e-05,
+      "loss": 0.6501,
+      "step": 3969
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.999294253973822e-05,
+      "loss": 0.8042,
+      "step": 3970
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992929784876815e-05,
+      "loss": 0.8571,
+      "step": 3971
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992917018504027e-05,
+      "loss": 0.6899,
+      "step": 3972
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992904240619886e-05,
+      "loss": 0.9957,
+      "step": 3973
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.99928914512244e-05,
+      "loss": 0.819,
+      "step": 3974
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992878650317588e-05,
+      "loss": 0.5324,
+      "step": 3975
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.999286583789946e-05,
+      "loss": 0.7271,
+      "step": 3976
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992853013970034e-05,
+      "loss": 0.6885,
+      "step": 3977
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992840178529326e-05,
+      "loss": 0.7642,
+      "step": 3978
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992827331577342e-05,
+      "loss": 0.7588,
+      "step": 3979
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992814473114106e-05,
+      "loss": 0.7853,
+      "step": 3980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.999280160313963e-05,
+      "loss": 0.9966,
+      "step": 3981
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.999278872165393e-05,
+      "loss": 0.9892,
+      "step": 3982
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.999277582865702e-05,
+      "loss": 0.8447,
+      "step": 3983
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992762924148913e-05,
+      "loss": 0.8376,
+      "step": 3984
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992750008129625e-05,
+      "loss": 0.8731,
+      "step": 3985
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992737080599173e-05,
+      "loss": 0.7269,
+      "step": 3986
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.999272414155757e-05,
+      "loss": 0.805,
+      "step": 3987
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992711191004828e-05,
+      "loss": 0.8749,
+      "step": 3988
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9992698228940965e-05,
+      "loss": 0.8617,
+      "step": 3989
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992685255366e-05,
+      "loss": 0.679,
+      "step": 3990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999267227027994e-05,
+      "loss": 0.8094,
+      "step": 3991
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992659273682807e-05,
+      "loss": 0.8359,
+      "step": 3992
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992646265574612e-05,
+      "loss": 0.8703,
+      "step": 3993
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999263324595537e-05,
+      "loss": 0.8909,
+      "step": 3994
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992620214825096e-05,
+      "loss": 0.9042,
+      "step": 3995
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992607172183805e-05,
+      "loss": 0.6699,
+      "step": 3996
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992594118031516e-05,
+      "loss": 0.9438,
+      "step": 3997
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992581052368236e-05,
+      "loss": 0.8526,
+      "step": 3998
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999256797519399e-05,
+      "loss": 0.7965,
+      "step": 3999
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992554886508782e-05,
+      "loss": 0.8797,
+      "step": 4000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999254178631264e-05,
+      "loss": 0.6969,
+      "step": 4001
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999252867460557e-05,
+      "loss": 0.7435,
+      "step": 4002
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999251555138759e-05,
+      "loss": 0.7849,
+      "step": 4003
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999250241665871e-05,
+      "loss": 0.8372,
+      "step": 4004
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999248927041895e-05,
+      "loss": 0.8563,
+      "step": 4005
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992476112668328e-05,
+      "loss": 0.8848,
+      "step": 4006
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992462943406853e-05,
+      "loss": 0.8819,
+      "step": 4007
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992449762634546e-05,
+      "loss": 0.8806,
+      "step": 4008
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992436570351414e-05,
+      "loss": 0.9235,
+      "step": 4009
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999242336655748e-05,
+      "loss": 0.7886,
+      "step": 4010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999241015125276e-05,
+      "loss": 0.7139,
+      "step": 4011
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992396924437262e-05,
+      "loss": 0.7476,
+      "step": 4012
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992383686111004e-05,
+      "loss": 0.8484,
+      "step": 4013
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992370436274006e-05,
+      "loss": 0.7203,
+      "step": 4014
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992357174926277e-05,
+      "loss": 0.7767,
+      "step": 4015
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992343902067835e-05,
+      "loss": 0.7123,
+      "step": 4016
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992330617698696e-05,
+      "loss": 0.6813,
+      "step": 4017
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992317321818874e-05,
+      "loss": 0.801,
+      "step": 4018
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992304014428382e-05,
+      "loss": 0.8597,
+      "step": 4019
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992290695527238e-05,
+      "loss": 0.753,
+      "step": 4020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999227736511546e-05,
+      "loss": 0.6832,
+      "step": 4021
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992264023193063e-05,
+      "loss": 0.7649,
+      "step": 4022
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992250669760052e-05,
+      "loss": 0.8403,
+      "step": 4023
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992237304816457e-05,
+      "loss": 0.6846,
+      "step": 4024
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992223928362287e-05,
+      "loss": 0.6766,
+      "step": 4025
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9992210540397556e-05,
+      "loss": 0.7659,
+      "step": 4026
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.999219714092228e-05,
+      "loss": 0.7816,
+      "step": 4027
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992183729936475e-05,
+      "loss": 0.9746,
+      "step": 4028
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.999217030744016e-05,
+      "loss": 0.74,
+      "step": 4029
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992156873433342e-05,
+      "loss": 0.8857,
+      "step": 4030
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992143427916043e-05,
+      "loss": 0.8075,
+      "step": 4031
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992129970888277e-05,
+      "loss": 0.7422,
+      "step": 4032
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992116502350062e-05,
+      "loss": 0.5861,
+      "step": 4033
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992103022301408e-05,
+      "loss": 0.7551,
+      "step": 4034
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992089530742335e-05,
+      "loss": 0.673,
+      "step": 4035
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992076027672857e-05,
+      "loss": 0.8073,
+      "step": 4036
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.999206251309299e-05,
+      "loss": 0.7393,
+      "step": 4037
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992048987002745e-05,
+      "loss": 0.7495,
+      "step": 4038
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992035449402146e-05,
+      "loss": 0.8402,
+      "step": 4039
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992021900291206e-05,
+      "loss": 0.7839,
+      "step": 4040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9992008339669935e-05,
+      "loss": 0.7436,
+      "step": 4041
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991994767538354e-05,
+      "loss": 0.7889,
+      "step": 4042
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.999198118389648e-05,
+      "loss": 0.758,
+      "step": 4043
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991967588744318e-05,
+      "loss": 0.6263,
+      "step": 4044
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.99919539820819e-05,
+      "loss": 0.7037,
+      "step": 4045
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.999194036390923e-05,
+      "loss": 0.7482,
+      "step": 4046
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991926734226324e-05,
+      "loss": 0.8974,
+      "step": 4047
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991913093033206e-05,
+      "loss": 0.8049,
+      "step": 4048
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991899440329883e-05,
+      "loss": 0.7893,
+      "step": 4049
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991885776116374e-05,
+      "loss": 0.6971,
+      "step": 4050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991872100392697e-05,
+      "loss": 0.7313,
+      "step": 4051
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991858413158863e-05,
+      "loss": 0.595,
+      "step": 4052
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991844714414894e-05,
+      "loss": 0.8438,
+      "step": 4053
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991831004160797e-05,
+      "loss": 0.7334,
+      "step": 4054
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991817282396594e-05,
+      "loss": 0.8224,
+      "step": 4055
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991803549122304e-05,
+      "loss": 0.5992,
+      "step": 4056
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991789804337937e-05,
+      "loss": 0.827,
+      "step": 4057
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991776048043507e-05,
+      "loss": 0.8,
+      "step": 4058
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991762280239035e-05,
+      "loss": 0.6456,
+      "step": 4059
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991748500924536e-05,
+      "loss": 0.7538,
+      "step": 4060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991734710100022e-05,
+      "loss": 0.6864,
+      "step": 4061
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991720907765516e-05,
+      "loss": 0.8587,
+      "step": 4062
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991707093921025e-05,
+      "loss": 0.6322,
+      "step": 4063
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991693268566573e-05,
+      "loss": 0.7063,
+      "step": 4064
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9991679431702173e-05,
+      "loss": 0.7466,
+      "step": 4065
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991665583327838e-05,
+      "loss": 0.6605,
+      "step": 4066
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999165172344359e-05,
+      "loss": 0.626,
+      "step": 4067
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991637852049437e-05,
+      "loss": 0.728,
+      "step": 4068
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.99916239691454e-05,
+      "loss": 0.8326,
+      "step": 4069
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.99916100747315e-05,
+      "loss": 0.7439,
+      "step": 4070
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999159616880774e-05,
+      "loss": 0.8154,
+      "step": 4071
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991582251374145e-05,
+      "loss": 0.5846,
+      "step": 4072
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991568322430734e-05,
+      "loss": 0.6854,
+      "step": 4073
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991554381977515e-05,
+      "loss": 0.7666,
+      "step": 4074
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999154043001451e-05,
+      "loss": 0.6066,
+      "step": 4075
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991526466541727e-05,
+      "loss": 0.74,
+      "step": 4076
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991512491559192e-05,
+      "loss": 0.6549,
+      "step": 4077
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991498505066916e-05,
+      "loss": 0.8315,
+      "step": 4078
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991484507064916e-05,
+      "loss": 0.7091,
+      "step": 4079
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991470497553206e-05,
+      "loss": 0.7532,
+      "step": 4080
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999145647653181e-05,
+      "loss": 0.9212,
+      "step": 4081
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991442444000733e-05,
+      "loss": 0.8682,
+      "step": 4082
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999142839996e-05,
+      "loss": 0.7281,
+      "step": 4083
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999141434440962e-05,
+      "loss": 0.6605,
+      "step": 4084
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991400277349612e-05,
+      "loss": 0.8369,
+      "step": 4085
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991386198779996e-05,
+      "loss": 0.7715,
+      "step": 4086
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991372108700785e-05,
+      "loss": 0.6933,
+      "step": 4087
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991358007111995e-05,
+      "loss": 0.6764,
+      "step": 4088
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991343894013644e-05,
+      "loss": 0.8621,
+      "step": 4089
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991329769405748e-05,
+      "loss": 0.8683,
+      "step": 4090
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999131563328832e-05,
+      "loss": 0.7303,
+      "step": 4091
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999130148566138e-05,
+      "loss": 0.7873,
+      "step": 4092
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999128732652494e-05,
+      "loss": 0.7547,
+      "step": 4093
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999127315587902e-05,
+      "loss": 0.8714,
+      "step": 4094
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991258973723636e-05,
+      "loss": 0.8077,
+      "step": 4095
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991244780058804e-05,
+      "loss": 0.758,
+      "step": 4096
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991230574884542e-05,
+      "loss": 0.8674,
+      "step": 4097
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.999121635820086e-05,
+      "loss": 0.9147,
+      "step": 4098
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991202130007785e-05,
+      "loss": 0.8606,
+      "step": 4099
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991187890305323e-05,
+      "loss": 0.6584,
+      "step": 4100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991173639093496e-05,
+      "loss": 0.7623,
+      "step": 4101
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991159376372318e-05,
+      "loss": 0.7591,
+      "step": 4102
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9991145102141808e-05,
+      "loss": 0.6961,
+      "step": 4103
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9991130816401984e-05,
+      "loss": 1.0336,
+      "step": 4104
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9991116519152852e-05,
+      "loss": 0.7942,
+      "step": 4105
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.999110221039444e-05,
+      "loss": 0.8814,
+      "step": 4106
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9991087890126765e-05,
+      "loss": 0.7646,
+      "step": 4107
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9991073558349833e-05,
+      "loss": 0.7876,
+      "step": 4108
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9991059215063665e-05,
+      "loss": 0.7845,
+      "step": 4109
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9991044860268283e-05,
+      "loss": 0.855,
+      "step": 4110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9991030493963697e-05,
+      "loss": 0.7068,
+      "step": 4111
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9991016116149928e-05,
+      "loss": 0.8623,
+      "step": 4112
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.999100172682699e-05,
+      "loss": 0.8402,
+      "step": 4113
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.99909873259949e-05,
+      "loss": 0.7974,
+      "step": 4114
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990972913653673e-05,
+      "loss": 0.708,
+      "step": 4115
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990958489803328e-05,
+      "loss": 0.6513,
+      "step": 4116
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990944054443885e-05,
+      "loss": 0.8977,
+      "step": 4117
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990929607575352e-05,
+      "loss": 0.6715,
+      "step": 4118
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990915149197755e-05,
+      "loss": 0.603,
+      "step": 4119
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.99909006793111e-05,
+      "loss": 0.7171,
+      "step": 4120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990886197915415e-05,
+      "loss": 0.8329,
+      "step": 4121
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990871705010705e-05,
+      "loss": 0.8887,
+      "step": 4122
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990857200596997e-05,
+      "loss": 0.7774,
+      "step": 4123
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990842684674305e-05,
+      "loss": 0.7393,
+      "step": 4124
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990828157242642e-05,
+      "loss": 0.8526,
+      "step": 4125
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990813618302028e-05,
+      "loss": 0.7004,
+      "step": 4126
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990799067852478e-05,
+      "loss": 0.831,
+      "step": 4127
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.999078450589401e-05,
+      "loss": 0.883,
+      "step": 4128
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.999076993242664e-05,
+      "loss": 0.8478,
+      "step": 4129
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990755347450386e-05,
+      "loss": 0.9247,
+      "step": 4130
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990740750965263e-05,
+      "loss": 0.7341,
+      "step": 4131
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990726142971287e-05,
+      "loss": 0.7332,
+      "step": 4132
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.999071152346848e-05,
+      "loss": 0.5933,
+      "step": 4133
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990696892456855e-05,
+      "loss": 0.683,
+      "step": 4134
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.999068224993643e-05,
+      "loss": 0.8946,
+      "step": 4135
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.999066759590722e-05,
+      "loss": 0.6396,
+      "step": 4136
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990652930369243e-05,
+      "loss": 0.7655,
+      "step": 4137
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990638253322515e-05,
+      "loss": 0.8671,
+      "step": 4138
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990623564767055e-05,
+      "loss": 0.7572,
+      "step": 4139
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.999060886470288e-05,
+      "loss": 0.6,
+      "step": 4140
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9990594153130006e-05,
+      "loss": 0.6644,
+      "step": 4141
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990579430048447e-05,
+      "loss": 0.7543,
+      "step": 4142
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990564695458224e-05,
+      "loss": 0.6644,
+      "step": 4143
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990549949359354e-05,
+      "loss": 0.7796,
+      "step": 4144
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990535191751847e-05,
+      "loss": 0.9645,
+      "step": 4145
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999052042263573e-05,
+      "loss": 0.8144,
+      "step": 4146
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990505642011016e-05,
+      "loss": 0.6781,
+      "step": 4147
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990490849877718e-05,
+      "loss": 0.6379,
+      "step": 4148
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999047604623586e-05,
+      "loss": 0.6793,
+      "step": 4149
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990461231085456e-05,
+      "loss": 0.6553,
+      "step": 4150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990446404426523e-05,
+      "loss": 0.9499,
+      "step": 4151
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990431566259078e-05,
+      "loss": 0.6539,
+      "step": 4152
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990416716583136e-05,
+      "loss": 0.7924,
+      "step": 4153
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990401855398716e-05,
+      "loss": 0.6364,
+      "step": 4154
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999038698270584e-05,
+      "loss": 0.7691,
+      "step": 4155
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990372098504515e-05,
+      "loss": 0.7219,
+      "step": 4156
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990357202794765e-05,
+      "loss": 0.711,
+      "step": 4157
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990342295576605e-05,
+      "loss": 0.8131,
+      "step": 4158
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990327376850052e-05,
+      "loss": 0.8116,
+      "step": 4159
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990312446615127e-05,
+      "loss": 0.8083,
+      "step": 4160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999029750487184e-05,
+      "loss": 0.7175,
+      "step": 4161
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990282551620213e-05,
+      "loss": 0.7834,
+      "step": 4162
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999026758686027e-05,
+      "loss": 0.7333,
+      "step": 4163
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999025261059201e-05,
+      "loss": 1.0466,
+      "step": 4164
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990237622815467e-05,
+      "loss": 0.6008,
+      "step": 4165
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990222623530653e-05,
+      "loss": 1.0715,
+      "step": 4166
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990207612737582e-05,
+      "loss": 0.7876,
+      "step": 4167
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990192590436278e-05,
+      "loss": 0.8181,
+      "step": 4168
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999017755662675e-05,
+      "loss": 0.7714,
+      "step": 4169
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990162511309023e-05,
+      "loss": 0.6125,
+      "step": 4170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990147454483106e-05,
+      "loss": 0.7522,
+      "step": 4171
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990132386149027e-05,
+      "loss": 0.7152,
+      "step": 4172
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990117306306795e-05,
+      "loss": 0.6967,
+      "step": 4173
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999010221495643e-05,
+      "loss": 0.8293,
+      "step": 4174
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990087112097948e-05,
+      "loss": 0.7066,
+      "step": 4175
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999007199773137e-05,
+      "loss": 0.8813,
+      "step": 4176
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999005687185671e-05,
+      "loss": 0.7936,
+      "step": 4177
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9990041734473987e-05,
+      "loss": 0.7292,
+      "step": 4178
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999002658558322e-05,
+      "loss": 0.776,
+      "step": 4179
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.999001142518442e-05,
+      "loss": 0.8293,
+      "step": 4180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998999625327761e-05,
+      "loss": 0.8325,
+      "step": 4181
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989981069862808e-05,
+      "loss": 0.7088,
+      "step": 4182
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998996587494003e-05,
+      "loss": 0.8784,
+      "step": 4183
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998995066850929e-05,
+      "loss": 0.8871,
+      "step": 4184
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989935450570613e-05,
+      "loss": 0.6762,
+      "step": 4185
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989920221124013e-05,
+      "loss": 0.7361,
+      "step": 4186
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989904980169505e-05,
+      "loss": 0.8355,
+      "step": 4187
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998988972770711e-05,
+      "loss": 0.6958,
+      "step": 4188
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989874463736842e-05,
+      "loss": 0.9321,
+      "step": 4189
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998985918825872e-05,
+      "loss": 0.9228,
+      "step": 4190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989843901272764e-05,
+      "loss": 0.8187,
+      "step": 4191
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989828602778992e-05,
+      "loss": 0.7486,
+      "step": 4192
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989813292777418e-05,
+      "loss": 0.8025,
+      "step": 4193
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998979797126806e-05,
+      "loss": 0.6383,
+      "step": 4194
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989782638250937e-05,
+      "loss": 0.7094,
+      "step": 4195
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998976729372607e-05,
+      "loss": 0.6751,
+      "step": 4196
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989751937693468e-05,
+      "loss": 0.8706,
+      "step": 4197
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998973657015316e-05,
+      "loss": 0.7071,
+      "step": 4198
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998972119110515e-05,
+      "loss": 0.7275,
+      "step": 4199
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998970580054947e-05,
+      "loss": 0.7644,
+      "step": 4200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989690398486127e-05,
+      "loss": 0.7504,
+      "step": 4201
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989674984915144e-05,
+      "loss": 0.9613,
+      "step": 4202
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989659559836537e-05,
+      "loss": 0.8278,
+      "step": 4203
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989644123250324e-05,
+      "loss": 0.6024,
+      "step": 4204
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989628675156524e-05,
+      "loss": 0.7284,
+      "step": 4205
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998961321555515e-05,
+      "loss": 0.7461,
+      "step": 4206
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998959774444623e-05,
+      "loss": 0.5907,
+      "step": 4207
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989582261829773e-05,
+      "loss": 0.7719,
+      "step": 4208
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.99895667677058e-05,
+      "loss": 0.8974,
+      "step": 4209
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989551262074323e-05,
+      "loss": 0.6354,
+      "step": 4210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998953574493537e-05,
+      "loss": 0.8819,
+      "step": 4211
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989520216288952e-05,
+      "loss": 0.761,
+      "step": 4212
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989504676135092e-05,
+      "loss": 0.872,
+      "step": 4213
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.99894891244738e-05,
+      "loss": 0.6968,
+      "step": 4214
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989473561305104e-05,
+      "loss": 0.8692,
+      "step": 4215
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9989457986629013e-05,
+      "loss": 0.7026,
+      "step": 4216
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998944240044555e-05,
+      "loss": 0.8636,
+      "step": 4217
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.998942680275473e-05,
+      "loss": 0.8361,
+      "step": 4218
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989411193556575e-05,
+      "loss": 0.9462,
+      "step": 4219
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989395572851094e-05,
+      "loss": 0.6793,
+      "step": 4220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998937994063832e-05,
+      "loss": 0.9443,
+      "step": 4221
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989364296918255e-05,
+      "loss": 0.8058,
+      "step": 4222
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989348641690927e-05,
+      "loss": 0.7406,
+      "step": 4223
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989332974956355e-05,
+      "loss": 0.6536,
+      "step": 4224
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998931729671455e-05,
+      "loss": 0.6389,
+      "step": 4225
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989301606965532e-05,
+      "loss": 0.9323,
+      "step": 4226
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989285905709323e-05,
+      "loss": 0.7344,
+      "step": 4227
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989270192945936e-05,
+      "loss": 0.698,
+      "step": 4228
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989254468675398e-05,
+      "loss": 0.902,
+      "step": 4229
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989238732897716e-05,
+      "loss": 0.77,
+      "step": 4230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989222985612913e-05,
+      "loss": 0.923,
+      "step": 4231
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998920722682101e-05,
+      "loss": 0.7844,
+      "step": 4232
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989191456522017e-05,
+      "loss": 0.6711,
+      "step": 4233
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989175674715963e-05,
+      "loss": 0.7133,
+      "step": 4234
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998915988140286e-05,
+      "loss": 0.5877,
+      "step": 4235
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989144076582724e-05,
+      "loss": 1.0404,
+      "step": 4236
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998912826025558e-05,
+      "loss": 0.8397,
+      "step": 4237
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998911243242144e-05,
+      "loss": 0.6659,
+      "step": 4238
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989096593080324e-05,
+      "loss": 0.6137,
+      "step": 4239
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998908074223225e-05,
+      "loss": 0.7744,
+      "step": 4240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998906487987724e-05,
+      "loss": 0.7223,
+      "step": 4241
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989049006015308e-05,
+      "loss": 0.8946,
+      "step": 4242
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989033120646472e-05,
+      "loss": 0.8483,
+      "step": 4243
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989017223770754e-05,
+      "loss": 0.8712,
+      "step": 4244
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9989001315388168e-05,
+      "loss": 0.6626,
+      "step": 4245
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9988985395498738e-05,
+      "loss": 0.7339,
+      "step": 4246
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9988969464102477e-05,
+      "loss": 0.6829,
+      "step": 4247
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9988953521199406e-05,
+      "loss": 0.7746,
+      "step": 4248
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998893756678954e-05,
+      "loss": 0.6579,
+      "step": 4249
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9988921600872903e-05,
+      "loss": 0.635,
+      "step": 4250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998890562344951e-05,
+      "loss": 0.6985,
+      "step": 4251
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9988889634519378e-05,
+      "loss": 0.8712,
+      "step": 4252
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.998887363408253e-05,
+      "loss": 0.6532,
+      "step": 4253
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9988857622138974e-05,
+      "loss": 0.8762,
+      "step": 4254
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9988841598688745e-05,
+      "loss": 0.7203,
+      "step": 4255
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9988825563731847e-05,
+      "loss": 0.631,
+      "step": 4256
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998880951726831e-05,
+      "loss": 0.9384,
+      "step": 4257
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988793459298142e-05,
+      "loss": 1.0733,
+      "step": 4258
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988777389821365e-05,
+      "loss": 0.7567,
+      "step": 4259
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988761308838e-05,
+      "loss": 0.7817,
+      "step": 4260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988745216348065e-05,
+      "loss": 0.7596,
+      "step": 4261
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988729112351578e-05,
+      "loss": 0.7795,
+      "step": 4262
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988712996848556e-05,
+      "loss": 0.9279,
+      "step": 4263
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988696869839018e-05,
+      "loss": 0.8311,
+      "step": 4264
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988680731322984e-05,
+      "loss": 0.8644,
+      "step": 4265
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998866458130047e-05,
+      "loss": 0.7431,
+      "step": 4266
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.99886484197715e-05,
+      "loss": 0.7238,
+      "step": 4267
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988632246736088e-05,
+      "loss": 0.7178,
+      "step": 4268
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988616062194252e-05,
+      "loss": 0.9295,
+      "step": 4269
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988599866146013e-05,
+      "loss": 0.7452,
+      "step": 4270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998858365859139e-05,
+      "loss": 0.8732,
+      "step": 4271
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988567439530397e-05,
+      "loss": 0.6513,
+      "step": 4272
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998855120896306e-05,
+      "loss": 0.7956,
+      "step": 4273
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988534966889393e-05,
+      "loss": 0.5825,
+      "step": 4274
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988518713309416e-05,
+      "loss": 0.7051,
+      "step": 4275
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998850244822315e-05,
+      "loss": 0.7758,
+      "step": 4276
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998848617163061e-05,
+      "loss": 0.7215,
+      "step": 4277
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988469883531812e-05,
+      "loss": 0.6548,
+      "step": 4278
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988453583926782e-05,
+      "loss": 0.7387,
+      "step": 4279
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988437272815534e-05,
+      "loss": 0.5942,
+      "step": 4280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998842095019809e-05,
+      "loss": 0.7639,
+      "step": 4281
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988404616074467e-05,
+      "loss": 0.8313,
+      "step": 4282
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988388270444685e-05,
+      "loss": 0.7909,
+      "step": 4283
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988371913308756e-05,
+      "loss": 0.7074,
+      "step": 4284
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998835554466671e-05,
+      "loss": 0.7355,
+      "step": 4285
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998833916451856e-05,
+      "loss": 0.8056,
+      "step": 4286
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988322772864323e-05,
+      "loss": 0.8798,
+      "step": 4287
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988306369704023e-05,
+      "loss": 0.6963,
+      "step": 4288
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998828995503767e-05,
+      "loss": 0.8112,
+      "step": 4289
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988273528865298e-05,
+      "loss": 1.0457,
+      "step": 4290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.998825709118691e-05,
+      "loss": 0.839,
+      "step": 4291
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988240642002533e-05,
+      "loss": 0.8902,
+      "step": 4292
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988224181312188e-05,
+      "loss": 0.8492,
+      "step": 4293
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9988207709115887e-05,
+      "loss": 0.6017,
+      "step": 4294
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9988191225413656e-05,
+      "loss": 0.858,
+      "step": 4295
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.998817473020551e-05,
+      "loss": 0.7674,
+      "step": 4296
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9988158223491467e-05,
+      "loss": 0.7568,
+      "step": 4297
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.998814170527155e-05,
+      "loss": 0.962,
+      "step": 4298
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9988125175545773e-05,
+      "loss": 0.7179,
+      "step": 4299
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.998810863431416e-05,
+      "loss": 0.7531,
+      "step": 4300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9988092081576727e-05,
+      "loss": 0.6259,
+      "step": 4301
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9988075517333495e-05,
+      "loss": 0.8921,
+      "step": 4302
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.998805894158448e-05,
+      "loss": 0.8558,
+      "step": 4303
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9988042354329705e-05,
+      "loss": 0.6539,
+      "step": 4304
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.998802575556919e-05,
+      "loss": 0.7146,
+      "step": 4305
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9988009145302946e-05,
+      "loss": 0.8419,
+      "step": 4306
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987992523531e-05,
+      "loss": 0.8973,
+      "step": 4307
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987975890253366e-05,
+      "loss": 0.7672,
+      "step": 4308
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987959245470067e-05,
+      "loss": 0.7124,
+      "step": 4309
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987942589181123e-05,
+      "loss": 0.8011,
+      "step": 4310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.998792592138655e-05,
+      "loss": 0.7495,
+      "step": 4311
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987909242086367e-05,
+      "loss": 0.8135,
+      "step": 4312
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987892551280596e-05,
+      "loss": 0.6002,
+      "step": 4313
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987875848969255e-05,
+      "loss": 0.9227,
+      "step": 4314
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987859135152364e-05,
+      "loss": 0.8936,
+      "step": 4315
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.998784240982994e-05,
+      "loss": 0.7613,
+      "step": 4316
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987825673002002e-05,
+      "loss": 0.7771,
+      "step": 4317
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987808924668572e-05,
+      "loss": 0.6908,
+      "step": 4318
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987792164829667e-05,
+      "loss": 0.865,
+      "step": 4319
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.998777539348531e-05,
+      "loss": 0.7868,
+      "step": 4320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.998775861063552e-05,
+      "loss": 0.6365,
+      "step": 4321
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987741816280305e-05,
+      "loss": 0.6199,
+      "step": 4322
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.99877250104197e-05,
+      "loss": 0.6985,
+      "step": 4323
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987708193053717e-05,
+      "loss": 0.5498,
+      "step": 4324
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987691364182375e-05,
+      "loss": 0.7252,
+      "step": 4325
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987674523805697e-05,
+      "loss": 0.766,
+      "step": 4326
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.99876576719237e-05,
+      "loss": 0.728,
+      "step": 4327
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.99876408085364e-05,
+      "loss": 0.6705,
+      "step": 4328
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987623933643822e-05,
+      "loss": 0.6613,
+      "step": 4329
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987607047245984e-05,
+      "loss": 0.8564,
+      "step": 4330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.99875901493429e-05,
+      "loss": 0.7818,
+      "step": 4331
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9987573239934602e-05,
+      "loss": 0.7661,
+      "step": 4332
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987556319021097e-05,
+      "loss": 0.709,
+      "step": 4333
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987539386602406e-05,
+      "loss": 0.7315,
+      "step": 4334
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987522442678556e-05,
+      "loss": 0.7269,
+      "step": 4335
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987505487249562e-05,
+      "loss": 0.8636,
+      "step": 4336
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998748852031544e-05,
+      "loss": 0.7521,
+      "step": 4337
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987471541876218e-05,
+      "loss": 0.6245,
+      "step": 4338
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998745455193191e-05,
+      "loss": 0.683,
+      "step": 4339
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987437550482534e-05,
+      "loss": 0.8526,
+      "step": 4340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987420537528114e-05,
+      "loss": 0.7767,
+      "step": 4341
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987403513068665e-05,
+      "loss": 0.8983,
+      "step": 4342
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987386477104212e-05,
+      "loss": 0.8133,
+      "step": 4343
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998736942963477e-05,
+      "loss": 0.6794,
+      "step": 4344
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998735237066036e-05,
+      "loss": 0.8285,
+      "step": 4345
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987335300181002e-05,
+      "loss": 0.6348,
+      "step": 4346
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987318218196717e-05,
+      "loss": 0.6932,
+      "step": 4347
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987301124707523e-05,
+      "loss": 0.7752,
+      "step": 4348
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998728401971344e-05,
+      "loss": 0.6569,
+      "step": 4349
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998726690321449e-05,
+      "loss": 0.5506,
+      "step": 4350
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987249775210687e-05,
+      "loss": 0.6738,
+      "step": 4351
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987232635702054e-05,
+      "loss": 0.7856,
+      "step": 4352
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987215484688612e-05,
+      "loss": 0.8665,
+      "step": 4353
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998719832217038e-05,
+      "loss": 0.725,
+      "step": 4354
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987181148147383e-05,
+      "loss": 0.6427,
+      "step": 4355
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987163962619625e-05,
+      "loss": 0.7294,
+      "step": 4356
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998714676558714e-05,
+      "loss": 1.1281,
+      "step": 4357
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987129557049947e-05,
+      "loss": 0.8285,
+      "step": 4358
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998711233700806e-05,
+      "loss": 0.6651,
+      "step": 4359
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987095105461504e-05,
+      "loss": 0.7142,
+      "step": 4360
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987077862410293e-05,
+      "loss": 0.7126,
+      "step": 4361
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987060607854452e-05,
+      "loss": 0.7129,
+      "step": 4362
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987043341793997e-05,
+      "loss": 0.7191,
+      "step": 4363
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987026064228955e-05,
+      "loss": 0.6771,
+      "step": 4364
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9987008775159337e-05,
+      "loss": 0.8229,
+      "step": 4365
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998699147458517e-05,
+      "loss": 0.68,
+      "step": 4366
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998697416250647e-05,
+      "loss": 0.8154,
+      "step": 4367
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9986956838923254e-05,
+      "loss": 0.6968,
+      "step": 4368
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998693950383555e-05,
+      "loss": 0.7739,
+      "step": 4369
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.998692215724337e-05,
+      "loss": 0.8634,
+      "step": 4370
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9986904799146737e-05,
+      "loss": 0.6266,
+      "step": 4371
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986887429545676e-05,
+      "loss": 0.6551,
+      "step": 4372
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986870048440203e-05,
+      "loss": 0.6577,
+      "step": 4373
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986852655830332e-05,
+      "loss": 0.7818,
+      "step": 4374
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986835251716093e-05,
+      "loss": 0.6824,
+      "step": 4375
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.99868178360975e-05,
+      "loss": 0.6001,
+      "step": 4376
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986800408974578e-05,
+      "loss": 0.6773,
+      "step": 4377
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998678297034734e-05,
+      "loss": 0.7953,
+      "step": 4378
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986765520215812e-05,
+      "loss": 0.8296,
+      "step": 4379
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986748058580013e-05,
+      "loss": 0.7899,
+      "step": 4380
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998673058543996e-05,
+      "loss": 0.9211,
+      "step": 4381
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986713100795672e-05,
+      "loss": 0.7923,
+      "step": 4382
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998669560464718e-05,
+      "loss": 0.7761,
+      "step": 4383
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998667809699449e-05,
+      "loss": 0.6949,
+      "step": 4384
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986660577837633e-05,
+      "loss": 0.8291,
+      "step": 4385
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986643047176625e-05,
+      "loss": 0.9151,
+      "step": 4386
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986625505011483e-05,
+      "loss": 0.82,
+      "step": 4387
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986607951342234e-05,
+      "loss": 0.8853,
+      "step": 4388
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986590386168892e-05,
+      "loss": 0.75,
+      "step": 4389
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998657280949148e-05,
+      "loss": 0.8512,
+      "step": 4390
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998655522131002e-05,
+      "loss": 0.764,
+      "step": 4391
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986537621624528e-05,
+      "loss": 0.7392,
+      "step": 4392
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998652001043503e-05,
+      "loss": 0.6081,
+      "step": 4393
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998650238774154e-05,
+      "loss": 0.745,
+      "step": 4394
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986484753544083e-05,
+      "loss": 0.6451,
+      "step": 4395
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986467107842675e-05,
+      "loss": 0.7267,
+      "step": 4396
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986449450637337e-05,
+      "loss": 0.6614,
+      "step": 4397
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986431781928095e-05,
+      "loss": 0.8893,
+      "step": 4398
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986414101714966e-05,
+      "loss": 0.784,
+      "step": 4399
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986396409997968e-05,
+      "loss": 0.7841,
+      "step": 4400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986378706777124e-05,
+      "loss": 0.7163,
+      "step": 4401
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986360992052455e-05,
+      "loss": 0.8961,
+      "step": 4402
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986343265823977e-05,
+      "loss": 0.8211,
+      "step": 4403
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986325528091717e-05,
+      "loss": 0.8877,
+      "step": 4404
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998630777885569e-05,
+      "loss": 0.9117,
+      "step": 4405
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998629001811592e-05,
+      "loss": 0.7066,
+      "step": 4406
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986272245872423e-05,
+      "loss": 0.654,
+      "step": 4407
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9986254462125224e-05,
+      "loss": 0.9357,
+      "step": 4408
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.998623666687434e-05,
+      "loss": 0.7815,
+      "step": 4409
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9986218860119794e-05,
+      "loss": 0.7734,
+      "step": 4410
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9986201041861606e-05,
+      "loss": 0.776,
+      "step": 4411
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9986183212099796e-05,
+      "loss": 0.757,
+      "step": 4412
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9986165370834385e-05,
+      "loss": 0.7806,
+      "step": 4413
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9986147518065395e-05,
+      "loss": 0.5709,
+      "step": 4414
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9986129653792844e-05,
+      "loss": 0.7564,
+      "step": 4415
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9986111778016752e-05,
+      "loss": 0.7607,
+      "step": 4416
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9986093890737143e-05,
+      "loss": 0.7273,
+      "step": 4417
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9986075991954037e-05,
+      "loss": 0.6907,
+      "step": 4418
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.998605808166745e-05,
+      "loss": 0.8085,
+      "step": 4419
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.998604015987741e-05,
+      "loss": 0.6204,
+      "step": 4420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.998602222658393e-05,
+      "loss": 0.5475,
+      "step": 4421
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9986004281787034e-05,
+      "loss": 0.6774,
+      "step": 4422
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.998598632548675e-05,
+      "loss": 0.6564,
+      "step": 4423
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985968357683085e-05,
+      "loss": 0.7758,
+      "step": 4424
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985950378376066e-05,
+      "loss": 0.6957,
+      "step": 4425
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985932387565717e-05,
+      "loss": 0.9366,
+      "step": 4426
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985914385252054e-05,
+      "loss": 0.761,
+      "step": 4427
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.99858963714351e-05,
+      "loss": 1.022,
+      "step": 4428
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985878346114875e-05,
+      "loss": 0.7181,
+      "step": 4429
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.99858603092914e-05,
+      "loss": 0.6369,
+      "step": 4430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985842260964697e-05,
+      "loss": 0.6925,
+      "step": 4431
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985824201134783e-05,
+      "loss": 0.9331,
+      "step": 4432
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985806129801684e-05,
+      "loss": 0.8886,
+      "step": 4433
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985788046965417e-05,
+      "loss": 0.7227,
+      "step": 4434
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985769952626005e-05,
+      "loss": 0.6216,
+      "step": 4435
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.998575184678347e-05,
+      "loss": 0.6734,
+      "step": 4436
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985733729437826e-05,
+      "loss": 0.9147,
+      "step": 4437
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985715600589103e-05,
+      "loss": 0.8953,
+      "step": 4438
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985697460237316e-05,
+      "loss": 0.8105,
+      "step": 4439
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985679308382487e-05,
+      "loss": 0.8365,
+      "step": 4440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985661145024635e-05,
+      "loss": 0.8039,
+      "step": 4441
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985642970163788e-05,
+      "loss": 0.8386,
+      "step": 4442
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.998562478379996e-05,
+      "loss": 0.883,
+      "step": 4443
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.998560658593317e-05,
+      "loss": 1.3165,
+      "step": 4444
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.998558837656345e-05,
+      "loss": 0.7313,
+      "step": 4445
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985570155690814e-05,
+      "loss": 0.5536,
+      "step": 4446
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985551923315277e-05,
+      "loss": 0.743,
+      "step": 4447
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.998553367943687e-05,
+      "loss": 0.9499,
+      "step": 4448
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985515424055608e-05,
+      "loss": 0.6029,
+      "step": 4449
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985497157171516e-05,
+      "loss": 0.9127,
+      "step": 4450
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.998547887878461e-05,
+      "loss": 0.7633,
+      "step": 4451
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985460588894918e-05,
+      "loss": 0.6523,
+      "step": 4452
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.998544228750246e-05,
+      "loss": 0.6317,
+      "step": 4453
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.998542397460725e-05,
+      "loss": 0.6089,
+      "step": 4454
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985405650209315e-05,
+      "loss": 0.8712,
+      "step": 4455
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.998538731430867e-05,
+      "loss": 0.8057,
+      "step": 4456
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985368966905344e-05,
+      "loss": 0.5356,
+      "step": 4457
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985350607999355e-05,
+      "loss": 0.8484,
+      "step": 4458
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985332237590723e-05,
+      "loss": 0.7771,
+      "step": 4459
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985313855679472e-05,
+      "loss": 0.812,
+      "step": 4460
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.998529546226562e-05,
+      "loss": 0.6868,
+      "step": 4461
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985277057349193e-05,
+      "loss": 0.8685,
+      "step": 4462
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985258640930202e-05,
+      "loss": 0.5768,
+      "step": 4463
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.998524021300868e-05,
+      "loss": 0.9348,
+      "step": 4464
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985221773584644e-05,
+      "loss": 0.8356,
+      "step": 4465
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.998520332265811e-05,
+      "loss": 0.6495,
+      "step": 4466
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985184860229107e-05,
+      "loss": 1.0243,
+      "step": 4467
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985166386297653e-05,
+      "loss": 0.6723,
+      "step": 4468
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985147900863766e-05,
+      "loss": 0.7137,
+      "step": 4469
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985129403927473e-05,
+      "loss": 0.7855,
+      "step": 4470
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985110895488792e-05,
+      "loss": 0.6878,
+      "step": 4471
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985092375547745e-05,
+      "loss": 0.6337,
+      "step": 4472
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985073844104353e-05,
+      "loss": 0.6489,
+      "step": 4473
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.998505530115864e-05,
+      "loss": 0.6548,
+      "step": 4474
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985036746710624e-05,
+      "loss": 0.7205,
+      "step": 4475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9985018180760326e-05,
+      "loss": 0.7401,
+      "step": 4476
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.998499960330777e-05,
+      "loss": 0.8323,
+      "step": 4477
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9984981014352977e-05,
+      "loss": 1.0681,
+      "step": 4478
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9984962413895967e-05,
+      "loss": 0.587,
+      "step": 4479
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9984943801936764e-05,
+      "loss": 0.8291,
+      "step": 4480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9984925178475386e-05,
+      "loss": 0.8378,
+      "step": 4481
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9984906543511856e-05,
+      "loss": 0.6454,
+      "step": 4482
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9984887897046194e-05,
+      "loss": 0.8251,
+      "step": 4483
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9984869239078425e-05,
+      "loss": 0.8207,
+      "step": 4484
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9984850569608568e-05,
+      "loss": 0.5304,
+      "step": 4485
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984831888636648e-05,
+      "loss": 0.679,
+      "step": 4486
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984813196162677e-05,
+      "loss": 0.6731,
+      "step": 4487
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984794492186687e-05,
+      "loss": 0.8894,
+      "step": 4488
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984775776708695e-05,
+      "loss": 0.708,
+      "step": 4489
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984757049728726e-05,
+      "loss": 0.809,
+      "step": 4490
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984738311246796e-05,
+      "loss": 0.6578,
+      "step": 4491
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984719561262927e-05,
+      "loss": 0.8311,
+      "step": 4492
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984700799777147e-05,
+      "loss": 0.7444,
+      "step": 4493
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998468202678947e-05,
+      "loss": 0.9177,
+      "step": 4494
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984663242299925e-05,
+      "loss": 0.976,
+      "step": 4495
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984644446308526e-05,
+      "loss": 0.596,
+      "step": 4496
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.99846256388153e-05,
+      "loss": 0.8076,
+      "step": 4497
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998460681982027e-05,
+      "loss": 0.7691,
+      "step": 4498
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998458798932345e-05,
+      "loss": 0.526,
+      "step": 4499
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998456914732487e-05,
+      "loss": 0.7758,
+      "step": 4500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984550293824547e-05,
+      "loss": 0.6355,
+      "step": 4501
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.99845314288225e-05,
+      "loss": 0.8004,
+      "step": 4502
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998451255231876e-05,
+      "loss": 0.8967,
+      "step": 4503
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984493664313343e-05,
+      "loss": 0.8771,
+      "step": 4504
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984474764806272e-05,
+      "loss": 0.5886,
+      "step": 4505
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984455853797568e-05,
+      "loss": 1.0616,
+      "step": 4506
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984436931287247e-05,
+      "loss": 0.8702,
+      "step": 4507
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998441799727534e-05,
+      "loss": 0.8286,
+      "step": 4508
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984399051761863e-05,
+      "loss": 0.8344,
+      "step": 4509
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984380094746846e-05,
+      "loss": 0.7477,
+      "step": 4510
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.99843611262303e-05,
+      "loss": 0.7994,
+      "step": 4511
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984342146212255e-05,
+      "loss": 0.8552,
+      "step": 4512
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984323154692727e-05,
+      "loss": 0.7669,
+      "step": 4513
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998430415167174e-05,
+      "loss": 0.7066,
+      "step": 4514
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998428513714932e-05,
+      "loss": 0.6793,
+      "step": 4515
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984266111125483e-05,
+      "loss": 0.7229,
+      "step": 4516
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984247073600254e-05,
+      "loss": 0.5809,
+      "step": 4517
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984228024573653e-05,
+      "loss": 0.8413,
+      "step": 4518
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984208964045703e-05,
+      "loss": 0.7772,
+      "step": 4519
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998418989201643e-05,
+      "loss": 0.8071,
+      "step": 4520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998417080848585e-05,
+      "loss": 0.8095,
+      "step": 4521
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9984151713453985e-05,
+      "loss": 0.7501,
+      "step": 4522
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.998413260692086e-05,
+      "loss": 0.6787,
+      "step": 4523
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9984113488886495e-05,
+      "loss": 0.9044,
+      "step": 4524
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9984094359350913e-05,
+      "loss": 0.6569,
+      "step": 4525
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998407521831414e-05,
+      "loss": 0.7363,
+      "step": 4526
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9984056065776188e-05,
+      "loss": 0.9859,
+      "step": 4527
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9984036901737092e-05,
+      "loss": 0.8761,
+      "step": 4528
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9984017726196864e-05,
+      "loss": 0.7564,
+      "step": 4529
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983998539155527e-05,
+      "loss": 0.6389,
+      "step": 4530
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998397934061311e-05,
+      "loss": 0.7627,
+      "step": 4531
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998396013056963e-05,
+      "loss": 0.6555,
+      "step": 4532
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983940909025106e-05,
+      "loss": 0.7602,
+      "step": 4533
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983921675979564e-05,
+      "loss": 0.932,
+      "step": 4534
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983902431433032e-05,
+      "loss": 0.7244,
+      "step": 4535
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998388317538552e-05,
+      "loss": 0.8207,
+      "step": 4536
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998386390783706e-05,
+      "loss": 0.7091,
+      "step": 4537
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998384462878767e-05,
+      "loss": 0.7575,
+      "step": 4538
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998382533823737e-05,
+      "loss": 0.8019,
+      "step": 4539
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998380603618619e-05,
+      "loss": 0.7489,
+      "step": 4540
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983786722634143e-05,
+      "loss": 0.6906,
+      "step": 4541
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983767397581256e-05,
+      "loss": 0.6146,
+      "step": 4542
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983748061027556e-05,
+      "loss": 0.7766,
+      "step": 4543
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983728712973053e-05,
+      "loss": 0.842,
+      "step": 4544
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998370935341778e-05,
+      "loss": 0.7096,
+      "step": 4545
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983689982361754e-05,
+      "loss": 0.8095,
+      "step": 4546
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983670599805e-05,
+      "loss": 0.8982,
+      "step": 4547
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998365120574754e-05,
+      "loss": 0.6808,
+      "step": 4548
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983631800189394e-05,
+      "loss": 0.6398,
+      "step": 4549
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983612383130585e-05,
+      "loss": 0.9225,
+      "step": 4550
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983592954571137e-05,
+      "loss": 0.7413,
+      "step": 4551
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983573514511074e-05,
+      "loss": 0.6395,
+      "step": 4552
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983554062950415e-05,
+      "loss": 0.8685,
+      "step": 4553
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983534599889186e-05,
+      "loss": 0.6751,
+      "step": 4554
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983515125327405e-05,
+      "loss": 0.7901,
+      "step": 4555
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983495639265095e-05,
+      "loss": 0.844,
+      "step": 4556
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983476141702277e-05,
+      "loss": 0.6361,
+      "step": 4557
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983456632638983e-05,
+      "loss": 0.724,
+      "step": 4558
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983437112075222e-05,
+      "loss": 0.6605,
+      "step": 4559
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9983417580011027e-05,
+      "loss": 0.6875,
+      "step": 4560
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.998339803644642e-05,
+      "loss": 0.6899,
+      "step": 4561
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983378481381413e-05,
+      "loss": 0.5787,
+      "step": 4562
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.998335891481604e-05,
+      "loss": 0.7426,
+      "step": 4563
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983339336750317e-05,
+      "loss": 0.7299,
+      "step": 4564
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983319747184274e-05,
+      "loss": 0.7963,
+      "step": 4565
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983300146117926e-05,
+      "loss": 0.8445,
+      "step": 4566
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983280533551296e-05,
+      "loss": 0.6226,
+      "step": 4567
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.998326090948441e-05,
+      "loss": 0.6035,
+      "step": 4568
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983241273917286e-05,
+      "loss": 0.7215,
+      "step": 4569
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983221626849954e-05,
+      "loss": 0.8353,
+      "step": 4570
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983201968282432e-05,
+      "loss": 0.7085,
+      "step": 4571
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983182298214744e-05,
+      "loss": 0.8239,
+      "step": 4572
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983162616646906e-05,
+      "loss": 0.887,
+      "step": 4573
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983142923578952e-05,
+      "loss": 0.8706,
+      "step": 4574
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983123219010897e-05,
+      "loss": 0.568,
+      "step": 4575
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983103502942767e-05,
+      "loss": 0.7301,
+      "step": 4576
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983083775374582e-05,
+      "loss": 0.598,
+      "step": 4577
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983064036306366e-05,
+      "loss": 0.9736,
+      "step": 4578
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983044285738144e-05,
+      "loss": 0.7081,
+      "step": 4579
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983024523669935e-05,
+      "loss": 0.7689,
+      "step": 4580
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9983004750101766e-05,
+      "loss": 0.8222,
+      "step": 4581
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982984965033657e-05,
+      "loss": 0.7701,
+      "step": 4582
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.998296516846563e-05,
+      "loss": 1.001,
+      "step": 4583
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.998294536039771e-05,
+      "loss": 0.8068,
+      "step": 4584
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982925540829917e-05,
+      "loss": 0.7787,
+      "step": 4585
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982905709762275e-05,
+      "loss": 0.6098,
+      "step": 4586
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.998288586719481e-05,
+      "loss": 0.6243,
+      "step": 4587
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.998286601312754e-05,
+      "loss": 0.6534,
+      "step": 4588
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.998284614756049e-05,
+      "loss": 0.7525,
+      "step": 4589
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982826270493682e-05,
+      "loss": 0.6616,
+      "step": 4590
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982806381927143e-05,
+      "loss": 0.7517,
+      "step": 4591
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982786481860892e-05,
+      "loss": 0.9201,
+      "step": 4592
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982766570294952e-05,
+      "loss": 0.7723,
+      "step": 4593
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982746647229344e-05,
+      "loss": 0.5923,
+      "step": 4594
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982726712664098e-05,
+      "loss": 0.6344,
+      "step": 4595
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.998270676659923e-05,
+      "loss": 0.6487,
+      "step": 4596
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982686809034767e-05,
+      "loss": 0.6581,
+      "step": 4597
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982666839970733e-05,
+      "loss": 0.8231,
+      "step": 4598
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9982646859407143e-05,
+      "loss": 0.5929,
+      "step": 4599
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.998262686734403e-05,
+      "loss": 0.754,
+      "step": 4600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998260686378141e-05,
+      "loss": 0.714,
+      "step": 4601
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998258684871931e-05,
+      "loss": 0.8364,
+      "step": 4602
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982566822157752e-05,
+      "loss": 0.7076,
+      "step": 4603
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998254678409676e-05,
+      "loss": 0.6972,
+      "step": 4604
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982526734536352e-05,
+      "loss": 0.7413,
+      "step": 4605
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982506673476557e-05,
+      "loss": 0.7374,
+      "step": 4606
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982486600917396e-05,
+      "loss": 0.7756,
+      "step": 4607
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982466516858892e-05,
+      "loss": 0.6787,
+      "step": 4608
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998244642130107e-05,
+      "loss": 0.7098,
+      "step": 4609
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982426314243946e-05,
+      "loss": 0.7327,
+      "step": 4610
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982406195687555e-05,
+      "loss": 0.7906,
+      "step": 4611
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982386065631913e-05,
+      "loss": 0.6498,
+      "step": 4612
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998236592407704e-05,
+      "loss": 0.7755,
+      "step": 4613
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982345771022965e-05,
+      "loss": 0.8122,
+      "step": 4614
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998232560646971e-05,
+      "loss": 0.7939,
+      "step": 4615
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982305430417296e-05,
+      "loss": 0.671,
+      "step": 4616
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982285242865748e-05,
+      "loss": 0.8856,
+      "step": 4617
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998226504381509e-05,
+      "loss": 0.6775,
+      "step": 4618
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998224483326535e-05,
+      "loss": 0.8644,
+      "step": 4619
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982224611216538e-05,
+      "loss": 0.7573,
+      "step": 4620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998220437766869e-05,
+      "loss": 0.7277,
+      "step": 4621
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998218413262182e-05,
+      "loss": 0.884,
+      "step": 4622
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982163876075957e-05,
+      "loss": 0.7145,
+      "step": 4623
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982143608031123e-05,
+      "loss": 0.9707,
+      "step": 4624
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998212332848734e-05,
+      "loss": 0.7669,
+      "step": 4625
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982103037444638e-05,
+      "loss": 0.6515,
+      "step": 4626
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998208273490303e-05,
+      "loss": 1.0048,
+      "step": 4627
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982062420862546e-05,
+      "loss": 0.8939,
+      "step": 4628
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998204209532321e-05,
+      "loss": 0.6936,
+      "step": 4629
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9982021758285037e-05,
+      "loss": 0.7295,
+      "step": 4630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998200140974806e-05,
+      "loss": 0.7812,
+      "step": 4631
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.99819810497123e-05,
+      "loss": 0.81,
+      "step": 4632
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998196067817778e-05,
+      "loss": 0.8553,
+      "step": 4633
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9981940295144523e-05,
+      "loss": 0.5256,
+      "step": 4634
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9981919900612553e-05,
+      "loss": 0.7033,
+      "step": 4635
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998189949458189e-05,
+      "loss": 0.7858,
+      "step": 4636
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9981879077052563e-05,
+      "loss": 0.6914,
+      "step": 4637
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.998185864802459e-05,
+      "loss": 0.8475,
+      "step": 4638
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981838207498e-05,
+      "loss": 0.7353,
+      "step": 4639
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981817755472815e-05,
+      "loss": 0.9749,
+      "step": 4640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981797291949056e-05,
+      "loss": 0.9044,
+      "step": 4641
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981776816926747e-05,
+      "loss": 0.6544,
+      "step": 4642
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981756330405915e-05,
+      "loss": 0.8073,
+      "step": 4643
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998173583238658e-05,
+      "loss": 0.5816,
+      "step": 4644
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998171532286877e-05,
+      "loss": 0.7167,
+      "step": 4645
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981694801852504e-05,
+      "loss": 0.497,
+      "step": 4646
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981674269337806e-05,
+      "loss": 0.7954,
+      "step": 4647
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.99816537253247e-05,
+      "loss": 0.7068,
+      "step": 4648
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998163316981321e-05,
+      "loss": 0.6737,
+      "step": 4649
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981612602803365e-05,
+      "loss": 0.671,
+      "step": 4650
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981592024295185e-05,
+      "loss": 0.999,
+      "step": 4651
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981571434288686e-05,
+      "loss": 0.7466,
+      "step": 4652
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981550832783903e-05,
+      "loss": 0.706,
+      "step": 4653
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981530219780852e-05,
+      "loss": 0.646,
+      "step": 4654
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981509595279563e-05,
+      "loss": 0.6976,
+      "step": 4655
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981488959280058e-05,
+      "loss": 0.6292,
+      "step": 4656
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981468311782356e-05,
+      "loss": 0.744,
+      "step": 4657
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981447652786486e-05,
+      "loss": 0.7652,
+      "step": 4658
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981426982292467e-05,
+      "loss": 0.8159,
+      "step": 4659
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981406300300327e-05,
+      "loss": 0.7083,
+      "step": 4660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998138560681009e-05,
+      "loss": 0.5663,
+      "step": 4661
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998136490182178e-05,
+      "loss": 0.6634,
+      "step": 4662
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981344185335415e-05,
+      "loss": 0.6842,
+      "step": 4663
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981323457351026e-05,
+      "loss": 0.5729,
+      "step": 4664
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981302717868632e-05,
+      "loss": 0.5481,
+      "step": 4665
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998128196688826e-05,
+      "loss": 0.8011,
+      "step": 4666
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981261204409933e-05,
+      "loss": 0.7372,
+      "step": 4667
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998124043043368e-05,
+      "loss": 0.5796,
+      "step": 4668
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981219644959512e-05,
+      "loss": 0.6843,
+      "step": 4669
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981198847987464e-05,
+      "loss": 0.7484,
+      "step": 4670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981178039517557e-05,
+      "loss": 0.6644,
+      "step": 4671
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981157219549812e-05,
+      "loss": 0.7859,
+      "step": 4672
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981136388084258e-05,
+      "loss": 0.7587,
+      "step": 4673
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981115545120916e-05,
+      "loss": 0.7028,
+      "step": 4674
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998109469065981e-05,
+      "loss": 0.793,
+      "step": 4675
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9981073824700965e-05,
+      "loss": 0.706,
+      "step": 4676
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9981052947244405e-05,
+      "loss": 0.6511,
+      "step": 4677
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9981032058290153e-05,
+      "loss": 0.7829,
+      "step": 4678
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9981011157838234e-05,
+      "loss": 0.7223,
+      "step": 4679
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998099024588867e-05,
+      "loss": 0.5894,
+      "step": 4680
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998096932244149e-05,
+      "loss": 0.8944,
+      "step": 4681
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998094838749671e-05,
+      "loss": 0.7608,
+      "step": 4682
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980927441054366e-05,
+      "loss": 0.7786,
+      "step": 4683
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980906483114474e-05,
+      "loss": 0.6872,
+      "step": 4684
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980885513677055e-05,
+      "loss": 0.9274,
+      "step": 4685
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998086453274214e-05,
+      "loss": 0.9674,
+      "step": 4686
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980843540309755e-05,
+      "loss": 0.5814,
+      "step": 4687
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980822536379914e-05,
+      "loss": 0.6835,
+      "step": 4688
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998080152095265e-05,
+      "loss": 0.5762,
+      "step": 4689
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998078049402798e-05,
+      "loss": 0.7178,
+      "step": 4690
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980759455605935e-05,
+      "loss": 0.8456,
+      "step": 4691
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998073840568654e-05,
+      "loss": 0.6953,
+      "step": 4692
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980717344269812e-05,
+      "loss": 0.8232,
+      "step": 4693
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998069627135578e-05,
+      "loss": 0.7328,
+      "step": 4694
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998067518694447e-05,
+      "loss": 0.7165,
+      "step": 4695
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980654091035904e-05,
+      "loss": 0.9366,
+      "step": 4696
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980632983630102e-05,
+      "loss": 0.864,
+      "step": 4697
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980611864727093e-05,
+      "loss": 0.6269,
+      "step": 4698
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980590734326905e-05,
+      "loss": 0.7713,
+      "step": 4699
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980569592429556e-05,
+      "loss": 0.7594,
+      "step": 4700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998054843903507e-05,
+      "loss": 0.6504,
+      "step": 4701
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980527274143474e-05,
+      "loss": 0.6182,
+      "step": 4702
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980506097754798e-05,
+      "loss": 0.6122,
+      "step": 4703
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980484909869056e-05,
+      "loss": 0.6889,
+      "step": 4704
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980463710486278e-05,
+      "loss": 0.698,
+      "step": 4705
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980442499606484e-05,
+      "loss": 0.7162,
+      "step": 4706
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998042127722971e-05,
+      "loss": 0.872,
+      "step": 4707
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980400043355967e-05,
+      "loss": 0.7411,
+      "step": 4708
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980378797985283e-05,
+      "loss": 0.9315,
+      "step": 4709
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980357541117686e-05,
+      "loss": 0.7148,
+      "step": 4710
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.99803362727532e-05,
+      "loss": 0.8347,
+      "step": 4711
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998031499289185e-05,
+      "loss": 0.8235,
+      "step": 4712
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980293701533653e-05,
+      "loss": 0.6569,
+      "step": 4713
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9980272398678644e-05,
+      "loss": 0.8638,
+      "step": 4714
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.998025108432684e-05,
+      "loss": 0.6467,
+      "step": 4715
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.998022975847827e-05,
+      "loss": 0.7293,
+      "step": 4716
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9980208421132958e-05,
+      "loss": 0.742,
+      "step": 4717
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9980187072290925e-05,
+      "loss": 0.7063,
+      "step": 4718
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.99801657119522e-05,
+      "loss": 0.6547,
+      "step": 4719
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9980144340116806e-05,
+      "loss": 0.7546,
+      "step": 4720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9980122956784767e-05,
+      "loss": 0.6319,
+      "step": 4721
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9980101561956106e-05,
+      "loss": 0.9598,
+      "step": 4722
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9980080155630853e-05,
+      "loss": 0.8324,
+      "step": 4723
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9980058737809025e-05,
+      "loss": 0.7003,
+      "step": 4724
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9980037308490653e-05,
+      "loss": 0.7717,
+      "step": 4725
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.998001586767576e-05,
+      "loss": 0.7351,
+      "step": 4726
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.997999441536437e-05,
+      "loss": 0.7236,
+      "step": 4727
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.997997295155651e-05,
+      "loss": 0.7163,
+      "step": 4728
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.99799514762522e-05,
+      "loss": 0.6112,
+      "step": 4729
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.997992998945147e-05,
+      "loss": 0.6321,
+      "step": 4730
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.997990849115434e-05,
+      "loss": 0.8658,
+      "step": 4731
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.997988698136084e-05,
+      "loss": 0.6235,
+      "step": 4732
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.997986546007099e-05,
+      "loss": 0.5595,
+      "step": 4733
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979843927284815e-05,
+      "loss": 0.6635,
+      "step": 4734
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979822383002344e-05,
+      "loss": 0.6372,
+      "step": 4735
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979800827223596e-05,
+      "loss": 0.6633,
+      "step": 4736
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.99797792599486e-05,
+      "loss": 0.8919,
+      "step": 4737
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979757681177385e-05,
+      "loss": 0.6256,
+      "step": 4738
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979736090909966e-05,
+      "loss": 0.7502,
+      "step": 4739
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979714489146375e-05,
+      "loss": 0.7382,
+      "step": 4740
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979692875886633e-05,
+      "loss": 0.6711,
+      "step": 4741
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979671251130763e-05,
+      "loss": 0.6385,
+      "step": 4742
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.99796496148788e-05,
+      "loss": 0.8612,
+      "step": 4743
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979627967130755e-05,
+      "loss": 0.7068,
+      "step": 4744
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979606307886665e-05,
+      "loss": 0.8218,
+      "step": 4745
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.997958463714655e-05,
+      "loss": 0.682,
+      "step": 4746
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979562954910434e-05,
+      "loss": 0.6797,
+      "step": 4747
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979541261178343e-05,
+      "loss": 0.7602,
+      "step": 4748
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9979519555950305e-05,
+      "loss": 0.5823,
+      "step": 4749
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.997949783922634e-05,
+      "loss": 0.8281,
+      "step": 4750
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.997947611100647e-05,
+      "loss": 0.8355,
+      "step": 4751
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.997945437129073e-05,
+      "loss": 0.7184,
+      "step": 4752
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997943262007914e-05,
+      "loss": 0.8734,
+      "step": 4753
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979410857371728e-05,
+      "loss": 0.7806,
+      "step": 4754
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979389083168512e-05,
+      "loss": 0.5859,
+      "step": 4755
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997936729746952e-05,
+      "loss": 0.8213,
+      "step": 4756
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979345500274786e-05,
+      "loss": 0.6459,
+      "step": 4757
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997932369158432e-05,
+      "loss": 0.6497,
+      "step": 4758
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997930187139816e-05,
+      "loss": 0.8489,
+      "step": 4759
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979280039716323e-05,
+      "loss": 0.6239,
+      "step": 4760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979258196538836e-05,
+      "loss": 0.6682,
+      "step": 4761
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997923634186573e-05,
+      "loss": 0.587,
+      "step": 4762
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997921447569702e-05,
+      "loss": 0.8275,
+      "step": 4763
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979192598032737e-05,
+      "loss": 0.8261,
+      "step": 4764
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979170708872907e-05,
+      "loss": 0.855,
+      "step": 4765
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979148808217555e-05,
+      "loss": 0.6934,
+      "step": 4766
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979126896066706e-05,
+      "loss": 0.7138,
+      "step": 4767
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997910497242038e-05,
+      "loss": 0.6515,
+      "step": 4768
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997908303727861e-05,
+      "loss": 0.6829,
+      "step": 4769
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979061090641416e-05,
+      "loss": 0.9055,
+      "step": 4770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979039132508828e-05,
+      "loss": 0.6546,
+      "step": 4771
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979017162880864e-05,
+      "loss": 0.8058,
+      "step": 4772
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978995181757556e-05,
+      "loss": 0.6938,
+      "step": 4773
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997897318913893e-05,
+      "loss": 0.7241,
+      "step": 4774
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978951185025005e-05,
+      "loss": 0.8056,
+      "step": 4775
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978929169415814e-05,
+      "loss": 0.8313,
+      "step": 4776
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978907142311373e-05,
+      "loss": 0.9008,
+      "step": 4777
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978885103711715e-05,
+      "loss": 0.8213,
+      "step": 4778
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978863053616862e-05,
+      "loss": 0.8892,
+      "step": 4779
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997884099202684e-05,
+      "loss": 0.573,
+      "step": 4780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978818918941676e-05,
+      "loss": 0.6751,
+      "step": 4781
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978796834361397e-05,
+      "loss": 0.8776,
+      "step": 4782
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997877473828602e-05,
+      "loss": 0.6759,
+      "step": 4783
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978752630715577e-05,
+      "loss": 0.5632,
+      "step": 4784
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978730511650094e-05,
+      "loss": 0.9064,
+      "step": 4785
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978708381089597e-05,
+      "loss": 0.7371,
+      "step": 4786
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978686239034108e-05,
+      "loss": 0.8208,
+      "step": 4787
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978664085483653e-05,
+      "loss": 0.6661,
+      "step": 4788
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997864192043826e-05,
+      "loss": 0.6263,
+      "step": 4789
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978619743897952e-05,
+      "loss": 0.6819,
+      "step": 4790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978597555862757e-05,
+      "loss": 0.6971,
+      "step": 4791
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978575356332697e-05,
+      "loss": 0.7045,
+      "step": 4792
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.99785531453078e-05,
+      "loss": 0.746,
+      "step": 4793
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978530922788092e-05,
+      "loss": 0.7717,
+      "step": 4794
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.99785086887736e-05,
+      "loss": 0.5336,
+      "step": 4795
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978486443264346e-05,
+      "loss": 0.574,
+      "step": 4796
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978464186260353e-05,
+      "loss": 0.5922,
+      "step": 4797
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978441917761656e-05,
+      "loss": 0.6755,
+      "step": 4798
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978419637768278e-05,
+      "loss": 0.7642,
+      "step": 4799
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978397346280235e-05,
+      "loss": 0.7452,
+      "step": 4800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978375043297563e-05,
+      "loss": 0.7163,
+      "step": 4801
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978352728820284e-05,
+      "loss": 0.5972,
+      "step": 4802
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978330402848426e-05,
+      "loss": 0.7055,
+      "step": 4803
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997830806538201e-05,
+      "loss": 0.7286,
+      "step": 4804
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978285716421067e-05,
+      "loss": 0.6884,
+      "step": 4805
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997826335596562e-05,
+      "loss": 0.7352,
+      "step": 4806
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978240984015692e-05,
+      "loss": 0.7375,
+      "step": 4807
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978218600571314e-05,
+      "loss": 0.724,
+      "step": 4808
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978196205632515e-05,
+      "loss": 0.6623,
+      "step": 4809
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978173799199307e-05,
+      "loss": 0.7544,
+      "step": 4810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978151381271733e-05,
+      "loss": 0.6752,
+      "step": 4811
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978128951849803e-05,
+      "loss": 0.6904,
+      "step": 4812
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997810651093355e-05,
+      "loss": 0.7689,
+      "step": 4813
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978084058523004e-05,
+      "loss": 0.6779,
+      "step": 4814
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978061594618184e-05,
+      "loss": 0.8785,
+      "step": 4815
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997803911921912e-05,
+      "loss": 0.6227,
+      "step": 4816
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9978016632325838e-05,
+      "loss": 0.5593,
+      "step": 4817
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997799413393836e-05,
+      "loss": 0.716,
+      "step": 4818
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9977971624056713e-05,
+      "loss": 0.9141,
+      "step": 4819
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9977949102680923e-05,
+      "loss": 0.8225,
+      "step": 4820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9977926569811022e-05,
+      "loss": 0.6437,
+      "step": 4821
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997790402544703e-05,
+      "loss": 0.8926,
+      "step": 4822
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9977881469588973e-05,
+      "loss": 0.8264,
+      "step": 4823
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997785890223688e-05,
+      "loss": 0.654,
+      "step": 4824
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9977836323390768e-05,
+      "loss": 0.6837,
+      "step": 4825
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9977813733050677e-05,
+      "loss": 0.8567,
+      "step": 4826
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9977791131216624e-05,
+      "loss": 0.8422,
+      "step": 4827
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997776851788864e-05,
+      "loss": 0.8632,
+      "step": 4828
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9977745893066745e-05,
+      "loss": 0.7843,
+      "step": 4829
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977723256750967e-05,
+      "loss": 0.7372,
+      "step": 4830
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977700608941334e-05,
+      "loss": 0.6543,
+      "step": 4831
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977677949637874e-05,
+      "loss": 0.6369,
+      "step": 4832
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977655278840607e-05,
+      "loss": 0.544,
+      "step": 4833
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977632596549568e-05,
+      "loss": 0.5711,
+      "step": 4834
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977609902764776e-05,
+      "loss": 0.7179,
+      "step": 4835
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977587197486255e-05,
+      "loss": 0.6047,
+      "step": 4836
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997756448071404e-05,
+      "loss": 0.9312,
+      "step": 4837
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977541752448148e-05,
+      "loss": 0.6797,
+      "step": 4838
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997751901268861e-05,
+      "loss": 0.7669,
+      "step": 4839
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977496261435454e-05,
+      "loss": 0.8706,
+      "step": 4840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977473498688705e-05,
+      "loss": 0.6251,
+      "step": 4841
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977450724448384e-05,
+      "loss": 0.7785,
+      "step": 4842
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977427938714523e-05,
+      "loss": 0.733,
+      "step": 4843
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977405141487147e-05,
+      "loss": 0.9264,
+      "step": 4844
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997738233276628e-05,
+      "loss": 0.593,
+      "step": 4845
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977359512551953e-05,
+      "loss": 1.076,
+      "step": 4846
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977336680844186e-05,
+      "loss": 0.6881,
+      "step": 4847
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997731383764301e-05,
+      "loss": 0.5791,
+      "step": 4848
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997729098294845e-05,
+      "loss": 0.8402,
+      "step": 4849
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977268116760532e-05,
+      "loss": 0.6451,
+      "step": 4850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977245239079283e-05,
+      "loss": 0.6338,
+      "step": 4851
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997722234990473e-05,
+      "loss": 0.7304,
+      "step": 4852
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977199449236896e-05,
+      "loss": 0.6369,
+      "step": 4853
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977176537075813e-05,
+      "loss": 0.6964,
+      "step": 4854
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.99771536134215e-05,
+      "loss": 0.6682,
+      "step": 4855
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997713067827399e-05,
+      "loss": 0.6599,
+      "step": 4856
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997710773163331e-05,
+      "loss": 0.7791,
+      "step": 4857
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977084773499478e-05,
+      "loss": 0.9318,
+      "step": 4858
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977061803872528e-05,
+      "loss": 0.6593,
+      "step": 4859
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977038822752483e-05,
+      "loss": 0.7502,
+      "step": 4860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9977015830139373e-05,
+      "loss": 0.6761,
+      "step": 4861
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9976992826033222e-05,
+      "loss": 0.5428,
+      "step": 4862
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9976969810434053e-05,
+      "loss": 0.8375,
+      "step": 4863
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9976946783341898e-05,
+      "loss": 0.8793,
+      "step": 4864
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9976923744756783e-05,
+      "loss": 0.6955,
+      "step": 4865
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9976900694678735e-05,
+      "loss": 0.6838,
+      "step": 4866
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997687763310778e-05,
+      "loss": 0.7958,
+      "step": 4867
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976854560043937e-05,
+      "loss": 0.7746,
+      "step": 4868
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976831475487244e-05,
+      "loss": 0.7562,
+      "step": 4869
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976808379437723e-05,
+      "loss": 0.8647,
+      "step": 4870
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976785271895395e-05,
+      "loss": 0.6423,
+      "step": 4871
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976762152860296e-05,
+      "loss": 0.6083,
+      "step": 4872
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.997673902233245e-05,
+      "loss": 0.7527,
+      "step": 4873
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976715880311878e-05,
+      "loss": 0.7249,
+      "step": 4874
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976692726798612e-05,
+      "loss": 0.6726,
+      "step": 4875
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976669561792678e-05,
+      "loss": 0.8146,
+      "step": 4876
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976646385294102e-05,
+      "loss": 0.7419,
+      "step": 4877
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976623197302912e-05,
+      "loss": 0.7644,
+      "step": 4878
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976599997819132e-05,
+      "loss": 0.8426,
+      "step": 4879
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976576786842788e-05,
+      "loss": 0.7978,
+      "step": 4880
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976553564373915e-05,
+      "loss": 0.7141,
+      "step": 4881
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976530330412533e-05,
+      "loss": 0.7503,
+      "step": 4882
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.997650708495866e-05,
+      "loss": 0.7337,
+      "step": 4883
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976483828012342e-05,
+      "loss": 0.7048,
+      "step": 4884
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976460559573595e-05,
+      "loss": 0.724,
+      "step": 4885
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976437279642447e-05,
+      "loss": 0.7122,
+      "step": 4886
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976413988218923e-05,
+      "loss": 0.7238,
+      "step": 4887
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.997639068530305e-05,
+      "loss": 0.6355,
+      "step": 4888
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976367370894857e-05,
+      "loss": 0.7655,
+      "step": 4889
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.997634404499437e-05,
+      "loss": 0.7106,
+      "step": 4890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.997632070760162e-05,
+      "loss": 0.6058,
+      "step": 4891
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976297358716626e-05,
+      "loss": 0.7955,
+      "step": 4892
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976273998339422e-05,
+      "loss": 0.9969,
+      "step": 4893
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976250626470027e-05,
+      "loss": 0.9332,
+      "step": 4894
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976227243108475e-05,
+      "loss": 0.7849,
+      "step": 4895
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976203848254794e-05,
+      "loss": 0.6996,
+      "step": 4896
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976180441909007e-05,
+      "loss": 0.792,
+      "step": 4897
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976157024071137e-05,
+      "loss": 0.6443,
+      "step": 4898
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.997613359474122e-05,
+      "loss": 0.799,
+      "step": 4899
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976110153919274e-05,
+      "loss": 0.6814,
+      "step": 4900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976086701605334e-05,
+      "loss": 0.5884,
+      "step": 4901
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976063237799423e-05,
+      "loss": 0.9224,
+      "step": 4902
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976039762501566e-05,
+      "loss": 0.7974,
+      "step": 4903
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9976016275711796e-05,
+      "loss": 0.7908,
+      "step": 4904
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9975992777430137e-05,
+      "loss": 0.8101,
+      "step": 4905
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975969267656613e-05,
+      "loss": 0.9113,
+      "step": 4906
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975945746391257e-05,
+      "loss": 0.8219,
+      "step": 4907
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997592221363409e-05,
+      "loss": 0.5709,
+      "step": 4908
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997589866938514e-05,
+      "loss": 0.7297,
+      "step": 4909
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997587511364444e-05,
+      "loss": 0.7465,
+      "step": 4910
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975851546412012e-05,
+      "loss": 0.8669,
+      "step": 4911
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975827967687888e-05,
+      "loss": 0.8794,
+      "step": 4912
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975804377472088e-05,
+      "loss": 0.734,
+      "step": 4913
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975780775764643e-05,
+      "loss": 0.715,
+      "step": 4914
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997575716256558e-05,
+      "loss": 0.6816,
+      "step": 4915
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997573353787493e-05,
+      "loss": 0.9295,
+      "step": 4916
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997570990169271e-05,
+      "loss": 0.7029,
+      "step": 4917
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975686254018954e-05,
+      "loss": 0.5975,
+      "step": 4918
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975662594853694e-05,
+      "loss": 0.8379,
+      "step": 4919
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975638924196947e-05,
+      "loss": 0.6992,
+      "step": 4920
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975615242048748e-05,
+      "loss": 0.8147,
+      "step": 4921
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975591548409125e-05,
+      "loss": 0.7579,
+      "step": 4922
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975567843278096e-05,
+      "loss": 0.9092,
+      "step": 4923
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975544126655697e-05,
+      "loss": 0.8379,
+      "step": 4924
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997552039854195e-05,
+      "loss": 0.6569,
+      "step": 4925
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997549665893689e-05,
+      "loss": 0.8179,
+      "step": 4926
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975472907840537e-05,
+      "loss": 0.7154,
+      "step": 4927
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997544914525292e-05,
+      "loss": 0.7487,
+      "step": 4928
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975425371174066e-05,
+      "loss": 0.516,
+      "step": 4929
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975401585604005e-05,
+      "loss": 0.7676,
+      "step": 4930
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975377788542763e-05,
+      "loss": 0.7919,
+      "step": 4931
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975353979990367e-05,
+      "loss": 0.7608,
+      "step": 4932
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997533015994684e-05,
+      "loss": 0.6426,
+      "step": 4933
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975306328412222e-05,
+      "loss": 0.8888,
+      "step": 4934
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975282485386527e-05,
+      "loss": 0.8181,
+      "step": 4935
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997525863086979e-05,
+      "loss": 0.9045,
+      "step": 4936
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975234764862035e-05,
+      "loss": 0.8128,
+      "step": 4937
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997521088736329e-05,
+      "loss": 0.8631,
+      "step": 4938
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975186998373588e-05,
+      "loss": 0.8004,
+      "step": 4939
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997516309789295e-05,
+      "loss": 0.6461,
+      "step": 4940
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975139185921406e-05,
+      "loss": 0.7172,
+      "step": 4941
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.997511526245898e-05,
+      "loss": 0.6599,
+      "step": 4942
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9975091327505702e-05,
+      "loss": 0.8374,
+      "step": 4943
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9975067381061604e-05,
+      "loss": 0.7112,
+      "step": 4944
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997504342312671e-05,
+      "loss": 0.6227,
+      "step": 4945
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9975019453701045e-05,
+      "loss": 0.7507,
+      "step": 4946
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997499547278464e-05,
+      "loss": 0.8841,
+      "step": 4947
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997497148037752e-05,
+      "loss": 0.7519,
+      "step": 4948
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974947476479715e-05,
+      "loss": 0.7899,
+      "step": 4949
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974923461091252e-05,
+      "loss": 0.6312,
+      "step": 4950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997489943421216e-05,
+      "loss": 0.7639,
+      "step": 4951
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974875395842463e-05,
+      "loss": 0.8255,
+      "step": 4952
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974851345982195e-05,
+      "loss": 0.7726,
+      "step": 4953
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974827284631376e-05,
+      "loss": 0.7103,
+      "step": 4954
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974803211790037e-05,
+      "loss": 0.6757,
+      "step": 4955
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974779127458207e-05,
+      "loss": 0.729,
+      "step": 4956
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974755031635916e-05,
+      "loss": 0.7629,
+      "step": 4957
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997473092432318e-05,
+      "loss": 0.5768,
+      "step": 4958
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974706805520046e-05,
+      "loss": 0.843,
+      "step": 4959
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974682675226525e-05,
+      "loss": 0.9528,
+      "step": 4960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974658533442653e-05,
+      "loss": 0.5862,
+      "step": 4961
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974634380168454e-05,
+      "loss": 0.6492,
+      "step": 4962
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997461021540396e-05,
+      "loss": 1.0038,
+      "step": 4963
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997458603914919e-05,
+      "loss": 0.7552,
+      "step": 4964
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974561851404183e-05,
+      "loss": 0.6331,
+      "step": 4965
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974537652168963e-05,
+      "loss": 0.7318,
+      "step": 4966
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974513441443555e-05,
+      "loss": 0.7597,
+      "step": 4967
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997448921922799e-05,
+      "loss": 0.7033,
+      "step": 4968
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974464985522294e-05,
+      "loss": 0.76,
+      "step": 4969
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974440740326497e-05,
+      "loss": 0.6155,
+      "step": 4970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974416483640627e-05,
+      "loss": 0.7596,
+      "step": 4971
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974392215464705e-05,
+      "loss": 0.7623,
+      "step": 4972
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997436793579877e-05,
+      "loss": 0.7918,
+      "step": 4973
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974343644642844e-05,
+      "loss": 0.6452,
+      "step": 4974
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974319341996954e-05,
+      "loss": 0.7161,
+      "step": 4975
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997429502786113e-05,
+      "loss": 0.769,
+      "step": 4976
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974270702235397e-05,
+      "loss": 0.613,
+      "step": 4977
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974246365119788e-05,
+      "loss": 0.7222,
+      "step": 4978
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997422201651433e-05,
+      "loss": 0.7135,
+      "step": 4979
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9974197656419046e-05,
+      "loss": 0.8035,
+      "step": 4980
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.997417328483397e-05,
+      "loss": 0.8433,
+      "step": 4981
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9974148901759126e-05,
+      "loss": 0.908,
+      "step": 4982
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9974124507194545e-05,
+      "loss": 0.6531,
+      "step": 4983
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9974100101140253e-05,
+      "loss": 0.6648,
+      "step": 4984
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997407568359628e-05,
+      "loss": 0.6922,
+      "step": 4985
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9974051254562652e-05,
+      "loss": 0.7453,
+      "step": 4986
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9974026814039398e-05,
+      "loss": 0.6254,
+      "step": 4987
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9974002362026548e-05,
+      "loss": 0.6596,
+      "step": 4988
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973977898524127e-05,
+      "loss": 0.682,
+      "step": 4989
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973953423532168e-05,
+      "loss": 0.7939,
+      "step": 4990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997392893705069e-05,
+      "loss": 0.7909,
+      "step": 4991
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973904439079735e-05,
+      "loss": 0.765,
+      "step": 4992
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973879929619318e-05,
+      "loss": 0.7847,
+      "step": 4993
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973855408669476e-05,
+      "loss": 0.6433,
+      "step": 4994
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997383087623023e-05,
+      "loss": 0.7499,
+      "step": 4995
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973806332301615e-05,
+      "loss": 0.5648,
+      "step": 4996
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973781776883657e-05,
+      "loss": 0.7334,
+      "step": 4997
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973757209976382e-05,
+      "loss": 0.4556,
+      "step": 4998
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997373263157982e-05,
+      "loss": 0.9645,
+      "step": 4999
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973708041694e-05,
+      "loss": 0.7787,
+      "step": 5000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997368344031895e-05,
+      "loss": 0.6766,
+      "step": 5001
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973658827454695e-05,
+      "loss": 0.6739,
+      "step": 5002
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997363420310127e-05,
+      "loss": 0.5897,
+      "step": 5003
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.99736095672587e-05,
+      "loss": 0.6951,
+      "step": 5004
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997358491992701e-05,
+      "loss": 0.821,
+      "step": 5005
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973560261106236e-05,
+      "loss": 0.6982,
+      "step": 5006
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973535590796397e-05,
+      "loss": 0.7006,
+      "step": 5007
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997351090899753e-05,
+      "loss": 0.6368,
+      "step": 5008
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973486215709655e-05,
+      "loss": 0.7663,
+      "step": 5009
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997346151093281e-05,
+      "loss": 0.9134,
+      "step": 5010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973436794667015e-05,
+      "loss": 0.7191,
+      "step": 5011
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973412066912307e-05,
+      "loss": 0.5595,
+      "step": 5012
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973387327668704e-05,
+      "loss": 0.8687,
+      "step": 5013
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997336257693624e-05,
+      "loss": 0.9333,
+      "step": 5014
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997333781471495e-05,
+      "loss": 0.5667,
+      "step": 5015
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997331304100485e-05,
+      "loss": 0.7607,
+      "step": 5016
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973288255805977e-05,
+      "loss": 0.762,
+      "step": 5017
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997326345911836e-05,
+      "loss": 0.6212,
+      "step": 5018
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.997323865094202e-05,
+      "loss": 0.7576,
+      "step": 5019
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9973213831276995e-05,
+      "loss": 0.8061,
+      "step": 5020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9973189000123305e-05,
+      "loss": 0.7477,
+      "step": 5021
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9973164157480984e-05,
+      "loss": 0.7035,
+      "step": 5022
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9973139303350058e-05,
+      "loss": 0.7409,
+      "step": 5023
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997311443773056e-05,
+      "loss": 0.8184,
+      "step": 5024
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9973089560622512e-05,
+      "loss": 0.6605,
+      "step": 5025
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9973064672025947e-05,
+      "loss": 0.743,
+      "step": 5026
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9973039771940895e-05,
+      "loss": 0.6322,
+      "step": 5027
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997301486036738e-05,
+      "loss": 0.7342,
+      "step": 5028
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972989937305433e-05,
+      "loss": 0.6255,
+      "step": 5029
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972965002755084e-05,
+      "loss": 0.6533,
+      "step": 5030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997294005671636e-05,
+      "loss": 0.6407,
+      "step": 5031
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997291509918929e-05,
+      "loss": 0.6346,
+      "step": 5032
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972890130173906e-05,
+      "loss": 0.6456,
+      "step": 5033
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997286514967023e-05,
+      "loss": 0.8226,
+      "step": 5034
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972840157678296e-05,
+      "loss": 0.9686,
+      "step": 5035
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972815154198134e-05,
+      "loss": 0.7918,
+      "step": 5036
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997279013922977e-05,
+      "loss": 0.5866,
+      "step": 5037
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997276511277323e-05,
+      "loss": 0.6764,
+      "step": 5038
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972740074828546e-05,
+      "loss": 0.5539,
+      "step": 5039
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972715025395746e-05,
+      "loss": 0.795,
+      "step": 5040
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997268996447486e-05,
+      "loss": 0.9737,
+      "step": 5041
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997266489206592e-05,
+      "loss": 0.7047,
+      "step": 5042
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972639808168947e-05,
+      "loss": 0.6681,
+      "step": 5043
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997261471278398e-05,
+      "loss": 0.584,
+      "step": 5044
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997258960591104e-05,
+      "loss": 0.7316,
+      "step": 5045
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972564487550153e-05,
+      "loss": 0.732,
+      "step": 5046
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972539357701355e-05,
+      "loss": 0.7984,
+      "step": 5047
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972514216364674e-05,
+      "loss": 0.6566,
+      "step": 5048
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972489063540138e-05,
+      "loss": 0.6607,
+      "step": 5049
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972463899227776e-05,
+      "loss": 0.6778,
+      "step": 5050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972438723427617e-05,
+      "loss": 0.6148,
+      "step": 5051
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972413536139686e-05,
+      "loss": 0.7349,
+      "step": 5052
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.997238833736402e-05,
+      "loss": 0.5852,
+      "step": 5053
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972363127100645e-05,
+      "loss": 0.7202,
+      "step": 5054
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972337905349587e-05,
+      "loss": 0.7217,
+      "step": 5055
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972312672110877e-05,
+      "loss": 0.5088,
+      "step": 5056
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972287427384547e-05,
+      "loss": 0.6755,
+      "step": 5057
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9972262171170618e-05,
+      "loss": 0.6578,
+      "step": 5058
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9972236903469126e-05,
+      "loss": 0.7132,
+      "step": 5059
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.99722116242801e-05,
+      "loss": 0.7766,
+      "step": 5060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9972186333603567e-05,
+      "loss": 0.6165,
+      "step": 5061
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9972161031439557e-05,
+      "loss": 0.7375,
+      "step": 5062
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9972135717788097e-05,
+      "loss": 0.8165,
+      "step": 5063
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9972110392649216e-05,
+      "loss": 0.6797,
+      "step": 5064
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997208505602295e-05,
+      "loss": 0.8788,
+      "step": 5065
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9972059707909323e-05,
+      "loss": 0.6849,
+      "step": 5066
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997203434830836e-05,
+      "loss": 0.8006,
+      "step": 5067
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9972008977220095e-05,
+      "loss": 0.812,
+      "step": 5068
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997198359464456e-05,
+      "loss": 0.7515,
+      "step": 5069
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971958200581782e-05,
+      "loss": 0.7607,
+      "step": 5070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971932795031787e-05,
+      "loss": 0.8013,
+      "step": 5071
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997190737799461e-05,
+      "loss": 0.8987,
+      "step": 5072
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997188194947027e-05,
+      "loss": 0.6555,
+      "step": 5073
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997185650945881e-05,
+      "loss": 1.0135,
+      "step": 5074
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971831057960248e-05,
+      "loss": 0.6936,
+      "step": 5075
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997180559497462e-05,
+      "loss": 0.8193,
+      "step": 5076
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971780120501956e-05,
+      "loss": 0.8818,
+      "step": 5077
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971754634542276e-05,
+      "loss": 0.6786,
+      "step": 5078
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997172913709562e-05,
+      "loss": 0.7255,
+      "step": 5079
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971703628162014e-05,
+      "loss": 0.7205,
+      "step": 5080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971678107741487e-05,
+      "loss": 0.8177,
+      "step": 5081
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971652575834064e-05,
+      "loss": 0.8014,
+      "step": 5082
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997162703243978e-05,
+      "loss": 0.6182,
+      "step": 5083
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971601477558663e-05,
+      "loss": 0.6288,
+      "step": 5084
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971575911190745e-05,
+      "loss": 0.5796,
+      "step": 5085
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997155033333605e-05,
+      "loss": 0.6898,
+      "step": 5086
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997152474399461e-05,
+      "loss": 0.8517,
+      "step": 5087
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971499143166455e-05,
+      "loss": 0.7784,
+      "step": 5088
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971473530851618e-05,
+      "loss": 0.7641,
+      "step": 5089
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997144790705012e-05,
+      "loss": 0.7334,
+      "step": 5090
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971422271761997e-05,
+      "loss": 0.5847,
+      "step": 5091
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971396624987277e-05,
+      "loss": 0.7368,
+      "step": 5092
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997137096672599e-05,
+      "loss": 0.7653,
+      "step": 5093
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971345296978163e-05,
+      "loss": 0.6403,
+      "step": 5094
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9971319615743826e-05,
+      "loss": 0.6772,
+      "step": 5095
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.997129392302301e-05,
+      "loss": 0.6976,
+      "step": 5096
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9971268218815748e-05,
+      "loss": 0.737,
+      "step": 5097
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9971242503122064e-05,
+      "loss": 0.6397,
+      "step": 5098
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9971216775941994e-05,
+      "loss": 0.7887,
+      "step": 5099
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997119103727556e-05,
+      "loss": 0.9679,
+      "step": 5100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9971165287122797e-05,
+      "loss": 0.7641,
+      "step": 5101
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997113952548373e-05,
+      "loss": 0.5996,
+      "step": 5102
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9971113752358392e-05,
+      "loss": 0.6397,
+      "step": 5103
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997108796774681e-05,
+      "loss": 0.5676,
+      "step": 5104
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997106217164902e-05,
+      "loss": 0.6002,
+      "step": 5105
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9971036364065045e-05,
+      "loss": 0.7504,
+      "step": 5106
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997101054499492e-05,
+      "loss": 0.5708,
+      "step": 5107
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997098471443867e-05,
+      "loss": 0.653,
+      "step": 5108
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997095887239633e-05,
+      "loss": 0.762,
+      "step": 5109
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970933018867922e-05,
+      "loss": 0.8037,
+      "step": 5110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970907153853484e-05,
+      "loss": 0.8348,
+      "step": 5111
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970881277353038e-05,
+      "loss": 0.8614,
+      "step": 5112
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997085538936662e-05,
+      "loss": 0.6023,
+      "step": 5113
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997082948989426e-05,
+      "loss": 0.5979,
+      "step": 5114
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997080357893598e-05,
+      "loss": 0.6581,
+      "step": 5115
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997077765649182e-05,
+      "loss": 0.7035,
+      "step": 5116
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970751722561806e-05,
+      "loss": 0.776,
+      "step": 5117
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970725777145967e-05,
+      "loss": 0.7794,
+      "step": 5118
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997069982024433e-05,
+      "loss": 0.6578,
+      "step": 5119
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997067385185693e-05,
+      "loss": 0.7052,
+      "step": 5120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970647871983794e-05,
+      "loss": 0.8415,
+      "step": 5121
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970621880624955e-05,
+      "loss": 0.649,
+      "step": 5122
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970595877780438e-05,
+      "loss": 0.7241,
+      "step": 5123
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997056986345028e-05,
+      "loss": 0.4885,
+      "step": 5124
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.99705438376345e-05,
+      "loss": 0.8262,
+      "step": 5125
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997051780033314e-05,
+      "loss": 0.6602,
+      "step": 5126
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970491751546222e-05,
+      "loss": 0.8782,
+      "step": 5127
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997046569127378e-05,
+      "loss": 0.8744,
+      "step": 5128
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970439619515843e-05,
+      "loss": 0.7876,
+      "step": 5129
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997041353627244e-05,
+      "loss": 0.679,
+      "step": 5130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.99703874415436e-05,
+      "loss": 0.6566,
+      "step": 5131
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970361335329357e-05,
+      "loss": 0.775,
+      "step": 5132
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.997033521762974e-05,
+      "loss": 0.6142,
+      "step": 5133
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9970309088444777e-05,
+      "loss": 0.8146,
+      "step": 5134
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9970282947774497e-05,
+      "loss": 0.6852,
+      "step": 5135
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9970256795618936e-05,
+      "loss": 0.6789,
+      "step": 5136
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9970230631978115e-05,
+      "loss": 0.5806,
+      "step": 5137
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9970204456852074e-05,
+      "loss": 0.5987,
+      "step": 5138
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9970178270240838e-05,
+      "loss": 0.6331,
+      "step": 5139
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9970152072144435e-05,
+      "loss": 0.7049,
+      "step": 5140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9970125862562902e-05,
+      "loss": 0.7295,
+      "step": 5141
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.997009964149626e-05,
+      "loss": 0.7972,
+      "step": 5142
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.997007340894455e-05,
+      "loss": 0.6792,
+      "step": 5143
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9970047164907793e-05,
+      "loss": 0.6797,
+      "step": 5144
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9970020909386022e-05,
+      "loss": 0.6104,
+      "step": 5145
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996999464237927e-05,
+      "loss": 0.6057,
+      "step": 5146
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969968363887566e-05,
+      "loss": 0.5868,
+      "step": 5147
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996994207391094e-05,
+      "loss": 0.9693,
+      "step": 5148
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996991577244942e-05,
+      "loss": 0.4938,
+      "step": 5149
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996988945950304e-05,
+      "loss": 0.7449,
+      "step": 5150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969863135071826e-05,
+      "loss": 0.7151,
+      "step": 5151
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996983679915581e-05,
+      "loss": 0.7394,
+      "step": 5152
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969810451755028e-05,
+      "loss": 0.6497,
+      "step": 5153
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.99697840928695e-05,
+      "loss": 0.8422,
+      "step": 5154
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996975772249927e-05,
+      "loss": 0.6628,
+      "step": 5155
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996973134064435e-05,
+      "loss": 0.644,
+      "step": 5156
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969704947304786e-05,
+      "loss": 0.6551,
+      "step": 5157
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.99696785424806e-05,
+      "loss": 0.748,
+      "step": 5158
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996965212617183e-05,
+      "loss": 0.7484,
+      "step": 5159
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.99696256983785e-05,
+      "loss": 0.7314,
+      "step": 5160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969599259100643e-05,
+      "loss": 0.6827,
+      "step": 5161
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996957280833829e-05,
+      "loss": 0.6445,
+      "step": 5162
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969546346091467e-05,
+      "loss": 0.7537,
+      "step": 5163
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996951987236021e-05,
+      "loss": 0.6744,
+      "step": 5164
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969493387144543e-05,
+      "loss": 0.7733,
+      "step": 5165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969466890444503e-05,
+      "loss": 0.5989,
+      "step": 5166
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996944038226012e-05,
+      "loss": 0.7553,
+      "step": 5167
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969413862591422e-05,
+      "loss": 0.7114,
+      "step": 5168
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969387331438443e-05,
+      "loss": 0.6578,
+      "step": 5169
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969360788801207e-05,
+      "loss": 0.6885,
+      "step": 5170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.996933423467975e-05,
+      "loss": 0.6879,
+      "step": 5171
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9969307669074098e-05,
+      "loss": 0.6068,
+      "step": 5172
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996928109198429e-05,
+      "loss": 0.7276,
+      "step": 5173
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9969254503410345e-05,
+      "loss": 0.6674,
+      "step": 5174
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9969227903352302e-05,
+      "loss": 0.8824,
+      "step": 5175
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9969201291810194e-05,
+      "loss": 0.7096,
+      "step": 5176
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9969174668784043e-05,
+      "loss": 0.7868,
+      "step": 5177
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9969148034273887e-05,
+      "loss": 0.715,
+      "step": 5178
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996912138827975e-05,
+      "loss": 0.6227,
+      "step": 5179
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996909473080167e-05,
+      "loss": 0.9253,
+      "step": 5180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996906806183967e-05,
+      "loss": 0.7574,
+      "step": 5181
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9969041381393786e-05,
+      "loss": 0.8186,
+      "step": 5182
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996901468946405e-05,
+      "loss": 0.8334,
+      "step": 5183
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968987986050487e-05,
+      "loss": 0.8307,
+      "step": 5184
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968961271153134e-05,
+      "loss": 0.6924,
+      "step": 5185
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968934544772016e-05,
+      "loss": 0.7123,
+      "step": 5186
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968907806907167e-05,
+      "loss": 0.7258,
+      "step": 5187
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996888105755862e-05,
+      "loss": 0.6834,
+      "step": 5188
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968854296726397e-05,
+      "loss": 0.6375,
+      "step": 5189
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996882752441054e-05,
+      "loss": 0.6483,
+      "step": 5190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968800740611075e-05,
+      "loss": 0.6117,
+      "step": 5191
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996877394532803e-05,
+      "loss": 0.6771,
+      "step": 5192
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996874713856144e-05,
+      "loss": 0.7916,
+      "step": 5193
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968720320311332e-05,
+      "loss": 0.731,
+      "step": 5194
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996869349057774e-05,
+      "loss": 0.6454,
+      "step": 5195
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968666649360697e-05,
+      "loss": 0.8238,
+      "step": 5196
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996863979666023e-05,
+      "loss": 0.6765,
+      "step": 5197
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996861293247637e-05,
+      "loss": 0.759,
+      "step": 5198
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996858605680915e-05,
+      "loss": 0.7622,
+      "step": 5199
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.99685591696586e-05,
+      "loss": 0.833,
+      "step": 5200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996853227102475e-05,
+      "loss": 0.8695,
+      "step": 5201
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968505360907634e-05,
+      "loss": 0.6894,
+      "step": 5202
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996847843930728e-05,
+      "loss": 0.6458,
+      "step": 5203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968451506223714e-05,
+      "loss": 0.6728,
+      "step": 5204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.996842456165698e-05,
+      "loss": 0.8181,
+      "step": 5205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968397605607102e-05,
+      "loss": 0.825,
+      "step": 5206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968370638074108e-05,
+      "loss": 0.6558,
+      "step": 5207
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968343659058034e-05,
+      "loss": 0.7426,
+      "step": 5208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968316668558908e-05,
+      "loss": 0.8127,
+      "step": 5209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9968289666576762e-05,
+      "loss": 0.6,
+      "step": 5210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9968262653111628e-05,
+      "loss": 0.6745,
+      "step": 5211
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9968235628163536e-05,
+      "loss": 0.7079,
+      "step": 5212
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996820859173252e-05,
+      "loss": 0.7591,
+      "step": 5213
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9968181543818606e-05,
+      "loss": 0.5438,
+      "step": 5214
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9968154484421826e-05,
+      "loss": 0.6683,
+      "step": 5215
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9968127413542217e-05,
+      "loss": 0.5272,
+      "step": 5216
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9968100331179806e-05,
+      "loss": 0.5903,
+      "step": 5217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9968073237334623e-05,
+      "loss": 0.7478,
+      "step": 5218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.99680461320067e-05,
+      "loss": 0.6251,
+      "step": 5219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996801901519607e-05,
+      "loss": 0.8303,
+      "step": 5220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967991886902762e-05,
+      "loss": 0.8123,
+      "step": 5221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996796474712681e-05,
+      "loss": 0.5171,
+      "step": 5222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967937595868244e-05,
+      "loss": 0.684,
+      "step": 5223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967910433127096e-05,
+      "loss": 0.707,
+      "step": 5224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967883258903392e-05,
+      "loss": 0.671,
+      "step": 5225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996785607319717e-05,
+      "loss": 0.7299,
+      "step": 5226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996782887600846e-05,
+      "loss": 0.7897,
+      "step": 5227
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996780166733729e-05,
+      "loss": 0.6284,
+      "step": 5228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967774447183693e-05,
+      "loss": 0.5807,
+      "step": 5229
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967747215547703e-05,
+      "loss": 0.6024,
+      "step": 5230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996771997242935e-05,
+      "loss": 0.6495,
+      "step": 5231
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967692717828662e-05,
+      "loss": 0.6798,
+      "step": 5232
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967665451745676e-05,
+      "loss": 0.6641,
+      "step": 5233
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996763817418042e-05,
+      "loss": 0.6041,
+      "step": 5234
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967610885132922e-05,
+      "loss": 0.8194,
+      "step": 5235
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996758358460322e-05,
+      "loss": 0.6926,
+      "step": 5236
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967556272591338e-05,
+      "loss": 0.5588,
+      "step": 5237
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967528949097322e-05,
+      "loss": 0.4754,
+      "step": 5238
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967501614121185e-05,
+      "loss": 0.6498,
+      "step": 5239
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996747426766297e-05,
+      "loss": 0.9553,
+      "step": 5240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967446909722707e-05,
+      "loss": 0.622,
+      "step": 5241
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967419540300426e-05,
+      "loss": 0.668,
+      "step": 5242
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996739215939616e-05,
+      "loss": 0.7416,
+      "step": 5243
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967364767009936e-05,
+      "loss": 0.9085,
+      "step": 5244
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996733736314179e-05,
+      "loss": 0.7318,
+      "step": 5245
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.996730994779175e-05,
+      "loss": 0.6912,
+      "step": 5246
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967282520959854e-05,
+      "loss": 0.8521,
+      "step": 5247
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967255082646127e-05,
+      "loss": 0.6242,
+      "step": 5248
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9967227632850605e-05,
+      "loss": 0.7077,
+      "step": 5249
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9967200171573314e-05,
+      "loss": 0.7524,
+      "step": 5250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9967172698814295e-05,
+      "loss": 0.6885,
+      "step": 5251
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.996714521457357e-05,
+      "loss": 0.7147,
+      "step": 5252
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9967117718851177e-05,
+      "loss": 0.6955,
+      "step": 5253
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9967090211647143e-05,
+      "loss": 0.5754,
+      "step": 5254
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9967062692961506e-05,
+      "loss": 0.5841,
+      "step": 5255
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.996703516279429e-05,
+      "loss": 0.7449,
+      "step": 5256
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.996700762114553e-05,
+      "loss": 0.7617,
+      "step": 5257
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966980068015264e-05,
+      "loss": 0.7536,
+      "step": 5258
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.996695250340351e-05,
+      "loss": 0.7552,
+      "step": 5259
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966924927310314e-05,
+      "loss": 0.9441,
+      "step": 5260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.99668973397357e-05,
+      "loss": 0.7041,
+      "step": 5261
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.99668697406797e-05,
+      "loss": 0.6946,
+      "step": 5262
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966842130142345e-05,
+      "loss": 0.6073,
+      "step": 5263
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966814508123672e-05,
+      "loss": 0.8422,
+      "step": 5264
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.996678687462371e-05,
+      "loss": 0.7982,
+      "step": 5265
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966759229642488e-05,
+      "loss": 0.6162,
+      "step": 5266
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.996673157318004e-05,
+      "loss": 0.6326,
+      "step": 5267
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.99667039052364e-05,
+      "loss": 0.7764,
+      "step": 5268
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966676225811602e-05,
+      "loss": 0.7645,
+      "step": 5269
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966648534905668e-05,
+      "loss": 0.7849,
+      "step": 5270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.996662083251864e-05,
+      "loss": 0.6943,
+      "step": 5271
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966593118650543e-05,
+      "loss": 0.5805,
+      "step": 5272
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966565393301413e-05,
+      "loss": 0.5588,
+      "step": 5273
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.996653765647128e-05,
+      "loss": 0.6943,
+      "step": 5274
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966509908160175e-05,
+      "loss": 0.7449,
+      "step": 5275
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966482148368134e-05,
+      "loss": 0.6616,
+      "step": 5276
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966454377095185e-05,
+      "loss": 0.6229,
+      "step": 5277
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966426594341362e-05,
+      "loss": 0.5729,
+      "step": 5278
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.99663988001067e-05,
+      "loss": 0.7258,
+      "step": 5279
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966370994391222e-05,
+      "loss": 0.773,
+      "step": 5280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.996634317719497e-05,
+      "loss": 0.7527,
+      "step": 5281
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966315348517967e-05,
+      "loss": 0.8602,
+      "step": 5282
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966287508360254e-05,
+      "loss": 0.747,
+      "step": 5283
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966259656721856e-05,
+      "loss": 0.8123,
+      "step": 5284
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966231793602807e-05,
+      "loss": 0.8114,
+      "step": 5285
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.996620391900314e-05,
+      "loss": 0.694,
+      "step": 5286
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9966176032922893e-05,
+      "loss": 0.749,
+      "step": 5287
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9966148135362085e-05,
+      "loss": 0.6062,
+      "step": 5288
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9966120226320763e-05,
+      "loss": 0.7018,
+      "step": 5289
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996609230579894e-05,
+      "loss": 0.6582,
+      "step": 5290
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9966064373796673e-05,
+      "loss": 0.6841,
+      "step": 5291
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996603643031397e-05,
+      "loss": 0.9115,
+      "step": 5292
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996600847535088e-05,
+      "loss": 0.608,
+      "step": 5293
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965980508907424e-05,
+      "loss": 0.7934,
+      "step": 5294
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965952530983647e-05,
+      "loss": 0.6057,
+      "step": 5295
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996592454157957e-05,
+      "loss": 0.8067,
+      "step": 5296
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965896540695225e-05,
+      "loss": 0.7595,
+      "step": 5297
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965868528330652e-05,
+      "loss": 0.7296,
+      "step": 5298
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965840504485882e-05,
+      "loss": 0.8664,
+      "step": 5299
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996581246916094e-05,
+      "loss": 0.5897,
+      "step": 5300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965784422355863e-05,
+      "loss": 0.5887,
+      "step": 5301
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965756364070686e-05,
+      "loss": 0.7401,
+      "step": 5302
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965728294305434e-05,
+      "loss": 0.9179,
+      "step": 5303
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965700213060148e-05,
+      "loss": 0.8439,
+      "step": 5304
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965672120334855e-05,
+      "loss": 0.6254,
+      "step": 5305
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965644016129592e-05,
+      "loss": 0.6204,
+      "step": 5306
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965615900444384e-05,
+      "loss": 0.7356,
+      "step": 5307
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996558777327927e-05,
+      "loss": 0.6664,
+      "step": 5308
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965559634634278e-05,
+      "loss": 0.581,
+      "step": 5309
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996553148450944e-05,
+      "loss": 0.5765,
+      "step": 5310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965503322904794e-05,
+      "loss": 0.6423,
+      "step": 5311
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965475149820368e-05,
+      "loss": 0.7715,
+      "step": 5312
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965446965256198e-05,
+      "loss": 0.8587,
+      "step": 5313
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996541876921231e-05,
+      "loss": 0.6623,
+      "step": 5314
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965390561688743e-05,
+      "loss": 0.6554,
+      "step": 5315
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965362342685525e-05,
+      "loss": 0.6845,
+      "step": 5316
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965334112202693e-05,
+      "loss": 0.4697,
+      "step": 5317
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965305870240278e-05,
+      "loss": 0.7609,
+      "step": 5318
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996527761679831e-05,
+      "loss": 0.6669,
+      "step": 5319
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996524935187682e-05,
+      "loss": 0.6713,
+      "step": 5320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9965221075475844e-05,
+      "loss": 0.6151,
+      "step": 5321
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996519278759542e-05,
+      "loss": 0.6894,
+      "step": 5322
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996516448823557e-05,
+      "loss": 0.6521,
+      "step": 5323
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996513617739633e-05,
+      "loss": 0.6351,
+      "step": 5324
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.996510785507774e-05,
+      "loss": 0.6732,
+      "step": 5325
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9965079521279823e-05,
+      "loss": 0.7818,
+      "step": 5326
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9965051176002617e-05,
+      "loss": 0.7139,
+      "step": 5327
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996502281924615e-05,
+      "loss": 0.8245,
+      "step": 5328
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996499445101046e-05,
+      "loss": 0.7204,
+      "step": 5329
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964966071295576e-05,
+      "loss": 0.6559,
+      "step": 5330
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964937680101534e-05,
+      "loss": 0.8395,
+      "step": 5331
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964909277428363e-05,
+      "loss": 0.7344,
+      "step": 5332
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964880863276096e-05,
+      "loss": 0.6186,
+      "step": 5333
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964852437644768e-05,
+      "loss": 0.6752,
+      "step": 5334
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996482400053441e-05,
+      "loss": 0.5364,
+      "step": 5335
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996479555194506e-05,
+      "loss": 0.6487,
+      "step": 5336
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964767091876744e-05,
+      "loss": 0.7392,
+      "step": 5337
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964738620329494e-05,
+      "loss": 0.686,
+      "step": 5338
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964710137303352e-05,
+      "loss": 0.7074,
+      "step": 5339
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964681642798342e-05,
+      "loss": 0.7154,
+      "step": 5340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964653136814497e-05,
+      "loss": 0.5443,
+      "step": 5341
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964624619351854e-05,
+      "loss": 0.7368,
+      "step": 5342
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964596090410448e-05,
+      "loss": 0.704,
+      "step": 5343
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964567549990302e-05,
+      "loss": 0.7995,
+      "step": 5344
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996453899809146e-05,
+      "loss": 0.6729,
+      "step": 5345
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996451043471395e-05,
+      "loss": 1.0096,
+      "step": 5346
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.99644818598578e-05,
+      "loss": 0.7821,
+      "step": 5347
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964453273523052e-05,
+      "loss": 0.8211,
+      "step": 5348
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964424675709733e-05,
+      "loss": 0.6497,
+      "step": 5349
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996439606641788e-05,
+      "loss": 0.5957,
+      "step": 5350
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996436744564752e-05,
+      "loss": 0.6288,
+      "step": 5351
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964338813398694e-05,
+      "loss": 0.6123,
+      "step": 5352
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964310169671428e-05,
+      "loss": 0.543,
+      "step": 5353
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964281514465758e-05,
+      "loss": 0.6593,
+      "step": 5354
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964252847781714e-05,
+      "loss": 0.5407,
+      "step": 5355
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964224169619334e-05,
+      "loss": 0.8708,
+      "step": 5356
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996419547997865e-05,
+      "loss": 0.5749,
+      "step": 5357
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996416677885969e-05,
+      "loss": 0.6194,
+      "step": 5358
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964138066262494e-05,
+      "loss": 0.8013,
+      "step": 5359
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996410934218709e-05,
+      "loss": 0.6724,
+      "step": 5360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.996408060663351e-05,
+      "loss": 0.6924,
+      "step": 5361
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964051859601795e-05,
+      "loss": 0.6987,
+      "step": 5362
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9964023101091973e-05,
+      "loss": 0.7935,
+      "step": 5363
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963994331104075e-05,
+      "loss": 0.6649,
+      "step": 5364
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996396554963814e-05,
+      "loss": 0.8157,
+      "step": 5365
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963936756694192e-05,
+      "loss": 0.8609,
+      "step": 5366
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996390795227227e-05,
+      "loss": 0.6878,
+      "step": 5367
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996387913637241e-05,
+      "loss": 0.5861,
+      "step": 5368
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963850308994645e-05,
+      "loss": 0.8546,
+      "step": 5369
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963821470139e-05,
+      "loss": 0.7068,
+      "step": 5370
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963792619805512e-05,
+      "loss": 0.7412,
+      "step": 5371
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963763757994222e-05,
+      "loss": 0.6622,
+      "step": 5372
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963734884705154e-05,
+      "loss": 0.7902,
+      "step": 5373
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963705999938342e-05,
+      "loss": 0.6599,
+      "step": 5374
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963677103693825e-05,
+      "loss": 0.7569,
+      "step": 5375
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996364819597163e-05,
+      "loss": 0.718,
+      "step": 5376
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963619276771795e-05,
+      "loss": 0.7984,
+      "step": 5377
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996359034609435e-05,
+      "loss": 0.6445,
+      "step": 5378
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963561403939333e-05,
+      "loss": 0.685,
+      "step": 5379
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963532450306774e-05,
+      "loss": 0.6599,
+      "step": 5380
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963503485196702e-05,
+      "loss": 0.5447,
+      "step": 5381
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963474508609158e-05,
+      "loss": 0.7615,
+      "step": 5382
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996344552054417e-05,
+      "loss": 0.7196,
+      "step": 5383
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963416521001776e-05,
+      "loss": 0.7871,
+      "step": 5384
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963387509982004e-05,
+      "loss": 0.6636,
+      "step": 5385
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996335848748489e-05,
+      "loss": 0.6085,
+      "step": 5386
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963329453510475e-05,
+      "loss": 0.8608,
+      "step": 5387
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996330040805878e-05,
+      "loss": 0.6251,
+      "step": 5388
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963271351129844e-05,
+      "loss": 0.638,
+      "step": 5389
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.99632422827237e-05,
+      "loss": 0.7292,
+      "step": 5390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996321320284038e-05,
+      "loss": 0.677,
+      "step": 5391
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963184111479922e-05,
+      "loss": 1.0373,
+      "step": 5392
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963155008642358e-05,
+      "loss": 0.832,
+      "step": 5393
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963125894327717e-05,
+      "loss": 0.6369,
+      "step": 5394
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963096768536037e-05,
+      "loss": 0.6163,
+      "step": 5395
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996306763126735e-05,
+      "loss": 0.5986,
+      "step": 5396
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996303848252169e-05,
+      "loss": 0.6916,
+      "step": 5397
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9963009322299092e-05,
+      "loss": 0.7243,
+      "step": 5398
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9962980150599585e-05,
+      "loss": 0.7976,
+      "step": 5399
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.996295096742321e-05,
+      "loss": 0.6218,
+      "step": 5400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9962921772769993e-05,
+      "loss": 0.7008,
+      "step": 5401
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.996289256663997e-05,
+      "loss": 0.8163,
+      "step": 5402
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962863349033177e-05,
+      "loss": 0.6686,
+      "step": 5403
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962834119949645e-05,
+      "loss": 0.8194,
+      "step": 5404
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.996280487938941e-05,
+      "loss": 0.6321,
+      "step": 5405
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962775627352506e-05,
+      "loss": 0.6555,
+      "step": 5406
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962746363838964e-05,
+      "loss": 0.6324,
+      "step": 5407
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.996271708884882e-05,
+      "loss": 0.634,
+      "step": 5408
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962687802382106e-05,
+      "loss": 0.6316,
+      "step": 5409
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962658504438857e-05,
+      "loss": 0.7733,
+      "step": 5410
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962629195019104e-05,
+      "loss": 0.7696,
+      "step": 5411
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962599874122885e-05,
+      "loss": 0.8151,
+      "step": 5412
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962570541750234e-05,
+      "loss": 0.8869,
+      "step": 5413
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962541197901177e-05,
+      "loss": 0.7408,
+      "step": 5414
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962511842575756e-05,
+      "loss": 0.8407,
+      "step": 5415
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962482475774e-05,
+      "loss": 0.6384,
+      "step": 5416
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962453097495947e-05,
+      "loss": 0.7561,
+      "step": 5417
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962423707741632e-05,
+      "loss": 0.6401,
+      "step": 5418
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.996239430651108e-05,
+      "loss": 0.7967,
+      "step": 5419
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962364893804333e-05,
+      "loss": 0.6807,
+      "step": 5420
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962335469621425e-05,
+      "loss": 0.7452,
+      "step": 5421
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962306033962386e-05,
+      "loss": 0.6636,
+      "step": 5422
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962276586827246e-05,
+      "loss": 0.5856,
+      "step": 5423
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.996224712821605e-05,
+      "loss": 0.6271,
+      "step": 5424
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962217658128825e-05,
+      "loss": 0.552,
+      "step": 5425
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962188176565604e-05,
+      "loss": 0.6993,
+      "step": 5426
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962158683526425e-05,
+      "loss": 0.5351,
+      "step": 5427
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.996212917901132e-05,
+      "loss": 0.5201,
+      "step": 5428
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962099663020322e-05,
+      "loss": 0.8118,
+      "step": 5429
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962070135553465e-05,
+      "loss": 0.5857,
+      "step": 5430
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962040596610786e-05,
+      "loss": 0.6581,
+      "step": 5431
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9962011046192315e-05,
+      "loss": 0.7243,
+      "step": 5432
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.996198148429809e-05,
+      "loss": 0.6294,
+      "step": 5433
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.996195191092814e-05,
+      "loss": 0.8521,
+      "step": 5434
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9961922326082507e-05,
+      "loss": 0.6588,
+      "step": 5435
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9961892729761216e-05,
+      "loss": 0.6829,
+      "step": 5436
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9961863121964306e-05,
+      "loss": 0.6834,
+      "step": 5437
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.996183350269181e-05,
+      "loss": 0.8254,
+      "step": 5438
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9961803871943767e-05,
+      "loss": 0.9432,
+      "step": 5439
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961774229720202e-05,
+      "loss": 0.8243,
+      "step": 5440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961744576021157e-05,
+      "loss": 0.6239,
+      "step": 5441
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.996171491084666e-05,
+      "loss": 0.6328,
+      "step": 5442
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961685234196747e-05,
+      "loss": 0.6562,
+      "step": 5443
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961655546071458e-05,
+      "loss": 0.6706,
+      "step": 5444
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.996162584647082e-05,
+      "loss": 0.6915,
+      "step": 5445
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.996159613539487e-05,
+      "loss": 0.8391,
+      "step": 5446
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961566412843644e-05,
+      "loss": 0.7003,
+      "step": 5447
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.996153667881717e-05,
+      "loss": 0.6372,
+      "step": 5448
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.996150693331549e-05,
+      "loss": 0.7336,
+      "step": 5449
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.996147717633863e-05,
+      "loss": 0.5723,
+      "step": 5450
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961447407886636e-05,
+      "loss": 0.7702,
+      "step": 5451
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961417627959532e-05,
+      "loss": 0.7251,
+      "step": 5452
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961387836557355e-05,
+      "loss": 0.7746,
+      "step": 5453
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961358033680144e-05,
+      "loss": 0.7783,
+      "step": 5454
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961328219327924e-05,
+      "loss": 0.7226,
+      "step": 5455
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961298393500736e-05,
+      "loss": 0.7269,
+      "step": 5456
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961268556198612e-05,
+      "loss": 0.7338,
+      "step": 5457
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961238707421587e-05,
+      "loss": 0.6613,
+      "step": 5458
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.99612088471697e-05,
+      "loss": 0.6553,
+      "step": 5459
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961178975442977e-05,
+      "loss": 0.6425,
+      "step": 5460
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.996114909224146e-05,
+      "loss": 0.5739,
+      "step": 5461
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961119197565176e-05,
+      "loss": 0.775,
+      "step": 5462
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961089291414165e-05,
+      "loss": 0.671,
+      "step": 5463
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9961059373788462e-05,
+      "loss": 0.6585,
+      "step": 5464
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.99610294446881e-05,
+      "loss": 0.6374,
+      "step": 5465
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9960999504113107e-05,
+      "loss": 0.6379,
+      "step": 5466
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9960969552063528e-05,
+      "loss": 0.7349,
+      "step": 5467
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.996093958853939e-05,
+      "loss": 0.6462,
+      "step": 5468
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9960909613540735e-05,
+      "loss": 0.6667,
+      "step": 5469
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9960879627067588e-05,
+      "loss": 0.6036,
+      "step": 5470
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.996084962911999e-05,
+      "loss": 0.6944,
+      "step": 5471
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9960819619697973e-05,
+      "loss": 0.8054,
+      "step": 5472
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9960789598801574e-05,
+      "loss": 0.5231,
+      "step": 5473
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9960759566430825e-05,
+      "loss": 0.7416,
+      "step": 5474
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9960729522585762e-05,
+      "loss": 0.6762,
+      "step": 5475
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.996069946726642e-05,
+      "loss": 0.5812,
+      "step": 5476
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9960669400472833e-05,
+      "loss": 0.7987,
+      "step": 5477
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9960639322205035e-05,
+      "loss": 0.6215,
+      "step": 5478
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.996060923246306e-05,
+      "loss": 0.6541,
+      "step": 5479
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960579131246945e-05,
+      "loss": 0.6802,
+      "step": 5480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960549018556724e-05,
+      "loss": 0.5381,
+      "step": 5481
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.996051889439243e-05,
+      "loss": 0.7173,
+      "step": 5482
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960488758754096e-05,
+      "loss": 0.6763,
+      "step": 5483
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960458611641768e-05,
+      "loss": 0.888,
+      "step": 5484
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960428453055466e-05,
+      "loss": 0.6809,
+      "step": 5485
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.996039828299523e-05,
+      "loss": 0.7842,
+      "step": 5486
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.99603681014611e-05,
+      "loss": 0.5903,
+      "step": 5487
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960337908453106e-05,
+      "loss": 0.6497,
+      "step": 5488
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960307703971283e-05,
+      "loss": 0.7912,
+      "step": 5489
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960277488015664e-05,
+      "loss": 0.6126,
+      "step": 5490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.996024726058629e-05,
+      "loss": 0.7326,
+      "step": 5491
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.996021702168319e-05,
+      "loss": 0.6258,
+      "step": 5492
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960186771306397e-05,
+      "loss": 0.781,
+      "step": 5493
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960156509455954e-05,
+      "loss": 0.6929,
+      "step": 5494
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960126236131888e-05,
+      "loss": 0.6727,
+      "step": 5495
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960095951334238e-05,
+      "loss": 0.7359,
+      "step": 5496
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.996006565506304e-05,
+      "loss": 0.7758,
+      "step": 5497
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960035347318326e-05,
+      "loss": 0.4905,
+      "step": 5498
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9960005028100134e-05,
+      "loss": 0.5347,
+      "step": 5499
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959974697408495e-05,
+      "loss": 0.929,
+      "step": 5500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.995994435524345e-05,
+      "loss": 0.7701,
+      "step": 5501
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959914001605022e-05,
+      "loss": 0.8176,
+      "step": 5502
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.995988363649326e-05,
+      "loss": 0.5669,
+      "step": 5503
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959853259908187e-05,
+      "loss": 0.7945,
+      "step": 5504
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.995982287184985e-05,
+      "loss": 0.8004,
+      "step": 5505
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959792472318273e-05,
+      "loss": 0.7953,
+      "step": 5506
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959762061313497e-05,
+      "loss": 0.7249,
+      "step": 5507
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959731638835555e-05,
+      "loss": 0.5454,
+      "step": 5508
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959701204884484e-05,
+      "loss": 0.7193,
+      "step": 5509
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959670759460318e-05,
+      "loss": 0.5752,
+      "step": 5510
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959640302563093e-05,
+      "loss": 0.7563,
+      "step": 5511
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.995960983419284e-05,
+      "loss": 0.7495,
+      "step": 5512
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.99595793543496e-05,
+      "loss": 0.8355,
+      "step": 5513
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959548863033405e-05,
+      "loss": 0.6394,
+      "step": 5514
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9959518360244288e-05,
+      "loss": 0.5866,
+      "step": 5515
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.995948784598229e-05,
+      "loss": 0.7266,
+      "step": 5516
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995945732024744e-05,
+      "loss": 0.725,
+      "step": 5517
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9959426783039776e-05,
+      "loss": 0.7165,
+      "step": 5518
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9959396234359332e-05,
+      "loss": 0.7045,
+      "step": 5519
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9959365674206145e-05,
+      "loss": 0.6634,
+      "step": 5520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995933510258025e-05,
+      "loss": 0.5888,
+      "step": 5521
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9959304519481683e-05,
+      "loss": 0.8756,
+      "step": 5522
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9959273924910478e-05,
+      "loss": 0.7993,
+      "step": 5523
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9959243318866666e-05,
+      "loss": 0.6831,
+      "step": 5524
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9959212701350287e-05,
+      "loss": 0.5252,
+      "step": 5525
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995918207236138e-05,
+      "loss": 0.8217,
+      "step": 5526
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995915143189997e-05,
+      "loss": 0.864,
+      "step": 5527
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9959120779966103e-05,
+      "loss": 0.8208,
+      "step": 5528
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9959090116559807e-05,
+      "loss": 0.8814,
+      "step": 5529
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995905944168112e-05,
+      "loss": 0.849,
+      "step": 5530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995902875533008e-05,
+      "loss": 0.7287,
+      "step": 5531
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958998057506714e-05,
+      "loss": 0.6921,
+      "step": 5532
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995896734821107e-05,
+      "loss": 0.5091,
+      "step": 5533
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995893662744317e-05,
+      "loss": 0.9006,
+      "step": 5534
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958905895203057e-05,
+      "loss": 0.6692,
+      "step": 5535
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958875151490766e-05,
+      "loss": 0.7716,
+      "step": 5536
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958844396306333e-05,
+      "loss": 0.6843,
+      "step": 5537
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995881362964979e-05,
+      "loss": 0.7665,
+      "step": 5538
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958782851521175e-05,
+      "loss": 0.6631,
+      "step": 5539
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995875206192052e-05,
+      "loss": 0.7035,
+      "step": 5540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958721260847865e-05,
+      "loss": 0.6722,
+      "step": 5541
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958690448303245e-05,
+      "loss": 0.6342,
+      "step": 5542
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995865962428669e-05,
+      "loss": 0.6606,
+      "step": 5543
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958628788798245e-05,
+      "loss": 0.727,
+      "step": 5544
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958597941837937e-05,
+      "loss": 0.6941,
+      "step": 5545
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.995856708340581e-05,
+      "loss": 0.9595,
+      "step": 5546
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958536213501885e-05,
+      "loss": 0.6018,
+      "step": 5547
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958505332126215e-05,
+      "loss": 0.8314,
+      "step": 5548
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958474439278826e-05,
+      "loss": 0.6881,
+      "step": 5549
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958443534959754e-05,
+      "loss": 0.6722,
+      "step": 5550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958412619169034e-05,
+      "loss": 0.7197,
+      "step": 5551
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958381691906706e-05,
+      "loss": 0.6741,
+      "step": 5552
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958350753172798e-05,
+      "loss": 0.627,
+      "step": 5553
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9958319802967353e-05,
+      "loss": 0.6234,
+      "step": 5554
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9958288841290408e-05,
+      "loss": 0.7149,
+      "step": 5555
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995825786814199e-05,
+      "loss": 0.7177,
+      "step": 5556
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9958226883522143e-05,
+      "loss": 0.6034,
+      "step": 5557
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9958195887430897e-05,
+      "loss": 0.8714,
+      "step": 5558
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995816487986829e-05,
+      "loss": 0.6886,
+      "step": 5559
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9958133860834358e-05,
+      "loss": 0.7559,
+      "step": 5560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995810283032914e-05,
+      "loss": 0.6777,
+      "step": 5561
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995807178835266e-05,
+      "loss": 0.7524,
+      "step": 5562
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995804073490497e-05,
+      "loss": 0.58,
+      "step": 5563
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9958009669986092e-05,
+      "loss": 0.697,
+      "step": 5564
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995797859359607e-05,
+      "loss": 0.8862,
+      "step": 5565
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995794750573494e-05,
+      "loss": 0.7648,
+      "step": 5566
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995791640640273e-05,
+      "loss": 0.5415,
+      "step": 5567
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957885295599482e-05,
+      "loss": 0.582,
+      "step": 5568
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995785417332523e-05,
+      "loss": 0.9196,
+      "step": 5569
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957823039580013e-05,
+      "loss": 0.7774,
+      "step": 5570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957791894363864e-05,
+      "loss": 0.773,
+      "step": 5571
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957760737676817e-05,
+      "loss": 0.7732,
+      "step": 5572
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957729569518912e-05,
+      "loss": 0.6303,
+      "step": 5573
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957698389890182e-05,
+      "loss": 0.6924,
+      "step": 5574
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957667198790664e-05,
+      "loss": 0.7427,
+      "step": 5575
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957635996220395e-05,
+      "loss": 0.5959,
+      "step": 5576
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957604782179413e-05,
+      "loss": 0.7303,
+      "step": 5577
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957573556667745e-05,
+      "loss": 0.6318,
+      "step": 5578
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957542319685435e-05,
+      "loss": 0.7738,
+      "step": 5579
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957511071232516e-05,
+      "loss": 0.6705,
+      "step": 5580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957479811309027e-05,
+      "loss": 0.6184,
+      "step": 5581
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957448539915e-05,
+      "loss": 0.6749,
+      "step": 5582
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957417257050474e-05,
+      "loss": 0.6023,
+      "step": 5583
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957385962715485e-05,
+      "loss": 0.7989,
+      "step": 5584
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957354656910064e-05,
+      "loss": 0.6244,
+      "step": 5585
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957323339634254e-05,
+      "loss": 0.7733,
+      "step": 5586
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957292010888088e-05,
+      "loss": 0.6753,
+      "step": 5587
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957260670671597e-05,
+      "loss": 0.7737,
+      "step": 5588
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995722931898483e-05,
+      "loss": 0.7386,
+      "step": 5589
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995719795582781e-05,
+      "loss": 0.6352,
+      "step": 5590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.995716658120058e-05,
+      "loss": 0.719,
+      "step": 5591
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9957135195103175e-05,
+      "loss": 0.8192,
+      "step": 5592
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9957103797535632e-05,
+      "loss": 0.6777,
+      "step": 5593
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9957072388497986e-05,
+      "loss": 0.5483,
+      "step": 5594
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995704096799027e-05,
+      "loss": 0.6426,
+      "step": 5595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9957009536012525e-05,
+      "loss": 0.826,
+      "step": 5596
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956978092564786e-05,
+      "loss": 0.6585,
+      "step": 5597
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995694663764709e-05,
+      "loss": 0.6647,
+      "step": 5598
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995691517125947e-05,
+      "loss": 0.7639,
+      "step": 5599
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956883693401966e-05,
+      "loss": 0.6298,
+      "step": 5600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995685220407461e-05,
+      "loss": 0.9038,
+      "step": 5601
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956820703277443e-05,
+      "loss": 0.7615,
+      "step": 5602
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956789191010495e-05,
+      "loss": 0.7784,
+      "step": 5603
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995675766727381e-05,
+      "loss": 0.7712,
+      "step": 5604
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995672613206742e-05,
+      "loss": 0.6105,
+      "step": 5605
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956694585391363e-05,
+      "loss": 0.9536,
+      "step": 5606
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956663027245672e-05,
+      "loss": 0.6331,
+      "step": 5607
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956631457630386e-05,
+      "loss": 0.5515,
+      "step": 5608
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956599876545542e-05,
+      "loss": 0.6929,
+      "step": 5609
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956568283991178e-05,
+      "loss": 0.7952,
+      "step": 5610
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956536679967326e-05,
+      "loss": 0.696,
+      "step": 5611
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995650506447402e-05,
+      "loss": 0.6903,
+      "step": 5612
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995647343751131e-05,
+      "loss": 0.5471,
+      "step": 5613
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956441799079217e-05,
+      "loss": 0.7999,
+      "step": 5614
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995641014917778e-05,
+      "loss": 0.7101,
+      "step": 5615
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956378487807044e-05,
+      "loss": 0.6831,
+      "step": 5616
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995634681496704e-05,
+      "loss": 0.8025,
+      "step": 5617
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956315130657802e-05,
+      "loss": 0.4416,
+      "step": 5618
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956283434879374e-05,
+      "loss": 0.9378,
+      "step": 5619
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956251727631785e-05,
+      "loss": 0.7784,
+      "step": 5620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956220008915074e-05,
+      "loss": 0.5354,
+      "step": 5621
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995618827872928e-05,
+      "loss": 0.8824,
+      "step": 5622
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956156537074437e-05,
+      "loss": 0.8765,
+      "step": 5623
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956124783950584e-05,
+      "loss": 0.9828,
+      "step": 5624
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956093019357753e-05,
+      "loss": 0.6711,
+      "step": 5625
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9956061243295984e-05,
+      "loss": 0.615,
+      "step": 5626
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995602945576531e-05,
+      "loss": 0.6196,
+      "step": 5627
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9955997656765775e-05,
+      "loss": 0.56,
+      "step": 5628
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995596584629741e-05,
+      "loss": 0.6625,
+      "step": 5629
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.995593402436025e-05,
+      "loss": 0.7309,
+      "step": 5630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.995590219095434e-05,
+      "loss": 0.7313,
+      "step": 5631
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955870346079706e-05,
+      "loss": 0.8343,
+      "step": 5632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.995583848973639e-05,
+      "loss": 0.6808,
+      "step": 5633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.995580662192443e-05,
+      "loss": 0.5571,
+      "step": 5634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955774742643867e-05,
+      "loss": 0.6699,
+      "step": 5635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.995574285189472e-05,
+      "loss": 0.8692,
+      "step": 5636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955710949677047e-05,
+      "loss": 0.8123,
+      "step": 5637
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955679035990872e-05,
+      "loss": 1.0141,
+      "step": 5638
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955647110836238e-05,
+      "loss": 0.965,
+      "step": 5639
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955615174213173e-05,
+      "loss": 0.5744,
+      "step": 5640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955583226121726e-05,
+      "loss": 0.7712,
+      "step": 5641
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955551266561924e-05,
+      "loss": 0.6234,
+      "step": 5642
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.995551929553381e-05,
+      "loss": 0.5912,
+      "step": 5643
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955487313037417e-05,
+      "loss": 0.721,
+      "step": 5644
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955455319072784e-05,
+      "loss": 0.7291,
+      "step": 5645
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955423313639944e-05,
+      "loss": 0.68,
+      "step": 5646
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.995539129673894e-05,
+      "loss": 0.7079,
+      "step": 5647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.99553592683698e-05,
+      "loss": 0.6266,
+      "step": 5648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.995532722853257e-05,
+      "loss": 0.7421,
+      "step": 5649
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955295177227287e-05,
+      "loss": 0.5216,
+      "step": 5650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.995526311445398e-05,
+      "loss": 0.6189,
+      "step": 5651
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955231040212696e-05,
+      "loss": 0.7336,
+      "step": 5652
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955198954503457e-05,
+      "loss": 0.6641,
+      "step": 5653
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955166857326317e-05,
+      "loss": 0.8148,
+      "step": 5654
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955134748681303e-05,
+      "loss": 0.7834,
+      "step": 5655
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955102628568453e-05,
+      "loss": 0.8344,
+      "step": 5656
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955070496987806e-05,
+      "loss": 0.6329,
+      "step": 5657
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955038353939397e-05,
+      "loss": 0.7128,
+      "step": 5658
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9955006199423266e-05,
+      "loss": 0.7333,
+      "step": 5659
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9954974033439448e-05,
+      "loss": 0.6551,
+      "step": 5660
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.995494185598798e-05,
+      "loss": 0.6406,
+      "step": 5661
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.99549096670689e-05,
+      "loss": 0.7057,
+      "step": 5662
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9954877466682243e-05,
+      "loss": 0.627,
+      "step": 5663
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9954845254828047e-05,
+      "loss": 0.6325,
+      "step": 5664
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9954813031506352e-05,
+      "loss": 0.6716,
+      "step": 5665
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9954780796717194e-05,
+      "loss": 0.7962,
+      "step": 5666
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9954748550460605e-05,
+      "loss": 0.7357,
+      "step": 5667
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9954716292736627e-05,
+      "loss": 0.7521,
+      "step": 5668
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954684023545297e-05,
+      "loss": 0.6465,
+      "step": 5669
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.995465174288665e-05,
+      "loss": 0.5631,
+      "step": 5670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954619450760728e-05,
+      "loss": 0.7674,
+      "step": 5671
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954587147167565e-05,
+      "loss": 0.593,
+      "step": 5672
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954554832107195e-05,
+      "loss": 0.762,
+      "step": 5673
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.995452250557966e-05,
+      "loss": 0.6023,
+      "step": 5674
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954490167584994e-05,
+      "loss": 0.7219,
+      "step": 5675
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954457818123238e-05,
+      "loss": 0.8062,
+      "step": 5676
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954425457194424e-05,
+      "loss": 0.8161,
+      "step": 5677
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954393084798593e-05,
+      "loss": 0.5611,
+      "step": 5678
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954360700935783e-05,
+      "loss": 0.6687,
+      "step": 5679
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954328305606028e-05,
+      "loss": 0.66,
+      "step": 5680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954295898809367e-05,
+      "loss": 0.6097,
+      "step": 5681
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.995426348054584e-05,
+      "loss": 0.7761,
+      "step": 5682
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.995423105081548e-05,
+      "loss": 0.7562,
+      "step": 5683
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954198609618327e-05,
+      "loss": 0.5289,
+      "step": 5684
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954166156954415e-05,
+      "loss": 0.6718,
+      "step": 5685
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.995413369282379e-05,
+      "loss": 0.6952,
+      "step": 5686
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954101217226477e-05,
+      "loss": 0.8379,
+      "step": 5687
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9954068730162524e-05,
+      "loss": 0.6362,
+      "step": 5688
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.995403623163196e-05,
+      "loss": 0.4809,
+      "step": 5689
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.995400372163483e-05,
+      "loss": 0.5933,
+      "step": 5690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953971200171168e-05,
+      "loss": 0.8966,
+      "step": 5691
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.995393866724101e-05,
+      "loss": 0.6786,
+      "step": 5692
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953906122844397e-05,
+      "loss": 0.6446,
+      "step": 5693
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953873566981363e-05,
+      "loss": 0.6911,
+      "step": 5694
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953840999651948e-05,
+      "loss": 0.6511,
+      "step": 5695
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953808420856188e-05,
+      "loss": 0.8903,
+      "step": 5696
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953775830594118e-05,
+      "loss": 0.5128,
+      "step": 5697
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953743228865784e-05,
+      "loss": 0.6287,
+      "step": 5698
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953710615671216e-05,
+      "loss": 0.7801,
+      "step": 5699
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953677991010454e-05,
+      "loss": 0.5335,
+      "step": 5700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953645354883533e-05,
+      "loss": 0.7097,
+      "step": 5701
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953612707290496e-05,
+      "loss": 0.6657,
+      "step": 5702
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953580048231374e-05,
+      "loss": 0.7374,
+      "step": 5703
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.995354737770621e-05,
+      "loss": 0.6875,
+      "step": 5704
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.995351469571504e-05,
+      "loss": 0.6929,
+      "step": 5705
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953482002257904e-05,
+      "loss": 0.6336,
+      "step": 5706
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9953449297334834e-05,
+      "loss": 0.8037,
+      "step": 5707
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9953416580945873e-05,
+      "loss": 0.7603,
+      "step": 5708
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9953383853091054e-05,
+      "loss": 0.7747,
+      "step": 5709
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.995335111377042e-05,
+      "loss": 0.7934,
+      "step": 5710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9953318362984003e-05,
+      "loss": 0.6176,
+      "step": 5711
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9953285600731843e-05,
+      "loss": 0.6346,
+      "step": 5712
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9953252827013977e-05,
+      "loss": 0.6942,
+      "step": 5713
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.995322004183045e-05,
+      "loss": 0.737,
+      "step": 5714
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9953187245181288e-05,
+      "loss": 0.8004,
+      "step": 5715
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.995315443706654e-05,
+      "loss": 0.6158,
+      "step": 5716
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9953121617486234e-05,
+      "loss": 0.5845,
+      "step": 5717
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9953088786440412e-05,
+      "loss": 0.7246,
+      "step": 5718
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9953055943929116e-05,
+      "loss": 0.6915,
+      "step": 5719
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9953023089952378e-05,
+      "loss": 0.845,
+      "step": 5720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952990224510236e-05,
+      "loss": 0.7005,
+      "step": 5721
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952957347602734e-05,
+      "loss": 0.6775,
+      "step": 5722
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.99529244592299e-05,
+      "loss": 0.748,
+      "step": 5723
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952891559391782e-05,
+      "loss": 0.7534,
+      "step": 5724
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.995285864808841e-05,
+      "loss": 0.8885,
+      "step": 5725
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952825725319823e-05,
+      "loss": 0.9389,
+      "step": 5726
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952792791086067e-05,
+      "loss": 0.5264,
+      "step": 5727
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.995275984538717e-05,
+      "loss": 0.7986,
+      "step": 5728
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952726888223174e-05,
+      "loss": 0.8029,
+      "step": 5729
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.995269391959412e-05,
+      "loss": 0.9091,
+      "step": 5730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.995266093950004e-05,
+      "loss": 0.8163,
+      "step": 5731
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952627947940978e-05,
+      "loss": 0.5487,
+      "step": 5732
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952594944916965e-05,
+      "loss": 0.5991,
+      "step": 5733
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952561930428046e-05,
+      "loss": 0.7444,
+      "step": 5734
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952528904474255e-05,
+      "loss": 0.6772,
+      "step": 5735
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952495867055633e-05,
+      "loss": 0.748,
+      "step": 5736
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952462818172212e-05,
+      "loss": 0.873,
+      "step": 5737
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952429757824035e-05,
+      "loss": 0.7685,
+      "step": 5738
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.995239668601114e-05,
+      "loss": 0.6114,
+      "step": 5739
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952363602733566e-05,
+      "loss": 0.8336,
+      "step": 5740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952330507991348e-05,
+      "loss": 0.7276,
+      "step": 5741
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952297401784526e-05,
+      "loss": 0.6705,
+      "step": 5742
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952264284113136e-05,
+      "loss": 0.6719,
+      "step": 5743
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.995223115497722e-05,
+      "loss": 0.7254,
+      "step": 5744
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9952198014376812e-05,
+      "loss": 0.7567,
+      "step": 5745
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9952164862311955e-05,
+      "loss": 0.8027,
+      "step": 5746
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9952131698782683e-05,
+      "loss": 0.6793,
+      "step": 5747
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9952098523789033e-05,
+      "loss": 0.7842,
+      "step": 5748
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9952065337331047e-05,
+      "loss": 0.8467,
+      "step": 5749
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9952032139408764e-05,
+      "loss": 0.8399,
+      "step": 5750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.995199893002222e-05,
+      "loss": 0.8099,
+      "step": 5751
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951965709171453e-05,
+      "loss": 0.8127,
+      "step": 5752
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.99519324768565e-05,
+      "loss": 0.7802,
+      "step": 5753
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.99518992330774e-05,
+      "loss": 0.6558,
+      "step": 5754
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951865977834196e-05,
+      "loss": 0.6713,
+      "step": 5755
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951832711126923e-05,
+      "loss": 0.6649,
+      "step": 5756
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951799432955617e-05,
+      "loss": 0.6387,
+      "step": 5757
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.995176614332032e-05,
+      "loss": 0.6277,
+      "step": 5758
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951732842221067e-05,
+      "loss": 0.6609,
+      "step": 5759
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951699529657896e-05,
+      "loss": 0.6012,
+      "step": 5760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951666205630852e-05,
+      "loss": 0.6482,
+      "step": 5761
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951632870139967e-05,
+      "loss": 0.7038,
+      "step": 5762
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.995159952318528e-05,
+      "loss": 0.7163,
+      "step": 5763
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951566164766832e-05,
+      "loss": 0.7984,
+      "step": 5764
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951532794884657e-05,
+      "loss": 0.7451,
+      "step": 5765
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.99514994135388e-05,
+      "loss": 0.6682,
+      "step": 5766
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951466020729296e-05,
+      "loss": 0.6476,
+      "step": 5767
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.995143261645618e-05,
+      "loss": 0.8865,
+      "step": 5768
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951399200719497e-05,
+      "loss": 0.5898,
+      "step": 5769
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.995136577351928e-05,
+      "loss": 0.6773,
+      "step": 5770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951332334855574e-05,
+      "loss": 0.8637,
+      "step": 5771
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951298884728408e-05,
+      "loss": 0.6978,
+      "step": 5772
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.995126542313783e-05,
+      "loss": 0.5895,
+      "step": 5773
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951231950083875e-05,
+      "loss": 0.7254,
+      "step": 5774
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951198465566578e-05,
+      "loss": 0.8085,
+      "step": 5775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.995116496958598e-05,
+      "loss": 0.6338,
+      "step": 5776
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951131462142122e-05,
+      "loss": 0.7313,
+      "step": 5777
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951097943235044e-05,
+      "loss": 0.626,
+      "step": 5778
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951064412864774e-05,
+      "loss": 0.6006,
+      "step": 5779
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9951030871031364e-05,
+      "loss": 0.7814,
+      "step": 5780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9950997317734843e-05,
+      "loss": 0.6181,
+      "step": 5781
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.995096375297526e-05,
+      "loss": 0.7753,
+      "step": 5782
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.995093017675264e-05,
+      "loss": 0.804,
+      "step": 5783
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950896589067032e-05,
+      "loss": 0.7098,
+      "step": 5784
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950862989918466e-05,
+      "loss": 0.7113,
+      "step": 5785
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.995082937930699e-05,
+      "loss": 0.7835,
+      "step": 5786
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.995079575723264e-05,
+      "loss": 0.7238,
+      "step": 5787
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950762123695452e-05,
+      "loss": 0.7979,
+      "step": 5788
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950728478695468e-05,
+      "loss": 0.6533,
+      "step": 5789
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950694822232723e-05,
+      "loss": 0.8237,
+      "step": 5790
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.995066115430726e-05,
+      "loss": 0.7721,
+      "step": 5791
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950627474919114e-05,
+      "loss": 0.7175,
+      "step": 5792
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950593784068325e-05,
+      "loss": 0.5963,
+      "step": 5793
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.995056008175493e-05,
+      "loss": 0.7335,
+      "step": 5794
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950526367978976e-05,
+      "loss": 0.7149,
+      "step": 5795
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950492642740492e-05,
+      "loss": 0.6409,
+      "step": 5796
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.995045890603952e-05,
+      "loss": 0.6905,
+      "step": 5797
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.99504251578761e-05,
+      "loss": 0.636,
+      "step": 5798
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.995039139825027e-05,
+      "loss": 1.0298,
+      "step": 5799
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.995035762716207e-05,
+      "loss": 0.6096,
+      "step": 5800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950323844611538e-05,
+      "loss": 0.7099,
+      "step": 5801
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950290050598716e-05,
+      "loss": 0.7356,
+      "step": 5802
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950256245123638e-05,
+      "loss": 0.6676,
+      "step": 5803
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950222428186343e-05,
+      "loss": 0.629,
+      "step": 5804
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950188599786874e-05,
+      "loss": 0.5657,
+      "step": 5805
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.995015475992527e-05,
+      "loss": 0.7474,
+      "step": 5806
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950120908601562e-05,
+      "loss": 0.6393,
+      "step": 5807
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950087045815797e-05,
+      "loss": 0.8354,
+      "step": 5808
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9950053171568016e-05,
+      "loss": 0.5729,
+      "step": 5809
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.995001928585825e-05,
+      "loss": 0.8105,
+      "step": 5810
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9949985388686543e-05,
+      "loss": 0.7518,
+      "step": 5811
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9949951480052934e-05,
+      "loss": 0.7758,
+      "step": 5812
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.994991755995746e-05,
+      "loss": 0.7872,
+      "step": 5813
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.994988362840016e-05,
+      "loss": 0.6538,
+      "step": 5814
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9949849685381077e-05,
+      "loss": 0.5478,
+      "step": 5815
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9949815730900245e-05,
+      "loss": 0.7635,
+      "step": 5816
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.994978176495771e-05,
+      "loss": 0.6525,
+      "step": 5817
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.99497477875535e-05,
+      "loss": 0.721,
+      "step": 5818
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9949713798687663e-05,
+      "loss": 0.7664,
+      "step": 5819
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9949679798360234e-05,
+      "loss": 0.6767,
+      "step": 5820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.994964578657126e-05,
+      "loss": 0.5869,
+      "step": 5821
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949611763320772e-05,
+      "loss": 0.8947,
+      "step": 5822
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949577728608807e-05,
+      "loss": 0.6322,
+      "step": 5823
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994954368243541e-05,
+      "loss": 0.7618,
+      "step": 5824
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949509624800624e-05,
+      "loss": 0.5988,
+      "step": 5825
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949475555704477e-05,
+      "loss": 0.738,
+      "step": 5826
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994944147514702e-05,
+      "loss": 0.8301,
+      "step": 5827
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994940738312828e-05,
+      "loss": 0.7626,
+      "step": 5828
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994937327964831e-05,
+      "loss": 0.5926,
+      "step": 5829
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949339164707138e-05,
+      "loss": 0.5992,
+      "step": 5830
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949305038304808e-05,
+      "loss": 0.6361,
+      "step": 5831
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949270900441356e-05,
+      "loss": 0.5948,
+      "step": 5832
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949236751116827e-05,
+      "loss": 0.5339,
+      "step": 5833
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949202590331258e-05,
+      "loss": 0.6145,
+      "step": 5834
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949168418084686e-05,
+      "loss": 0.5957,
+      "step": 5835
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949134234377152e-05,
+      "loss": 0.6199,
+      "step": 5836
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949100039208696e-05,
+      "loss": 0.5795,
+      "step": 5837
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949065832579356e-05,
+      "loss": 0.9118,
+      "step": 5838
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9949031614489176e-05,
+      "loss": 0.7216,
+      "step": 5839
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994899738493819e-05,
+      "loss": 0.6764,
+      "step": 5840
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994896314392644e-05,
+      "loss": 0.654,
+      "step": 5841
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9948928891453962e-05,
+      "loss": 0.6673,
+      "step": 5842
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.99488946275208e-05,
+      "loss": 0.5786,
+      "step": 5843
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9948860352126987e-05,
+      "loss": 0.5613,
+      "step": 5844
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994882606527257e-05,
+      "loss": 0.6626,
+      "step": 5845
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9948791766957586e-05,
+      "loss": 0.5891,
+      "step": 5846
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9948757457182076e-05,
+      "loss": 0.6544,
+      "step": 5847
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9948723135946078e-05,
+      "loss": 0.5815,
+      "step": 5848
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9948688803249626e-05,
+      "loss": 0.6247,
+      "step": 5849
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9948654459092767e-05,
+      "loss": 0.6815,
+      "step": 5850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994862010347554e-05,
+      "loss": 0.6939,
+      "step": 5851
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994858573639798e-05,
+      "loss": 0.6094,
+      "step": 5852
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994855135786013e-05,
+      "loss": 0.8442,
+      "step": 5853
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9948516967862032e-05,
+      "loss": 0.6969,
+      "step": 5854
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994848256640372e-05,
+      "loss": 0.6387,
+      "step": 5855
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9948448153485235e-05,
+      "loss": 0.5031,
+      "step": 5856
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994841372910662e-05,
+      "loss": 0.629,
+      "step": 5857
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994837929326791e-05,
+      "loss": 0.8612,
+      "step": 5858
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.994834484596915e-05,
+      "loss": 0.8148,
+      "step": 5859
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9948310387210376e-05,
+      "loss": 0.6538,
+      "step": 5860
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9948275916991626e-05,
+      "loss": 0.7768,
+      "step": 5861
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9948241435312944e-05,
+      "loss": 0.5604,
+      "step": 5862
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9948206942174368e-05,
+      "loss": 0.784,
+      "step": 5863
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.994817243757594e-05,
+      "loss": 0.6067,
+      "step": 5864
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9948137921517695e-05,
+      "loss": 0.7448,
+      "step": 5865
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9948103393999677e-05,
+      "loss": 0.5825,
+      "step": 5866
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.994806885502192e-05,
+      "loss": 0.7304,
+      "step": 5867
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9948034304584468e-05,
+      "loss": 0.6128,
+      "step": 5868
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947999742687365e-05,
+      "loss": 0.7925,
+      "step": 5869
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947965169330645e-05,
+      "loss": 0.645,
+      "step": 5870
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947930584514348e-05,
+      "loss": 0.7764,
+      "step": 5871
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947895988238515e-05,
+      "loss": 0.5371,
+      "step": 5872
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947861380503188e-05,
+      "loss": 0.7032,
+      "step": 5873
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947826761308402e-05,
+      "loss": 0.7333,
+      "step": 5874
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.99477921306542e-05,
+      "loss": 0.7905,
+      "step": 5875
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947757488540623e-05,
+      "loss": 0.6822,
+      "step": 5876
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947722834967707e-05,
+      "loss": 0.7696,
+      "step": 5877
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947688169935496e-05,
+      "loss": 0.7201,
+      "step": 5878
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.994765349344403e-05,
+      "loss": 0.8397,
+      "step": 5879
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947618805493345e-05,
+      "loss": 0.7772,
+      "step": 5880
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947584106083485e-05,
+      "loss": 0.8203,
+      "step": 5881
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947549395214486e-05,
+      "loss": 0.7155,
+      "step": 5882
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947514672886392e-05,
+      "loss": 0.5911,
+      "step": 5883
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947479939099243e-05,
+      "loss": 0.6779,
+      "step": 5884
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947445193853073e-05,
+      "loss": 0.6462,
+      "step": 5885
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947410437147926e-05,
+      "loss": 0.8444,
+      "step": 5886
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947375668983843e-05,
+      "loss": 0.7018,
+      "step": 5887
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947340889360865e-05,
+      "loss": 0.5708,
+      "step": 5888
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.994730609827903e-05,
+      "loss": 0.701,
+      "step": 5889
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947271295738377e-05,
+      "loss": 0.7471,
+      "step": 5890
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947236481738946e-05,
+      "loss": 0.7233,
+      "step": 5891
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947201656280782e-05,
+      "loss": 0.6312,
+      "step": 5892
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947166819363923e-05,
+      "loss": 0.7247,
+      "step": 5893
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947131970988405e-05,
+      "loss": 0.6902,
+      "step": 5894
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947097111154273e-05,
+      "loss": 1.0046,
+      "step": 5895
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947062239861564e-05,
+      "loss": 0.795,
+      "step": 5896
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9947027357110318e-05,
+      "loss": 0.7772,
+      "step": 5897
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9946992462900577e-05,
+      "loss": 0.7622,
+      "step": 5898
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.994695755723238e-05,
+      "loss": 0.6689,
+      "step": 5899
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946922640105774e-05,
+      "loss": 0.6179,
+      "step": 5900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946887711520786e-05,
+      "loss": 0.6819,
+      "step": 5901
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946852771477465e-05,
+      "loss": 0.7447,
+      "step": 5902
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.994681781997585e-05,
+      "loss": 0.6967,
+      "step": 5903
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946782857015984e-05,
+      "loss": 0.8658,
+      "step": 5904
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946747882597897e-05,
+      "loss": 0.7172,
+      "step": 5905
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946712896721644e-05,
+      "loss": 0.4458,
+      "step": 5906
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946677899387255e-05,
+      "loss": 0.897,
+      "step": 5907
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946642890594772e-05,
+      "loss": 0.5719,
+      "step": 5908
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946607870344238e-05,
+      "loss": 0.6524,
+      "step": 5909
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.994657283863569e-05,
+      "loss": 0.7385,
+      "step": 5910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.994653779546917e-05,
+      "loss": 0.7306,
+      "step": 5911
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.994650274084472e-05,
+      "loss": 0.6653,
+      "step": 5912
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.994646767476238e-05,
+      "loss": 0.7121,
+      "step": 5913
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946432597222186e-05,
+      "loss": 0.6153,
+      "step": 5914
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946397508224184e-05,
+      "loss": 0.6263,
+      "step": 5915
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946362407768412e-05,
+      "loss": 0.6178,
+      "step": 5916
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.994632729585491e-05,
+      "loss": 0.7209,
+      "step": 5917
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946292172483715e-05,
+      "loss": 0.8575,
+      "step": 5918
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946257037654878e-05,
+      "loss": 0.6047,
+      "step": 5919
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.994622189136843e-05,
+      "loss": 0.6993,
+      "step": 5920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946186733624415e-05,
+      "loss": 0.8072,
+      "step": 5921
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946151564422872e-05,
+      "loss": 0.6234,
+      "step": 5922
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946116383763844e-05,
+      "loss": 0.8263,
+      "step": 5923
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946081191647367e-05,
+      "loss": 0.5795,
+      "step": 5924
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9946045988073486e-05,
+      "loss": 0.8153,
+      "step": 5925
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.994601077304224e-05,
+      "loss": 0.5717,
+      "step": 5926
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9945975546553667e-05,
+      "loss": 0.6235,
+      "step": 5927
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9945940308607816e-05,
+      "loss": 0.6943,
+      "step": 5928
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9945905059204718e-05,
+      "loss": 0.736,
+      "step": 5929
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9945869798344416e-05,
+      "loss": 0.5002,
+      "step": 5930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9945834526026953e-05,
+      "loss": 0.6134,
+      "step": 5931
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9945799242252374e-05,
+      "loss": 0.5702,
+      "step": 5932
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9945763947020708e-05,
+      "loss": 0.5233,
+      "step": 5933
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9945728640332006e-05,
+      "loss": 0.6145,
+      "step": 5934
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.99456933221863e-05,
+      "loss": 0.7737,
+      "step": 5935
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9945657992583637e-05,
+      "loss": 0.7172,
+      "step": 5936
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9945622651524056e-05,
+      "loss": 0.7071,
+      "step": 5937
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.99455872990076e-05,
+      "loss": 0.631,
+      "step": 5938
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9945551935034304e-05,
+      "loss": 0.6986,
+      "step": 5939
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9945516559604213e-05,
+      "loss": 0.5948,
+      "step": 5940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994548117271737e-05,
+      "loss": 0.7501,
+      "step": 5941
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994544577437381e-05,
+      "loss": 0.587,
+      "step": 5942
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9945410364573576e-05,
+      "loss": 0.7431,
+      "step": 5943
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994537494331671e-05,
+      "loss": 0.806,
+      "step": 5944
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994533951060325e-05,
+      "loss": 0.644,
+      "step": 5945
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9945304066433243e-05,
+      "loss": 0.7392,
+      "step": 5946
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9945268610806723e-05,
+      "loss": 0.5797,
+      "step": 5947
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994523314372373e-05,
+      "loss": 0.6462,
+      "step": 5948
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9945197665184316e-05,
+      "loss": 0.6399,
+      "step": 5949
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994516217518851e-05,
+      "loss": 0.6598,
+      "step": 5950
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9945126673736357e-05,
+      "loss": 0.6488,
+      "step": 5951
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.99450911608279e-05,
+      "loss": 0.7557,
+      "step": 5952
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9945055636463176e-05,
+      "loss": 0.6677,
+      "step": 5953
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9945020100642226e-05,
+      "loss": 0.7075,
+      "step": 5954
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944984553365096e-05,
+      "loss": 0.6428,
+      "step": 5955
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944948994631824e-05,
+      "loss": 0.7035,
+      "step": 5956
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994491342444245e-05,
+      "loss": 0.8112,
+      "step": 5957
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944877842797013e-05,
+      "loss": 0.6925,
+      "step": 5958
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994484224969556e-05,
+      "loss": 0.758,
+      "step": 5959
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944806645138127e-05,
+      "loss": 0.5595,
+      "step": 5960
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994477102912476e-05,
+      "loss": 0.5581,
+      "step": 5961
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994473540165549e-05,
+      "loss": 0.6421,
+      "step": 5962
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994469976273037e-05,
+      "loss": 0.691,
+      "step": 5963
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944664112349434e-05,
+      "loss": 0.8506,
+      "step": 5964
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944628450512727e-05,
+      "loss": 0.8059,
+      "step": 5965
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944592777220285e-05,
+      "loss": 0.7025,
+      "step": 5966
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944557092472153e-05,
+      "loss": 0.7927,
+      "step": 5967
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944521396268372e-05,
+      "loss": 0.7079,
+      "step": 5968
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944485688608978e-05,
+      "loss": 0.5637,
+      "step": 5969
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944449969494022e-05,
+      "loss": 0.6588,
+      "step": 5970
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994441423892354e-05,
+      "loss": 0.6628,
+      "step": 5971
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.994437849689757e-05,
+      "loss": 0.7667,
+      "step": 5972
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944342743416155e-05,
+      "loss": 0.7925,
+      "step": 5973
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9944306978479336e-05,
+      "loss": 0.7901,
+      "step": 5974
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.994427120208716e-05,
+      "loss": 0.6981,
+      "step": 5975
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.994423541423966e-05,
+      "loss": 0.5791,
+      "step": 5976
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9944199614936882e-05,
+      "loss": 0.7677,
+      "step": 5977
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9944163804178864e-05,
+      "loss": 0.5845,
+      "step": 5978
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.994412798196565e-05,
+      "loss": 0.6207,
+      "step": 5979
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9944092148297284e-05,
+      "loss": 0.6915,
+      "step": 5980
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.99440563031738e-05,
+      "loss": 0.5657,
+      "step": 5981
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9944020446595244e-05,
+      "loss": 0.7025,
+      "step": 5982
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943984578561654e-05,
+      "loss": 0.6325,
+      "step": 5983
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943948699073078e-05,
+      "loss": 0.6568,
+      "step": 5984
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.994391280812955e-05,
+      "loss": 0.6871,
+      "step": 5985
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943876905731116e-05,
+      "loss": 0.5774,
+      "step": 5986
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943840991877814e-05,
+      "loss": 0.6899,
+      "step": 5987
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.994380506656969e-05,
+      "loss": 0.7736,
+      "step": 5988
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943769129806778e-05,
+      "loss": 0.6155,
+      "step": 5989
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943733181589126e-05,
+      "loss": 0.6402,
+      "step": 5990
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943697221916776e-05,
+      "loss": 0.7008,
+      "step": 5991
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943661250789762e-05,
+      "loss": 0.3961,
+      "step": 5992
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943625268208133e-05,
+      "loss": 0.6232,
+      "step": 5993
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943589274171927e-05,
+      "loss": 0.7571,
+      "step": 5994
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943553268681187e-05,
+      "loss": 0.6336,
+      "step": 5995
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943517251735952e-05,
+      "loss": 0.5749,
+      "step": 5996
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943481223336264e-05,
+      "loss": 0.6117,
+      "step": 5997
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.994344518348217e-05,
+      "loss": 0.6673,
+      "step": 5998
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943409132173702e-05,
+      "loss": 0.6035,
+      "step": 5999
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9943373069410907e-05,
+      "loss": 0.5988,
+      "step": 6000
+    }
+  ],
+  "max_steps": 95425,
+  "num_train_epochs": 25,
+  "total_flos": 4.593793318413926e+16,
+  "trial_name": null,
+  "trial_params": null
+}