diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16216 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7072691552062869,
+  "global_step": 2700,
+  "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": 1.4644,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3971358714634999e-08,
+      "loss": 1.5629,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.09570380719525e-08,
+      "loss": 2.0134,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7942717429269998e-08,
+      "loss": 1.401,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.49283967865875e-08,
+      "loss": 1.4192,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1914076143905e-08,
+      "loss": 1.7393,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8899755501222494e-08,
+      "loss": 1.3847,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.5885434858539996e-08,
+      "loss": 1.3837,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.28711142158575e-08,
+      "loss": 1.5526,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.9856793573175e-08,
+      "loss": 1.4067,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.684247293049249e-08,
+      "loss": 1.674,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.382815228781e-08,
+      "loss": 1.4864,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.08138316451275e-08,
+      "loss": 1.488,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.779951100244499e-08,
+      "loss": 1.6202,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.047851903597625e-07,
+      "loss": 1.409,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1177086971707999e-07,
+      "loss": 1.4326,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1875654907439749e-07,
+      "loss": 1.6823,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25742228431715e-07,
+      "loss": 1.4071,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.327279077890325e-07,
+      "loss": 1.4533,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3971358714635e-07,
+      "loss": 1.3791,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.466992665036675e-07,
+      "loss": 1.6526,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5368494586098498e-07,
+      "loss": 1.3785,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.606706252183025e-07,
+      "loss": 1.2634,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6765630457562e-07,
+      "loss": 1.5199,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.746419839329375e-07,
+      "loss": 1.3937,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.81627663290255e-07,
+      "loss": 1.3435,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8861334264757248e-07,
+      "loss": 1.5302,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9559902200488998e-07,
+      "loss": 1.4573,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.025847013622075e-07,
+      "loss": 1.5519,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.09570380719525e-07,
+      "loss": 1.3345,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.165560600768425e-07,
+      "loss": 1.4614,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2354173943415998e-07,
+      "loss": 1.38,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3052741879147748e-07,
+      "loss": 1.4828,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3751309814879497e-07,
+      "loss": 1.3173,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.444987775061125e-07,
+      "loss": 1.3707,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5148445686343e-07,
+      "loss": 1.5127,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.584701362207475e-07,
+      "loss": 1.4279,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.65455815578065e-07,
+      "loss": 1.5132,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.724414949353825e-07,
+      "loss": 1.5465,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.794271742927e-07,
+      "loss": 1.4457,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.864128536500175e-07,
+      "loss": 1.4795,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.93398533007335e-07,
+      "loss": 1.3745,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.003842123646525e-07,
+      "loss": 1.3534,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0736989172196997e-07,
+      "loss": 1.2348,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1435557107928746e-07,
+      "loss": 1.4859,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.21341250436605e-07,
+      "loss": 1.4924,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2832692979392245e-07,
+      "loss": 1.5308,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3531260915124e-07,
+      "loss": 1.4909,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4229828850855744e-07,
+      "loss": 1.2494,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.49283967865875e-07,
+      "loss": 1.3319,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.562696472231925e-07,
+      "loss": 1.2278,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6325532658051e-07,
+      "loss": 1.3761,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7024100593782747e-07,
+      "loss": 1.3013,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7722668529514497e-07,
+      "loss": 1.4193,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8421236465246246e-07,
+      "loss": 1.3547,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9119804400977996e-07,
+      "loss": 1.6448,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.981837233670975e-07,
+      "loss": 1.4847,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.05169402724415e-07,
+      "loss": 1.3958,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.121550820817325e-07,
+      "loss": 1.2839,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1914076143905e-07,
+      "loss": 1.5101,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.261264407963675e-07,
+      "loss": 1.2937,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.33112120153685e-07,
+      "loss": 1.2479,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4009779951100247e-07,
+      "loss": 1.2389,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4708347886831996e-07,
+      "loss": 1.4648,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.540691582256375e-07,
+      "loss": 1.3969,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6105483758295495e-07,
+      "loss": 1.2556,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.680405169402725e-07,
+      "loss": 1.3961,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7502619629758994e-07,
+      "loss": 1.2743,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.820118756549074e-07,
+      "loss": 1.3667,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88997555012225e-07,
+      "loss": 1.4766,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.959832343695424e-07,
+      "loss": 1.3527,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0296891372686e-07,
+      "loss": 1.5932,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.099545930841775e-07,
+      "loss": 1.3812,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.16940272441495e-07,
+      "loss": 1.411,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.239259517988125e-07,
+      "loss": 1.4679,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3091163115613e-07,
+      "loss": 1.2466,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.378973105134475e-07,
+      "loss": 1.3309,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.44882989870765e-07,
+      "loss": 1.2908,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.518686692280825e-07,
+      "loss": 1.3423,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.588543485854e-07,
+      "loss": 1.2479,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.658400279427175e-07,
+      "loss": 1.3398,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.72825707300035e-07,
+      "loss": 1.3885,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.798113866573525e-07,
+      "loss": 1.4727,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8679706601467e-07,
+      "loss": 1.3895,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.937827453719875e-07,
+      "loss": 1.3682,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.00768424729305e-07,
+      "loss": 1.5171,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.077541040866225e-07,
+      "loss": 1.3397,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.147397834439399e-07,
+      "loss": 1.4369,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.217254628012575e-07,
+      "loss": 1.4479,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.287111421585749e-07,
+      "loss": 1.2839,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.356968215158925e-07,
+      "loss": 1.4499,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4268250087321e-07,
+      "loss": 1.2228,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.496681802305274e-07,
+      "loss": 1.3141,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.566538595878449e-07,
+      "loss": 1.3574,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.636395389451625e-07,
+      "loss": 1.6254,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.7062521830248e-07,
+      "loss": 1.2116,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.776108976597975e-07,
+      "loss": 1.3485,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.845965770171149e-07,
+      "loss": 1.313,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.915822563744325e-07,
+      "loss": 1.6239,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9856793573175e-07,
+      "loss": 1.8277,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.055536150890676e-07,
+      "loss": 1.3085,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.12539294446385e-07,
+      "loss": 1.3989,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.195249738037025e-07,
+      "loss": 1.1528,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2651065316102e-07,
+      "loss": 1.3736,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.334963325183376e-07,
+      "loss": 1.3287,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.404820118756549e-07,
+      "loss": 1.3853,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.474676912329724e-07,
+      "loss": 1.5243,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.544533705902899e-07,
+      "loss": 1.2002,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.614390499476075e-07,
+      "loss": 1.4667,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.684247293049249e-07,
+      "loss": 1.4795,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.754104086622424e-07,
+      "loss": 1.3397,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.823960880195599e-07,
+      "loss": 1.3796,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.893817673768775e-07,
+      "loss": 1.3595,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.96367446734195e-07,
+      "loss": 1.4804,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.033531260915124e-07,
+      "loss": 1.3565,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.1033880544883e-07,
+      "loss": 1.4348,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.173244848061475e-07,
+      "loss": 1.4343,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.24310164163465e-07,
+      "loss": 1.5168,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.312958435207824e-07,
+      "loss": 1.3738,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.382815228781e-07,
+      "loss": 1.3617,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.452672022354175e-07,
+      "loss": 1.3616,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.52252881592735e-07,
+      "loss": 1.4238,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.592385609500524e-07,
+      "loss": 1.309,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.6622424030737e-07,
+      "loss": 1.3402,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.732099196646874e-07,
+      "loss": 1.3035,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.801955990220049e-07,
+      "loss": 1.424,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.871812783793225e-07,
+      "loss": 1.3428,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.941669577366399e-07,
+      "loss": 1.3888,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.011526370939574e-07,
+      "loss": 1.2346,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.08138316451275e-07,
+      "loss": 1.3996,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.151239958085925e-07,
+      "loss": 1.389,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.221096751659099e-07,
+      "loss": 1.3885,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.290953545232274e-07,
+      "loss": 1.479,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.36081033880545e-07,
+      "loss": 1.2741,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.430667132378625e-07,
+      "loss": 1.4442,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.500523925951799e-07,
+      "loss": 1.5092,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.570380719524975e-07,
+      "loss": 1.462,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.640237513098149e-07,
+      "loss": 1.2827,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.710094306671325e-07,
+      "loss": 1.555,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.7799511002445e-07,
+      "loss": 1.3258,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.849807893817675e-07,
+      "loss": 1.1876,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.919664687390849e-07,
+      "loss": 1.384,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989521480964025e-07,
+      "loss": 1.2253,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.00593782745372e-06,
+      "loss": 1.3511,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0129235068110374e-06,
+      "loss": 1.1971,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.019909186168355e-06,
+      "loss": 1.5177,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0268948655256724e-06,
+      "loss": 1.3264,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.03388054488299e-06,
+      "loss": 1.4016,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0408662242403074e-06,
+      "loss": 1.2735,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.047851903597625e-06,
+      "loss": 1.3612,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0548375829549424e-06,
+      "loss": 1.3161,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.06182326231226e-06,
+      "loss": 1.3338,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0688089416695776e-06,
+      "loss": 1.2991,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.075794621026895e-06,
+      "loss": 1.3923,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0827803003842124e-06,
+      "loss": 1.3338,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.08976597974153e-06,
+      "loss": 1.2683,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0967516590988476e-06,
+      "loss": 1.4257,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.103737338456165e-06,
+      "loss": 1.3869,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1107230178134824e-06,
+      "loss": 1.2689,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1177086971708e-06,
+      "loss": 1.2631,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1246943765281176e-06,
+      "loss": 1.6753,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.131680055885435e-06,
+      "loss": 1.2705,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1386657352427523e-06,
+      "loss": 1.3548,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.14565141460007e-06,
+      "loss": 1.4222,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1526370939573875e-06,
+      "loss": 1.2803,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.159622773314705e-06,
+      "loss": 1.2818,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1666084526720223e-06,
+      "loss": 1.3757,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.17359413202934e-06,
+      "loss": 1.3327,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1805798113866575e-06,
+      "loss": 1.2702,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.187565490743975e-06,
+      "loss": 1.3148,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1945511701012923e-06,
+      "loss": 1.342,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.20153684945861e-06,
+      "loss": 1.2641,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2085225288159275e-06,
+      "loss": 1.42,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.215508208173245e-06,
+      "loss": 1.2203,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2224938875305625e-06,
+      "loss": 1.4888,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2294795668878799e-06,
+      "loss": 1.2838,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2364652462451975e-06,
+      "loss": 1.3158,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.243450925602515e-06,
+      "loss": 1.3616,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2504366049598325e-06,
+      "loss": 1.3966,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2574222843171499e-06,
+      "loss": 1.3425,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2644079636744675e-06,
+      "loss": 1.2154,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.271393643031785e-06,
+      "loss": 1.3445,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2783793223891024e-06,
+      "loss": 1.4734,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.28536500174642e-06,
+      "loss": 1.3823,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2923506811037376e-06,
+      "loss": 1.156,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2993363604610548e-06,
+      "loss": 1.2277,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3063220398183724e-06,
+      "loss": 1.5267,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3133077191756898e-06,
+      "loss": 1.2093,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3202933985330074e-06,
+      "loss": 1.4447,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.327279077890325e-06,
+      "loss": 1.2998,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3342647572476424e-06,
+      "loss": 1.5573,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.34125043660496e-06,
+      "loss": 1.2754,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3482361159622776e-06,
+      "loss": 1.4213,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.355221795319595e-06,
+      "loss": 1.4138,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3622074746769124e-06,
+      "loss": 1.3404,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3691931540342298e-06,
+      "loss": 1.1863,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3761788333915474e-06,
+      "loss": 1.1644,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.383164512748865e-06,
+      "loss": 1.4019,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3901501921061826e-06,
+      "loss": 1.4631,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3971358714635e-06,
+      "loss": 1.2853,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4041215508208176e-06,
+      "loss": 1.428,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4111072301781352e-06,
+      "loss": 1.2117,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4180929095354523e-06,
+      "loss": 1.2431,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.42507858889277e-06,
+      "loss": 1.2978,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4320642682500873e-06,
+      "loss": 1.3639,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.439049947607405e-06,
+      "loss": 1.3784,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4460356269647225e-06,
+      "loss": 1.1566,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.45302130632204e-06,
+      "loss": 1.4212,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4600069856793575e-06,
+      "loss": 1.4901,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4669926650366751e-06,
+      "loss": 1.2414,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4739783443939925e-06,
+      "loss": 1.2219,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4809640237513099e-06,
+      "loss": 1.8129,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4879497031086273e-06,
+      "loss": 1.3614,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4949353824659449e-06,
+      "loss": 1.1905,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5019210618232625e-06,
+      "loss": 1.241,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5089067411805799e-06,
+      "loss": 1.3266,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5158924205378975e-06,
+      "loss": 1.3816,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.522878099895215e-06,
+      "loss": 1.2888,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5298637792525325e-06,
+      "loss": 1.1424,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5368494586098498e-06,
+      "loss": 1.3706,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5438351379671672e-06,
+      "loss": 1.3192,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5508208173244848e-06,
+      "loss": 1.4127,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5578064966818024e-06,
+      "loss": 1.3136,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5647921760391198e-06,
+      "loss": 1.2164,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5717778553964374e-06,
+      "loss": 1.3042,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.578763534753755e-06,
+      "loss": 1.3313,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5857492141110726e-06,
+      "loss": 1.3361,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.59273489346839e-06,
+      "loss": 1.2423,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5997205728257074e-06,
+      "loss": 1.3384,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6067062521830248e-06,
+      "loss": 1.6595,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6136919315403424e-06,
+      "loss": 1.3548,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.62067761089766e-06,
+      "loss": 1.278,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6276632902549774e-06,
+      "loss": 1.3328,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.634648969612295e-06,
+      "loss": 1.2607,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6416346489696126e-06,
+      "loss": 1.3282,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.64862032832693e-06,
+      "loss": 1.2518,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6556060076842476e-06,
+      "loss": 1.338,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6625916870415647e-06,
+      "loss": 1.5349,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6695773663988823e-06,
+      "loss": 1.2162,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6765630457562e-06,
+      "loss": 1.4837,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6835487251135173e-06,
+      "loss": 1.2767,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.690534404470835e-06,
+      "loss": 1.1707,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6975200838281525e-06,
+      "loss": 1.3064,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.70450576318547e-06,
+      "loss": 1.527,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7114914425427875e-06,
+      "loss": 1.2547,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7184771219001047e-06,
+      "loss": 1.4005,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7254628012574223e-06,
+      "loss": 1.3422,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.73244848061474e-06,
+      "loss": 1.3515,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7394341599720573e-06,
+      "loss": 1.3831,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7464198393293749e-06,
+      "loss": 1.2165,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7534055186866925e-06,
+      "loss": 1.3322,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7603911980440099e-06,
+      "loss": 1.284,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7673768774013275e-06,
+      "loss": 1.3417,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.774362556758645e-06,
+      "loss": 1.321,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7813482361159623e-06,
+      "loss": 1.3982,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7883339154732799e-06,
+      "loss": 1.3393,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7953195948305975e-06,
+      "loss": 1.3325,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8023052741879148e-06,
+      "loss": 1.2982,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8092909535452324e-06,
+      "loss": 1.2627,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.81627663290255e-06,
+      "loss": 1.2171,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8232623122598674e-06,
+      "loss": 1.2088,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.830247991617185e-06,
+      "loss": 1.3465,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8372336709745026e-06,
+      "loss": 1.373,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8442193503318198e-06,
+      "loss": 1.6293,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8512050296891374e-06,
+      "loss": 1.3064,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8581907090464548e-06,
+      "loss": 1.3163,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8651763884037724e-06,
+      "loss": 1.3881,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.87216206776109e-06,
+      "loss": 1.2884,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8791477471184074e-06,
+      "loss": 1.2382,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.886133426475725e-06,
+      "loss": 1.4332,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8931191058330426e-06,
+      "loss": 1.3417,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9001047851903598e-06,
+      "loss": 1.8241,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9070904645476774e-06,
+      "loss": 1.2564,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.914076143904995e-06,
+      "loss": 1.2285,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9210618232623124e-06,
+      "loss": 1.1477,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9280475026196297e-06,
+      "loss": 1.3258,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9350331819769476e-06,
+      "loss": 1.2289,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.942018861334265e-06,
+      "loss": 1.3994,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9490045406915823e-06,
+      "loss": 1.3537,
+      "step": 279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9559902200489e-06,
+      "loss": 1.3002,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.962975899406217e-06,
+      "loss": 1.5378,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.969961578763535e-06,
+      "loss": 1.1495,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769472581208523e-06,
+      "loss": 1.3313,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9839329374781697e-06,
+      "loss": 1.3397,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909186168354875e-06,
+      "loss": 1.3009,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997904296192805e-06,
+      "loss": 1.3032,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0048899755501223e-06,
+      "loss": 1.2767,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.01187565490744e-06,
+      "loss": 1.3213,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.018861334264757e-06,
+      "loss": 1.1973,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.025847013622075e-06,
+      "loss": 1.3507,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0328326929793923e-06,
+      "loss": 1.2732,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.03981837233671e-06,
+      "loss": 1.291,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0468040516940275e-06,
+      "loss": 1.4363,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.053789731051345e-06,
+      "loss": 1.2825,
+      "step": 294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0607754104086627e-06,
+      "loss": 1.3656,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.06776108976598e-06,
+      "loss": 1.4008,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0747467691232974e-06,
+      "loss": 1.1777,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.081732448480615e-06,
+      "loss": 1.4392,
+      "step": 298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0887181278379322e-06,
+      "loss": 1.3268,
+      "step": 299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.09570380719525e-06,
+      "loss": 1.3801,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1026894865525674e-06,
+      "loss": 1.2682,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.109675165909885e-06,
+      "loss": 1.1743,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1166608452672026e-06,
+      "loss": 1.5576,
+      "step": 303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.12364652462452e-06,
+      "loss": 1.3402,
+      "step": 304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1306322039818374e-06,
+      "loss": 1.341,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1376178833391552e-06,
+      "loss": 1.3446,
+      "step": 306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.144603562696472e-06,
+      "loss": 1.2684,
+      "step": 307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.15158924205379e-06,
+      "loss": 1.107,
+      "step": 308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1585749214111074e-06,
+      "loss": 1.3769,
+      "step": 309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1655606007684248e-06,
+      "loss": 1.3554,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1725462801257426e-06,
+      "loss": 1.3256,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.17953195948306e-06,
+      "loss": 1.2122,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865176388403774e-06,
+      "loss": 1.3682,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.193503318197695e-06,
+      "loss": 1.4437,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.200488997555012e-06,
+      "loss": 1.4511,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.20747467691233e-06,
+      "loss": 1.2353,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2144603562696473e-06,
+      "loss": 1.4098,
+      "step": 317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2214460356269647e-06,
+      "loss": 1.2732,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2284317149842825e-06,
+      "loss": 1.2734,
+      "step": 319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2354173943416e-06,
+      "loss": 1.2168,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2424030736989173e-06,
+      "loss": 1.2488,
+      "step": 321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.249388753056235e-06,
+      "loss": 1.2265,
+      "step": 322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.2563744324135525e-06,
+      "loss": 1.2238,
+      "step": 323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.26336011177087e-06,
+      "loss": 1.0909,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2703457911281873e-06,
+      "loss": 1.292,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2773314704855047e-06,
+      "loss": 1.2028,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2843171498428225e-06,
+      "loss": 1.2571,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.29130282920014e-06,
+      "loss": 1.4452,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2982885085574573e-06,
+      "loss": 1.3969,
+      "step": 329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.305274187914775e-06,
+      "loss": 1.306,
+      "step": 330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3122598672720925e-06,
+      "loss": 1.4062,
+      "step": 331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.31924554662941e-06,
+      "loss": 1.4036,
+      "step": 332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3262312259867273e-06,
+      "loss": 1.1844,
+      "step": 333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3332169053440446e-06,
+      "loss": 1.1258,
+      "step": 334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3402025847013625e-06,
+      "loss": 1.2227,
+      "step": 335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.34718826405868e-06,
+      "loss": 1.4109,
+      "step": 336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3541739434159972e-06,
+      "loss": 1.2339,
+      "step": 337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.361159622773315e-06,
+      "loss": 1.3657,
+      "step": 338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3681453021306324e-06,
+      "loss": 1.2707,
+      "step": 339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.37513098148795e-06,
+      "loss": 1.3647,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.382116660845267e-06,
+      "loss": 1.175,
+      "step": 341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3891023402025846e-06,
+      "loss": 1.54,
+      "step": 342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3960880195599024e-06,
+      "loss": 1.2487,
+      "step": 343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.40307369891722e-06,
+      "loss": 1.1668,
+      "step": 344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4100593782745376e-06,
+      "loss": 1.3026,
+      "step": 345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.417045057631855e-06,
+      "loss": 1.2983,
+      "step": 346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4240307369891724e-06,
+      "loss": 1.2269,
+      "step": 347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.43101641634649e-06,
+      "loss": 1.4308,
+      "step": 348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4380020957038076e-06,
+      "loss": 1.231,
+      "step": 349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.444987775061125e-06,
+      "loss": 1.2368,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4519734544184424e-06,
+      "loss": 1.2974,
+      "step": 351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4589591337757598e-06,
+      "loss": 1.2897,
+      "step": 352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4659448131330776e-06,
+      "loss": 1.2159,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.472930492490395e-06,
+      "loss": 1.3513,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4799161718477123e-06,
+      "loss": 1.4255,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.48690185120503e-06,
+      "loss": 1.2668,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4938875305623475e-06,
+      "loss": 1.3188,
+      "step": 357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.500873209919665e-06,
+      "loss": 1.3351,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5078588892769823e-06,
+      "loss": 1.3464,
+      "step": 359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5148445686342997e-06,
+      "loss": 1.2194,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.5218302479916175e-06,
+      "loss": 1.4928,
+      "step": 361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.528815927348935e-06,
+      "loss": 1.5063,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5358016067062523e-06,
+      "loss": 1.2409,
+      "step": 363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.54278728606357e-06,
+      "loss": 1.2165,
+      "step": 364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5497729654208875e-06,
+      "loss": 1.3867,
+      "step": 365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.556758644778205e-06,
+      "loss": 1.374,
+      "step": 366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5637443241355227e-06,
+      "loss": 1.4215,
+      "step": 367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.57073000349284e-06,
+      "loss": 1.1848,
+      "step": 368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5777156828501575e-06,
+      "loss": 1.2198,
+      "step": 369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5847013622074753e-06,
+      "loss": 1.3343,
+      "step": 370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5916870415647923e-06,
+      "loss": 1.3236,
+      "step": 371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5986727209221096e-06,
+      "loss": 1.1926,
+      "step": 372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.605658400279427e-06,
+      "loss": 1.0712,
+      "step": 373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.612644079636745e-06,
+      "loss": 1.3582,
+      "step": 374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6196297589940622e-06,
+      "loss": 1.2098,
+      "step": 375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6266154383513796e-06,
+      "loss": 1.2608,
+      "step": 376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6336011177086974e-06,
+      "loss": 1.2425,
+      "step": 377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.640586797066015e-06,
+      "loss": 1.2339,
+      "step": 378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.647572476423332e-06,
+      "loss": 1.5014,
+      "step": 379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.65455815578065e-06,
+      "loss": 1.4406,
+      "step": 380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6615438351379674e-06,
+      "loss": 1.4684,
+      "step": 381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.668529514495285e-06,
+      "loss": 1.2937,
+      "step": 382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6755151938526026e-06,
+      "loss": 1.2451,
+      "step": 383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.68250087320992e-06,
+      "loss": 1.2639,
+      "step": 384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6894865525672374e-06,
+      "loss": 1.196,
+      "step": 385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.696472231924555e-06,
+      "loss": 1.2429,
+      "step": 386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7034579112818726e-06,
+      "loss": 1.2918,
+      "step": 387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.71044359063919e-06,
+      "loss": 1.4149,
+      "step": 388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.717429269996507e-06,
+      "loss": 1.2046,
+      "step": 389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7244149493538248e-06,
+      "loss": 1.3563,
+      "step": 390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.731400628711142e-06,
+      "loss": 1.2347,
+      "step": 391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7383863080684595e-06,
+      "loss": 1.254,
+      "step": 392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7453719874257773e-06,
+      "loss": 1.283,
+      "step": 393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7523576667830947e-06,
+      "loss": 1.225,
+      "step": 394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.759343346140412e-06,
+      "loss": 1.1809,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.76632902549773e-06,
+      "loss": 1.163,
+      "step": 396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7733147048550473e-06,
+      "loss": 1.1603,
+      "step": 397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.780300384212365e-06,
+      "loss": 1.2677,
+      "step": 398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7872860635696825e-06,
+      "loss": 1.0816,
+      "step": 399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.794271742927e-06,
+      "loss": 1.5494,
+      "step": 400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8012574222843177e-06,
+      "loss": 1.38,
+      "step": 401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.808243101641635e-06,
+      "loss": 1.2651,
+      "step": 402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8152287809989525e-06,
+      "loss": 1.1507,
+      "step": 403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8222144603562703e-06,
+      "loss": 1.2135,
+      "step": 404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8292001397135877e-06,
+      "loss": 1.2346,
+      "step": 405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8361858190709047e-06,
+      "loss": 1.2925,
+      "step": 406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.843171498428222e-06,
+      "loss": 1.1637,
+      "step": 407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.85015717778554e-06,
+      "loss": 1.3602,
+      "step": 408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.2232,
+      "step": 409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8641285365001746e-06,
+      "loss": 1.2772,
+      "step": 410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8711142158574925e-06,
+      "loss": 1.2089,
+      "step": 411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.87809989521481e-06,
+      "loss": 1.314,
+      "step": 412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8850855745721272e-06,
+      "loss": 1.2341,
+      "step": 413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.892071253929445e-06,
+      "loss": 1.196,
+      "step": 414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8990569332867624e-06,
+      "loss": 1.2875,
+      "step": 415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.90604261264408e-06,
+      "loss": 1.3488,
+      "step": 416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9130282920013976e-06,
+      "loss": 1.2751,
+      "step": 417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.920013971358715e-06,
+      "loss": 1.1703,
+      "step": 418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9269996507160324e-06,
+      "loss": 1.3268,
+      "step": 419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9339853300733502e-06,
+      "loss": 1.2537,
+      "step": 420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9409710094306676e-06,
+      "loss": 1.3621,
+      "step": 421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.947956688787985e-06,
+      "loss": 1.2002,
+      "step": 422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.954942368145302e-06,
+      "loss": 1.1934,
+      "step": 423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9619280475026198e-06,
+      "loss": 1.2746,
+      "step": 424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.968913726859937e-06,
+      "loss": 1.0521,
+      "step": 425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9758994062172546e-06,
+      "loss": 1.2748,
+      "step": 426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9828850855745724e-06,
+      "loss": 1.2627,
+      "step": 427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9898707649318898e-06,
+      "loss": 1.1476,
+      "step": 428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.996856444289207e-06,
+      "loss": 1.2516,
+      "step": 429
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.003842123646525e-06,
+      "loss": 1.3148,
+      "step": 430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0108278030038423e-06,
+      "loss": 1.2211,
+      "step": 431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0178134823611597e-06,
+      "loss": 1.2071,
+      "step": 432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0247991617184775e-06,
+      "loss": 1.2921,
+      "step": 433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.031784841075795e-06,
+      "loss": 1.4297,
+      "step": 434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0387705204331123e-06,
+      "loss": 1.324,
+      "step": 435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.04575619979043e-06,
+      "loss": 1.3862,
+      "step": 436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0527418791477475e-06,
+      "loss": 1.4925,
+      "step": 437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.059727558505065e-06,
+      "loss": 1.1665,
+      "step": 438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0667132378623827e-06,
+      "loss": 1.2136,
+      "step": 439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0736989172196997e-06,
+      "loss": 1.3217,
+      "step": 440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.080684596577017e-06,
+      "loss": 1.3639,
+      "step": 441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0876702759343345e-06,
+      "loss": 1.4001,
+      "step": 442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0946559552916523e-06,
+      "loss": 1.2435,
+      "step": 443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1016416346489697e-06,
+      "loss": 1.3056,
+      "step": 444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.108627314006287e-06,
+      "loss": 1.4026,
+      "step": 445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.115612993363605e-06,
+      "loss": 1.3803,
+      "step": 446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1225986727209223e-06,
+      "loss": 1.162,
+      "step": 447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1295843520782396e-06,
+      "loss": 1.2558,
+      "step": 448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1365700314355575e-06,
+      "loss": 1.291,
+      "step": 449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.143555710792875e-06,
+      "loss": 1.4406,
+      "step": 450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1505413901501922e-06,
+      "loss": 1.2193,
+      "step": 451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.15752706950751e-06,
+      "loss": 1.1955,
+      "step": 452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1645127488648274e-06,
+      "loss": 1.2235,
+      "step": 453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1714984282221452e-06,
+      "loss": 1.2063,
+      "step": 454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1784841075794626e-06,
+      "loss": 1.29,
+      "step": 455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.18546978693678e-06,
+      "loss": 1.3176,
+      "step": 456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.192455466294098e-06,
+      "loss": 1.2532,
+      "step": 457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.199441145651415e-06,
+      "loss": 1.1878,
+      "step": 458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.206426825008732e-06,
+      "loss": 1.3781,
+      "step": 459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2134125043660496e-06,
+      "loss": 1.2837,
+      "step": 460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2203981837233674e-06,
+      "loss": 1.2598,
+      "step": 461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2273838630806848e-06,
+      "loss": 1.3522,
+      "step": 462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.234369542438002e-06,
+      "loss": 1.1574,
+      "step": 463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.24135522179532e-06,
+      "loss": 1.257,
+      "step": 464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2483409011526374e-06,
+      "loss": 1.2176,
+      "step": 465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2553265805099548e-06,
+      "loss": 1.2465,
+      "step": 466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2623122598672726e-06,
+      "loss": 1.2284,
+      "step": 467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.26929793922459e-06,
+      "loss": 1.3767,
+      "step": 468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2762836185819073e-06,
+      "loss": 1.168,
+      "step": 469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.283269297939225e-06,
+      "loss": 1.4175,
+      "step": 470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2902549772965425e-06,
+      "loss": 1.1837,
+      "step": 471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.29724065665386e-06,
+      "loss": 1.1059,
+      "step": 472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3042263360111777e-06,
+      "loss": 1.2092,
+      "step": 473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.311212015368495e-06,
+      "loss": 1.3162,
+      "step": 474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.318197694725812e-06,
+      "loss": 1.4319,
+      "step": 475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3251833740831295e-06,
+      "loss": 1.2159,
+      "step": 476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3321690534404473e-06,
+      "loss": 1.3014,
+      "step": 477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3391547327977647e-06,
+      "loss": 1.3485,
+      "step": 478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.346140412155082e-06,
+      "loss": 1.2528,
+      "step": 479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3531260915124e-06,
+      "loss": 1.223,
+      "step": 480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3601117708697173e-06,
+      "loss": 1.1752,
+      "step": 481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3670974502270347e-06,
+      "loss": 1.2226,
+      "step": 482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3740831295843525e-06,
+      "loss": 1.2906,
+      "step": 483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.38106880894167e-06,
+      "loss": 1.2588,
+      "step": 484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3880544882989873e-06,
+      "loss": 1.1977,
+      "step": 485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.395040167656305e-06,
+      "loss": 1.1682,
+      "step": 486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4020258470136225e-06,
+      "loss": 1.2568,
+      "step": 487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.40901152637094e-06,
+      "loss": 1.2153,
+      "step": 488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4159972057282577e-06,
+      "loss": 1.2709,
+      "step": 489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.422982885085575e-06,
+      "loss": 1.2835,
+      "step": 490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4299685644428924e-06,
+      "loss": 1.2627,
+      "step": 491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4369542438002094e-06,
+      "loss": 1.497,
+      "step": 492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4439399231575272e-06,
+      "loss": 1.1623,
+      "step": 493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4509256025148446e-06,
+      "loss": 1.3123,
+      "step": 494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.457911281872162e-06,
+      "loss": 1.2124,
+      "step": 495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.46489696122948e-06,
+      "loss": 1.3035,
+      "step": 496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.471882640586797e-06,
+      "loss": 1.2432,
+      "step": 497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4788683199441146e-06,
+      "loss": 1.2726,
+      "step": 498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4858539993014324e-06,
+      "loss": 1.2151,
+      "step": 499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4928396786587498e-06,
+      "loss": 1.2115,
+      "step": 500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.499825358016067e-06,
+      "loss": 1.1762,
+      "step": 501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.506811037373385e-06,
+      "loss": 1.2209,
+      "step": 502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5137967167307024e-06,
+      "loss": 1.4392,
+      "step": 503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5207823960880198e-06,
+      "loss": 1.1947,
+      "step": 504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5277680754453376e-06,
+      "loss": 1.1519,
+      "step": 505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.534753754802655e-06,
+      "loss": 1.1486,
+      "step": 506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5417394341599728e-06,
+      "loss": 1.2385,
+      "step": 507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.54872511351729e-06,
+      "loss": 1.2886,
+      "step": 508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.555710792874607e-06,
+      "loss": 1.6084,
+      "step": 509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5626964722319245e-06,
+      "loss": 1.1554,
+      "step": 510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5696821515892423e-06,
+      "loss": 1.2259,
+      "step": 511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5766678309465597e-06,
+      "loss": 0.9766,
+      "step": 512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.583653510303877e-06,
+      "loss": 1.6162,
+      "step": 513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.590639189661195e-06,
+      "loss": 1.2443,
+      "step": 514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5976248690185123e-06,
+      "loss": 1.1609,
+      "step": 515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6046105483758297e-06,
+      "loss": 1.206,
+      "step": 516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6115962277331475e-06,
+      "loss": 1.2447,
+      "step": 517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.618581907090465e-06,
+      "loss": 1.2622,
+      "step": 518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6255675864477823e-06,
+      "loss": 1.4215,
+      "step": 519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6325532658051e-06,
+      "loss": 1.2489,
+      "step": 520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6395389451624175e-06,
+      "loss": 1.4436,
+      "step": 521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.646524624519735e-06,
+      "loss": 1.2579,
+      "step": 522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6535103038770527e-06,
+      "loss": 1.3572,
+      "step": 523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.66049598323437e-06,
+      "loss": 1.1069,
+      "step": 524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6674816625916875e-06,
+      "loss": 1.2016,
+      "step": 525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6744673419490053e-06,
+      "loss": 1.1949,
+      "step": 526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6814530213063222e-06,
+      "loss": 1.4201,
+      "step": 527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6884387006636396e-06,
+      "loss": 1.3197,
+      "step": 528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.695424380020957e-06,
+      "loss": 1.0983,
+      "step": 529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.702410059378275e-06,
+      "loss": 1.3323,
+      "step": 530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7093957387355922e-06,
+      "loss": 1.3459,
+      "step": 531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7163814180929096e-06,
+      "loss": 1.2299,
+      "step": 532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7233670974502274e-06,
+      "loss": 1.2551,
+      "step": 533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.730352776807545e-06,
+      "loss": 1.3945,
+      "step": 534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.737338456164862e-06,
+      "loss": 1.0986,
+      "step": 535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.74432413552218e-06,
+      "loss": 1.2036,
+      "step": 536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7513098148794974e-06,
+      "loss": 1.2644,
+      "step": 537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7582954942368148e-06,
+      "loss": 1.2566,
+      "step": 538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7652811735941326e-06,
+      "loss": 1.1791,
+      "step": 539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.77226685295145e-06,
+      "loss": 1.1867,
+      "step": 540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7792525323087674e-06,
+      "loss": 1.1577,
+      "step": 541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.786238211666085e-06,
+      "loss": 1.4074,
+      "step": 542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7932238910234026e-06,
+      "loss": 1.2275,
+      "step": 543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8002095703807195e-06,
+      "loss": 1.1854,
+      "step": 544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.807195249738037e-06,
+      "loss": 1.1697,
+      "step": 545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8141809290953547e-06,
+      "loss": 1.175,
+      "step": 546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.821166608452672e-06,
+      "loss": 1.0617,
+      "step": 547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.82815228780999e-06,
+      "loss": 1.3273,
+      "step": 548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.835137967167307e-06,
+      "loss": 1.2573,
+      "step": 549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.842123646524625e-06,
+      "loss": 1.0765,
+      "step": 550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8491093258819425e-06,
+      "loss": 1.1218,
+      "step": 551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8560950052392595e-06,
+      "loss": 1.2118,
+      "step": 552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.863080684596577e-06,
+      "loss": 1.1864,
+      "step": 553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.870066363953895e-06,
+      "loss": 1.1781,
+      "step": 554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.877052043311212e-06,
+      "loss": 1.2775,
+      "step": 555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.88403772266853e-06,
+      "loss": 1.2078,
+      "step": 556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.891023402025848e-06,
+      "loss": 1.139,
+      "step": 557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.898009081383165e-06,
+      "loss": 1.2855,
+      "step": 558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9049947607404825e-06,
+      "loss": 1.1867,
+      "step": 559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9119804400978e-06,
+      "loss": 1.3904,
+      "step": 560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.918966119455117e-06,
+      "loss": 1.2053,
+      "step": 561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.925951798812434e-06,
+      "loss": 1.3836,
+      "step": 562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.932937478169752e-06,
+      "loss": 1.3129,
+      "step": 563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.93992315752707e-06,
+      "loss": 1.2527,
+      "step": 564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.946908836884387e-06,
+      "loss": 1.2648,
+      "step": 565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.953894516241705e-06,
+      "loss": 1.349,
+      "step": 566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9608801955990224e-06,
+      "loss": 1.4556,
+      "step": 567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.967865874956339e-06,
+      "loss": 1.3431,
+      "step": 568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.974851554313657e-06,
+      "loss": 1.2544,
+      "step": 569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.981837233670975e-06,
+      "loss": 1.3661,
+      "step": 570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.988822913028292e-06,
+      "loss": 1.1458,
+      "step": 571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.99580859238561e-06,
+      "loss": 1.3597,
+      "step": 572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.002794271742928e-06,
+      "loss": 1.22,
+      "step": 573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.009779951100245e-06,
+      "loss": 1.2552,
+      "step": 574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.016765630457562e-06,
+      "loss": 1.2191,
+      "step": 575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.02375130981488e-06,
+      "loss": 1.1788,
+      "step": 576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.030736989172197e-06,
+      "loss": 1.0703,
+      "step": 577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.037722668529514e-06,
+      "loss": 0.9953,
+      "step": 578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.044708347886832e-06,
+      "loss": 1.1021,
+      "step": 579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.05169402724415e-06,
+      "loss": 1.3247,
+      "step": 580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.058679706601467e-06,
+      "loss": 1.2331,
+      "step": 581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.0656653859587845e-06,
+      "loss": 1.323,
+      "step": 582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.072651065316102e-06,
+      "loss": 1.1804,
+      "step": 583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.07963674467342e-06,
+      "loss": 1.1965,
+      "step": 584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.086622424030737e-06,
+      "loss": 1.0263,
+      "step": 585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.093608103388055e-06,
+      "loss": 1.3432,
+      "step": 586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.100593782745373e-06,
+      "loss": 1.2635,
+      "step": 587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.10757946210269e-06,
+      "loss": 1.1327,
+      "step": 588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.1145651414600075e-06,
+      "loss": 1.1782,
+      "step": 589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.121550820817325e-06,
+      "loss": 1.1654,
+      "step": 590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.128536500174642e-06,
+      "loss": 1.2928,
+      "step": 591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.13552217953196e-06,
+      "loss": 1.1544,
+      "step": 592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.142507858889278e-06,
+      "loss": 1.3031,
+      "step": 593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.149493538246595e-06,
+      "loss": 1.2857,
+      "step": 594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.156479217603913e-06,
+      "loss": 1.2265,
+      "step": 595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.16346489696123e-06,
+      "loss": 1.2309,
+      "step": 596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1704505763185475e-06,
+      "loss": 1.3665,
+      "step": 597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1774362556758645e-06,
+      "loss": 1.2652,
+      "step": 598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.184421935033182e-06,
+      "loss": 1.1758,
+      "step": 599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1914076143905e-06,
+      "loss": 1.2317,
+      "step": 600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.198393293747817e-06,
+      "loss": 1.3792,
+      "step": 601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.205378973105135e-06,
+      "loss": 1.2175,
+      "step": 602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.212364652462453e-06,
+      "loss": 1.4268,
+      "step": 603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.21935033181977e-06,
+      "loss": 1.2031,
+      "step": 604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2263360111770874e-06,
+      "loss": 1.2576,
+      "step": 605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.233321690534405e-06,
+      "loss": 1.2396,
+      "step": 606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.240307369891722e-06,
+      "loss": 1.2749,
+      "step": 607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.24729304924904e-06,
+      "loss": 1.2173,
+      "step": 608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.254278728606358e-06,
+      "loss": 1.121,
+      "step": 609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.261264407963675e-06,
+      "loss": 1.1922,
+      "step": 610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.268250087320993e-06,
+      "loss": 1.1941,
+      "step": 611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2752357666783104e-06,
+      "loss": 1.5811,
+      "step": 612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.282221446035627e-06,
+      "loss": 1.2003,
+      "step": 613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.289207125392944e-06,
+      "loss": 1.0548,
+      "step": 614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.296192804750262e-06,
+      "loss": 1.305,
+      "step": 615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.30317848410758e-06,
+      "loss": 1.2374,
+      "step": 616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.310164163464897e-06,
+      "loss": 1.2431,
+      "step": 617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.317149842822215e-06,
+      "loss": 1.2095,
+      "step": 618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.324135522179533e-06,
+      "loss": 1.2794,
+      "step": 619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3311212015368495e-06,
+      "loss": 1.1643,
+      "step": 620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.338106880894167e-06,
+      "loss": 1.1302,
+      "step": 621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.345092560251485e-06,
+      "loss": 1.3542,
+      "step": 622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.352078239608802e-06,
+      "loss": 1.3347,
+      "step": 623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.35906391896612e-06,
+      "loss": 1.3979,
+      "step": 624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.366049598323438e-06,
+      "loss": 1.32,
+      "step": 625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.373035277680755e-06,
+      "loss": 1.0995,
+      "step": 626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3800209570380725e-06,
+      "loss": 1.2139,
+      "step": 627
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.38700663639539e-06,
+      "loss": 1.0287,
+      "step": 628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.393992315752707e-06,
+      "loss": 1.177,
+      "step": 629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.400977995110024e-06,
+      "loss": 1.2411,
+      "step": 630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.407963674467342e-06,
+      "loss": 1.1907,
+      "step": 631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.41494935382466e-06,
+      "loss": 1.2686,
+      "step": 632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.421935033181977e-06,
+      "loss": 1.4292,
+      "step": 633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.428920712539295e-06,
+      "loss": 1.2346,
+      "step": 634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4359063918966125e-06,
+      "loss": 1.1944,
+      "step": 635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4428920712539295e-06,
+      "loss": 1.2674,
+      "step": 636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.449877750611247e-06,
+      "loss": 1.0882,
+      "step": 637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.456863429968565e-06,
+      "loss": 1.2107,
+      "step": 638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.463849109325882e-06,
+      "loss": 1.3579,
+      "step": 639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4708347886832e-06,
+      "loss": 1.2806,
+      "step": 640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.477820468040518e-06,
+      "loss": 1.288,
+      "step": 641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.484806147397835e-06,
+      "loss": 0.945,
+      "step": 642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4917918267551524e-06,
+      "loss": 1.133,
+      "step": 643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.49877750611247e-06,
+      "loss": 1.3206,
+      "step": 644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.505763185469787e-06,
+      "loss": 1.2048,
+      "step": 645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.512748864827105e-06,
+      "loss": 1.1437,
+      "step": 646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.519734544184422e-06,
+      "loss": 1.1796,
+      "step": 647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.52672022354174e-06,
+      "loss": 1.2763,
+      "step": 648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.533705902899057e-06,
+      "loss": 1.1633,
+      "step": 649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.540691582256375e-06,
+      "loss": 1.2775,
+      "step": 650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.547677261613692e-06,
+      "loss": 1.0949,
+      "step": 651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.554662940971009e-06,
+      "loss": 1.5183,
+      "step": 652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.561648620328327e-06,
+      "loss": 1.1068,
+      "step": 653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.568634299685645e-06,
+      "loss": 1.3135,
+      "step": 654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.575619979042962e-06,
+      "loss": 1.1423,
+      "step": 655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.58260565840028e-06,
+      "loss": 1.2507,
+      "step": 656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.589591337757598e-06,
+      "loss": 1.1952,
+      "step": 657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5965770171149145e-06,
+      "loss": 1.1532,
+      "step": 658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.603562696472232e-06,
+      "loss": 1.2089,
+      "step": 659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.61054837582955e-06,
+      "loss": 1.0891,
+      "step": 660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.617534055186867e-06,
+      "loss": 1.271,
+      "step": 661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.624519734544185e-06,
+      "loss": 1.1797,
+      "step": 662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.631505413901503e-06,
+      "loss": 1.1186,
+      "step": 663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.63849109325882e-06,
+      "loss": 1.2259,
+      "step": 664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.645476772616137e-06,
+      "loss": 1.289,
+      "step": 665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6524624519734545e-06,
+      "loss": 1.0929,
+      "step": 666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.659448131330772e-06,
+      "loss": 1.409,
+      "step": 667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.666433810688089e-06,
+      "loss": 1.0597,
+      "step": 668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.673419490045407e-06,
+      "loss": 1.3641,
+      "step": 669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.680405169402725e-06,
+      "loss": 1.0831,
+      "step": 670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.687390848760042e-06,
+      "loss": 1.4106,
+      "step": 671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.69437652811736e-06,
+      "loss": 1.2839,
+      "step": 672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7013622074746775e-06,
+      "loss": 1.2671,
+      "step": 673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7083478868319945e-06,
+      "loss": 1.2784,
+      "step": 674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715333566189312e-06,
+      "loss": 1.1424,
+      "step": 675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.72231924554663e-06,
+      "loss": 1.3538,
+      "step": 676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.729304924903947e-06,
+      "loss": 1.1437,
+      "step": 677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.736290604261265e-06,
+      "loss": 1.2437,
+      "step": 678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.743276283618583e-06,
+      "loss": 1.0785,
+      "step": 679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7502619629759e-06,
+      "loss": 1.1346,
+      "step": 680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7572476423332174e-06,
+      "loss": 1.187,
+      "step": 681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.764233321690534e-06,
+      "loss": 1.0974,
+      "step": 682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.771219001047852e-06,
+      "loss": 1.2624,
+      "step": 683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.778204680405169e-06,
+      "loss": 1.0777,
+      "step": 684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.785190359762487e-06,
+      "loss": 1.1281,
+      "step": 685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.792176039119805e-06,
+      "loss": 1.1031,
+      "step": 686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.799161718477122e-06,
+      "loss": 1.3027,
+      "step": 687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.80614739783444e-06,
+      "loss": 1.1678,
+      "step": 688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.813133077191757e-06,
+      "loss": 1.111,
+      "step": 689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.820118756549075e-06,
+      "loss": 1.1402,
+      "step": 690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.827104435906392e-06,
+      "loss": 1.2071,
+      "step": 691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.83409011526371e-06,
+      "loss": 1.3287,
+      "step": 692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.841075794621028e-06,
+      "loss": 1.2684,
+      "step": 693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.848061473978345e-06,
+      "loss": 1.4211,
+      "step": 694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.855047153335663e-06,
+      "loss": 1.2781,
+      "step": 695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.86203283269298e-06,
+      "loss": 1.1884,
+      "step": 696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.869018512050297e-06,
+      "loss": 1.2678,
+      "step": 697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.876004191407615e-06,
+      "loss": 1.3356,
+      "step": 698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.882989870764932e-06,
+      "loss": 0.9879,
+      "step": 699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.88997555012225e-06,
+      "loss": 1.222,
+      "step": 700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.896961229479567e-06,
+      "loss": 1.1416,
+      "step": 701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.903946908836885e-06,
+      "loss": 1.2469,
+      "step": 702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9109325881942025e-06,
+      "loss": 1.1206,
+      "step": 703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9179182675515195e-06,
+      "loss": 0.9749,
+      "step": 704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.924903946908837e-06,
+      "loss": 1.1637,
+      "step": 705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.931889626266155e-06,
+      "loss": 1.0508,
+      "step": 706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.938875305623472e-06,
+      "loss": 1.3032,
+      "step": 707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.94586098498079e-06,
+      "loss": 1.0385,
+      "step": 708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.952846664338108e-06,
+      "loss": 1.1504,
+      "step": 709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.959832343695425e-06,
+      "loss": 1.2026,
+      "step": 710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9668180230527425e-06,
+      "loss": 1.2101,
+      "step": 711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.97380370241006e-06,
+      "loss": 1.3972,
+      "step": 712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.980789381767377e-06,
+      "loss": 1.0323,
+      "step": 713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.987775061124695e-06,
+      "loss": 1.0506,
+      "step": 714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.994760740482013e-06,
+      "loss": 1.1856,
+      "step": 715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.00174641983933e-06,
+      "loss": 1.2089,
+      "step": 716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.008732099196647e-06,
+      "loss": 1.1084,
+      "step": 717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.015717778553965e-06,
+      "loss": 1.3077,
+      "step": 718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.0227034579112825e-06,
+      "loss": 1.403,
+      "step": 719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.029689137268599e-06,
+      "loss": 1.2933,
+      "step": 720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.036674816625917e-06,
+      "loss": 0.9993,
+      "step": 721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.043660495983235e-06,
+      "loss": 1.1422,
+      "step": 722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.050646175340552e-06,
+      "loss": 1.2196,
+      "step": 723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.05763185469787e-06,
+      "loss": 1.166,
+      "step": 724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.064617534055188e-06,
+      "loss": 1.4361,
+      "step": 725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.071603213412505e-06,
+      "loss": 1.2541,
+      "step": 726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.078588892769822e-06,
+      "loss": 1.2617,
+      "step": 727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.08557457212714e-06,
+      "loss": 0.9318,
+      "step": 728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.092560251484457e-06,
+      "loss": 1.2386,
+      "step": 729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.099545930841775e-06,
+      "loss": 1.1946,
+      "step": 730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.106531610199093e-06,
+      "loss": 1.3171,
+      "step": 731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.11351728955641e-06,
+      "loss": 1.2193,
+      "step": 732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.120502968913728e-06,
+      "loss": 1.2257,
+      "step": 733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.127488648271045e-06,
+      "loss": 1.1721,
+      "step": 734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.134474327628362e-06,
+      "loss": 1.1262,
+      "step": 735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.14146000698568e-06,
+      "loss": 1.2794,
+      "step": 736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.148445686342998e-06,
+      "loss": 1.1304,
+      "step": 737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.155431365700315e-06,
+      "loss": 1.2097,
+      "step": 738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.162417045057633e-06,
+      "loss": 1.3242,
+      "step": 739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.169402724414951e-06,
+      "loss": 1.1717,
+      "step": 740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1763884037722675e-06,
+      "loss": 1.0866,
+      "step": 741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1833740831295845e-06,
+      "loss": 1.4706,
+      "step": 742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1903597624869015e-06,
+      "loss": 1.2836,
+      "step": 743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.197345441844219e-06,
+      "loss": 1.1805,
+      "step": 744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.204331121201537e-06,
+      "loss": 1.005,
+      "step": 745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.211316800558854e-06,
+      "loss": 1.0867,
+      "step": 746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.218302479916172e-06,
+      "loss": 1.3015,
+      "step": 747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.22528815927349e-06,
+      "loss": 1.1342,
+      "step": 748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.232273838630807e-06,
+      "loss": 1.1384,
+      "step": 749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.2392595179881245e-06,
+      "loss": 1.5399,
+      "step": 750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.246245197345442e-06,
+      "loss": 1.2507,
+      "step": 751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.253230876702759e-06,
+      "loss": 1.1714,
+      "step": 752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.260216556060077e-06,
+      "loss": 1.1798,
+      "step": 753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.267202235417395e-06,
+      "loss": 1.3546,
+      "step": 754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.274187914774712e-06,
+      "loss": 1.3872,
+      "step": 755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.28117359413203e-06,
+      "loss": 1.189,
+      "step": 756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.2881592734893475e-06,
+      "loss": 1.2183,
+      "step": 757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.295144952846664e-06,
+      "loss": 1.2033,
+      "step": 758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.302130632203982e-06,
+      "loss": 1.251,
+      "step": 759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.3091163115613e-06,
+      "loss": 1.2819,
+      "step": 760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.316101990918617e-06,
+      "loss": 1.1765,
+      "step": 761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.323087670275935e-06,
+      "loss": 1.1932,
+      "step": 762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.330073349633253e-06,
+      "loss": 1.149,
+      "step": 763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.33705902899057e-06,
+      "loss": 1.378,
+      "step": 764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.344044708347887e-06,
+      "loss": 1.0783,
+      "step": 765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.351030387705205e-06,
+      "loss": 1.0535,
+      "step": 766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.358016067062522e-06,
+      "loss": 1.3241,
+      "step": 767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.36500174641984e-06,
+      "loss": 1.3772,
+      "step": 768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.371987425777158e-06,
+      "loss": 1.1262,
+      "step": 769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.378973105134475e-06,
+      "loss": 1.2204,
+      "step": 770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.385958784491793e-06,
+      "loss": 1.0666,
+      "step": 771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.39294446384911e-06,
+      "loss": 1.2636,
+      "step": 772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.399930143206427e-06,
+      "loss": 1.2552,
+      "step": 773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.406915822563745e-06,
+      "loss": 1.1373,
+      "step": 774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.413901501921063e-06,
+      "loss": 1.226,
+      "step": 775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.42088718127838e-06,
+      "loss": 1.1935,
+      "step": 776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.427872860635697e-06,
+      "loss": 1.1298,
+      "step": 777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.434858539993014e-06,
+      "loss": 1.2157,
+      "step": 778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.441844219350332e-06,
+      "loss": 1.1801,
+      "step": 779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4488298987076495e-06,
+      "loss": 1.2301,
+      "step": 780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4558155780649665e-06,
+      "loss": 1.2856,
+      "step": 781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.462801257422284e-06,
+      "loss": 1.138,
+      "step": 782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.469786936779602e-06,
+      "loss": 1.1586,
+      "step": 783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.476772616136919e-06,
+      "loss": 1.2551,
+      "step": 784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.483758295494237e-06,
+      "loss": 1.0712,
+      "step": 785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.490743974851555e-06,
+      "loss": 1.2839,
+      "step": 786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.497729654208872e-06,
+      "loss": 1.2142,
+      "step": 787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.5047153335661895e-06,
+      "loss": 0.9646,
+      "step": 788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.511701012923507e-06,
+      "loss": 1.0963,
+      "step": 789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.518686692280824e-06,
+      "loss": 1.2565,
+      "step": 790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.525672371638142e-06,
+      "loss": 1.2516,
+      "step": 791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.53265805099546e-06,
+      "loss": 1.0565,
+      "step": 792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.539643730352777e-06,
+      "loss": 1.0419,
+      "step": 793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.546629409710095e-06,
+      "loss": 1.2846,
+      "step": 794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.5536150890674125e-06,
+      "loss": 1.0049,
+      "step": 795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.56060076842473e-06,
+      "loss": 1.1608,
+      "step": 796
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.567586447782047e-06,
+      "loss": 1.0424,
+      "step": 797
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.574572127139365e-06,
+      "loss": 1.3698,
+      "step": 798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.581557806496683e-06,
+      "loss": 1.1117,
+      "step": 799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.588543485854e-06,
+      "loss": 1.4456,
+      "step": 800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.595529165211318e-06,
+      "loss": 1.1183,
+      "step": 801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.6025148445686354e-06,
+      "loss": 1.1519,
+      "step": 802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.609500523925952e-06,
+      "loss": 1.1653,
+      "step": 803
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.61648620328327e-06,
+      "loss": 1.4462,
+      "step": 804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.623471882640588e-06,
+      "loss": 1.2738,
+      "step": 805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.630457561997905e-06,
+      "loss": 1.0793,
+      "step": 806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.637443241355223e-06,
+      "loss": 1.2195,
+      "step": 807
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.644428920712541e-06,
+      "loss": 1.2201,
+      "step": 808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.651414600069858e-06,
+      "loss": 1.1214,
+      "step": 809
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.658400279427175e-06,
+      "loss": 1.0654,
+      "step": 810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.6653859587844915e-06,
+      "loss": 1.1208,
+      "step": 811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.672371638141809e-06,
+      "loss": 1.3212,
+      "step": 812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.679357317499127e-06,
+      "loss": 1.1547,
+      "step": 813
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.686342996856444e-06,
+      "loss": 0.8745,
+      "step": 814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.693328676213762e-06,
+      "loss": 1.187,
+      "step": 815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.70031435557108e-06,
+      "loss": 1.2274,
+      "step": 816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.707300034928397e-06,
+      "loss": 1.0349,
+      "step": 817
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.1544,
+      "step": 818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.721271393643032e-06,
+      "loss": 1.1633,
+      "step": 819
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.728257073000349e-06,
+      "loss": 1.1157,
+      "step": 820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.735242752357667e-06,
+      "loss": 1.2058,
+      "step": 821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.742228431714985e-06,
+      "loss": 1.0577,
+      "step": 822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.749214111072302e-06,
+      "loss": 1.1497,
+      "step": 823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.75619979042962e-06,
+      "loss": 1.213,
+      "step": 824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.7631854697869375e-06,
+      "loss": 1.3249,
+      "step": 825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.7701711491442545e-06,
+      "loss": 1.1116,
+      "step": 826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.777156828501572e-06,
+      "loss": 1.1107,
+      "step": 827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.78414250785889e-06,
+      "loss": 1.352,
+      "step": 828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.791128187216207e-06,
+      "loss": 1.1543,
+      "step": 829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.798113866573525e-06,
+      "loss": 1.2153,
+      "step": 830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.805099545930843e-06,
+      "loss": 1.0923,
+      "step": 831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.81208522528816e-06,
+      "loss": 1.2996,
+      "step": 832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8190709046454775e-06,
+      "loss": 1.1103,
+      "step": 833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.826056584002795e-06,
+      "loss": 1.1818,
+      "step": 834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.833042263360112e-06,
+      "loss": 1.1551,
+      "step": 835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.84002794271743e-06,
+      "loss": 1.2193,
+      "step": 836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.847013622074748e-06,
+      "loss": 1.0822,
+      "step": 837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.853999301432065e-06,
+      "loss": 1.1538,
+      "step": 838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.860984980789383e-06,
+      "loss": 1.2204,
+      "step": 839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8679706601467004e-06,
+      "loss": 1.2482,
+      "step": 840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.874956339504017e-06,
+      "loss": 1.2514,
+      "step": 841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.881942018861335e-06,
+      "loss": 1.0991,
+      "step": 842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.888927698218653e-06,
+      "loss": 0.9813,
+      "step": 843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.89591337757597e-06,
+      "loss": 1.3975,
+      "step": 844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.902899056933288e-06,
+      "loss": 1.5285,
+      "step": 845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.909884736290604e-06,
+      "loss": 1.2435,
+      "step": 846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.916870415647922e-06,
+      "loss": 1.0332,
+      "step": 847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9238560950052396e-06,
+      "loss": 1.1475,
+      "step": 848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9308417743625565e-06,
+      "loss": 1.202,
+      "step": 849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.937827453719874e-06,
+      "loss": 1.104,
+      "step": 850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.944813133077192e-06,
+      "loss": 1.0098,
+      "step": 851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.951798812434509e-06,
+      "loss": 1.1578,
+      "step": 852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958784491791827e-06,
+      "loss": 1.0925,
+      "step": 853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.965770171149145e-06,
+      "loss": 1.0446,
+      "step": 854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.972755850506462e-06,
+      "loss": 1.3806,
+      "step": 855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9797415298637795e-06,
+      "loss": 1.1031,
+      "step": 856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.986727209221097e-06,
+      "loss": 1.1692,
+      "step": 857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.993712888578414e-06,
+      "loss": 1.0681,
+      "step": 858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.000698567935732e-06,
+      "loss": 1.0006,
+      "step": 859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.00768424729305e-06,
+      "loss": 1.2067,
+      "step": 860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.014669926650367e-06,
+      "loss": 1.1632,
+      "step": 861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.021655606007685e-06,
+      "loss": 1.331,
+      "step": 862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.0286412853650025e-06,
+      "loss": 0.9699,
+      "step": 863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.0356269647223195e-06,
+      "loss": 1.1125,
+      "step": 864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.042612644079637e-06,
+      "loss": 1.1074,
+      "step": 865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.049598323436955e-06,
+      "loss": 1.1325,
+      "step": 866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.056584002794272e-06,
+      "loss": 1.1433,
+      "step": 867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.06356968215159e-06,
+      "loss": 1.045,
+      "step": 868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.070555361508908e-06,
+      "loss": 1.1362,
+      "step": 869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.077541040866225e-06,
+      "loss": 1.2943,
+      "step": 870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.0845267202235425e-06,
+      "loss": 1.1792,
+      "step": 871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.09151239958086e-06,
+      "loss": 1.1376,
+      "step": 872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.098498078938177e-06,
+      "loss": 1.1155,
+      "step": 873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.105483758295495e-06,
+      "loss": 1.2495,
+      "step": 874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.112469437652813e-06,
+      "loss": 1.1281,
+      "step": 875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.11945511701013e-06,
+      "loss": 1.1998,
+      "step": 876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.126440796367448e-06,
+      "loss": 1.419,
+      "step": 877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1334264757247655e-06,
+      "loss": 1.1571,
+      "step": 878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.140412155082082e-06,
+      "loss": 1.052,
+      "step": 879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.147397834439399e-06,
+      "loss": 1.1916,
+      "step": 880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.154383513796716e-06,
+      "loss": 1.165,
+      "step": 881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.161369193154034e-06,
+      "loss": 1.0186,
+      "step": 882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.168354872511352e-06,
+      "loss": 1.2104,
+      "step": 883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.175340551868669e-06,
+      "loss": 1.0695,
+      "step": 884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.182326231225987e-06,
+      "loss": 1.1033,
+      "step": 885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1893119105833046e-06,
+      "loss": 1.2374,
+      "step": 886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1962975899406215e-06,
+      "loss": 1.2813,
+      "step": 887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.203283269297939e-06,
+      "loss": 1.1762,
+      "step": 888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.210268948655257e-06,
+      "loss": 1.0616,
+      "step": 889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.217254628012574e-06,
+      "loss": 1.119,
+      "step": 890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.224240307369892e-06,
+      "loss": 1.2382,
+      "step": 891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.23122598672721e-06,
+      "loss": 1.0184,
+      "step": 892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.238211666084527e-06,
+      "loss": 1.2857,
+      "step": 893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.2451973454418445e-06,
+      "loss": 1.1157,
+      "step": 894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.252183024799162e-06,
+      "loss": 1.2043,
+      "step": 895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.259168704156479e-06,
+      "loss": 1.1229,
+      "step": 896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.266154383513797e-06,
+      "loss": 1.1474,
+      "step": 897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.273140062871115e-06,
+      "loss": 1.1204,
+      "step": 898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.280125742228432e-06,
+      "loss": 0.9603,
+      "step": 899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.28711142158575e-06,
+      "loss": 1.3524,
+      "step": 900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.2940971009430675e-06,
+      "loss": 0.99,
+      "step": 901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.3010827803003845e-06,
+      "loss": 1.2116,
+      "step": 902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.308068459657702e-06,
+      "loss": 1.2969,
+      "step": 903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.31505413901502e-06,
+      "loss": 1.1199,
+      "step": 904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.322039818372338e-06,
+      "loss": 1.0282,
+      "step": 905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.329025497729655e-06,
+      "loss": 1.0333,
+      "step": 906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.336011177086973e-06,
+      "loss": 0.9353,
+      "step": 907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.3429968564442905e-06,
+      "loss": 1.1313,
+      "step": 908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.3499825358016075e-06,
+      "loss": 1.1402,
+      "step": 909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.356968215158925e-06,
+      "loss": 1.3329,
+      "step": 910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.363953894516243e-06,
+      "loss": 1.0399,
+      "step": 911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.37093957387356e-06,
+      "loss": 1.1791,
+      "step": 912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.377925253230878e-06,
+      "loss": 1.1584,
+      "step": 913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.384910932588196e-06,
+      "loss": 1.0398,
+      "step": 914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.391896611945512e-06,
+      "loss": 1.1755,
+      "step": 915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.39888229130283e-06,
+      "loss": 1.1507,
+      "step": 916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.4058679706601466e-06,
+      "loss": 1.097,
+      "step": 917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.412853650017464e-06,
+      "loss": 1.3074,
+      "step": 918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.419839329374782e-06,
+      "loss": 1.1532,
+      "step": 919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.426825008732099e-06,
+      "loss": 1.0429,
+      "step": 920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.433810688089417e-06,
+      "loss": 1.2806,
+      "step": 921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.440796367446735e-06,
+      "loss": 1.264,
+      "step": 922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.447782046804052e-06,
+      "loss": 1.0385,
+      "step": 923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.4547677261613696e-06,
+      "loss": 1.3229,
+      "step": 924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.461753405518687e-06,
+      "loss": 1.1883,
+      "step": 925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.468739084876004e-06,
+      "loss": 0.9235,
+      "step": 926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.475724764233322e-06,
+      "loss": 0.9917,
+      "step": 927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.48271044359064e-06,
+      "loss": 1.3053,
+      "step": 928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.489696122947957e-06,
+      "loss": 1.4032,
+      "step": 929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.496681802305275e-06,
+      "loss": 1.1723,
+      "step": 930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.5036674816625926e-06,
+      "loss": 1.1063,
+      "step": 931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.5106531610199095e-06,
+      "loss": 1.1866,
+      "step": 932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.517638840377227e-06,
+      "loss": 1.381,
+      "step": 933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.524624519734545e-06,
+      "loss": 1.2354,
+      "step": 934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.531610199091862e-06,
+      "loss": 1.3167,
+      "step": 935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.53859587844918e-06,
+      "loss": 1.1899,
+      "step": 936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.545581557806498e-06,
+      "loss": 1.3399,
+      "step": 937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.552567237163815e-06,
+      "loss": 1.1074,
+      "step": 938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.5595529165211325e-06,
+      "loss": 1.2021,
+      "step": 939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.56653859587845e-06,
+      "loss": 1.0447,
+      "step": 940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.573524275235767e-06,
+      "loss": 1.2719,
+      "step": 941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.580509954593085e-06,
+      "loss": 1.1637,
+      "step": 942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.587495633950403e-06,
+      "loss": 1.0243,
+      "step": 943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.59448131330772e-06,
+      "loss": 1.2944,
+      "step": 944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.601466992665038e-06,
+      "loss": 1.1669,
+      "step": 945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.6084526720223555e-06,
+      "loss": 1.2278,
+      "step": 946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.6154383513796725e-06,
+      "loss": 1.2385,
+      "step": 947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.62242403073699e-06,
+      "loss": 1.0392,
+      "step": 948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.629409710094306e-06,
+      "loss": 1.1452,
+      "step": 949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.636395389451624e-06,
+      "loss": 1.1221,
+      "step": 950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.643381068808942e-06,
+      "loss": 1.1026,
+      "step": 951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.650366748166259e-06,
+      "loss": 0.9779,
+      "step": 952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.657352427523577e-06,
+      "loss": 1.0261,
+      "step": 953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.664338106880895e-06,
+      "loss": 1.1145,
+      "step": 954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.6713237862382116e-06,
+      "loss": 1.0964,
+      "step": 955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.678309465595529e-06,
+      "loss": 1.1691,
+      "step": 956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.685295144952847e-06,
+      "loss": 1.1215,
+      "step": 957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.692280824310164e-06,
+      "loss": 1.2765,
+      "step": 958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.699266503667482e-06,
+      "loss": 1.1446,
+      "step": 959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7062521830248e-06,
+      "loss": 1.0898,
+      "step": 960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.713237862382117e-06,
+      "loss": 1.2727,
+      "step": 961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7202235417394346e-06,
+      "loss": 1.0946,
+      "step": 962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.727209221096752e-06,
+      "loss": 1.0904,
+      "step": 963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.734194900454069e-06,
+      "loss": 1.2476,
+      "step": 964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.741180579811387e-06,
+      "loss": 1.3953,
+      "step": 965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.748166259168705e-06,
+      "loss": 1.3515,
+      "step": 966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.755151938526022e-06,
+      "loss": 1.2337,
+      "step": 967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.76213761788334e-06,
+      "loss": 1.1184,
+      "step": 968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7691232972406576e-06,
+      "loss": 1.0021,
+      "step": 969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.7761089765979745e-06,
+      "loss": 1.1852,
+      "step": 970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.783094655955292e-06,
+      "loss": 1.1164,
+      "step": 971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.79008033531261e-06,
+      "loss": 1.2504,
+      "step": 972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.797066014669927e-06,
+      "loss": 1.1296,
+      "step": 973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.804051694027245e-06,
+      "loss": 1.0935,
+      "step": 974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.811037373384563e-06,
+      "loss": 1.144,
+      "step": 975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.81802305274188e-06,
+      "loss": 1.4347,
+      "step": 976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.8250087320991975e-06,
+      "loss": 1.2506,
+      "step": 977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.831994411456515e-06,
+      "loss": 1.1219,
+      "step": 978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.838980090813832e-06,
+      "loss": 1.1409,
+      "step": 979
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.84596577017115e-06,
+      "loss": 1.0163,
+      "step": 980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.852951449528468e-06,
+      "loss": 0.9943,
+      "step": 981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.859937128885785e-06,
+      "loss": 0.9352,
+      "step": 982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.866922808243103e-06,
+      "loss": 1.1765,
+      "step": 983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.873908487600419e-06,
+      "loss": 0.9656,
+      "step": 984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.880894166957737e-06,
+      "loss": 1.0486,
+      "step": 985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.8878798463150544e-06,
+      "loss": 1.2281,
+      "step": 986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.894865525672371e-06,
+      "loss": 1.1197,
+      "step": 987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.901851205029689e-06,
+      "loss": 1.1611,
+      "step": 988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.908836884387007e-06,
+      "loss": 1.083,
+      "step": 989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.915822563744324e-06,
+      "loss": 0.935,
+      "step": 990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.922808243101642e-06,
+      "loss": 1.2106,
+      "step": 991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.92979392245896e-06,
+      "loss": 1.0436,
+      "step": 992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.9367796018162766e-06,
+      "loss": 1.2052,
+      "step": 993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.943765281173594e-06,
+      "loss": 1.292,
+      "step": 994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.950750960530912e-06,
+      "loss": 1.0803,
+      "step": 995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.957736639888229e-06,
+      "loss": 1.1578,
+      "step": 996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.964722319245547e-06,
+      "loss": 1.1687,
+      "step": 997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.971707998602865e-06,
+      "loss": 1.288,
+      "step": 998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.978693677960182e-06,
+      "loss": 0.9752,
+      "step": 999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.9856793573174996e-06,
+      "loss": 1.5032,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.992665036674817e-06,
+      "loss": 1.3342,
+      "step": 1001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.999650716032134e-06,
+      "loss": 1.1567,
+      "step": 1002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.006636395389452e-06,
+      "loss": 1.1398,
+      "step": 1003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.01362207474677e-06,
+      "loss": 1.128,
+      "step": 1004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.020607754104087e-06,
+      "loss": 1.1144,
+      "step": 1005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.027593433461405e-06,
+      "loss": 1.1478,
+      "step": 1006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.0345791128187226e-06,
+      "loss": 1.059,
+      "step": 1007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.0415647921760395e-06,
+      "loss": 1.1418,
+      "step": 1008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.048550471533357e-06,
+      "loss": 1.0526,
+      "step": 1009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.055536150890675e-06,
+      "loss": 1.288,
+      "step": 1010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.062521830247993e-06,
+      "loss": 1.4917,
+      "step": 1011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.06950750960531e-06,
+      "loss": 1.1351,
+      "step": 1012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.076493188962628e-06,
+      "loss": 1.1997,
+      "step": 1013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.0834788683199455e-06,
+      "loss": 1.1299,
+      "step": 1014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.0904645476772625e-06,
+      "loss": 1.0651,
+      "step": 1015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.09745022703458e-06,
+      "loss": 0.992,
+      "step": 1016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.104435906391898e-06,
+      "loss": 1.008,
+      "step": 1017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.111421585749214e-06,
+      "loss": 1.0187,
+      "step": 1018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.118407265106532e-06,
+      "loss": 1.1841,
+      "step": 1019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.125392944463849e-06,
+      "loss": 1.0931,
+      "step": 1020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.132378623821167e-06,
+      "loss": 1.1382,
+      "step": 1021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.139364303178485e-06,
+      "loss": 1.0903,
+      "step": 1022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.146349982535802e-06,
+      "loss": 1.2279,
+      "step": 1023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.1533356618931194e-06,
+      "loss": 1.0579,
+      "step": 1024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.160321341250437e-06,
+      "loss": 1.0372,
+      "step": 1025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.167307020607754e-06,
+      "loss": 1.1,
+      "step": 1026
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.174292699965072e-06,
+      "loss": 1.1196,
+      "step": 1027
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.18127837932239e-06,
+      "loss": 1.2486,
+      "step": 1028
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.188264058679707e-06,
+      "loss": 1.2372,
+      "step": 1029
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.195249738037025e-06,
+      "loss": 1.0585,
+      "step": 1030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2022354173943424e-06,
+      "loss": 1.0094,
+      "step": 1031
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.209221096751659e-06,
+      "loss": 1.6137,
+      "step": 1032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.216206776108977e-06,
+      "loss": 0.9742,
+      "step": 1033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.223192455466295e-06,
+      "loss": 1.1653,
+      "step": 1034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.230178134823612e-06,
+      "loss": 1.1713,
+      "step": 1035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.23716381418093e-06,
+      "loss": 0.9943,
+      "step": 1036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.244149493538248e-06,
+      "loss": 1.1164,
+      "step": 1037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2511351728955646e-06,
+      "loss": 1.0571,
+      "step": 1038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.258120852252882e-06,
+      "loss": 1.1798,
+      "step": 1039
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2651065316102e-06,
+      "loss": 1.0979,
+      "step": 1040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.272092210967517e-06,
+      "loss": 1.0675,
+      "step": 1041
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.279077890324835e-06,
+      "loss": 1.0953,
+      "step": 1042
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.286063569682153e-06,
+      "loss": 1.0734,
+      "step": 1043
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.29304924903947e-06,
+      "loss": 1.2579,
+      "step": 1044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.3000349283967876e-06,
+      "loss": 1.2512,
+      "step": 1045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.307020607754105e-06,
+      "loss": 1.1508,
+      "step": 1046
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.314006287111422e-06,
+      "loss": 1.0102,
+      "step": 1047
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.32099196646874e-06,
+      "loss": 1.1452,
+      "step": 1048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.327977645826058e-06,
+      "loss": 1.0518,
+      "step": 1049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.334963325183375e-06,
+      "loss": 1.1784,
+      "step": 1050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.341949004540693e-06,
+      "loss": 1.036,
+      "step": 1051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.3489346838980105e-06,
+      "loss": 1.2704,
+      "step": 1052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.355920363255327e-06,
+      "loss": 1.2063,
+      "step": 1053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.3629060426126445e-06,
+      "loss": 1.0722,
+      "step": 1054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.3698917219699614e-06,
+      "loss": 1.1833,
+      "step": 1055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.376877401327279e-06,
+      "loss": 1.2346,
+      "step": 1056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.383863080684597e-06,
+      "loss": 1.0261,
+      "step": 1057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.390848760041914e-06,
+      "loss": 1.176,
+      "step": 1058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.397834439399232e-06,
+      "loss": 1.0949,
+      "step": 1059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.40482011875655e-06,
+      "loss": 1.0962,
+      "step": 1060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.411805798113867e-06,
+      "loss": 1.1554,
+      "step": 1061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.4187914774711844e-06,
+      "loss": 1.0744,
+      "step": 1062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.425777156828502e-06,
+      "loss": 1.075,
+      "step": 1063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.432762836185819e-06,
+      "loss": 1.231,
+      "step": 1064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.439748515543137e-06,
+      "loss": 1.0542,
+      "step": 1065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.446734194900455e-06,
+      "loss": 1.2992,
+      "step": 1066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.453719874257772e-06,
+      "loss": 0.9897,
+      "step": 1067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.46070555361509e-06,
+      "loss": 1.0643,
+      "step": 1068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.4676912329724074e-06,
+      "loss": 1.2612,
+      "step": 1069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.474676912329724e-06,
+      "loss": 1.1967,
+      "step": 1070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.481662591687042e-06,
+      "loss": 1.1616,
+      "step": 1071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.48864827104436e-06,
+      "loss": 1.1507,
+      "step": 1072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.495633950401677e-06,
+      "loss": 1.1456,
+      "step": 1073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.502619629758995e-06,
+      "loss": 1.1746,
+      "step": 1074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.509605309116313e-06,
+      "loss": 1.3332,
+      "step": 1075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.5165909884736296e-06,
+      "loss": 1.2264,
+      "step": 1076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.523576667830947e-06,
+      "loss": 1.0202,
+      "step": 1077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.530562347188265e-06,
+      "loss": 1.2055,
+      "step": 1078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.537548026545582e-06,
+      "loss": 0.9621,
+      "step": 1079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.5445337059029e-06,
+      "loss": 1.1232,
+      "step": 1080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.551519385260218e-06,
+      "loss": 1.018,
+      "step": 1081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.558505064617535e-06,
+      "loss": 1.1228,
+      "step": 1082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.5654907439748526e-06,
+      "loss": 1.1268,
+      "step": 1083
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.57247642333217e-06,
+      "loss": 0.9849,
+      "step": 1084
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.579462102689487e-06,
+      "loss": 1.0941,
+      "step": 1085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.586447782046805e-06,
+      "loss": 1.1823,
+      "step": 1086
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.593433461404121e-06,
+      "loss": 1.0545,
+      "step": 1087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.600419140761439e-06,
+      "loss": 0.9786,
+      "step": 1088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.607404820118757e-06,
+      "loss": 1.0185,
+      "step": 1089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.614390499476074e-06,
+      "loss": 1.2097,
+      "step": 1090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.621376178833392e-06,
+      "loss": 1.1979,
+      "step": 1091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.6283618581907095e-06,
+      "loss": 1.0641,
+      "step": 1092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.635347537548026e-06,
+      "loss": 1.1115,
+      "step": 1093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.642333216905344e-06,
+      "loss": 1.032,
+      "step": 1094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.649318896262662e-06,
+      "loss": 1.1019,
+      "step": 1095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.65630457561998e-06,
+      "loss": 1.0662,
+      "step": 1096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.663290254977298e-06,
+      "loss": 1.0884,
+      "step": 1097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.670275934334614e-06,
+      "loss": 1.0933,
+      "step": 1098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.677261613691932e-06,
+      "loss": 1.2929,
+      "step": 1099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.68424729304925e-06,
+      "loss": 1.0885,
+      "step": 1100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.691232972406567e-06,
+      "loss": 1.0783,
+      "step": 1101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.698218651763885e-06,
+      "loss": 1.0992,
+      "step": 1102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.705204331121203e-06,
+      "loss": 1.1385,
+      "step": 1103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.712190010478519e-06,
+      "loss": 0.98,
+      "step": 1104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.719175689835837e-06,
+      "loss": 1.2732,
+      "step": 1105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.726161369193155e-06,
+      "loss": 1.1467,
+      "step": 1106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.733147048550472e-06,
+      "loss": 1.197,
+      "step": 1107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.74013272790779e-06,
+      "loss": 1.0204,
+      "step": 1108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.747118407265108e-06,
+      "loss": 1.0919,
+      "step": 1109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.754104086622424e-06,
+      "loss": 1.1739,
+      "step": 1110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.761089765979742e-06,
+      "loss": 1.0658,
+      "step": 1111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.76807544533706e-06,
+      "loss": 0.9233,
+      "step": 1112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.775061124694378e-06,
+      "loss": 0.9483,
+      "step": 1113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.782046804051695e-06,
+      "loss": 1.0076,
+      "step": 1114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.789032483409013e-06,
+      "loss": 1.1336,
+      "step": 1115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.79601816276633e-06,
+      "loss": 1.0262,
+      "step": 1116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.803003842123647e-06,
+      "loss": 1.1436,
+      "step": 1117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.809989521480965e-06,
+      "loss": 1.1792,
+      "step": 1118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.816975200838283e-06,
+      "loss": 1.1331,
+      "step": 1119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.8239608801956e-06,
+      "loss": 1.1212,
+      "step": 1120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.830946559552918e-06,
+      "loss": 1.1186,
+      "step": 1121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.837932238910235e-06,
+      "loss": 0.9593,
+      "step": 1122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.844917918267552e-06,
+      "loss": 1.3248,
+      "step": 1123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.851903597624868e-06,
+      "loss": 1.3368,
+      "step": 1124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.858889276982186e-06,
+      "loss": 1.0949,
+      "step": 1125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.865874956339504e-06,
+      "loss": 1.0494,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.872860635696822e-06,
+      "loss": 1.1705,
+      "step": 1127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.87984631505414e-06,
+      "loss": 1.1463,
+      "step": 1128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.886831994411458e-06,
+      "loss": 1.1179,
+      "step": 1129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.893817673768774e-06,
+      "loss": 1.0927,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.900803353126091e-06,
+      "loss": 1.0583,
+      "step": 1131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.90778903248341e-06,
+      "loss": 1.0555,
+      "step": 1132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.914774711840727e-06,
+      "loss": 1.2337,
+      "step": 1133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.921760391198045e-06,
+      "loss": 1.1616,
+      "step": 1134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.928746070555363e-06,
+      "loss": 1.1895,
+      "step": 1135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.935731749912679e-06,
+      "loss": 1.1674,
+      "step": 1136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.942717429269997e-06,
+      "loss": 1.1334,
+      "step": 1137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.949703108627314e-06,
+      "loss": 1.0459,
+      "step": 1138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.956688787984632e-06,
+      "loss": 1.2724,
+      "step": 1139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.96367446734195e-06,
+      "loss": 1.2724,
+      "step": 1140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.970660146699268e-06,
+      "loss": 1.1618,
+      "step": 1141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.977645826056584e-06,
+      "loss": 1.0233,
+      "step": 1142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.984631505413902e-06,
+      "loss": 0.9336,
+      "step": 1143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.99161718477122e-06,
+      "loss": 1.2032,
+      "step": 1144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.998602864128537e-06,
+      "loss": 1.0465,
+      "step": 1145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.005588543485855e-06,
+      "loss": 1.1631,
+      "step": 1146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.012574222843173e-06,
+      "loss": 1.01,
+      "step": 1147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.01955990220049e-06,
+      "loss": 1.2605,
+      "step": 1148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.026545581557807e-06,
+      "loss": 1.1456,
+      "step": 1149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.033531260915125e-06,
+      "loss": 1.2014,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.040516940272443e-06,
+      "loss": 1.0737,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.04750261962976e-06,
+      "loss": 1.0272,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.054488298987078e-06,
+      "loss": 1.0898,
+      "step": 1153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.061473978344394e-06,
+      "loss": 1.0582,
+      "step": 1154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.068459657701712e-06,
+      "loss": 1.0905,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.075445337059028e-06,
+      "loss": 1.0317,
+      "step": 1156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.082431016416346e-06,
+      "loss": 1.087,
+      "step": 1157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.089416695773664e-06,
+      "loss": 1.0749,
+      "step": 1158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.096402375130982e-06,
+      "loss": 1.1155,
+      "step": 1159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.1033880544883e-06,
+      "loss": 1.0622,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.110373733845617e-06,
+      "loss": 1.2093,
+      "step": 1161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.117359413202933e-06,
+      "loss": 1.0226,
+      "step": 1162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.124345092560251e-06,
+      "loss": 0.9815,
+      "step": 1163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.131330771917569e-06,
+      "loss": 1.2893,
+      "step": 1164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138316451274887e-06,
+      "loss": 1.0881,
+      "step": 1165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.145302130632205e-06,
+      "loss": 0.9864,
+      "step": 1166
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.152287809989523e-06,
+      "loss": 1.1257,
+      "step": 1167
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.15927348934684e-06,
+      "loss": 1.2403,
+      "step": 1168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.166259168704156e-06,
+      "loss": 0.9657,
+      "step": 1169
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.173244848061474e-06,
+      "loss": 1.1294,
+      "step": 1170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.180230527418792e-06,
+      "loss": 0.9674,
+      "step": 1171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.18721620677611e-06,
+      "loss": 1.0488,
+      "step": 1172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.194201886133428e-06,
+      "loss": 1.238,
+      "step": 1173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.201187565490746e-06,
+      "loss": 1.1877,
+      "step": 1174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.208173244848062e-06,
+      "loss": 1.1652,
+      "step": 1175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.21515892420538e-06,
+      "loss": 1.232,
+      "step": 1176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.222144603562697e-06,
+      "loss": 0.9844,
+      "step": 1177
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.229130282920015e-06,
+      "loss": 1.2389,
+      "step": 1178
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.236115962277333e-06,
+      "loss": 1.1745,
+      "step": 1179
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.24310164163465e-06,
+      "loss": 1.2209,
+      "step": 1180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.250087320991967e-06,
+      "loss": 1.0518,
+      "step": 1181
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.257073000349285e-06,
+      "loss": 1.1275,
+      "step": 1182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.264058679706602e-06,
+      "loss": 1.0481,
+      "step": 1183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.27104435906392e-06,
+      "loss": 1.1519,
+      "step": 1184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.278030038421238e-06,
+      "loss": 1.0291,
+      "step": 1185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.285015717778556e-06,
+      "loss": 1.0545,
+      "step": 1186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.292001397135872e-06,
+      "loss": 1.0372,
+      "step": 1187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.29898707649319e-06,
+      "loss": 0.9726,
+      "step": 1188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.305972755850508e-06,
+      "loss": 1.0872,
+      "step": 1189
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.312958435207825e-06,
+      "loss": 0.9531,
+      "step": 1190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.319944114565142e-06,
+      "loss": 1.014,
+      "step": 1191
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.32692979392246e-06,
+      "loss": 1.0835,
+      "step": 1192
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.333915473279777e-06,
+      "loss": 1.1675,
+      "step": 1193
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.340901152637095e-06,
+      "loss": 0.8927,
+      "step": 1194
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.347886831994411e-06,
+      "loss": 1.1054,
+      "step": 1195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.354872511351729e-06,
+      "loss": 1.1131,
+      "step": 1196
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.361858190709047e-06,
+      "loss": 1.1176,
+      "step": 1197
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.368843870066365e-06,
+      "loss": 1.0598,
+      "step": 1198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.375829549423682e-06,
+      "loss": 1.1726,
+      "step": 1199
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.382815228781e-06,
+      "loss": 1.0077,
+      "step": 1200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.389800908138316e-06,
+      "loss": 1.2768,
+      "step": 1201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.396786587495634e-06,
+      "loss": 1.1536,
+      "step": 1202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.403772266852952e-06,
+      "loss": 0.9183,
+      "step": 1203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.41075794621027e-06,
+      "loss": 1.0931,
+      "step": 1204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.417743625567588e-06,
+      "loss": 1.1309,
+      "step": 1205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.424729304924905e-06,
+      "loss": 1.3125,
+      "step": 1206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.431714984282221e-06,
+      "loss": 1.3194,
+      "step": 1207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.43870066363954e-06,
+      "loss": 1.0414,
+      "step": 1208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.445686342996857e-06,
+      "loss": 0.9496,
+      "step": 1209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.452672022354175e-06,
+      "loss": 1.3288,
+      "step": 1210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.459657701711493e-06,
+      "loss": 1.1115,
+      "step": 1211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.46664338106881e-06,
+      "loss": 1.031,
+      "step": 1212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.473629060426127e-06,
+      "loss": 1.1829,
+      "step": 1213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.480614739783444e-06,
+      "loss": 1.1602,
+      "step": 1214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.487600419140762e-06,
+      "loss": 1.1719,
+      "step": 1215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.49458609849808e-06,
+      "loss": 1.1742,
+      "step": 1216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.501571777855398e-06,
+      "loss": 1.0681,
+      "step": 1217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.508557457212716e-06,
+      "loss": 1.0843,
+      "step": 1218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.515543136570032e-06,
+      "loss": 1.0102,
+      "step": 1219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.52252881592735e-06,
+      "loss": 1.1325,
+      "step": 1220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.529514495284667e-06,
+      "loss": 0.8971,
+      "step": 1221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.536500174641985e-06,
+      "loss": 1.1351,
+      "step": 1222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.543485853999303e-06,
+      "loss": 1.0044,
+      "step": 1223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.550471533356621e-06,
+      "loss": 1.1571,
+      "step": 1224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.557457212713937e-06,
+      "loss": 1.0874,
+      "step": 1225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.564442892071255e-06,
+      "loss": 0.9949,
+      "step": 1226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.0235,
+      "step": 1227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.578414250785889e-06,
+      "loss": 1.0215,
+      "step": 1228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.585399930143207e-06,
+      "loss": 1.0029,
+      "step": 1229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.592385609500524e-06,
+      "loss": 1.1173,
+      "step": 1230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.599371288857842e-06,
+      "loss": 1.4503,
+      "step": 1231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.60635696821516e-06,
+      "loss": 1.1276,
+      "step": 1232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.613342647572476e-06,
+      "loss": 1.2122,
+      "step": 1233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.620328326929794e-06,
+      "loss": 1.0707,
+      "step": 1234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.627314006287112e-06,
+      "loss": 0.9391,
+      "step": 1235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.63429968564443e-06,
+      "loss": 0.9688,
+      "step": 1236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.641285365001747e-06,
+      "loss": 1.2267,
+      "step": 1237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.648271044359065e-06,
+      "loss": 1.1148,
+      "step": 1238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.655256723716381e-06,
+      "loss": 1.2895,
+      "step": 1239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.662242403073699e-06,
+      "loss": 1.1895,
+      "step": 1240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.669228082431017e-06,
+      "loss": 1.2144,
+      "step": 1241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.676213761788335e-06,
+      "loss": 1.1339,
+      "step": 1242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.683199441145653e-06,
+      "loss": 1.1933,
+      "step": 1243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.69018512050297e-06,
+      "loss": 1.1978,
+      "step": 1244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.697170799860286e-06,
+      "loss": 0.9573,
+      "step": 1245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.704156479217604e-06,
+      "loss": 1.0281,
+      "step": 1246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.711142158574922e-06,
+      "loss": 1.238,
+      "step": 1247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.71812783793224e-06,
+      "loss": 1.0401,
+      "step": 1248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.725113517289558e-06,
+      "loss": 1.265,
+      "step": 1249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.732099196646876e-06,
+      "loss": 1.1825,
+      "step": 1250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.739084876004192e-06,
+      "loss": 1.0674,
+      "step": 1251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.74607055536151e-06,
+      "loss": 1.0542,
+      "step": 1252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.753056234718827e-06,
+      "loss": 1.0835,
+      "step": 1253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.760041914076145e-06,
+      "loss": 1.1671,
+      "step": 1254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.767027593433463e-06,
+      "loss": 1.2462,
+      "step": 1255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.77401327279078e-06,
+      "loss": 1.0717,
+      "step": 1256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.780998952148097e-06,
+      "loss": 1.1599,
+      "step": 1257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.787984631505415e-06,
+      "loss": 0.9979,
+      "step": 1258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.794970310862732e-06,
+      "loss": 1.3583,
+      "step": 1259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.801955990220049e-06,
+      "loss": 0.9003,
+      "step": 1260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.808941669577366e-06,
+      "loss": 1.1911,
+      "step": 1261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.815927348934684e-06,
+      "loss": 1.072,
+      "step": 1262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.822913028292002e-06,
+      "loss": 0.9582,
+      "step": 1263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.82989870764932e-06,
+      "loss": 1.2193,
+      "step": 1264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.836884387006636e-06,
+      "loss": 1.0088,
+      "step": 1265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.843870066363954e-06,
+      "loss": 1.0281,
+      "step": 1266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.850855745721272e-06,
+      "loss": 1.1879,
+      "step": 1267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.85784142507859e-06,
+      "loss": 1.2787,
+      "step": 1268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.864827104435907e-06,
+      "loss": 1.0613,
+      "step": 1269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.871812783793225e-06,
+      "loss": 0.9409,
+      "step": 1270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.878798463150541e-06,
+      "loss": 1.1345,
+      "step": 1271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.885784142507859e-06,
+      "loss": 1.0748,
+      "step": 1272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.892769821865177e-06,
+      "loss": 1.1857,
+      "step": 1273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.899755501222495e-06,
+      "loss": 0.9922,
+      "step": 1274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.906741180579812e-06,
+      "loss": 1.0708,
+      "step": 1275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.91372685993713e-06,
+      "loss": 1.1384,
+      "step": 1276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.920712539294448e-06,
+      "loss": 1.0631,
+      "step": 1277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.927698218651764e-06,
+      "loss": 1.0598,
+      "step": 1278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.934683898009082e-06,
+      "loss": 1.2737,
+      "step": 1279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.9416695773664e-06,
+      "loss": 1.1352,
+      "step": 1280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.948655256723718e-06,
+      "loss": 1.1337,
+      "step": 1281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.955640936081035e-06,
+      "loss": 1.1975,
+      "step": 1282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.962626615438353e-06,
+      "loss": 0.9168,
+      "step": 1283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.96961229479567e-06,
+      "loss": 1.0583,
+      "step": 1284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.976597974152987e-06,
+      "loss": 0.9166,
+      "step": 1285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.983583653510305e-06,
+      "loss": 1.0582,
+      "step": 1286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.990569332867623e-06,
+      "loss": 1.0916,
+      "step": 1287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.99755501222494e-06,
+      "loss": 1.1743,
+      "step": 1288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.004540691582258e-06,
+      "loss": 1.2916,
+      "step": 1289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.011526370939574e-06,
+      "loss": 1.1683,
+      "step": 1290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.018512050296892e-06,
+      "loss": 1.106,
+      "step": 1291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.02549772965421e-06,
+      "loss": 1.0634,
+      "step": 1292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.032483409011528e-06,
+      "loss": 1.1322,
+      "step": 1293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.039469088368844e-06,
+      "loss": 1.0493,
+      "step": 1294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.046454767726162e-06,
+      "loss": 0.9508,
+      "step": 1295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.05344044708348e-06,
+      "loss": 1.1075,
+      "step": 1296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.060426126440797e-06,
+      "loss": 1.0509,
+      "step": 1297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.067411805798114e-06,
+      "loss": 0.9329,
+      "step": 1298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.074397485155431e-06,
+      "loss": 1.1608,
+      "step": 1299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.08138316451275e-06,
+      "loss": 0.9956,
+      "step": 1300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.088368843870067e-06,
+      "loss": 1.0777,
+      "step": 1301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.095354523227385e-06,
+      "loss": 1.0073,
+      "step": 1302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.102340202584703e-06,
+      "loss": 1.0797,
+      "step": 1303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.109325881942019e-06,
+      "loss": 1.0939,
+      "step": 1304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.116311561299337e-06,
+      "loss": 0.9926,
+      "step": 1305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.123297240656654e-06,
+      "loss": 1.1009,
+      "step": 1306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.130282920013972e-06,
+      "loss": 1.1013,
+      "step": 1307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.13726859937129e-06,
+      "loss": 1.0474,
+      "step": 1308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.144254278728608e-06,
+      "loss": 1.154,
+      "step": 1309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.151239958085924e-06,
+      "loss": 1.2683,
+      "step": 1310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.158225637443242e-06,
+      "loss": 1.1514,
+      "step": 1311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.16521131680056e-06,
+      "loss": 0.9129,
+      "step": 1312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.172196996157877e-06,
+      "loss": 1.1316,
+      "step": 1313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.179182675515195e-06,
+      "loss": 1.1011,
+      "step": 1314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.186168354872513e-06,
+      "loss": 0.9938,
+      "step": 1315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.193154034229829e-06,
+      "loss": 1.0595,
+      "step": 1316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.200139713587147e-06,
+      "loss": 0.996,
+      "step": 1317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.207125392944465e-06,
+      "loss": 0.9833,
+      "step": 1318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.214111072301783e-06,
+      "loss": 1.1867,
+      "step": 1319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.2210967516591e-06,
+      "loss": 1.0445,
+      "step": 1320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.228082431016418e-06,
+      "loss": 1.147,
+      "step": 1321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.235068110373734e-06,
+      "loss": 1.0228,
+      "step": 1322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.242053789731052e-06,
+      "loss": 1.1047,
+      "step": 1323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.24903946908837e-06,
+      "loss": 1.1273,
+      "step": 1324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.256025148445688e-06,
+      "loss": 1.0085,
+      "step": 1325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.263010827803006e-06,
+      "loss": 0.9171,
+      "step": 1326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.269996507160323e-06,
+      "loss": 1.0853,
+      "step": 1327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.27698218651764e-06,
+      "loss": 1.1818,
+      "step": 1328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.283967865874957e-06,
+      "loss": 1.1491,
+      "step": 1329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.290953545232273e-06,
+      "loss": 0.9669,
+      "step": 1330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.297939224589591e-06,
+      "loss": 1.1718,
+      "step": 1331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.304924903946909e-06,
+      "loss": 1.1272,
+      "step": 1332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.311910583304227e-06,
+      "loss": 1.2165,
+      "step": 1333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.318896262661545e-06,
+      "loss": 0.9827,
+      "step": 1334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.325881942018862e-06,
+      "loss": 1.2002,
+      "step": 1335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.332867621376179e-06,
+      "loss": 1.2118,
+      "step": 1336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.339853300733496e-06,
+      "loss": 0.9357,
+      "step": 1337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.346838980090814e-06,
+      "loss": 1.0638,
+      "step": 1338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.353824659448132e-06,
+      "loss": 0.8744,
+      "step": 1339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.36081033880545e-06,
+      "loss": 1.0782,
+      "step": 1340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.367796018162768e-06,
+      "loss": 1.2025,
+      "step": 1341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.374781697520084e-06,
+      "loss": 1.0189,
+      "step": 1342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.381767376877402e-06,
+      "loss": 1.1497,
+      "step": 1343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.38875305623472e-06,
+      "loss": 0.9429,
+      "step": 1344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.395738735592037e-06,
+      "loss": 1.0946,
+      "step": 1345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.402724414949355e-06,
+      "loss": 1.1634,
+      "step": 1346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.409710094306673e-06,
+      "loss": 0.9825,
+      "step": 1347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.416695773663989e-06,
+      "loss": 1.18,
+      "step": 1348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.423681453021307e-06,
+      "loss": 1.0037,
+      "step": 1349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.430667132378625e-06,
+      "loss": 1.0522,
+      "step": 1350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.437652811735942e-06,
+      "loss": 0.9321,
+      "step": 1351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.44463849109326e-06,
+      "loss": 1.0602,
+      "step": 1352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.451624170450578e-06,
+      "loss": 1.0641,
+      "step": 1353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.458609849807894e-06,
+      "loss": 1.0993,
+      "step": 1354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.465595529165212e-06,
+      "loss": 1.1208,
+      "step": 1355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.47258120852253e-06,
+      "loss": 1.0662,
+      "step": 1356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.479566887879848e-06,
+      "loss": 0.8066,
+      "step": 1357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.486552567237165e-06,
+      "loss": 1.0506,
+      "step": 1358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.493538246594483e-06,
+      "loss": 1.0831,
+      "step": 1359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.5005239259518e-06,
+      "loss": 0.9536,
+      "step": 1360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.507509605309117e-06,
+      "loss": 1.1979,
+      "step": 1361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.514495284666435e-06,
+      "loss": 1.393,
+      "step": 1362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.521480964023751e-06,
+      "loss": 1.1692,
+      "step": 1363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.528466643381069e-06,
+      "loss": 1.2024,
+      "step": 1364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.535452322738387e-06,
+      "loss": 1.2624,
+      "step": 1365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.542438002095704e-06,
+      "loss": 1.2106,
+      "step": 1366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.549423681453022e-06,
+      "loss": 1.3134,
+      "step": 1367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.556409360810338e-06,
+      "loss": 1.0774,
+      "step": 1368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.563395040167656e-06,
+      "loss": 1.2582,
+      "step": 1369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.570380719524974e-06,
+      "loss": 0.9471,
+      "step": 1370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.577366398882292e-06,
+      "loss": 1.0068,
+      "step": 1371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.58435207823961e-06,
+      "loss": 0.853,
+      "step": 1372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.591337757596927e-06,
+      "loss": 1.1052,
+      "step": 1373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.598323436954244e-06,
+      "loss": 1.6152,
+      "step": 1374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.605309116311561e-06,
+      "loss": 1.4012,
+      "step": 1375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.61229479566888e-06,
+      "loss": 1.0267,
+      "step": 1376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.619280475026197e-06,
+      "loss": 1.1062,
+      "step": 1377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.626266154383515e-06,
+      "loss": 1.1004,
+      "step": 1378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.633251833740833e-06,
+      "loss": 1.2613,
+      "step": 1379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.64023751309815e-06,
+      "loss": 1.2263,
+      "step": 1380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.647223192455467e-06,
+      "loss": 1.0281,
+      "step": 1381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.654208871812784e-06,
+      "loss": 1.0058,
+      "step": 1382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.661194551170102e-06,
+      "loss": 0.979,
+      "step": 1383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.66818023052742e-06,
+      "loss": 1.0384,
+      "step": 1384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.675165909884738e-06,
+      "loss": 1.1956,
+      "step": 1385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.682151589242056e-06,
+      "loss": 1.1617,
+      "step": 1386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.689137268599372e-06,
+      "loss": 1.1773,
+      "step": 1387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.69612294795669e-06,
+      "loss": 1.1486,
+      "step": 1388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.703108627314007e-06,
+      "loss": 1.0996,
+      "step": 1389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.710094306671325e-06,
+      "loss": 0.9651,
+      "step": 1390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.717079986028643e-06,
+      "loss": 0.9183,
+      "step": 1391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.72406566538596e-06,
+      "loss": 0.9914,
+      "step": 1392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.731051344743277e-06,
+      "loss": 1.0412,
+      "step": 1393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.738037024100595e-06,
+      "loss": 1.061,
+      "step": 1394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.745022703457913e-06,
+      "loss": 1.1067,
+      "step": 1395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.75200838281523e-06,
+      "loss": 1.1579,
+      "step": 1396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.758994062172548e-06,
+      "loss": 1.1843,
+      "step": 1397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.765979741529864e-06,
+      "loss": 1.1019,
+      "step": 1398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.772965420887182e-06,
+      "loss": 0.9872,
+      "step": 1399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.7799511002445e-06,
+      "loss": 1.1367,
+      "step": 1400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.786936779601816e-06,
+      "loss": 1.3796,
+      "step": 1401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.793922458959134e-06,
+      "loss": 1.0939,
+      "step": 1402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.800908138316452e-06,
+      "loss": 1.0932,
+      "step": 1403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.80789381767377e-06,
+      "loss": 1.0609,
+      "step": 1404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.814879497031087e-06,
+      "loss": 1.0778,
+      "step": 1405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.821865176388405e-06,
+      "loss": 1.0679,
+      "step": 1406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.828850855745721e-06,
+      "loss": 1.0602,
+      "step": 1407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.835836535103039e-06,
+      "loss": 1.0975,
+      "step": 1408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.842822214460357e-06,
+      "loss": 0.8918,
+      "step": 1409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.849807893817675e-06,
+      "loss": 1.187,
+      "step": 1410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.856793573174992e-06,
+      "loss": 1.0198,
+      "step": 1411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.86377925253231e-06,
+      "loss": 0.9439,
+      "step": 1412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.870764931889626e-06,
+      "loss": 1.1261,
+      "step": 1413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.877750611246944e-06,
+      "loss": 1.1199,
+      "step": 1414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.884736290604262e-06,
+      "loss": 0.8975,
+      "step": 1415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.89172196996158e-06,
+      "loss": 1.1318,
+      "step": 1416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.898707649318898e-06,
+      "loss": 1.0484,
+      "step": 1417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.905693328676215e-06,
+      "loss": 1.0315,
+      "step": 1418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.912679008033532e-06,
+      "loss": 1.2538,
+      "step": 1419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.91966468739085e-06,
+      "loss": 0.9708,
+      "step": 1420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.926650366748167e-06,
+      "loss": 0.9226,
+      "step": 1421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.933636046105485e-06,
+      "loss": 0.9726,
+      "step": 1422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.940621725462803e-06,
+      "loss": 1.0015,
+      "step": 1423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.94760740482012e-06,
+      "loss": 1.0625,
+      "step": 1424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.954593084177437e-06,
+      "loss": 0.9147,
+      "step": 1425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.961578763534755e-06,
+      "loss": 1.1255,
+      "step": 1426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.968564442892072e-06,
+      "loss": 1.1714,
+      "step": 1427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.97555012224939e-06,
+      "loss": 1.0504,
+      "step": 1428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.982535801606708e-06,
+      "loss": 1.0082,
+      "step": 1429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.989521480964026e-06,
+      "loss": 1.011,
+      "step": 1430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.996507160321342e-06,
+      "loss": 0.8751,
+      "step": 1431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.000349283967866e-05,
+      "loss": 1.1306,
+      "step": 1432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0010478519035978e-05,
+      "loss": 0.9531,
+      "step": 1433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0017464198393294e-05,
+      "loss": 1.1743,
+      "step": 1434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0024449877750613e-05,
+      "loss": 1.2668,
+      "step": 1435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.003143555710793e-05,
+      "loss": 0.9396,
+      "step": 1436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0038421236465247e-05,
+      "loss": 1.1883,
+      "step": 1437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0045406915822565e-05,
+      "loss": 1.0595,
+      "step": 1438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0052392595179883e-05,
+      "loss": 0.9397,
+      "step": 1439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0059378274537199e-05,
+      "loss": 1.2096,
+      "step": 1440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0066363953894518e-05,
+      "loss": 1.1181,
+      "step": 1441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0073349633251834e-05,
+      "loss": 1.2638,
+      "step": 1442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0080335312609152e-05,
+      "loss": 1.1396,
+      "step": 1443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.008732099196647e-05,
+      "loss": 0.9786,
+      "step": 1444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0094306671323788e-05,
+      "loss": 1.0217,
+      "step": 1445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0101292350681104e-05,
+      "loss": 1.0535,
+      "step": 1446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0108278030038424e-05,
+      "loss": 0.9962,
+      "step": 1447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.011526370939574e-05,
+      "loss": 1.1826,
+      "step": 1448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0122249388753056e-05,
+      "loss": 0.9132,
+      "step": 1449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0129235068110375e-05,
+      "loss": 1.1274,
+      "step": 1450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0136220747467691e-05,
+      "loss": 0.93,
+      "step": 1451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.014320642682501e-05,
+      "loss": 1.0576,
+      "step": 1452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0150192106182325e-05,
+      "loss": 1.1563,
+      "step": 1453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0157177785539645e-05,
+      "loss": 0.9775,
+      "step": 1454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0164163464896961e-05,
+      "loss": 1.0392,
+      "step": 1455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.017114914425428e-05,
+      "loss": 1.0205,
+      "step": 1456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0178134823611597e-05,
+      "loss": 1.0907,
+      "step": 1457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0185120502968914e-05,
+      "loss": 1.23,
+      "step": 1458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.019210618232623e-05,
+      "loss": 0.9967,
+      "step": 1459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.019909186168355e-05,
+      "loss": 0.9485,
+      "step": 1460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0206077541040866e-05,
+      "loss": 1.1415,
+      "step": 1461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0213063220398186e-05,
+      "loss": 0.9139,
+      "step": 1462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0220048899755502e-05,
+      "loss": 1.2178,
+      "step": 1463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.022703457911282e-05,
+      "loss": 1.0001,
+      "step": 1464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0234020258470136e-05,
+      "loss": 1.2067,
+      "step": 1465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0241005937827455e-05,
+      "loss": 1.1561,
+      "step": 1466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0247991617184771e-05,
+      "loss": 1.0657,
+      "step": 1467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.025497729654209e-05,
+      "loss": 1.0644,
+      "step": 1468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0261962975899407e-05,
+      "loss": 1.1669,
+      "step": 1469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0268948655256725e-05,
+      "loss": 0.9496,
+      "step": 1470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.027593433461404e-05,
+      "loss": 1.0992,
+      "step": 1471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.028292001397136e-05,
+      "loss": 1.0673,
+      "step": 1472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0289905693328676e-05,
+      "loss": 0.9447,
+      "step": 1473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0296891372685996e-05,
+      "loss": 1.2079,
+      "step": 1474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0303877052043312e-05,
+      "loss": 0.8651,
+      "step": 1475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.031086273140063e-05,
+      "loss": 1.1485,
+      "step": 1476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0317848410757946e-05,
+      "loss": 1.0316,
+      "step": 1477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0324834090115266e-05,
+      "loss": 0.9504,
+      "step": 1478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0331819769472582e-05,
+      "loss": 0.9212,
+      "step": 1479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0338805448829901e-05,
+      "loss": 1.0072,
+      "step": 1480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0345791128187217e-05,
+      "loss": 1.1456,
+      "step": 1481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0352776807544535e-05,
+      "loss": 0.9117,
+      "step": 1482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0359762486901851e-05,
+      "loss": 1.0945,
+      "step": 1483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0366748166259169e-05,
+      "loss": 1.1104,
+      "step": 1484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0373733845616487e-05,
+      "loss": 1.0161,
+      "step": 1485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0380719524973803e-05,
+      "loss": 0.9676,
+      "step": 1486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0387705204331122e-05,
+      "loss": 1.0239,
+      "step": 1487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0394690883688439e-05,
+      "loss": 1.0969,
+      "step": 1488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0401676563045758e-05,
+      "loss": 1.2598,
+      "step": 1489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0408662242403074e-05,
+      "loss": 1.137,
+      "step": 1490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0415647921760392e-05,
+      "loss": 1.0068,
+      "step": 1491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0422633601117708e-05,
+      "loss": 0.9835,
+      "step": 1492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0429619280475028e-05,
+      "loss": 0.8974,
+      "step": 1493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0436604959832344e-05,
+      "loss": 1.1191,
+      "step": 1494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0443590639189663e-05,
+      "loss": 0.8587,
+      "step": 1495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.045057631854698e-05,
+      "loss": 0.9782,
+      "step": 1496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0457561997904297e-05,
+      "loss": 0.884,
+      "step": 1497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0464547677261613e-05,
+      "loss": 1.2904,
+      "step": 1498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0471533356618933e-05,
+      "loss": 1.0247,
+      "step": 1499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0478519035976249e-05,
+      "loss": 1.0963,
+      "step": 1500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0485504715333568e-05,
+      "loss": 1.1014,
+      "step": 1501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0492490394690885e-05,
+      "loss": 1.0133,
+      "step": 1502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0499476074048202e-05,
+      "loss": 1.2023,
+      "step": 1503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0506461753405518e-05,
+      "loss": 1.1629,
+      "step": 1504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0513447432762838e-05,
+      "loss": 1.0549,
+      "step": 1505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0520433112120154e-05,
+      "loss": 1.0406,
+      "step": 1506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0527418791477474e-05,
+      "loss": 1.0679,
+      "step": 1507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.053440447083479e-05,
+      "loss": 0.9833,
+      "step": 1508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0541390150192108e-05,
+      "loss": 0.9907,
+      "step": 1509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0548375829549424e-05,
+      "loss": 0.963,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0555361508906743e-05,
+      "loss": 1.0737,
+      "step": 1511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.056234718826406e-05,
+      "loss": 1.1732,
+      "step": 1512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0569332867621379e-05,
+      "loss": 1.1849,
+      "step": 1513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0576318546978695e-05,
+      "loss": 1.0609,
+      "step": 1514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0583304226336013e-05,
+      "loss": 1.0615,
+      "step": 1515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0590289905693329e-05,
+      "loss": 1.2054,
+      "step": 1516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0597275585050648e-05,
+      "loss": 0.9128,
+      "step": 1517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0604261264407964e-05,
+      "loss": 1.0502,
+      "step": 1518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.061124694376528e-05,
+      "loss": 1.2223,
+      "step": 1519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.06182326231226e-05,
+      "loss": 0.9883,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0625218302479916e-05,
+      "loss": 1.1408,
+      "step": 1521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0632203981837234e-05,
+      "loss": 1.0661,
+      "step": 1522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0639189661194552e-05,
+      "loss": 1.0941,
+      "step": 1523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.064617534055187e-05,
+      "loss": 0.8207,
+      "step": 1524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0653161019909186e-05,
+      "loss": 0.9227,
+      "step": 1525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0660146699266505e-05,
+      "loss": 0.8871,
+      "step": 1526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0667132378623821e-05,
+      "loss": 1.061,
+      "step": 1527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.067411805798114e-05,
+      "loss": 0.9345,
+      "step": 1528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0681103737338457e-05,
+      "loss": 1.0206,
+      "step": 1529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0688089416695775e-05,
+      "loss": 1.1528,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0695075096053091e-05,
+      "loss": 1.0481,
+      "step": 1531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.070206077541041e-05,
+      "loss": 0.994,
+      "step": 1532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0709046454767727e-05,
+      "loss": 0.9943,
+      "step": 1533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0716032134125044e-05,
+      "loss": 0.9651,
+      "step": 1534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0723017813482362e-05,
+      "loss": 1.0576,
+      "step": 1535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.073000349283968e-05,
+      "loss": 0.9382,
+      "step": 1536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0736989172196996e-05,
+      "loss": 0.9351,
+      "step": 1537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0743974851554316e-05,
+      "loss": 1.039,
+      "step": 1538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0750960530911632e-05,
+      "loss": 1.158,
+      "step": 1539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.075794621026895e-05,
+      "loss": 1.0398,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0764931889626267e-05,
+      "loss": 0.9159,
+      "step": 1541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0771917568983585e-05,
+      "loss": 1.0163,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0778903248340901e-05,
+      "loss": 1.0973,
+      "step": 1543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.078588892769822e-05,
+      "loss": 1.0116,
+      "step": 1544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0792874607055537e-05,
+      "loss": 1.0726,
+      "step": 1545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0799860286412855e-05,
+      "loss": 1.0222,
+      "step": 1546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0806845965770173e-05,
+      "loss": 1.0636,
+      "step": 1547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.081383164512749e-05,
+      "loss": 0.8799,
+      "step": 1548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0820817324484806e-05,
+      "loss": 1.1073,
+      "step": 1549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0827803003842126e-05,
+      "loss": 1.011,
+      "step": 1550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0834788683199442e-05,
+      "loss": 1.4292,
+      "step": 1551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.084177436255676e-05,
+      "loss": 0.9635,
+      "step": 1552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0848760041914078e-05,
+      "loss": 0.9564,
+      "step": 1553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0855745721271394e-05,
+      "loss": 1.1133,
+      "step": 1554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0862731400628712e-05,
+      "loss": 1.0079,
+      "step": 1555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0869717079986028e-05,
+      "loss": 0.9991,
+      "step": 1556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0876702759343347e-05,
+      "loss": 1.0255,
+      "step": 1557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0883688438700663e-05,
+      "loss": 1.0157,
+      "step": 1558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0890674118057983e-05,
+      "loss": 1.0676,
+      "step": 1559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0897659797415299e-05,
+      "loss": 1.0263,
+      "step": 1560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0904645476772617e-05,
+      "loss": 1.0042,
+      "step": 1561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0911631156129933e-05,
+      "loss": 0.9696,
+      "step": 1562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0918616835487252e-05,
+      "loss": 0.9297,
+      "step": 1563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0925602514844569e-05,
+      "loss": 1.0354,
+      "step": 1564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0932588194201888e-05,
+      "loss": 0.7537,
+      "step": 1565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0939573873559204e-05,
+      "loss": 1.1741,
+      "step": 1566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0946559552916522e-05,
+      "loss": 0.8998,
+      "step": 1567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0953545232273838e-05,
+      "loss": 0.8511,
+      "step": 1568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0960530911631158e-05,
+      "loss": 0.9627,
+      "step": 1569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0967516590988474e-05,
+      "loss": 1.0505,
+      "step": 1570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0974502270345793e-05,
+      "loss": 0.9845,
+      "step": 1571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.098148794970311e-05,
+      "loss": 1.0834,
+      "step": 1572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0988473629060427e-05,
+      "loss": 1.1102,
+      "step": 1573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.0995459308417743e-05,
+      "loss": 1.1312,
+      "step": 1574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1002444987775063e-05,
+      "loss": 0.9818,
+      "step": 1575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1009430667132379e-05,
+      "loss": 0.8969,
+      "step": 1576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1016416346489698e-05,
+      "loss": 1.0346,
+      "step": 1577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1023402025847015e-05,
+      "loss": 0.9362,
+      "step": 1578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1030387705204332e-05,
+      "loss": 1.019,
+      "step": 1579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1037373384561648e-05,
+      "loss": 0.9429,
+      "step": 1580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1044359063918968e-05,
+      "loss": 0.8854,
+      "step": 1581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1051344743276284e-05,
+      "loss": 0.9939,
+      "step": 1582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1058330422633604e-05,
+      "loss": 1.023,
+      "step": 1583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.106531610199092e-05,
+      "loss": 0.9281,
+      "step": 1584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1072301781348238e-05,
+      "loss": 0.9578,
+      "step": 1585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1079287460705554e-05,
+      "loss": 1.0617,
+      "step": 1586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1086273140062871e-05,
+      "loss": 0.9053,
+      "step": 1587
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.109325881942019e-05,
+      "loss": 1.0589,
+      "step": 1588
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1100244498777505e-05,
+      "loss": 1.0509,
+      "step": 1589
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1107230178134825e-05,
+      "loss": 1.045,
+      "step": 1590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1114215857492141e-05,
+      "loss": 1.1537,
+      "step": 1591
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.112120153684946e-05,
+      "loss": 1.1389,
+      "step": 1592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1128187216206777e-05,
+      "loss": 0.9954,
+      "step": 1593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1135172895564094e-05,
+      "loss": 1.0368,
+      "step": 1594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.114215857492141e-05,
+      "loss": 1.043,
+      "step": 1595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.114914425427873e-05,
+      "loss": 0.8971,
+      "step": 1596
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1156129933636046e-05,
+      "loss": 0.9446,
+      "step": 1597
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1163115612993366e-05,
+      "loss": 0.9204,
+      "step": 1598
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1170101292350682e-05,
+      "loss": 1.0721,
+      "step": 1599
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1177086971708e-05,
+      "loss": 1.0904,
+      "step": 1600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1184072651065316e-05,
+      "loss": 0.864,
+      "step": 1601
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1191058330422635e-05,
+      "loss": 0.8654,
+      "step": 1602
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1198044009779951e-05,
+      "loss": 0.8471,
+      "step": 1603
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1205029689137271e-05,
+      "loss": 0.8374,
+      "step": 1604
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1212015368494587e-05,
+      "loss": 1.2393,
+      "step": 1605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1219001047851905e-05,
+      "loss": 0.9979,
+      "step": 1606
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1225986727209221e-05,
+      "loss": 1.1722,
+      "step": 1607
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.123297240656654e-05,
+      "loss": 1.0808,
+      "step": 1608
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1239958085923857e-05,
+      "loss": 1.0511,
+      "step": 1609
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1246943765281176e-05,
+      "loss": 0.9425,
+      "step": 1610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1253929444638492e-05,
+      "loss": 1.0119,
+      "step": 1611
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.126091512399581e-05,
+      "loss": 1.0071,
+      "step": 1612
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1267900803353126e-05,
+      "loss": 1.1457,
+      "step": 1613
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1274886482710446e-05,
+      "loss": 0.9917,
+      "step": 1614
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1281872162067762e-05,
+      "loss": 1.0651,
+      "step": 1615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1288857841425081e-05,
+      "loss": 1.138,
+      "step": 1616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1295843520782397e-05,
+      "loss": 0.9955,
+      "step": 1617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1302829200139715e-05,
+      "loss": 0.9846,
+      "step": 1618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1309814879497031e-05,
+      "loss": 0.8374,
+      "step": 1619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.131680055885435e-05,
+      "loss": 0.9731,
+      "step": 1620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1323786238211667e-05,
+      "loss": 1.0979,
+      "step": 1621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1330771917568983e-05,
+      "loss": 1.0788,
+      "step": 1622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1337757596926303e-05,
+      "loss": 1.0927,
+      "step": 1623
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1344743276283619e-05,
+      "loss": 1.0631,
+      "step": 1624
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1351728955640936e-05,
+      "loss": 1.0882,
+      "step": 1625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1358714634998254e-05,
+      "loss": 0.9414,
+      "step": 1626
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1365700314355572e-05,
+      "loss": 0.9841,
+      "step": 1627
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1372685993712888e-05,
+      "loss": 0.8572,
+      "step": 1628
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1379671673070208e-05,
+      "loss": 1.1166,
+      "step": 1629
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1386657352427524e-05,
+      "loss": 1.393,
+      "step": 1630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1393643031784842e-05,
+      "loss": 0.9971,
+      "step": 1631
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.140062871114216e-05,
+      "loss": 0.9217,
+      "step": 1632
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1407614390499477e-05,
+      "loss": 1.2548,
+      "step": 1633
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1414600069856793e-05,
+      "loss": 1.1248,
+      "step": 1634
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1421585749214113e-05,
+      "loss": 0.9901,
+      "step": 1635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.0755,
+      "step": 1636
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1435557107928747e-05,
+      "loss": 0.9015,
+      "step": 1637
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1442542787286065e-05,
+      "loss": 0.8064,
+      "step": 1638
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1449528466643382e-05,
+      "loss": 0.9816,
+      "step": 1639
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1456514146000699e-05,
+      "loss": 0.9864,
+      "step": 1640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1463499825358018e-05,
+      "loss": 0.8482,
+      "step": 1641
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1470485504715334e-05,
+      "loss": 1.0543,
+      "step": 1642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1477471184072652e-05,
+      "loss": 0.9752,
+      "step": 1643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.148445686342997e-05,
+      "loss": 0.8003,
+      "step": 1644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1491442542787288e-05,
+      "loss": 1.0649,
+      "step": 1645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1498428222144604e-05,
+      "loss": 1.0693,
+      "step": 1646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1505413901501923e-05,
+      "loss": 1.0872,
+      "step": 1647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.151239958085924e-05,
+      "loss": 1.0834,
+      "step": 1648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1519385260216557e-05,
+      "loss": 1.0798,
+      "step": 1649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1526370939573875e-05,
+      "loss": 1.0125,
+      "step": 1650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1533356618931193e-05,
+      "loss": 1.0548,
+      "step": 1651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1540342298288509e-05,
+      "loss": 1.1586,
+      "step": 1652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1547327977645828e-05,
+      "loss": 1.0462,
+      "step": 1653
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1554313657003145e-05,
+      "loss": 1.0055,
+      "step": 1654
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1561299336360462e-05,
+      "loss": 1.015,
+      "step": 1655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.156828501571778e-05,
+      "loss": 0.9727,
+      "step": 1656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1575270695075096e-05,
+      "loss": 1.1215,
+      "step": 1657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1582256374432414e-05,
+      "loss": 1.0307,
+      "step": 1658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.158924205378973e-05,
+      "loss": 0.9978,
+      "step": 1659
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.159622773314705e-05,
+      "loss": 1.0775,
+      "step": 1660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1603213412504366e-05,
+      "loss": 0.9,
+      "step": 1661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1610199091861685e-05,
+      "loss": 0.8869,
+      "step": 1662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1617184771219001e-05,
+      "loss": 1.1277,
+      "step": 1663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.162417045057632e-05,
+      "loss": 1.0027,
+      "step": 1664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1631156129933635e-05,
+      "loss": 1.0699,
+      "step": 1665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1638141809290955e-05,
+      "loss": 1.1349,
+      "step": 1666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1645127488648271e-05,
+      "loss": 1.0522,
+      "step": 1667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.165211316800559e-05,
+      "loss": 1.1409,
+      "step": 1668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1659098847362907e-05,
+      "loss": 0.8886,
+      "step": 1669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1666084526720224e-05,
+      "loss": 1.0059,
+      "step": 1670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.167307020607754e-05,
+      "loss": 1.0799,
+      "step": 1671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.168005588543486e-05,
+      "loss": 0.9038,
+      "step": 1672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1687041564792176e-05,
+      "loss": 0.9478,
+      "step": 1673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1694027244149496e-05,
+      "loss": 0.9928,
+      "step": 1674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1701012923506812e-05,
+      "loss": 0.9528,
+      "step": 1675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.170799860286413e-05,
+      "loss": 0.9598,
+      "step": 1676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1714984282221446e-05,
+      "loss": 0.8511,
+      "step": 1677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1721969961578765e-05,
+      "loss": 1.1121,
+      "step": 1678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1728955640936081e-05,
+      "loss": 0.8672,
+      "step": 1679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1735941320293401e-05,
+      "loss": 1.0642,
+      "step": 1680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1742926999650717e-05,
+      "loss": 1.1018,
+      "step": 1681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1749912679008035e-05,
+      "loss": 1.1037,
+      "step": 1682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1756898358365351e-05,
+      "loss": 1.0016,
+      "step": 1683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.176388403772267e-05,
+      "loss": 1.0266,
+      "step": 1684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1770869717079987e-05,
+      "loss": 0.9398,
+      "step": 1685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1777855396437306e-05,
+      "loss": 1.1086,
+      "step": 1686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1784841075794622e-05,
+      "loss": 1.1712,
+      "step": 1687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.179182675515194e-05,
+      "loss": 0.9868,
+      "step": 1688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1798812434509256e-05,
+      "loss": 0.9838,
+      "step": 1689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1805798113866576e-05,
+      "loss": 1.0592,
+      "step": 1690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1812783793223892e-05,
+      "loss": 1.1197,
+      "step": 1691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1819769472581208e-05,
+      "loss": 0.9909,
+      "step": 1692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1826755151938527e-05,
+      "loss": 1.0774,
+      "step": 1693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1833740831295843e-05,
+      "loss": 0.9028,
+      "step": 1694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1840726510653161e-05,
+      "loss": 1.034,
+      "step": 1695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1847712190010479e-05,
+      "loss": 0.9438,
+      "step": 1696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1854697869367797e-05,
+      "loss": 0.9825,
+      "step": 1697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1861683548725113e-05,
+      "loss": 0.8949,
+      "step": 1698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1868669228082433e-05,
+      "loss": 0.8628,
+      "step": 1699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1875654907439749e-05,
+      "loss": 0.9779,
+      "step": 1700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1882640586797068e-05,
+      "loss": 1.0238,
+      "step": 1701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1889626266154384e-05,
+      "loss": 1.0107,
+      "step": 1702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1896611945511702e-05,
+      "loss": 1.1522,
+      "step": 1703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1903597624869018e-05,
+      "loss": 1.1483,
+      "step": 1704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1910583304226338e-05,
+      "loss": 1.095,
+      "step": 1705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1917568983583654e-05,
+      "loss": 1.1202,
+      "step": 1706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1924554662940973e-05,
+      "loss": 0.948,
+      "step": 1707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193154034229829e-05,
+      "loss": 1.1629,
+      "step": 1708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938526021655607e-05,
+      "loss": 0.982,
+      "step": 1709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1945511701012923e-05,
+      "loss": 1.0023,
+      "step": 1710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952497380370243e-05,
+      "loss": 0.9209,
+      "step": 1711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959483059727559e-05,
+      "loss": 0.9482,
+      "step": 1712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966468739084879e-05,
+      "loss": 0.8947,
+      "step": 1713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973454418442195e-05,
+      "loss": 0.9725,
+      "step": 1714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1980440097799512e-05,
+      "loss": 0.8653,
+      "step": 1715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987425777156829e-05,
+      "loss": 0.947,
+      "step": 1716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994411456514148e-05,
+      "loss": 1.0596,
+      "step": 1717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001397135871464e-05,
+      "loss": 1.2678,
+      "step": 1718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2008382815228784e-05,
+      "loss": 0.9927,
+      "step": 1719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20153684945861e-05,
+      "loss": 1.0853,
+      "step": 1720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022354173943418e-05,
+      "loss": 0.9694,
+      "step": 1721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029339853300734e-05,
+      "loss": 1.1678,
+      "step": 1722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036325532658053e-05,
+      "loss": 1.2789,
+      "step": 1723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204331121201537e-05,
+      "loss": 0.9074,
+      "step": 1724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050296891372686e-05,
+      "loss": 1.0588,
+      "step": 1725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2057282570730005e-05,
+      "loss": 1.2284,
+      "step": 1726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064268250087321e-05,
+      "loss": 0.8626,
+      "step": 1727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071253929444639e-05,
+      "loss": 0.9408,
+      "step": 1728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078239608801957e-05,
+      "loss": 0.9995,
+      "step": 1729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085225288159275e-05,
+      "loss": 1.0291,
+      "step": 1730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209221096751659e-05,
+      "loss": 0.9874,
+      "step": 1731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209919664687391e-05,
+      "loss": 1.0312,
+      "step": 1732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106182326231226e-05,
+      "loss": 0.8882,
+      "step": 1733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2113168005588544e-05,
+      "loss": 1.0176,
+      "step": 1734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120153684945862e-05,
+      "loss": 1.197,
+      "step": 1735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212713936430318e-05,
+      "loss": 0.9338,
+      "step": 1736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2134125043660496e-05,
+      "loss": 0.9995,
+      "step": 1737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2141110723017815e-05,
+      "loss": 1.0326,
+      "step": 1738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2148096402375131e-05,
+      "loss": 0.935,
+      "step": 1739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.215508208173245e-05,
+      "loss": 1.2418,
+      "step": 1740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2162067761089767e-05,
+      "loss": 0.9934,
+      "step": 1741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2169053440447085e-05,
+      "loss": 0.8763,
+      "step": 1742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2176039119804401e-05,
+      "loss": 1.0389,
+      "step": 1743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.218302479916172e-05,
+      "loss": 0.8938,
+      "step": 1744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2190010478519037e-05,
+      "loss": 1.1036,
+      "step": 1745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2196996157876354e-05,
+      "loss": 0.9089,
+      "step": 1746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2203981837233672e-05,
+      "loss": 0.8809,
+      "step": 1747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.221096751659099e-05,
+      "loss": 0.9898,
+      "step": 1748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2217953195948306e-05,
+      "loss": 0.8364,
+      "step": 1749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2224938875305626e-05,
+      "loss": 1.0095,
+      "step": 1750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2231924554662942e-05,
+      "loss": 0.9427,
+      "step": 1751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.223891023402026e-05,
+      "loss": 0.9141,
+      "step": 1752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2245895913377577e-05,
+      "loss": 1.1257,
+      "step": 1753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2252881592734895e-05,
+      "loss": 1.0888,
+      "step": 1754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2259867272092211e-05,
+      "loss": 1.0273,
+      "step": 1755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2266852951449531e-05,
+      "loss": 1.2611,
+      "step": 1756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2273838630806847e-05,
+      "loss": 0.8865,
+      "step": 1757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2280824310164165e-05,
+      "loss": 0.9259,
+      "step": 1758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2287809989521483e-05,
+      "loss": 0.8003,
+      "step": 1759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2294795668878799e-05,
+      "loss": 0.8465,
+      "step": 1760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2301781348236117e-05,
+      "loss": 1.0174,
+      "step": 1761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2308767027593433e-05,
+      "loss": 0.9638,
+      "step": 1762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2315752706950752e-05,
+      "loss": 0.8903,
+      "step": 1763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2322738386308068e-05,
+      "loss": 0.9677,
+      "step": 1764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2329724065665388e-05,
+      "loss": 0.9167,
+      "step": 1765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2336709745022704e-05,
+      "loss": 1.0178,
+      "step": 1766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2343695424380022e-05,
+      "loss": 1.1,
+      "step": 1767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2350681103737338e-05,
+      "loss": 1.1639,
+      "step": 1768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2357666783094657e-05,
+      "loss": 0.9769,
+      "step": 1769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2364652462451973e-05,
+      "loss": 1.0896,
+      "step": 1770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2371638141809293e-05,
+      "loss": 1.1619,
+      "step": 1771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2378623821166609e-05,
+      "loss": 0.919,
+      "step": 1772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2385609500523927e-05,
+      "loss": 1.0042,
+      "step": 1773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2392595179881243e-05,
+      "loss": 1.169,
+      "step": 1774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2399580859238563e-05,
+      "loss": 0.9733,
+      "step": 1775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2406566538595879e-05,
+      "loss": 0.9816,
+      "step": 1776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2413552217953198e-05,
+      "loss": 1.0213,
+      "step": 1777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2420537897310514e-05,
+      "loss": 0.8967,
+      "step": 1778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2427523576667832e-05,
+      "loss": 0.9534,
+      "step": 1779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2434509256025148e-05,
+      "loss": 1.3033,
+      "step": 1780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2441494935382468e-05,
+      "loss": 1.0449,
+      "step": 1781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2448480614739784e-05,
+      "loss": 0.9884,
+      "step": 1782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2455466294097103e-05,
+      "loss": 0.9603,
+      "step": 1783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.246245197345442e-05,
+      "loss": 0.978,
+      "step": 1784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2469437652811737e-05,
+      "loss": 0.9415,
+      "step": 1785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2476423332169053e-05,
+      "loss": 1.2105,
+      "step": 1786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2483409011526373e-05,
+      "loss": 1.0923,
+      "step": 1787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2490394690883689e-05,
+      "loss": 0.9629,
+      "step": 1788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2497380370241009e-05,
+      "loss": 0.9914,
+      "step": 1789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2504366049598325e-05,
+      "loss": 1.0843,
+      "step": 1790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2511351728955642e-05,
+      "loss": 0.896,
+      "step": 1791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2518337408312959e-05,
+      "loss": 0.7889,
+      "step": 1792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2525323087670278e-05,
+      "loss": 0.9095,
+      "step": 1793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2532308767027594e-05,
+      "loss": 0.8815,
+      "step": 1794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.253929444638491e-05,
+      "loss": 1.0325,
+      "step": 1795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.254628012574223e-05,
+      "loss": 1.0461,
+      "step": 1796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2553265805099546e-05,
+      "loss": 1.037,
+      "step": 1797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2560251484456864e-05,
+      "loss": 1.0695,
+      "step": 1798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2567237163814182e-05,
+      "loss": 1.1122,
+      "step": 1799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.25742228431715e-05,
+      "loss": 0.954,
+      "step": 1800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2581208522528816e-05,
+      "loss": 0.9564,
+      "step": 1801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2588194201886135e-05,
+      "loss": 1.0235,
+      "step": 1802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2595179881243451e-05,
+      "loss": 0.9823,
+      "step": 1803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2602165560600769e-05,
+      "loss": 1.0508,
+      "step": 1804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2609151239958087e-05,
+      "loss": 1.0475,
+      "step": 1805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2616136919315405e-05,
+      "loss": 1.0282,
+      "step": 1806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.262312259867272e-05,
+      "loss": 1.0957,
+      "step": 1807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.263010827803004e-05,
+      "loss": 0.9379,
+      "step": 1808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2637093957387356e-05,
+      "loss": 1.0348,
+      "step": 1809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2644079636744676e-05,
+      "loss": 1.3118,
+      "step": 1810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2651065316101992e-05,
+      "loss": 0.8936,
+      "step": 1811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.265805099545931e-05,
+      "loss": 1.0119,
+      "step": 1812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2665036674816626e-05,
+      "loss": 1.0861,
+      "step": 1813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2672022354173945e-05,
+      "loss": 1.0548,
+      "step": 1814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2679008033531261e-05,
+      "loss": 0.9096,
+      "step": 1815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2685993712888581e-05,
+      "loss": 0.995,
+      "step": 1816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2692979392245897e-05,
+      "loss": 1.1608,
+      "step": 1817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2699965071603215e-05,
+      "loss": 1.0267,
+      "step": 1818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2706950750960531e-05,
+      "loss": 1.0017,
+      "step": 1819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.271393643031785e-05,
+      "loss": 0.869,
+      "step": 1820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2720922109675167e-05,
+      "loss": 1.004,
+      "step": 1821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2727907789032486e-05,
+      "loss": 0.9387,
+      "step": 1822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2734893468389802e-05,
+      "loss": 1.0207,
+      "step": 1823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.274187914774712e-05,
+      "loss": 0.8973,
+      "step": 1824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2748864827104436e-05,
+      "loss": 1.3035,
+      "step": 1825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2755850506461756e-05,
+      "loss": 0.9747,
+      "step": 1826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2762836185819072e-05,
+      "loss": 0.9368,
+      "step": 1827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2769821865176391e-05,
+      "loss": 0.9042,
+      "step": 1828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2776807544533707e-05,
+      "loss": 1.051,
+      "step": 1829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2783793223891024e-05,
+      "loss": 0.9748,
+      "step": 1830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2790778903248341e-05,
+      "loss": 1.0821,
+      "step": 1831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.279776458260566e-05,
+      "loss": 0.8172,
+      "step": 1832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2804750261962977e-05,
+      "loss": 1.0098,
+      "step": 1833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2811735941320293e-05,
+      "loss": 0.953,
+      "step": 1834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2818721620677613e-05,
+      "loss": 0.9797,
+      "step": 1835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2825707300034929e-05,
+      "loss": 1.1826,
+      "step": 1836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2832692979392247e-05,
+      "loss": 1.0566,
+      "step": 1837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2839678658749564e-05,
+      "loss": 0.9422,
+      "step": 1838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2846664338106882e-05,
+      "loss": 0.7977,
+      "step": 1839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2853650017464198e-05,
+      "loss": 0.9024,
+      "step": 1840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2860635696821518e-05,
+      "loss": 1.1929,
+      "step": 1841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2867621376178834e-05,
+      "loss": 1.0073,
+      "step": 1842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2874607055536152e-05,
+      "loss": 0.8727,
+      "step": 1843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.288159273489347e-05,
+      "loss": 0.9757,
+      "step": 1844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2888578414250787e-05,
+      "loss": 1.083,
+      "step": 1845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2895564093608103e-05,
+      "loss": 1.1983,
+      "step": 1846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2902549772965423e-05,
+      "loss": 1.1351,
+      "step": 1847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2909535452322739e-05,
+      "loss": 1.0222,
+      "step": 1848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2916521131680057e-05,
+      "loss": 0.9382,
+      "step": 1849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2923506811037375e-05,
+      "loss": 0.8503,
+      "step": 1850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2930492490394693e-05,
+      "loss": 1.0273,
+      "step": 1851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2937478169752009e-05,
+      "loss": 0.9363,
+      "step": 1852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2944463849109328e-05,
+      "loss": 1.1164,
+      "step": 1853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2951449528466644e-05,
+      "loss": 0.9698,
+      "step": 1854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2958435207823962e-05,
+      "loss": 1.0903,
+      "step": 1855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.296542088718128e-05,
+      "loss": 1.1552,
+      "step": 1856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2972406566538598e-05,
+      "loss": 1.0091,
+      "step": 1857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2979392245895914e-05,
+      "loss": 0.8077,
+      "step": 1858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2986377925253233e-05,
+      "loss": 1.0259,
+      "step": 1859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.299336360461055e-05,
+      "loss": 0.9758,
+      "step": 1860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3000349283967867e-05,
+      "loss": 1.0396,
+      "step": 1861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3007334963325185e-05,
+      "loss": 0.9849,
+      "step": 1862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3014320642682501e-05,
+      "loss": 1.1596,
+      "step": 1863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3021306322039819e-05,
+      "loss": 1.0396,
+      "step": 1864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3028292001397135e-05,
+      "loss": 0.9915,
+      "step": 1865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3035277680754455e-05,
+      "loss": 0.9775,
+      "step": 1866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.304226336011177e-05,
+      "loss": 1.0054,
+      "step": 1867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.304924903946909e-05,
+      "loss": 0.9324,
+      "step": 1868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3056234718826406e-05,
+      "loss": 0.9972,
+      "step": 1869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3063220398183724e-05,
+      "loss": 1.2139,
+      "step": 1870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.307020607754104e-05,
+      "loss": 1.0882,
+      "step": 1871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.307719175689836e-05,
+      "loss": 1.2363,
+      "step": 1872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3084177436255676e-05,
+      "loss": 0.9993,
+      "step": 1873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3091163115612995e-05,
+      "loss": 0.993,
+      "step": 1874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3098148794970312e-05,
+      "loss": 1.0854,
+      "step": 1875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.310513447432763e-05,
+      "loss": 0.9977,
+      "step": 1876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3112120153684946e-05,
+      "loss": 1.0833,
+      "step": 1877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3119105833042265e-05,
+      "loss": 1.1266,
+      "step": 1878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3126091512399581e-05,
+      "loss": 0.9638,
+      "step": 1879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.31330771917569e-05,
+      "loss": 0.9284,
+      "step": 1880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3140062871114217e-05,
+      "loss": 1.1146,
+      "step": 1881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3147048550471535e-05,
+      "loss": 0.955,
+      "step": 1882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.315403422982885e-05,
+      "loss": 0.8549,
+      "step": 1883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.316101990918617e-05,
+      "loss": 0.9113,
+      "step": 1884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3168005588543486e-05,
+      "loss": 1.0524,
+      "step": 1885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3174991267900806e-05,
+      "loss": 1.0046,
+      "step": 1886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3181976947258122e-05,
+      "loss": 1.0358,
+      "step": 1887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.318896262661544e-05,
+      "loss": 0.9487,
+      "step": 1888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3195948305972756e-05,
+      "loss": 0.9971,
+      "step": 1889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3202933985330075e-05,
+      "loss": 1.1213,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3209919664687391e-05,
+      "loss": 1.0695,
+      "step": 1891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3216905344044711e-05,
+      "loss": 0.9852,
+      "step": 1892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3223891023402027e-05,
+      "loss": 1.0853,
+      "step": 1893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3230876702759345e-05,
+      "loss": 0.9879,
+      "step": 1894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3237862382116661e-05,
+      "loss": 1.0243,
+      "step": 1895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.324484806147398e-05,
+      "loss": 1.0324,
+      "step": 1896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3251833740831297e-05,
+      "loss": 0.7936,
+      "step": 1897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3258819420188613e-05,
+      "loss": 0.7824,
+      "step": 1898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3265805099545932e-05,
+      "loss": 1.1115,
+      "step": 1899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3272790778903248e-05,
+      "loss": 0.9193,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3279776458260566e-05,
+      "loss": 0.9463,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3286762137617884e-05,
+      "loss": 0.9829,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3293747816975202e-05,
+      "loss": 0.9284,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3300733496332518e-05,
+      "loss": 0.9444,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3307719175689837e-05,
+      "loss": 0.8519,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3314704855047154e-05,
+      "loss": 0.9223,
+      "step": 1906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3321690534404471e-05,
+      "loss": 0.9775,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.332867621376179e-05,
+      "loss": 0.9173,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3335661893119107e-05,
+      "loss": 1.0308,
+      "step": 1909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3342647572476423e-05,
+      "loss": 0.9626,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3349633251833743e-05,
+      "loss": 1.0402,
+      "step": 1911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3356618931191059e-05,
+      "loss": 1.0656,
+      "step": 1912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3363604610548378e-05,
+      "loss": 0.8932,
+      "step": 1913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3370590289905694e-05,
+      "loss": 0.9221,
+      "step": 1914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3377575969263012e-05,
+      "loss": 0.8429,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3384561648620328e-05,
+      "loss": 1.1848,
+      "step": 1916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3391547327977648e-05,
+      "loss": 1.0199,
+      "step": 1917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3398533007334964e-05,
+      "loss": 0.8549,
+      "step": 1918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3405518686692283e-05,
+      "loss": 0.8852,
+      "step": 1919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.34125043660496e-05,
+      "loss": 0.9444,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3419490045406917e-05,
+      "loss": 0.9475,
+      "step": 1921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3426475724764234e-05,
+      "loss": 0.9301,
+      "step": 1922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3433461404121553e-05,
+      "loss": 0.9435,
+      "step": 1923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3440447083478869e-05,
+      "loss": 0.9543,
+      "step": 1924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3447432762836189e-05,
+      "loss": 0.9523,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3454418442193505e-05,
+      "loss": 0.8611,
+      "step": 1926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3461404121550823e-05,
+      "loss": 0.9561,
+      "step": 1927
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3468389800908139e-05,
+      "loss": 1.2411,
+      "step": 1928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3475375480265458e-05,
+      "loss": 1.0304,
+      "step": 1929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3482361159622774e-05,
+      "loss": 0.9833,
+      "step": 1930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3489346838980094e-05,
+      "loss": 0.8832,
+      "step": 1931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.349633251833741e-05,
+      "loss": 0.873,
+      "step": 1932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3503318197694726e-05,
+      "loss": 1.0621,
+      "step": 1933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3510303877052044e-05,
+      "loss": 1.0545,
+      "step": 1934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3517289556409362e-05,
+      "loss": 0.9328,
+      "step": 1935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.352427523576668e-05,
+      "loss": 0.947,
+      "step": 1936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3531260915123996e-05,
+      "loss": 0.9207,
+      "step": 1937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3538246594481315e-05,
+      "loss": 1.0365,
+      "step": 1938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3545232273838631e-05,
+      "loss": 0.9035,
+      "step": 1939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3552217953195949e-05,
+      "loss": 1.0611,
+      "step": 1940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3559203632553267e-05,
+      "loss": 1.1557,
+      "step": 1941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3566189311910585e-05,
+      "loss": 0.9136,
+      "step": 1942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.35731749912679e-05,
+      "loss": 1.0559,
+      "step": 1943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.358016067062522e-05,
+      "loss": 1.0316,
+      "step": 1944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3587146349982536e-05,
+      "loss": 0.9129,
+      "step": 1945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3594132029339854e-05,
+      "loss": 1.1034,
+      "step": 1946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3601117708697172e-05,
+      "loss": 0.8594,
+      "step": 1947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.360810338805449e-05,
+      "loss": 0.9087,
+      "step": 1948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3615089067411806e-05,
+      "loss": 0.9184,
+      "step": 1949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3622074746769125e-05,
+      "loss": 1.078,
+      "step": 1950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3629060426126442e-05,
+      "loss": 0.8661,
+      "step": 1951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.363604610548376e-05,
+      "loss": 1.0172,
+      "step": 1952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3643031784841077e-05,
+      "loss": 1.032,
+      "step": 1953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3650017464198395e-05,
+      "loss": 0.9848,
+      "step": 1954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3657003143555711e-05,
+      "loss": 0.9812,
+      "step": 1955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.366398882291303e-05,
+      "loss": 0.8548,
+      "step": 1956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3670974502270347e-05,
+      "loss": 0.9317,
+      "step": 1957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3677960181627665e-05,
+      "loss": 0.8863,
+      "step": 1958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3684945860984982e-05,
+      "loss": 1.0064,
+      "step": 1959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.36919315403423e-05,
+      "loss": 1.0571,
+      "step": 1960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3698917219699616e-05,
+      "loss": 0.9794,
+      "step": 1961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3705902899056936e-05,
+      "loss": 0.9816,
+      "step": 1962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3712888578414252e-05,
+      "loss": 1.0535,
+      "step": 1963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.371987425777157e-05,
+      "loss": 0.8491,
+      "step": 1964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3726859937128888e-05,
+      "loss": 0.9091,
+      "step": 1965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3733845616486205e-05,
+      "loss": 0.8897,
+      "step": 1966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3740831295843521e-05,
+      "loss": 1.0481,
+      "step": 1967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3747816975200838e-05,
+      "loss": 1.0652,
+      "step": 1968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3754802654558157e-05,
+      "loss": 0.8617,
+      "step": 1969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3761788333915473e-05,
+      "loss": 1.0335,
+      "step": 1970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3768774013272793e-05,
+      "loss": 0.9234,
+      "step": 1971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3775759692630109e-05,
+      "loss": 0.9496,
+      "step": 1972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3782745371987427e-05,
+      "loss": 1.1684,
+      "step": 1973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3789731051344743e-05,
+      "loss": 1.1114,
+      "step": 1974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3796716730702062e-05,
+      "loss": 1.0674,
+      "step": 1975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3803702410059378e-05,
+      "loss": 0.8891,
+      "step": 1976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3810688089416698e-05,
+      "loss": 0.9967,
+      "step": 1977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3817673768774014e-05,
+      "loss": 0.9303,
+      "step": 1978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3824659448131332e-05,
+      "loss": 0.821,
+      "step": 1979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3831645127488648e-05,
+      "loss": 1.0444,
+      "step": 1980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3838630806845967e-05,
+      "loss": 1.0033,
+      "step": 1981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3845616486203284e-05,
+      "loss": 0.989,
+      "step": 1982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3852602165560603e-05,
+      "loss": 1.0029,
+      "step": 1983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.385958784491792e-05,
+      "loss": 0.9828,
+      "step": 1984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3866573524275237e-05,
+      "loss": 0.9734,
+      "step": 1985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3873559203632553e-05,
+      "loss": 1.0154,
+      "step": 1986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3880544882989873e-05,
+      "loss": 1.0141,
+      "step": 1987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3887530562347189e-05,
+      "loss": 0.8739,
+      "step": 1988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3894516241704508e-05,
+      "loss": 0.9612,
+      "step": 1989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3901501921061824e-05,
+      "loss": 1.0475,
+      "step": 1990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3908487600419142e-05,
+      "loss": 0.881,
+      "step": 1991
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3915473279776458e-05,
+      "loss": 1.0457,
+      "step": 1992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3922458959133778e-05,
+      "loss": 0.8257,
+      "step": 1993
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3929444638491094e-05,
+      "loss": 0.9615,
+      "step": 1994
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3936430317848413e-05,
+      "loss": 0.8876,
+      "step": 1995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.394341599720573e-05,
+      "loss": 0.9076,
+      "step": 1996
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3950401676563047e-05,
+      "loss": 1.0192,
+      "step": 1997
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3957387355920364e-05,
+      "loss": 1.0913,
+      "step": 1998
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3964373035277683e-05,
+      "loss": 1.0193,
+      "step": 1999
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3971358714634999e-05,
+      "loss": 1.0983,
+      "step": 2000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3978344393992315e-05,
+      "loss": 1.0265,
+      "step": 2001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3985330073349635e-05,
+      "loss": 0.9277,
+      "step": 2002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3992315752706951e-05,
+      "loss": 0.9228,
+      "step": 2003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3999301432064269e-05,
+      "loss": 0.8402,
+      "step": 2004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4006287111421586e-05,
+      "loss": 0.8931,
+      "step": 2005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4013272790778904e-05,
+      "loss": 0.975,
+      "step": 2006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.402025847013622e-05,
+      "loss": 1.0674,
+      "step": 2007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.402724414949354e-05,
+      "loss": 0.9627,
+      "step": 2008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4034229828850856e-05,
+      "loss": 1.0022,
+      "step": 2009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4041215508208174e-05,
+      "loss": 0.9202,
+      "step": 2010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4048201187565492e-05,
+      "loss": 0.9243,
+      "step": 2011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.405518686692281e-05,
+      "loss": 0.9814,
+      "step": 2012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4062172546280126e-05,
+      "loss": 1.1024,
+      "step": 2013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4069158225637445e-05,
+      "loss": 0.9372,
+      "step": 2014
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4076143904994761e-05,
+      "loss": 0.9763,
+      "step": 2015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4083129584352079e-05,
+      "loss": 0.8389,
+      "step": 2016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4090115263709397e-05,
+      "loss": 1.0918,
+      "step": 2017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4097100943066715e-05,
+      "loss": 0.9607,
+      "step": 2018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.410408662242403e-05,
+      "loss": 0.8761,
+      "step": 2019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.411107230178135e-05,
+      "loss": 1.0537,
+      "step": 2020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4118057981138666e-05,
+      "loss": 0.96,
+      "step": 2021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4125043660495986e-05,
+      "loss": 0.951,
+      "step": 2022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4132029339853302e-05,
+      "loss": 0.838,
+      "step": 2023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.413901501921062e-05,
+      "loss": 0.8953,
+      "step": 2024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4146000698567936e-05,
+      "loss": 0.9934,
+      "step": 2025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4152986377925255e-05,
+      "loss": 1.0746,
+      "step": 2026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4159972057282572e-05,
+      "loss": 0.9639,
+      "step": 2027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4166957736639891e-05,
+      "loss": 0.8557,
+      "step": 2028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4173943415997207e-05,
+      "loss": 0.8749,
+      "step": 2029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4180929095354525e-05,
+      "loss": 1.0099,
+      "step": 2030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4187914774711841e-05,
+      "loss": 0.8209,
+      "step": 2031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.419490045406916e-05,
+      "loss": 0.9577,
+      "step": 2032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4201886133426477e-05,
+      "loss": 0.8891,
+      "step": 2033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4208871812783796e-05,
+      "loss": 0.9765,
+      "step": 2034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4215857492141112e-05,
+      "loss": 0.8047,
+      "step": 2035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4222843171498429e-05,
+      "loss": 0.9745,
+      "step": 2036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4229828850855746e-05,
+      "loss": 1.0406,
+      "step": 2037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4236814530213064e-05,
+      "loss": 1.0569,
+      "step": 2038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4243800209570382e-05,
+      "loss": 0.95,
+      "step": 2039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4250785888927698e-05,
+      "loss": 0.9289,
+      "step": 2040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4257771568285018e-05,
+      "loss": 0.8519,
+      "step": 2041
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4264757247642334e-05,
+      "loss": 1.0338,
+      "step": 2042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4271742926999652e-05,
+      "loss": 0.9848,
+      "step": 2043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.427872860635697e-05,
+      "loss": 0.9995,
+      "step": 2044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.9505,
+      "step": 2045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4292699965071603e-05,
+      "loss": 1.0104,
+      "step": 2046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4299685644428923e-05,
+      "loss": 0.9403,
+      "step": 2047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4306671323786239e-05,
+      "loss": 0.9265,
+      "step": 2048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4313657003143557e-05,
+      "loss": 0.9967,
+      "step": 2049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4320642682500874e-05,
+      "loss": 0.9785,
+      "step": 2050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4327628361858192e-05,
+      "loss": 0.9446,
+      "step": 2051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4334614041215508e-05,
+      "loss": 1.085,
+      "step": 2052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4341599720572828e-05,
+      "loss": 0.8882,
+      "step": 2053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4348585399930144e-05,
+      "loss": 0.9834,
+      "step": 2054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4355571079287462e-05,
+      "loss": 0.8775,
+      "step": 2055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.436255675864478e-05,
+      "loss": 0.7659,
+      "step": 2056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4369542438002097e-05,
+      "loss": 0.997,
+      "step": 2057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4376528117359414e-05,
+      "loss": 1.0098,
+      "step": 2058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4383513796716733e-05,
+      "loss": 1.0593,
+      "step": 2059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.439049947607405e-05,
+      "loss": 0.8014,
+      "step": 2060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4397485155431367e-05,
+      "loss": 1.111,
+      "step": 2061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4404470834788685e-05,
+      "loss": 0.8626,
+      "step": 2062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4411456514146003e-05,
+      "loss": 0.9615,
+      "step": 2063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4418442193503319e-05,
+      "loss": 0.9742,
+      "step": 2064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4425427872860638e-05,
+      "loss": 0.8842,
+      "step": 2065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4432413552217954e-05,
+      "loss": 1.0117,
+      "step": 2066
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4439399231575272e-05,
+      "loss": 0.9837,
+      "step": 2067
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.444638491093259e-05,
+      "loss": 0.9929,
+      "step": 2068
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4453370590289908e-05,
+      "loss": 0.7972,
+      "step": 2069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4460356269647224e-05,
+      "loss": 0.8489,
+      "step": 2070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.446734194900454e-05,
+      "loss": 1.1843,
+      "step": 2071
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.447432762836186e-05,
+      "loss": 0.7832,
+      "step": 2072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4481313307719176e-05,
+      "loss": 1.0657,
+      "step": 2073
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4488298987076495e-05,
+      "loss": 0.7748,
+      "step": 2074
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4495284666433811e-05,
+      "loss": 1.0645,
+      "step": 2075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4502270345791129e-05,
+      "loss": 0.9703,
+      "step": 2076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4509256025148445e-05,
+      "loss": 0.9872,
+      "step": 2077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4516241704505765e-05,
+      "loss": 0.8435,
+      "step": 2078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4523227383863081e-05,
+      "loss": 0.9953,
+      "step": 2079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.45302130632204e-05,
+      "loss": 0.9053,
+      "step": 2080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4537198742577717e-05,
+      "loss": 1.0324,
+      "step": 2081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4544184421935034e-05,
+      "loss": 0.9468,
+      "step": 2082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.455117010129235e-05,
+      "loss": 1.0307,
+      "step": 2083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.455815578064967e-05,
+      "loss": 0.8789,
+      "step": 2084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4565141460006986e-05,
+      "loss": 0.8621,
+      "step": 2085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4572127139364306e-05,
+      "loss": 0.8497,
+      "step": 2086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4579112818721622e-05,
+      "loss": 1.0751,
+      "step": 2087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.458609849807894e-05,
+      "loss": 1.0203,
+      "step": 2088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4593084177436256e-05,
+      "loss": 0.9348,
+      "step": 2089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4600069856793575e-05,
+      "loss": 0.833,
+      "step": 2090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4607055536150891e-05,
+      "loss": 1.0112,
+      "step": 2091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.461404121550821e-05,
+      "loss": 0.865,
+      "step": 2092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4621026894865527e-05,
+      "loss": 0.8346,
+      "step": 2093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4628012574222845e-05,
+      "loss": 0.9225,
+      "step": 2094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.463499825358016e-05,
+      "loss": 1.0381,
+      "step": 2095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.464198393293748e-05,
+      "loss": 0.8128,
+      "step": 2096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4648969612294796e-05,
+      "loss": 1.0151,
+      "step": 2097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4655955291652116e-05,
+      "loss": 0.927,
+      "step": 2098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4662940971009432e-05,
+      "loss": 0.9909,
+      "step": 2099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.466992665036675e-05,
+      "loss": 1.0237,
+      "step": 2100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4676912329724066e-05,
+      "loss": 0.9664,
+      "step": 2101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4683898009081385e-05,
+      "loss": 1.0306,
+      "step": 2102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4690883688438702e-05,
+      "loss": 1.0162,
+      "step": 2103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4697869367796021e-05,
+      "loss": 0.896,
+      "step": 2104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4704855047153337e-05,
+      "loss": 0.8544,
+      "step": 2105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4711840726510653e-05,
+      "loss": 1.0196,
+      "step": 2106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4718826405867971e-05,
+      "loss": 0.9903,
+      "step": 2107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4725812085225289e-05,
+      "loss": 1.0699,
+      "step": 2108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4732797764582607e-05,
+      "loss": 0.7811,
+      "step": 2109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4739783443939923e-05,
+      "loss": 0.9277,
+      "step": 2110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4746769123297242e-05,
+      "loss": 0.9566,
+      "step": 2111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4753754802654559e-05,
+      "loss": 0.9598,
+      "step": 2112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4760740482011876e-05,
+      "loss": 0.9371,
+      "step": 2113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4767726161369194e-05,
+      "loss": 0.9702,
+      "step": 2114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4774711840726512e-05,
+      "loss": 0.9636,
+      "step": 2115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4781697520083828e-05,
+      "loss": 0.9614,
+      "step": 2116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4788683199441148e-05,
+      "loss": 0.9261,
+      "step": 2117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4795668878798464e-05,
+      "loss": 1.0855,
+      "step": 2118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4802654558155782e-05,
+      "loss": 0.6485,
+      "step": 2119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.48096402375131e-05,
+      "loss": 0.9691,
+      "step": 2120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4816625916870417e-05,
+      "loss": 1.1639,
+      "step": 2121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4823611596227733e-05,
+      "loss": 1.0318,
+      "step": 2122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4830597275585053e-05,
+      "loss": 1.1302,
+      "step": 2123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4837582954942369e-05,
+      "loss": 0.8094,
+      "step": 2124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4844568634299687e-05,
+      "loss": 1.0205,
+      "step": 2125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4851554313657004e-05,
+      "loss": 0.9724,
+      "step": 2126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4858539993014322e-05,
+      "loss": 0.829,
+      "step": 2127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4865525672371638e-05,
+      "loss": 1.0795,
+      "step": 2128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4872511351728958e-05,
+      "loss": 0.8878,
+      "step": 2129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4879497031086274e-05,
+      "loss": 1.0205,
+      "step": 2130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4886482710443594e-05,
+      "loss": 0.9462,
+      "step": 2131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.489346838980091e-05,
+      "loss": 1.1714,
+      "step": 2132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4900454069158227e-05,
+      "loss": 1.0022,
+      "step": 2133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4907439748515544e-05,
+      "loss": 0.9391,
+      "step": 2134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4914425427872863e-05,
+      "loss": 0.9375,
+      "step": 2135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.492141110723018e-05,
+      "loss": 0.964,
+      "step": 2136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4928396786587499e-05,
+      "loss": 0.8691,
+      "step": 2137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4935382465944815e-05,
+      "loss": 1.221,
+      "step": 2138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4942368145302131e-05,
+      "loss": 0.945,
+      "step": 2139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4949353824659449e-05,
+      "loss": 1.0006,
+      "step": 2140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4956339504016767e-05,
+      "loss": 1.0633,
+      "step": 2141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4963325183374084e-05,
+      "loss": 1.1848,
+      "step": 2142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.49703108627314e-05,
+      "loss": 0.9888,
+      "step": 2143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.497729654208872e-05,
+      "loss": 0.8024,
+      "step": 2144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4984282221446036e-05,
+      "loss": 1.0859,
+      "step": 2145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4991267900803354e-05,
+      "loss": 0.8941,
+      "step": 2146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4998253580160672e-05,
+      "loss": 0.9817,
+      "step": 2147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.500523925951799e-05,
+      "loss": 0.9162,
+      "step": 2148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5012224938875306e-05,
+      "loss": 0.9803,
+      "step": 2149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5019210618232625e-05,
+      "loss": 0.8399,
+      "step": 2150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5026196297589941e-05,
+      "loss": 0.9281,
+      "step": 2151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5033181976947259e-05,
+      "loss": 0.9786,
+      "step": 2152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5040167656304577e-05,
+      "loss": 0.8936,
+      "step": 2153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5047153335661895e-05,
+      "loss": 1.2731,
+      "step": 2154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5054139015019211e-05,
+      "loss": 0.753,
+      "step": 2155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.506112469437653e-05,
+      "loss": 0.928,
+      "step": 2156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5068110373733847e-05,
+      "loss": 0.908,
+      "step": 2157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5075096053091164e-05,
+      "loss": 0.8713,
+      "step": 2158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5082081732448482e-05,
+      "loss": 0.9742,
+      "step": 2159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.50890674118058e-05,
+      "loss": 0.8687,
+      "step": 2160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5096053091163116e-05,
+      "loss": 1.1819,
+      "step": 2161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5103038770520436e-05,
+      "loss": 0.9923,
+      "step": 2162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5110024449877752e-05,
+      "loss": 0.9334,
+      "step": 2163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.511701012923507e-05,
+      "loss": 0.9862,
+      "step": 2164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5123995808592387e-05,
+      "loss": 0.8294,
+      "step": 2165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5130981487949705e-05,
+      "loss": 1.2494,
+      "step": 2166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5137967167307021e-05,
+      "loss": 1.2011,
+      "step": 2167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.514495284666434e-05,
+      "loss": 0.9264,
+      "step": 2168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5151938526021657e-05,
+      "loss": 0.9094,
+      "step": 2169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5158924205378975e-05,
+      "loss": 1.0655,
+      "step": 2170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5165909884736292e-05,
+      "loss": 0.7167,
+      "step": 2171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.517289556409361e-05,
+      "loss": 0.9546,
+      "step": 2172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5179881243450926e-05,
+      "loss": 0.9803,
+      "step": 2173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5186866922808243e-05,
+      "loss": 1.0759,
+      "step": 2174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5193852602165562e-05,
+      "loss": 0.9043,
+      "step": 2175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5200838281522878e-05,
+      "loss": 1.0284,
+      "step": 2176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5207823960880198e-05,
+      "loss": 1.1472,
+      "step": 2177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5214809640237514e-05,
+      "loss": 0.9367,
+      "step": 2178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5221795319594832e-05,
+      "loss": 0.8852,
+      "step": 2179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5228780998952148e-05,
+      "loss": 1.0421,
+      "step": 2180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5235766678309467e-05,
+      "loss": 0.9908,
+      "step": 2181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5242752357666783e-05,
+      "loss": 0.962,
+      "step": 2182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5249738037024103e-05,
+      "loss": 1.0189,
+      "step": 2183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5256723716381419e-05,
+      "loss": 0.995,
+      "step": 2184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.526370939573874e-05,
+      "loss": 1.0818,
+      "step": 2185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5270695075096053e-05,
+      "loss": 0.8834,
+      "step": 2186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.527768075445337e-05,
+      "loss": 0.9986,
+      "step": 2187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.528466643381069e-05,
+      "loss": 0.8599,
+      "step": 2188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5291652113168006e-05,
+      "loss": 0.9742,
+      "step": 2189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5298637792525324e-05,
+      "loss": 0.865,
+      "step": 2190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5305623471882642e-05,
+      "loss": 0.9881,
+      "step": 2191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.531260915123996e-05,
+      "loss": 1.1727,
+      "step": 2192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5319594830597278e-05,
+      "loss": 0.9425,
+      "step": 2193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5326580509954595e-05,
+      "loss": 0.9006,
+      "step": 2194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5333566189311913e-05,
+      "loss": 0.8593,
+      "step": 2195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5340551868669228e-05,
+      "loss": 1.0084,
+      "step": 2196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.534753754802655e-05,
+      "loss": 0.9964,
+      "step": 2197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5354523227383863e-05,
+      "loss": 1.0237,
+      "step": 2198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.536150890674118e-05,
+      "loss": 0.9556,
+      "step": 2199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.53684945860985e-05,
+      "loss": 0.9249,
+      "step": 2200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5375480265455817e-05,
+      "loss": 0.8166,
+      "step": 2201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5382465944813135e-05,
+      "loss": 0.9853,
+      "step": 2202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5389451624170452e-05,
+      "loss": 1.0667,
+      "step": 2203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.539643730352777e-05,
+      "loss": 0.9498,
+      "step": 2204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5403422982885088e-05,
+      "loss": 0.9548,
+      "step": 2205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5410408662242406e-05,
+      "loss": 1.0182,
+      "step": 2206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5417394341599724e-05,
+      "loss": 0.8497,
+      "step": 2207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5424380020957038e-05,
+      "loss": 0.9054,
+      "step": 2208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5431365700314356e-05,
+      "loss": 0.8966,
+      "step": 2209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5438351379671674e-05,
+      "loss": 0.9924,
+      "step": 2210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.544533705902899e-05,
+      "loss": 0.9223,
+      "step": 2211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.545232273838631e-05,
+      "loss": 1.1439,
+      "step": 2212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5459308417743627e-05,
+      "loss": 0.9012,
+      "step": 2213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5466294097100945e-05,
+      "loss": 1.0702,
+      "step": 2214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.547327977645826e-05,
+      "loss": 0.8691,
+      "step": 2215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.548026545581558e-05,
+      "loss": 0.8996,
+      "step": 2216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5487251135172895e-05,
+      "loss": 0.8553,
+      "step": 2217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5494236814530216e-05,
+      "loss": 0.8692,
+      "step": 2218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.550122249388753e-05,
+      "loss": 0.8205,
+      "step": 2219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.550820817324485e-05,
+      "loss": 0.8988,
+      "step": 2220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5515193852602166e-05,
+      "loss": 1.0549,
+      "step": 2221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5522179531959484e-05,
+      "loss": 0.907,
+      "step": 2222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5529165211316802e-05,
+      "loss": 0.9299,
+      "step": 2223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.553615089067412e-05,
+      "loss": 1.0366,
+      "step": 2224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5543136570031437e-05,
+      "loss": 0.79,
+      "step": 2225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5550122249388755e-05,
+      "loss": 1.095,
+      "step": 2226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5557107928746073e-05,
+      "loss": 0.9595,
+      "step": 2227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.556409360810339e-05,
+      "loss": 0.9742,
+      "step": 2228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5571079287460705e-05,
+      "loss": 0.9393,
+      "step": 2229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5578064966818026e-05,
+      "loss": 0.8875,
+      "step": 2230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.558505064617534e-05,
+      "loss": 1.1092,
+      "step": 2231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.559203632553266e-05,
+      "loss": 1.0302,
+      "step": 2232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5599022004889977e-05,
+      "loss": 0.924,
+      "step": 2233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5606007684247294e-05,
+      "loss": 0.8876,
+      "step": 2234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5612993363604612e-05,
+      "loss": 0.8774,
+      "step": 2235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.561997904296193e-05,
+      "loss": 0.8776,
+      "step": 2236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5626964722319248e-05,
+      "loss": 1.0123,
+      "step": 2237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5633950401676566e-05,
+      "loss": 0.8602,
+      "step": 2238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5640936081033883e-05,
+      "loss": 1.1388,
+      "step": 2239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.56479217603912e-05,
+      "loss": 1.0284,
+      "step": 2240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5654907439748516e-05,
+      "loss": 0.9937,
+      "step": 2241
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5661893119105837e-05,
+      "loss": 0.8789,
+      "step": 2242
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.566887879846315e-05,
+      "loss": 0.9876,
+      "step": 2243
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.567586447782047e-05,
+      "loss": 0.8634,
+      "step": 2244
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5682850157177787e-05,
+      "loss": 1.0562,
+      "step": 2245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5689835836535105e-05,
+      "loss": 0.8561,
+      "step": 2246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5696821515892422e-05,
+      "loss": 0.9984,
+      "step": 2247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5703807195249737e-05,
+      "loss": 0.9228,
+      "step": 2248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5710792874607058e-05,
+      "loss": 0.9794,
+      "step": 2249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5717778553964373e-05,
+      "loss": 0.8837,
+      "step": 2250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5724764233321694e-05,
+      "loss": 0.858,
+      "step": 2251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5731749912679008e-05,
+      "loss": 0.7955,
+      "step": 2252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5738735592036326e-05,
+      "loss": 0.9885,
+      "step": 2253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5745721271393644e-05,
+      "loss": 0.9986,
+      "step": 2254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.575270695075096e-05,
+      "loss": 1.0084,
+      "step": 2255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.575969263010828e-05,
+      "loss": 0.8876,
+      "step": 2256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5766678309465597e-05,
+      "loss": 0.9354,
+      "step": 2257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5773663988822915e-05,
+      "loss": 0.8953,
+      "step": 2258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5780649668180233e-05,
+      "loss": 0.864,
+      "step": 2259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5787635347537547e-05,
+      "loss": 0.8305,
+      "step": 2260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.579462102689487e-05,
+      "loss": 1.0016,
+      "step": 2261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5801606706252183e-05,
+      "loss": 1.0514,
+      "step": 2262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5808592385609504e-05,
+      "loss": 0.9111,
+      "step": 2263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.581557806496682e-05,
+      "loss": 1.0234,
+      "step": 2264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5822563744324136e-05,
+      "loss": 0.7444,
+      "step": 2265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5829549423681454e-05,
+      "loss": 0.996,
+      "step": 2266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5836535103038772e-05,
+      "loss": 0.7225,
+      "step": 2267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.584352078239609e-05,
+      "loss": 0.9417,
+      "step": 2268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5850506461753408e-05,
+      "loss": 0.9856,
+      "step": 2269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5857492141110725e-05,
+      "loss": 0.8966,
+      "step": 2270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5864477820468043e-05,
+      "loss": 1.091,
+      "step": 2271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5871463499825358e-05,
+      "loss": 0.9595,
+      "step": 2272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.587844917918268e-05,
+      "loss": 0.9844,
+      "step": 2273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5885434858539993e-05,
+      "loss": 0.8932,
+      "step": 2274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5892420537897314e-05,
+      "loss": 0.9115,
+      "step": 2275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.589940621725463e-05,
+      "loss": 0.9623,
+      "step": 2276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5906391896611947e-05,
+      "loss": 0.9535,
+      "step": 2277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5913377575969265e-05,
+      "loss": 1.0645,
+      "step": 2278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5920363255326582e-05,
+      "loss": 0.9736,
+      "step": 2279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.59273489346839e-05,
+      "loss": 0.8295,
+      "step": 2280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5934334614041215e-05,
+      "loss": 0.8928,
+      "step": 2281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5941320293398536e-05,
+      "loss": 0.9002,
+      "step": 2282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.594830597275585e-05,
+      "loss": 0.9294,
+      "step": 2283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5955291652113168e-05,
+      "loss": 0.8048,
+      "step": 2284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5962277331470486e-05,
+      "loss": 1.0106,
+      "step": 2285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5969263010827804e-05,
+      "loss": 0.9236,
+      "step": 2286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.597624869018512e-05,
+      "loss": 0.8339,
+      "step": 2287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.598323436954244e-05,
+      "loss": 1.0701,
+      "step": 2288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5990220048899757e-05,
+      "loss": 0.8615,
+      "step": 2289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5997205728257075e-05,
+      "loss": 0.9269,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6004191407614393e-05,
+      "loss": 0.9738,
+      "step": 2291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.601117708697171e-05,
+      "loss": 0.8901,
+      "step": 2292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6018162766329025e-05,
+      "loss": 0.9487,
+      "step": 2293
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6025148445686346e-05,
+      "loss": 0.8033,
+      "step": 2294
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.603213412504366e-05,
+      "loss": 1.0437,
+      "step": 2295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.603911980440098e-05,
+      "loss": 0.9878,
+      "step": 2296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6046105483758296e-05,
+      "loss": 0.9582,
+      "step": 2297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6053091163115614e-05,
+      "loss": 1.1453,
+      "step": 2298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6060076842472932e-05,
+      "loss": 1.0896,
+      "step": 2299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.606706252183025e-05,
+      "loss": 0.7617,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6074048201187567e-05,
+      "loss": 0.8805,
+      "step": 2301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6081033880544885e-05,
+      "loss": 0.8936,
+      "step": 2302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6088019559902203e-05,
+      "loss": 0.9535,
+      "step": 2303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.609500523925952e-05,
+      "loss": 0.7445,
+      "step": 2304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6101990918616835e-05,
+      "loss": 0.9752,
+      "step": 2305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6108976597974156e-05,
+      "loss": 0.7766,
+      "step": 2306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.611596227733147e-05,
+      "loss": 0.8841,
+      "step": 2307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.612294795668879e-05,
+      "loss": 0.8658,
+      "step": 2308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6129933636046107e-05,
+      "loss": 0.7859,
+      "step": 2309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6136919315403424e-05,
+      "loss": 0.7821,
+      "step": 2310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6143904994760742e-05,
+      "loss": 0.9063,
+      "step": 2311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6150890674118057e-05,
+      "loss": 0.8463,
+      "step": 2312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6157876353475378e-05,
+      "loss": 0.9455,
+      "step": 2313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6164862032832692e-05,
+      "loss": 1.1783,
+      "step": 2314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6171847712190013e-05,
+      "loss": 0.9386,
+      "step": 2315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6178833391547328e-05,
+      "loss": 0.8953,
+      "step": 2316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6185819070904646e-05,
+      "loss": 0.9498,
+      "step": 2317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6192804750261963e-05,
+      "loss": 1.0543,
+      "step": 2318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.619979042961928e-05,
+      "loss": 0.832,
+      "step": 2319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.62067761089766e-05,
+      "loss": 0.9764,
+      "step": 2320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6213761788333917e-05,
+      "loss": 0.9897,
+      "step": 2321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6220747467691235e-05,
+      "loss": 0.7535,
+      "step": 2322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6227733147048552e-05,
+      "loss": 0.96,
+      "step": 2323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6234718826405867e-05,
+      "loss": 1.1963,
+      "step": 2324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6241704505763188e-05,
+      "loss": 0.8175,
+      "step": 2325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6248690185120503e-05,
+      "loss": 1.0104,
+      "step": 2326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6255675864477824e-05,
+      "loss": 0.8113,
+      "step": 2327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6262661543835138e-05,
+      "loss": 0.9437,
+      "step": 2328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6269647223192456e-05,
+      "loss": 0.971,
+      "step": 2329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6276632902549774e-05,
+      "loss": 0.9546,
+      "step": 2330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628361858190709e-05,
+      "loss": 0.9302,
+      "step": 2331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629060426126441e-05,
+      "loss": 0.9774,
+      "step": 2332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6297589940621727e-05,
+      "loss": 1.02,
+      "step": 2333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6304575619979045e-05,
+      "loss": 0.7554,
+      "step": 2334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6311561299336363e-05,
+      "loss": 1.2909,
+      "step": 2335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.631854697869368e-05,
+      "loss": 1.105,
+      "step": 2336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6325532658051e-05,
+      "loss": 0.8478,
+      "step": 2337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6332518337408313e-05,
+      "loss": 0.8846,
+      "step": 2338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6339504016765634e-05,
+      "loss": 0.8922,
+      "step": 2339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634648969612295e-05,
+      "loss": 0.9068,
+      "step": 2340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6353475375480266e-05,
+      "loss": 0.8672,
+      "step": 2341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6360461054837584e-05,
+      "loss": 0.989,
+      "step": 2342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6367446734194902e-05,
+      "loss": 0.8175,
+      "step": 2343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.637443241355222e-05,
+      "loss": 1.1754,
+      "step": 2344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6381418092909538e-05,
+      "loss": 0.8701,
+      "step": 2345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6388403772266855e-05,
+      "loss": 1.1464,
+      "step": 2346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.639538945162417e-05,
+      "loss": 0.9169,
+      "step": 2347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.640237513098149e-05,
+      "loss": 0.8902,
+      "step": 2348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6409360810338805e-05,
+      "loss": 1.0353,
+      "step": 2349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6416346489696123e-05,
+      "loss": 0.8424,
+      "step": 2350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.642333216905344e-05,
+      "loss": 0.9581,
+      "step": 2351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.643031784841076e-05,
+      "loss": 1.0047,
+      "step": 2352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6437303527768077e-05,
+      "loss": 0.821,
+      "step": 2353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6444289207125395e-05,
+      "loss": 1.0431,
+      "step": 2354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6451274886482712e-05,
+      "loss": 0.8548,
+      "step": 2355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.645826056584003e-05,
+      "loss": 1.0014,
+      "step": 2356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6465246245197345e-05,
+      "loss": 0.8705,
+      "step": 2357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6472231924554666e-05,
+      "loss": 0.9084,
+      "step": 2358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.647921760391198e-05,
+      "loss": 0.9567,
+      "step": 2359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.64862032832693e-05,
+      "loss": 0.8895,
+      "step": 2360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6493188962626616e-05,
+      "loss": 1.1381,
+      "step": 2361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6500174641983934e-05,
+      "loss": 0.8899,
+      "step": 2362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.650716032134125e-05,
+      "loss": 0.8536,
+      "step": 2363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.651414600069857e-05,
+      "loss": 0.931,
+      "step": 2364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6521131680055887e-05,
+      "loss": 1.1645,
+      "step": 2365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6528117359413205e-05,
+      "loss": 0.8826,
+      "step": 2366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6535103038770523e-05,
+      "loss": 1.1107,
+      "step": 2367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.654208871812784e-05,
+      "loss": 0.8564,
+      "step": 2368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6549074397485155e-05,
+      "loss": 1.0218,
+      "step": 2369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6556060076842476e-05,
+      "loss": 0.8742,
+      "step": 2370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.656304575619979e-05,
+      "loss": 0.8571,
+      "step": 2371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6570031435557112e-05,
+      "loss": 1.1475,
+      "step": 2372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6577017114914426e-05,
+      "loss": 0.981,
+      "step": 2373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6584002794271744e-05,
+      "loss": 1.045,
+      "step": 2374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6590988473629062e-05,
+      "loss": 0.7213,
+      "step": 2375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.659797415298638e-05,
+      "loss": 0.8116,
+      "step": 2376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6604959832343697e-05,
+      "loss": 1.0389,
+      "step": 2377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6611945511701015e-05,
+      "loss": 1.0246,
+      "step": 2378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6618931191058333e-05,
+      "loss": 0.8773,
+      "step": 2379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.662591687041565e-05,
+      "loss": 0.771,
+      "step": 2380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6632902549772965e-05,
+      "loss": 0.9322,
+      "step": 2381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6639888229130283e-05,
+      "loss": 0.7962,
+      "step": 2382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.66468739084876e-05,
+      "loss": 1.0017,
+      "step": 2383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.665385958784492e-05,
+      "loss": 0.9395,
+      "step": 2384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6660845267202237e-05,
+      "loss": 0.9837,
+      "step": 2385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6667830946559554e-05,
+      "loss": 0.9656,
+      "step": 2386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6674816625916872e-05,
+      "loss": 0.9803,
+      "step": 2387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.668180230527419e-05,
+      "loss": 0.8298,
+      "step": 2388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6688787984631508e-05,
+      "loss": 0.8737,
+      "step": 2389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6695773663988822e-05,
+      "loss": 0.9226,
+      "step": 2390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6702759343346143e-05,
+      "loss": 0.9589,
+      "step": 2391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6709745022703458e-05,
+      "loss": 0.9001,
+      "step": 2392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6716730702060776e-05,
+      "loss": 0.7893,
+      "step": 2393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6723716381418093e-05,
+      "loss": 1.0642,
+      "step": 2394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.673070206077541e-05,
+      "loss": 0.9666,
+      "step": 2395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.673768774013273e-05,
+      "loss": 1.1376,
+      "step": 2396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6744673419490047e-05,
+      "loss": 1.1848,
+      "step": 2397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6751659098847365e-05,
+      "loss": 1.0479,
+      "step": 2398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6758644778204683e-05,
+      "loss": 1.0747,
+      "step": 2399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6765630457562e-05,
+      "loss": 0.907,
+      "step": 2400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6772616136919318e-05,
+      "loss": 0.9131,
+      "step": 2401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6779601816276633e-05,
+      "loss": 0.7562,
+      "step": 2402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6786587495633954e-05,
+      "loss": 0.8121,
+      "step": 2403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6793573174991268e-05,
+      "loss": 0.9386,
+      "step": 2404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6800558854348586e-05,
+      "loss": 1.0831,
+      "step": 2405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6807544533705904e-05,
+      "loss": 0.8754,
+      "step": 2406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.681453021306322e-05,
+      "loss": 0.992,
+      "step": 2407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.682151589242054e-05,
+      "loss": 0.9037,
+      "step": 2408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6828501571777857e-05,
+      "loss": 0.8999,
+      "step": 2409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6835487251135175e-05,
+      "loss": 0.8136,
+      "step": 2410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6842472930492493e-05,
+      "loss": 0.8533,
+      "step": 2411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.684945860984981e-05,
+      "loss": 1.0555,
+      "step": 2412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.685644428920713e-05,
+      "loss": 0.9132,
+      "step": 2413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6863429968564443e-05,
+      "loss": 0.9074,
+      "step": 2414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.687041564792176e-05,
+      "loss": 0.7994,
+      "step": 2415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.687740132727908e-05,
+      "loss": 0.9312,
+      "step": 2416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6884387006636396e-05,
+      "loss": 0.6584,
+      "step": 2417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6891372685993714e-05,
+      "loss": 0.8011,
+      "step": 2418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6898358365351032e-05,
+      "loss": 0.815,
+      "step": 2419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.690534404470835e-05,
+      "loss": 0.9112,
+      "step": 2420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6912329724065664e-05,
+      "loss": 0.9825,
+      "step": 2421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6919315403422985e-05,
+      "loss": 0.957,
+      "step": 2422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.69263010827803e-05,
+      "loss": 0.9924,
+      "step": 2423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.693328676213762e-05,
+      "loss": 0.8295,
+      "step": 2424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6940272441494935e-05,
+      "loss": 0.7602,
+      "step": 2425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6947258120852253e-05,
+      "loss": 0.9899,
+      "step": 2426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.695424380020957e-05,
+      "loss": 0.9066,
+      "step": 2427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.696122947956689e-05,
+      "loss": 0.9755,
+      "step": 2428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6968215158924207e-05,
+      "loss": 0.9257,
+      "step": 2429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6975200838281525e-05,
+      "loss": 1.0231,
+      "step": 2430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6982186517638842e-05,
+      "loss": 0.868,
+      "step": 2431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.698917219699616e-05,
+      "loss": 0.8715,
+      "step": 2432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6996157876353475e-05,
+      "loss": 0.816,
+      "step": 2433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7003143555710796e-05,
+      "loss": 0.9929,
+      "step": 2434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.701012923506811e-05,
+      "loss": 0.8635,
+      "step": 2435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.701711491442543e-05,
+      "loss": 0.8343,
+      "step": 2436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7024100593782746e-05,
+      "loss": 0.9981,
+      "step": 2437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7031086273140064e-05,
+      "loss": 1.0662,
+      "step": 2438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.703807195249738e-05,
+      "loss": 1.0752,
+      "step": 2439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.70450576318547e-05,
+      "loss": 0.8406,
+      "step": 2440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7052043311212017e-05,
+      "loss": 1.0056,
+      "step": 2441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7059028990569335e-05,
+      "loss": 0.8354,
+      "step": 2442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7066014669926653e-05,
+      "loss": 0.8604,
+      "step": 2443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.707300034928397e-05,
+      "loss": 0.9834,
+      "step": 2444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.707998602864129e-05,
+      "loss": 0.9352,
+      "step": 2445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7086971707998606e-05,
+      "loss": 1.0334,
+      "step": 2446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.709395738735592e-05,
+      "loss": 0.9212,
+      "step": 2447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7100943066713242e-05,
+      "loss": 0.9603,
+      "step": 2448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7107928746070556e-05,
+      "loss": 0.9483,
+      "step": 2449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7114914425427874e-05,
+      "loss": 1.0603,
+      "step": 2450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7121900104785192e-05,
+      "loss": 0.9919,
+      "step": 2451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.712888578414251e-05,
+      "loss": 0.9018,
+      "step": 2452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7135871463499827e-05,
+      "loss": 0.923,
+      "step": 2453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.959,
+      "step": 2454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7149842822214463e-05,
+      "loss": 0.8944,
+      "step": 2455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7156828501571777e-05,
+      "loss": 0.8386,
+      "step": 2456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.71638141809291e-05,
+      "loss": 0.7737,
+      "step": 2457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7170799860286413e-05,
+      "loss": 1.0269,
+      "step": 2458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.717778553964373e-05,
+      "loss": 0.9451,
+      "step": 2459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.718477121900105e-05,
+      "loss": 0.8387,
+      "step": 2460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7191756898358367e-05,
+      "loss": 0.9491,
+      "step": 2461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7198742577715684e-05,
+      "loss": 0.9001,
+      "step": 2462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7205728257073002e-05,
+      "loss": 0.981,
+      "step": 2463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.721271393643032e-05,
+      "loss": 0.825,
+      "step": 2464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7219699615787638e-05,
+      "loss": 0.8356,
+      "step": 2465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7226685295144952e-05,
+      "loss": 0.8932,
+      "step": 2466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7233670974502273e-05,
+      "loss": 1.0039,
+      "step": 2467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7240656653859588e-05,
+      "loss": 0.9844,
+      "step": 2468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.724764233321691e-05,
+      "loss": 0.9653,
+      "step": 2469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7254628012574223e-05,
+      "loss": 0.8449,
+      "step": 2470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.726161369193154e-05,
+      "loss": 0.9883,
+      "step": 2471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.726859937128886e-05,
+      "loss": 0.7642,
+      "step": 2472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7275585050646177e-05,
+      "loss": 0.9098,
+      "step": 2473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7282570730003495e-05,
+      "loss": 0.9289,
+      "step": 2474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7289556409360813e-05,
+      "loss": 0.8951,
+      "step": 2475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.729654208871813e-05,
+      "loss": 0.914,
+      "step": 2476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7303527768075448e-05,
+      "loss": 0.9195,
+      "step": 2477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7310513447432763e-05,
+      "loss": 1.1805,
+      "step": 2478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7317499126790084e-05,
+      "loss": 0.8935,
+      "step": 2479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7324484806147398e-05,
+      "loss": 0.8267,
+      "step": 2480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.733147048550472e-05,
+      "loss": 0.9862,
+      "step": 2481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7338456164862034e-05,
+      "loss": 0.8629,
+      "step": 2482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.734544184421935e-05,
+      "loss": 0.8562,
+      "step": 2483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.735242752357667e-05,
+      "loss": 1.0425,
+      "step": 2484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7359413202933987e-05,
+      "loss": 0.8693,
+      "step": 2485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7366398882291305e-05,
+      "loss": 0.906,
+      "step": 2486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.737338456164862e-05,
+      "loss": 0.827,
+      "step": 2487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.738037024100594e-05,
+      "loss": 1.1381,
+      "step": 2488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7387355920363255e-05,
+      "loss": 0.941,
+      "step": 2489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7394341599720573e-05,
+      "loss": 1.0169,
+      "step": 2490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.740132727907789e-05,
+      "loss": 0.7809,
+      "step": 2491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.740831295843521e-05,
+      "loss": 0.7343,
+      "step": 2492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7415298637792526e-05,
+      "loss": 0.8844,
+      "step": 2493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7422284317149844e-05,
+      "loss": 0.8272,
+      "step": 2494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7429269996507162e-05,
+      "loss": 0.7732,
+      "step": 2495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.743625567586448e-05,
+      "loss": 0.8632,
+      "step": 2496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7443241355221798e-05,
+      "loss": 0.9825,
+      "step": 2497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7450227034579115e-05,
+      "loss": 0.9807,
+      "step": 2498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.745721271393643e-05,
+      "loss": 0.987,
+      "step": 2499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.746419839329375e-05,
+      "loss": 0.8505,
+      "step": 2500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7471184072651065e-05,
+      "loss": 0.9366,
+      "step": 2501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7478169752008383e-05,
+      "loss": 0.7995,
+      "step": 2502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.74851554313657e-05,
+      "loss": 0.9503,
+      "step": 2503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.749214111072302e-05,
+      "loss": 0.9261,
+      "step": 2504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7499126790080337e-05,
+      "loss": 0.9401,
+      "step": 2505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7506112469437655e-05,
+      "loss": 1.0078,
+      "step": 2506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7513098148794972e-05,
+      "loss": 0.9921,
+      "step": 2507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.752008382815229e-05,
+      "loss": 0.9144,
+      "step": 2508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7527069507509608e-05,
+      "loss": 0.9485,
+      "step": 2509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7534055186866926e-05,
+      "loss": 0.9997,
+      "step": 2510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.754104086622424e-05,
+      "loss": 0.9151,
+      "step": 2511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.754802654558156e-05,
+      "loss": 0.8535,
+      "step": 2512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7555012224938876e-05,
+      "loss": 0.7586,
+      "step": 2513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7561997904296194e-05,
+      "loss": 1.1726,
+      "step": 2514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.756898358365351e-05,
+      "loss": 0.7472,
+      "step": 2515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.757596926301083e-05,
+      "loss": 0.8006,
+      "step": 2516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7582954942368147e-05,
+      "loss": 0.8844,
+      "step": 2517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7589940621725465e-05,
+      "loss": 0.9845,
+      "step": 2518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7596926301082783e-05,
+      "loss": 0.835,
+      "step": 2519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7603911980440097e-05,
+      "loss": 0.8621,
+      "step": 2520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.761089765979742e-05,
+      "loss": 0.8039,
+      "step": 2521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7617883339154733e-05,
+      "loss": 0.9633,
+      "step": 2522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.762486901851205e-05,
+      "loss": 1.0132,
+      "step": 2523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.763185469786937e-05,
+      "loss": 1.1725,
+      "step": 2524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7638840377226686e-05,
+      "loss": 0.8145,
+      "step": 2525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7645826056584004e-05,
+      "loss": 0.9179,
+      "step": 2526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7652811735941322e-05,
+      "loss": 0.7606,
+      "step": 2527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.765979741529864e-05,
+      "loss": 0.9004,
+      "step": 2528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7666783094655957e-05,
+      "loss": 1.0455,
+      "step": 2529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7673768774013272e-05,
+      "loss": 0.9272,
+      "step": 2530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7680754453370593e-05,
+      "loss": 1.1148,
+      "step": 2531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7687740132727907e-05,
+      "loss": 0.8242,
+      "step": 2532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.769472581208523e-05,
+      "loss": 0.7652,
+      "step": 2533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7701711491442543e-05,
+      "loss": 0.98,
+      "step": 2534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.770869717079986e-05,
+      "loss": 0.8267,
+      "step": 2535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.771568285015718e-05,
+      "loss": 1.0131,
+      "step": 2536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7722668529514497e-05,
+      "loss": 0.944,
+      "step": 2537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7729654208871814e-05,
+      "loss": 1.1155,
+      "step": 2538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7736639888229132e-05,
+      "loss": 0.8264,
+      "step": 2539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.774362556758645e-05,
+      "loss": 0.7866,
+      "step": 2540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7750611246943768e-05,
+      "loss": 0.6935,
+      "step": 2541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7757596926301082e-05,
+      "loss": 0.9766,
+      "step": 2542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7764582605658403e-05,
+      "loss": 0.9464,
+      "step": 2543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7771568285015718e-05,
+      "loss": 0.8253,
+      "step": 2544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.777855396437304e-05,
+      "loss": 0.9228,
+      "step": 2545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7785539643730353e-05,
+      "loss": 0.8612,
+      "step": 2546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.779252532308767e-05,
+      "loss": 0.7726,
+      "step": 2547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.779951100244499e-05,
+      "loss": 1.107,
+      "step": 2548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7806496681802307e-05,
+      "loss": 0.7896,
+      "step": 2549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7813482361159625e-05,
+      "loss": 0.8676,
+      "step": 2550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7820468040516943e-05,
+      "loss": 1.1502,
+      "step": 2551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.782745371987426e-05,
+      "loss": 0.9277,
+      "step": 2552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7834439399231575e-05,
+      "loss": 0.8158,
+      "step": 2553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7841425078588896e-05,
+      "loss": 1.1581,
+      "step": 2554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.784841075794621e-05,
+      "loss": 1.1101,
+      "step": 2555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7855396437303528e-05,
+      "loss": 1.1626,
+      "step": 2556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7862382116660846e-05,
+      "loss": 0.8567,
+      "step": 2557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7869367796018164e-05,
+      "loss": 1.0041,
+      "step": 2558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.787635347537548e-05,
+      "loss": 1.0514,
+      "step": 2559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.78833391547328e-05,
+      "loss": 0.8237,
+      "step": 2560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7890324834090117e-05,
+      "loss": 1.0196,
+      "step": 2561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7897310513447435e-05,
+      "loss": 0.9619,
+      "step": 2562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.790429619280475e-05,
+      "loss": 0.8386,
+      "step": 2563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.791128187216207e-05,
+      "loss": 0.971,
+      "step": 2564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7918267551519385e-05,
+      "loss": 0.9465,
+      "step": 2565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7925253230876706e-05,
+      "loss": 0.9477,
+      "step": 2566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.793223891023402e-05,
+      "loss": 0.9101,
+      "step": 2567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.793922458959134e-05,
+      "loss": 1.0386,
+      "step": 2568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7946210268948656e-05,
+      "loss": 1.0082,
+      "step": 2569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7953195948305974e-05,
+      "loss": 1.0661,
+      "step": 2570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7960181627663292e-05,
+      "loss": 0.9065,
+      "step": 2571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.796716730702061e-05,
+      "loss": 0.9957,
+      "step": 2572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7974152986377928e-05,
+      "loss": 1.0089,
+      "step": 2573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7981138665735245e-05,
+      "loss": 0.8035,
+      "step": 2574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.798812434509256e-05,
+      "loss": 1.0871,
+      "step": 2575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.799511002444988e-05,
+      "loss": 1.1785,
+      "step": 2576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8002095703807195e-05,
+      "loss": 0.873,
+      "step": 2577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8009081383164517e-05,
+      "loss": 0.9512,
+      "step": 2578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801606706252183e-05,
+      "loss": 0.8936,
+      "step": 2579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802305274187915e-05,
+      "loss": 0.9485,
+      "step": 2580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030038421236467e-05,
+      "loss": 0.9624,
+      "step": 2581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037024100593785e-05,
+      "loss": 0.9238,
+      "step": 2582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8044009779951102e-05,
+      "loss": 0.9663,
+      "step": 2583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805099545930842e-05,
+      "loss": 0.9016,
+      "step": 2584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8057981138665738e-05,
+      "loss": 0.9617,
+      "step": 2585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8064966818023056e-05,
+      "loss": 0.9352,
+      "step": 2586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.807195249738037e-05,
+      "loss": 0.714,
+      "step": 2587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8078938176737688e-05,
+      "loss": 0.7258,
+      "step": 2588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8085923856095006e-05,
+      "loss": 1.0136,
+      "step": 2589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8092909535452324e-05,
+      "loss": 1.084,
+      "step": 2590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.809989521480964e-05,
+      "loss": 1.0061,
+      "step": 2591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.810688089416696e-05,
+      "loss": 1.0031,
+      "step": 2592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8113866573524277e-05,
+      "loss": 0.8458,
+      "step": 2593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8120852252881595e-05,
+      "loss": 1.0542,
+      "step": 2594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8127837932238913e-05,
+      "loss": 1.0895,
+      "step": 2595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8134823611596227e-05,
+      "loss": 0.8718,
+      "step": 2596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.814180929095355e-05,
+      "loss": 0.8139,
+      "step": 2597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8148794970310863e-05,
+      "loss": 0.864,
+      "step": 2598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.815578064966818e-05,
+      "loss": 0.8496,
+      "step": 2599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.81627663290255e-05,
+      "loss": 0.9272,
+      "step": 2600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8169752008382816e-05,
+      "loss": 0.9468,
+      "step": 2601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8176737687740134e-05,
+      "loss": 1.098,
+      "step": 2602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8183723367097452e-05,
+      "loss": 0.8525,
+      "step": 2603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.819070904645477e-05,
+      "loss": 0.9931,
+      "step": 2604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8197694725812087e-05,
+      "loss": 0.8495,
+      "step": 2605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8204680405169405e-05,
+      "loss": 0.8589,
+      "step": 2606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8211666084526723e-05,
+      "loss": 0.8395,
+      "step": 2607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8218651763884037e-05,
+      "loss": 0.7138,
+      "step": 2608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.822563744324136e-05,
+      "loss": 0.8514,
+      "step": 2609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8232623122598673e-05,
+      "loss": 0.8576,
+      "step": 2610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.823960880195599e-05,
+      "loss": 1.0106,
+      "step": 2611
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.824659448131331e-05,
+      "loss": 0.9155,
+      "step": 2612
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8253580160670627e-05,
+      "loss": 0.8937,
+      "step": 2613
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8260565840027944e-05,
+      "loss": 0.9155,
+      "step": 2614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8267551519385262e-05,
+      "loss": 1.0837,
+      "step": 2615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.827453719874258e-05,
+      "loss": 0.8482,
+      "step": 2616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8281522878099898e-05,
+      "loss": 0.9452,
+      "step": 2617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8288508557457216e-05,
+      "loss": 0.807,
+      "step": 2618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8295494236814533e-05,
+      "loss": 0.9466,
+      "step": 2619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8302479916171848e-05,
+      "loss": 0.8556,
+      "step": 2620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.830946559552917e-05,
+      "loss": 0.8001,
+      "step": 2621
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8316451274886483e-05,
+      "loss": 0.751,
+      "step": 2622
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.83234369542438e-05,
+      "loss": 0.7687,
+      "step": 2623
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.833042263360112e-05,
+      "loss": 0.9913,
+      "step": 2624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8337408312958437e-05,
+      "loss": 0.9497,
+      "step": 2625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8344393992315755e-05,
+      "loss": 0.8123,
+      "step": 2626
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.835137967167307e-05,
+      "loss": 0.9629,
+      "step": 2627
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.835836535103039e-05,
+      "loss": 0.9406,
+      "step": 2628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8365351030387705e-05,
+      "loss": 0.9398,
+      "step": 2629
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8372336709745026e-05,
+      "loss": 0.8194,
+      "step": 2630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.837932238910234e-05,
+      "loss": 0.9373,
+      "step": 2631
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8386308068459658e-05,
+      "loss": 0.841,
+      "step": 2632
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8393293747816976e-05,
+      "loss": 1.036,
+      "step": 2633
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8400279427174294e-05,
+      "loss": 0.8155,
+      "step": 2634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.840726510653161e-05,
+      "loss": 0.8947,
+      "step": 2635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.841425078588893e-05,
+      "loss": 1.0505,
+      "step": 2636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8421236465246247e-05,
+      "loss": 0.9194,
+      "step": 2637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8428222144603565e-05,
+      "loss": 0.9511,
+      "step": 2638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.843520782396088e-05,
+      "loss": 0.897,
+      "step": 2639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.84421935033182e-05,
+      "loss": 0.7893,
+      "step": 2640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8449179182675515e-05,
+      "loss": 0.8937,
+      "step": 2641
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8456164862032836e-05,
+      "loss": 0.9254,
+      "step": 2642
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.846315054139015e-05,
+      "loss": 0.832,
+      "step": 2643
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.847013622074747e-05,
+      "loss": 0.9661,
+      "step": 2644
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8477121900104786e-05,
+      "loss": 0.87,
+      "step": 2645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8484107579462104e-05,
+      "loss": 0.8561,
+      "step": 2646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8491093258819422e-05,
+      "loss": 1.1395,
+      "step": 2647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.849807893817674e-05,
+      "loss": 0.7856,
+      "step": 2648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8505064617534058e-05,
+      "loss": 0.8588,
+      "step": 2649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8512050296891375e-05,
+      "loss": 0.9929,
+      "step": 2650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8519035976248693e-05,
+      "loss": 0.7899,
+      "step": 2651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.852602165560601e-05,
+      "loss": 0.9226,
+      "step": 2652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8533007334963325e-05,
+      "loss": 0.8676,
+      "step": 2653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8539993014320647e-05,
+      "loss": 1.0156,
+      "step": 2654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.854697869367796e-05,
+      "loss": 0.8408,
+      "step": 2655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.855396437303528e-05,
+      "loss": 1.0508,
+      "step": 2656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8560950052392597e-05,
+      "loss": 0.771,
+      "step": 2657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8567935731749915e-05,
+      "loss": 1.1744,
+      "step": 2658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8574921411107232e-05,
+      "loss": 0.9361,
+      "step": 2659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8581907090464547e-05,
+      "loss": 1.0266,
+      "step": 2660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8588892769821868e-05,
+      "loss": 0.7195,
+      "step": 2661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8595878449179182e-05,
+      "loss": 0.8907,
+      "step": 2662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8602864128536504e-05,
+      "loss": 0.7889,
+      "step": 2663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8609849807893818e-05,
+      "loss": 0.9399,
+      "step": 2664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8616835487251136e-05,
+      "loss": 0.8243,
+      "step": 2665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8623821166608454e-05,
+      "loss": 0.8514,
+      "step": 2666
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.863080684596577e-05,
+      "loss": 0.8916,
+      "step": 2667
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.863779252532309e-05,
+      "loss": 0.8657,
+      "step": 2668
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8644778204680407e-05,
+      "loss": 1.1876,
+      "step": 2669
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8651763884037725e-05,
+      "loss": 1.0234,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8658749563395043e-05,
+      "loss": 0.8531,
+      "step": 2671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8665735242752357e-05,
+      "loss": 1.0595,
+      "step": 2672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.867272092210968e-05,
+      "loss": 0.9583,
+      "step": 2673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8679706601466993e-05,
+      "loss": 0.8324,
+      "step": 2674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8686692280824314e-05,
+      "loss": 0.7002,
+      "step": 2675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.869367796018163e-05,
+      "loss": 1.0184,
+      "step": 2676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8700663639538946e-05,
+      "loss": 0.822,
+      "step": 2677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8707649318896264e-05,
+      "loss": 1.0754,
+      "step": 2678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8714634998253582e-05,
+      "loss": 0.7568,
+      "step": 2679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.87216206776109e-05,
+      "loss": 0.9801,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8728606356968217e-05,
+      "loss": 0.9768,
+      "step": 2681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8735592036325535e-05,
+      "loss": 1.0125,
+      "step": 2682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8742577715682853e-05,
+      "loss": 0.9018,
+      "step": 2683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8749563395040167e-05,
+      "loss": 0.8503,
+      "step": 2684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.875654907439749e-05,
+      "loss": 0.8526,
+      "step": 2685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8763534753754803e-05,
+      "loss": 0.7278,
+      "step": 2686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8770520433112124e-05,
+      "loss": 0.8249,
+      "step": 2687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.877750611246944e-05,
+      "loss": 1.0569,
+      "step": 2688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8784491791826757e-05,
+      "loss": 0.8638,
+      "step": 2689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8791477471184074e-05,
+      "loss": 0.8853,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8798463150541392e-05,
+      "loss": 0.7552,
+      "step": 2691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.880544882989871e-05,
+      "loss": 0.9638,
+      "step": 2692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8812434509256024e-05,
+      "loss": 1.0877,
+      "step": 2693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8819420188613346e-05,
+      "loss": 0.9437,
+      "step": 2694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.882640586797066e-05,
+      "loss": 0.8781,
+      "step": 2695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8833391547327978e-05,
+      "loss": 0.8295,
+      "step": 2696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8840377226685296e-05,
+      "loss": 0.8285,
+      "step": 2697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8847362906042613e-05,
+      "loss": 1.036,
+      "step": 2698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.885434858539993e-05,
+      "loss": 0.977,
+      "step": 2699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.886133426475725e-05,
+      "loss": 0.9238,
+      "step": 2700
+    }
+  ],
+  "max_steps": 95425,
+  "num_train_epochs": 25,
+  "total_flos": 2.065163321868288e+16,
+  "trial_name": null,
+  "trial_params": null
+}